Gut microbial production of imidazole propionate drives Parkinson's pathologies

Abstract

Parkinson's disease (PD) is characterized by the selective degeneration of midbrain dopaminergic neurons and aggregation of α-synuclein. Emerging evidence implicates the gut microbiome in PD, with microbial metabolites proposed as potential pathological mediators. However, the specific microbes and metabolites involved, and whether gut-derived metabolites can reach the brain to directly induce neurodegeneration, remain unclear. Here we show that elevated levels of Streptococcus mutans (S. mutans) and its enzyme urocanate reductase (UrdA), which produces imidazole propionate (ImP), in the gut microbiome of patients with PD, along with increased plasma ImP. Colonization of mice with S. mutans harboring UrdA or Escherichia coli expressing UrdA from S. mutans increases systemic and brain ImP levels, inducing PD-like symptoms including dopaminergic neuronal loss, astrogliosis, microgliosis, and motor impairment. Additionally, S. mutans exacerbates α-synuclein pathology in a mouse model. ImP administration alone recapitulates key PD features, supporting the UrdA-ImP axis as a microbial driver of PD pathology. Mechanistically, mTORC1 activation is crucial for both S. mutans- and ImP-induced PD pathology. Together, these findings identify microbial ImP, produced via UrdA, as a direct pathological mediator of the gut-brain axis in PD.

Fig. 1. Increased urdA enzyme abundance in the gut microbiome of patients with Parkinson’s disease (PD) and PD key pathologies induced by gut-colonization of imidazole propionate (ImP)-producing Streptococcus mutans (S. mutans).

a Relative abundance of S. mutans in the gut microbiome of patients with PD (n = 491) and age-matched healthy controls (n = 234). Data were log₁₀-transformed and are presented as box plots showing minimum, 25% quartile, median, 75% quartile, and maximum. b, c Normalized gene copy numbers of urocanate reductase (urdA, b) and histidine ammonia lyase (hutH, c) in the gut microbiome from patients with PD and age-matched healthy controls. Box plots represent the median (centre line), 25%–75% quartile (bounds of box), minima and maxima (whiskers), with all individual data points shown. d Schematic illustration of the experimental procedure showing colonization of germ-free (GF) mice with S. mutans or pasteurized S. mutans (P-S. mutans) via oral gavage. Fecal samples were collected from GF mice before (D0) and 28 days after S. mutans colonization. Created in BioRender. Kim, J. (https://BioRender.com/xsj2j6c). e Quantification of live S. mutans in the feces by assessing viable colony formation from GF mice gavaged with vehicle (Veh), S. mutans, or P-S. mutans (n = 7 mice per Veh and P-S. mutans group, n = 8 mice per S. mutans group). CFU, colony forming unit. f (Left panel) Representative anti-TH (tyrosine hydroxylase) immunohistochemistry images of ventral midbrains from GF mice colonized with S. mutans or P-S. mutans, or vehicle control. Brain sections were counterstained with Nissl. Magnified images showing dopaminergic processes in the substantia nigra pars reticulata (SNr) are also presented below. (Right panel) Stereological counts of TH-positive and Nissl-stained dopaminergic neurons in the substantia nigra pars compacta (SNpc). Relative TH fiber density and Nissl-stained neuronal counts in the SNr (n = 7 mice per group). Scale bar = 100 μm, and 25 μm for enlarged images, respectively. g (Left panel) Representative images of inflammatory astrogliosis in ventral midbrains of GF mice colonized with S. mutans, P-S. mutans, or vehicle control, assessed via GFAP immunohistochemistry. Magnified images of the SNpc were presented on the right. (Right panel) Quantification of GFAP-positive area as percentage of total area (% area fraction) (n = 7 mice per group). Scale bar = 100 μm. h (Left panel) Representative images of inflammatory microgliosis in ventral midbrains of GF mice colonized with S. mutans, P-S. mutans, or vehicle control, assessed via Iba1 immunohistochemistry. Magnified images of the SNpc subregions on the right highlight inflammatory microgliosis. (Right panel) Quantification of Iba1-positive area fraction (%; n = 3 mice per group) and microglial soma size, measured from 10 cells per mouse (n = 3 mice per group). Scale bar = 100 μm. i Assessment of bradykinesia in GF mice colonized with S. mutans or P-S. mutans monitored by measuring latency to reach the base in the pole test (n = 7 mice per Veh and P-S. mutans group, n = 8 mice per S. mutans group). j Quantification of ImP concentration in the plasma from GF mice gavaged with Veh, S. mutans, or P-S. mutans. (n = 7 mice per Veh and P-S. mutans group, n = 8 mice per S. mutans group). k Quantification of ImP concentration in the brains from GF mice gavaged with Veh, S. mutans, or P-S. mutans. (n = 4 mice per Veh and P-S. mutans group, n = 6 mice per S. mutans group). Quantitative data are expressed as the mean ± standard error of the mean (S.E.M). **P < 0.01, ***P < 0.001; ns, not significant. P values were determined by the two-sided Wilcoxon rank-sum test (a-c), two-way ANOVA with Tukey’s post hoc test (e), and one-way ANOVA with Tukey’s post hoc test (f–k). Source data and test statistics are provided as a Source Data file. See also Supplementary Fig. 1.

Fig. 2. PD pathologies induced by gut colonized Escherichia coli (E. coli) heterologously expressing S. mutans urdA.

a Quantification of ImP concentration in the supernatants from the culture media of E. coli MG1655 strains carrying either the empty pUC19 vector (E. coli (Con)) or the recombinant plasmid pUC19-urdA (E. coli (urdA)), and S. mutans after incubation with 1 mM urocanate for 24 h (n = 3 per group). b Schematic illustration of the experimental procedure showing colonization of GF mice with commensal E. coli MG1655 carrying either the empty pUC19 vector (E. coli (Con)) or urdA from S. mutans (E. coli (urdA)) via oral gavage. Created in BioRender. Kim, J. (https://BioRender.com/9pc9rgz). c Quantification of live E. coli in feces by assessing viable colony formation (n = 6 mice per E. coli (Con) group, n = 7 mice per E. coli (urdA) group). CFU, colony forming unit. d, e Quantification of ImP concentration in the plasma (d) and brains (e) from GF mice gavaged with E. coli (Con) or E. coli (urdA) (plasma: n = 6 for E. coli (Con) group, n = 7 for E. coli (urdA) group; brain: n = 6 per group). Box plots represent the median (centre line), 25%–75% quartile (bounds of box), minima and maxima (whiskers), with all individual data points shown.f (Left panel) Representative anti-TH immunohistochemistry of ventral midbrains from GF mice colonized with E. coli (Con) or E. coli (urdA). Magnified images showing dopaminergic processes in SNr are also presented below. (Right panel) Stereological quantification of TH-stained and Nissl-stained dopaminergic neurons in the SNpc, Nissl-stained neuronal counts, and relative TH fiber density in the SNr (n = 6 mice per E. coli (Con) group, n = 7 mice per E. coli (urdA) group). Scale bar = 100 μm. g (Upper panel) Representative images of inflammatory astrogliosis in the ventral midbrains of GF mice colonized with E. coli (Con) or E. coli (urdA) determined by anti-GFAP immunohistochemistry. Magnified images of the SNpc subregions are shown below. (Bottom panel) Quantified % area fraction of GFAP signals (n = 5 mice per E. coli (Con) group, n = 4 mice per E. coli (urdA) group). Scale bar = 100 μm. h (Upper panel) Representative images of inflammatory microgliosis in the ventral midbrains of GF mice colonized with E. coli (Con) or E. coli (urdA) determined by anti-Iba1 immunohistochemistry. Magnified images of the SNpc subregions are shown below. (Bottom panel) The Iba1-positive area fraction (%) (n = 5 mice per E. coli (Con) group, n = 7 mice per E. coli (urdA) group) and microglial soma size were quantified (n = 50 microglia from 5 mice per E. coli (Con) group, n = 70 microglia from 7 mice per E. coli (urdA) group). Scale bar = 100 μm. i Assessment of bradykinesia in GF mice colonized with E. coli (Con) or E. coli (urdA) monitored by measuring latency to reach the base in the pole test (n = 6 mice per E. coli (Con) group, n = 7 mice per E. coli (urdA) group). Quantitative data are expressed as the mean ± standard error of the mean (S.E.M). *P < 0.05, **P < 0.01, ***P < 0.001; ns, not significant. P values were determined by one-way ANOVA with Tukey’s post hoc test (a), two-way ANOVA with Tukey’s post hoc test (c), and unpaired two-tailed Student’s t-tests (d–i). Source data and test statistics are provided as a Source Data file. See also Supplementary Fig. 2.

Fig. 3. mTORC1-dependent selective dopaminergic neurotoxicity, astrogliosis, and motor impairments by gut-colonized S. mutans.

a (Upper panel) Representative immunofluorescence images of S235/S236 phosphorylated S6 ribosomal protein (pS6) and TH in the ventral midbrain coronal sections from GF mice gavaged with Veh, S. mutans, or pasteurized S. mutans (P-S. mutans). pS6 signals originally captured in the red channel are pseudocolored in white for visualization. (Bottom panel) Quantification of relative pS6 signal intensities in TH-positive dopamine neurons of the SNpc (n = 7 mice per group). Scale bar = 100 μm. b (Upper panel) Representative immunofluorescence images of pS6 and TH in the ventral midbrain coronal sections from GF mice gavaged with E. coli (Con) or E. coli (urdA). Original red pS6 immunofluorescence signals are pseudocolored in white. (Bottom panel) Quantification of relative pS6 signal intensities in TH-positive dopamine neurons of the SNpc (n = 6 mice per E. coli (Con) and n = 7 mice per E. coli (urdA) group). Scale bar = 100 μm. c (Upper panel) Representative Western blots of pIRS1 (S636/S639), IRS1, pS6K1 (T389), and S6K1 in mouse primary cortical neurons treated with ImP (100 μM, 24 h) in the presence or absence of the mTORC1 inhibitor rapamycin (Rap, 20 nM, 24 h) or the p38γ inhibitor pirfenidone (Pirf, 1 mM, 24 h). (Bottom panel) Quantification of relative expression levels of pIRS1 (S636/S639) and pS6K1 (T389) (n = 4 per group). d Schematic illustration of the experimental procedure showing antibiotic cocktail treatment to sterilize the gut, cessation of antibiotics (Abx) treatment, and subsequent S. mutans colonization with or without rapamycin intraperitoneal (i.p.) injection. Fourteen days after colonization, mice were subjected to behavior tests and immunohistochemical analysis. Created in BioRender. Kim, J. (https://BioRender.com/yjm4i12). e, f Quantification of imidazole propionate (ImP) concentration in the plasma (e) and brains (f) from the antibiotic-treated mice colonized with S. mutans or Veh and treated with rapamycin (Rap, 4 mg/kg/day) or vehicle (n = 7 mice per Veh group, n = 8 mice per S. mutans group, n = 6 mice per S. mutans+Rap group). Box plots represent the median (centre line), 25%–75% quartile (bounds of box), minima and maxima (whiskers), with all individual data points shown. g (Upper panel) Representative immunofluorescence images of pS6 and TH in the ventral midbrain coronal sections from the antibiotic-treated mice colonized with S. mutans or Veh and administered with rapamycin or vehicle. Original red pS6 immunofluorescence signals are pseudocolored in white. (Bottom panel) Quantification of relative pS6 signal intensities in TH-positive dopamine neurons of the SNpc is shown below (n = 6 mice per group). Scale bar = 100 μm. SNpc and SNr subregions of the VM are indicated by the white dashed line. h (Left panel) Representative anti-TH immunohistochemistry of ventral midbrain coronal sections from the antibiotic-treated mice colonized with S. mutans or vehicle (Veh) and administered with rapamycin or vehicle. Magnified images showing dopaminergic processes in the SNr are presented on the right. (Right panel) Stereological counts of TH-stained and Nissl-stained dopamine neurons in the SNpc. Quantification of Nissl-stained neuronal counts and TH fiber density in the SNr of the indicated experimental groups (n = 5 mice per group). Scale bar = 100 μm, and 25 μm for enlarged images, respectively. i Inflammatory astrogliosis in the ventral midbrains of the indicated experimental antibiotic-treated mouse groups determined by anti-GFAP immunohistochemistry. Magnified images of the SNpc are shown below. Quantified % area fraction of GFAP signals is shown in the right panel (n = 5 mice per group). Scale bar = 100 μm. j Assessment of bradykinesia in the antibiotic-treated mice colonized with S. mutans or Veh and administered with rapamycin or vehicle monitored by measuring latency to reach the base in the pole test (n = 10 mice per Veh and S. mutans group, n = 9 mice per S. mutans+Rap group). Quantitative data are expressed as the mean ± S.E.M. *P < 0.05, **P < 0.01, ***P < 0.001; ns, not significant. P values were determined using one-way ANOVA with Tukey’s post hoc test (a, c, e–j) and unpaired two-tailed Student’s t-tests (b). Source data and test statistics are provided as a Source Data file. See also Supplementary Fig. 3.

Fig. 4. α-synuclein preformed fibril (PFF)-induced mTORC1 activation, α-synuclein aggregation, dopaminergic neurotoxicity, and motor impairment promoted by gut-colonized S. mutans.

a Representative immunofluorescence images showing the α-synuclein (α-Syn) aggregation marker S129 phosphorylated α-Syn (pS129-α-Syn) and neuronal marker microtubule-associated protein 2 (MAP2) in mouse primary cultured cortical neurons treated with the indicated combinations of ImP (1, 10, 100, 1000 nM every three days for six days) and α-Syn PFF (0.1 μg/mL treated once and incubated for seven days), counterstained with DAPI. Scale bar = 20 μm. b Quantification of relative immunofluorescence intensities of pS129-α-Syn in the indicated experimental groups (n = 4 per vehicle (DMSO) group, n = 3 per ImP group). c Assessment of neuronal viability in each experimental group, determined by counting MAP2-positive neuritic beads with a pathological phenotype of fragmentation (n = 4 per vehicle (DMSO) group, n = 3 per ImP group). d Schematic illustration of the experimental procedure showing antibiotic cocktail treatment to sterilize the gut, cessation of Abx treatment, and subsequent intranigral stereotaxic injections (inj.) of α-Syn PFF (10 μg in 2 μL) to model sporadic PD in mice followed by S. mutans colonization. Created in BioRender. Kim, J. (https://BioRender.com/x59j418). e Representative immunofluorescence images of pS6 and TH in the ventral midbrain coronal sections from the PBS-injected control or nigral α-Syn PFF-injected mice colonized with S. mutans or vehicle. Original red pS6 immunofluorescence signals are pseudocolored in white. Nuclei were counterstained with DAPI. Quantification of relative pS6 signal intensities in TH-positive dopamine neurons of the SNpc is shown in the right panel (n = 8 mice per group). Scale bar = 100 μm. f Lewy-like inclusions in the ventral midbrains of the PBS-injected control or nigral α-Syn PFF-injected mice colonized with S. mutans or vehicle, determined by anti-pS129-α-Syn (Lewy body marker) immunohistochemistry. Quantified % area fraction of pS129-α-Syn signals in the SNpc subregion is shown in the bottom panel (n = 6 mice per group). Scale bar = 100 μm, and 30 μm for enlarged images, respectively. g (Upper panel) Representative anti-TH immunohistochemistry of ventral midbrains from the indicated experimental mouse groups. Magnified images showing dopaminergic processes in SNr are also presented on the right. (Bottom panel) Stereological counts of TH-stained and Nissl-stained dopamine neurons in the lesioned hemisphere of the SNpc. Nissl-stained neuronal counts and TH fiber density in the SNr of the indicated experimental mouse group (n = 6 mice per group). Scale bar = 100 μm, and 25 μm for enlarged images, respectively. h Assessment of bradykinesia in the PBS-injected control or nigral α-Syn PFF-injected mice colonized with S. mutans or vehicle monitored by measuring latency to reach the base in the pole test (n = 8 mice per PBS-injected and PFF-injected group, n = 9 mice per PFF + S. mutans group). i Assessment of motor coordination in the indicated experimental mouse groups monitored by measuring latency to fall off in the accelerating rotarod test (n = 8 mice per PBS-injected and PFF-injected group, n = 9 mice per PFF + S. mutans group). Quantitative data are expressed as the mean ± S.E.M. *P < 0.05, **P < 0.01, ***P < 0.001; ns, not significant. P values were determined using one-way ANOVA with Tukey’s post hoc test. Source data and test statistics are provided as a Source Data file. See also Supplementary Fig. 4.

Fig. 5. PD-associated ImP induces mTORC1-dependent selective loss of dopaminergic neurons, inflammatory astrogliosis, and motor impairment.

a Schematic illustration of the experimental procedure showing systemic intraperitoneal (i.p.) administration of ImP (20 μg per mouse for three weeks). Created in BioRender. Kim, J. (https://BioRender.com/32jcwkt). b Representative anti-TH immunohistochemistry (Nissl counterstained) of ventral midbrain coronal sections from mice injected with ImP or vehicle (1% DMSO, Veh). Magnified images showing dopaminergic processes in the SNr are also presented on the right. Stereological counting of TH-positive dopaminergic neurons in the SNpc of each experimental mouse group is shown in the right panel (n = 5 mice per Veh group, n = 4 mice per ImP group). Scale bar = 400, and 100 μm for enlarged images, respectively. c, d Pole assessment of bradykinesia phenotype (c) and rotarod assessment of motor coordination (d) in mice injected with ImP or vehicle (1% DMSO) (n = 5 mice per Veh group, n = 4 mice per ImP group). e Schematic illustration of the experimental procedure showing intranigral stereotaxic injection of ImP (single injection of 7 μg of ImP in 2 μL) where rapamycin was injected intraperitoneally one day before ImP injection at 1 mg/kg per mouse daily for a total of four days. Created in BioRender. Kim, J. (https://BioRender.com/rxfm6ng). f Representative immunofluorescence images of pS6 and TH in the ventral midbrain coronal sections from mice with intranigral brain injections of PBS or ImP with or without mTORC1 inhibitor, rapamycin. Original red pS6 immunofluorescence signals are pseudocolored in white. Quantification of relative pS6 signal intensities in TH-positive dopamine neurons of the SNpc is shown in the right panel (n = 7 mice per group). Scale bar = 100. SNpc and SNr subregions of the VM are indicated by the white dashed line. g (Upper panel) Representative anti-TH immunohistochemistry images of ventral midbrains from mice with nigral ImP or PBS injections with or without rapamycin treatment. Magnified images showing dopaminergic processes in the SNr are also presented on the right. (Bottom panel) Stereological counts of TH-stained and Nissl-stained dopamine neurons in the lesioned hemisphere of the SNpc. Nissl-stained neuronal counts and TH fiber density in the SNr of the indicated experimental mouse group (n = 5 mice per group). Scale bar = 100 μm, and 50 μm for enlarged images, respectively. h Inflammatory astrogliosis in the ventral midbrains of the indicated experimental mouse groups determined by anti-GFAP immunohistochemistry. Quantified % area fraction of GFAP signals is shown below (n = 7 mice per group). Scale bar = 100 μm. i, j Pole assessment of bradykinesia (i) and rotarod assessment of motor coordination (j) in the indicated experimental groups (n = 7 mice per group). k Measurement of ImP concentrations in the plasma samples from patients with PD (n = 65) and age-matched neurologically healthy control participants (n = 65). Data transformed to log₁₀ scale and presented as box plots show minimum, 25% quartile, median, 75% quartile, maximum. Quantitative data are expressed as the mean ± S.E.M. *P < 0.05, **P < 0.01, ***P < 0.001; ns not significant. P values were determined using unpaired two-tailed Student’s t-tests (b–d), two-way ANOVA with Tukey’s post hoc test (f–j), and two-sided Wilcoxon rank-sum tests (k). Source data and test statistics are provided as a Source Data file. See also Supplementary Fig. 5.

Fig. 6. Schematic depiction of the gut-brain axis in PD via S. mutans-UrdA-ImP.

The graphical abstract illustrates that the UrdA-ImP axis is elevated in the gut microbiome of patients with Parkinson’s disease (PD). Using mouse models, we show that this contributes causally to PD pathology through mTORC1 activation. These findings identify UrdA as a potential gut microbiome–based therapeutic target for gut-brain axis disorders such as PD. Created in BioRender. Kim, J. (https://BioRender.com/vg7s2sy).