Main

There is growing awareness that the ENS may represent the initial site of αS pathology in PD. Constipation is one of the earliest and most common symptoms in patients with PD and probably reflects ENS dysfunction1,10. The concept of body-first PD is strongly supported by postmortem and multimodal imaging studies showing early Lewy pathology in the ENS of patients with PD, suggesting a caudo-rostral distribution from the intestine to the brainstem and other brain regions2,11. Further, intestinal injections of αS in mice result in the hierarchical spread of Lewy pathology to the brain, strongly implicating the ENS in early stages of disease3,4. However, little is known about the cellular and molecular mechanisms that could trigger the onset of αS pathology in the ENS and progression to the brain.

Tissue-resident macrophages are specialized phagocytes that orchestrate numerous niche-specific functions critical for tissue homeostasis12. In the intestine, macrophages facing the microbiota are continually replaced by blood monocytes and are strategically positioned in the murine lamina propria to engulf penetrating pathogens13. By contrast, ME-Macs reside in the muscularis externa (ME) and support the functional integrity of the myenteric plexus, a network of enteric neurons integral to gastrointestinal motility8,9,14. ME-Macs colonize the murine intestine before birth, are self-maintained but gradually replaced throughout life and maintain enteric neuronal health through the production of neurotrophic factors and the clearance of debris. In the ageing murine intestine, ME-Macs accumulate αS and express PD-associated genes Gba1 and Lrrk2, raising the intriguing question of whether ME-Macs could confer risk in body-first PD pathology15. Whereas central nervous system (CNS)-resident macrophages have been implicated in αS pathology, the role of ENS-resident ME-Macs remains unclear7,16,17.

Here we show that ME-Macs are necessary for the formation and distribution of αS pathology in the intestine and the CNS. In mouse models of PD, including αS transgenic mice and ME injections of patient-derived-αS, ME-Macs, but not enteric neurons, contain misfolded, aggregated αS that coincides with endolysosomal activation in ME-Macs. Mechanistically, we found that ME-Macs in PD mouse models modulate T cell expansion in the ENS. We further show that T cells invade the CNS as αS pathology progresses, and preventing T cell egression ameliorated neurodegeneration. Functionally, we show that injection of anti-CSF1R with anti-CCR2 into the ME, targeting ME-Macs, ameliorated αS pathology in both the ENS and CNS, abolished T cell migration along the gut–brain axis and improved motor defects and neurodegeneration in mouse models of synucleinopathy. Together, our results uncover the role of ME-Macs in the onset and progression of αS pathology and motor impairments along the gut–brain axis in body-first PD.

ME-Macs engulf and modify αS pathology

We first investigated the onset of αS pathology in the ENS versus CNS in 3KL αS transgenic mice, in which αS expression is driven by Thy1 and the 3K construct introduces phenotypic amplification of familial E46K mutations promoting αS tetramer destabilization and aggregation18. 3KL mice demonstrate key features of PD pathology including progressive cortical pathological αS, hereafter defined as s129p+ αS, selective neuronal loss and l-DOPA responsive motor impairments at 8 months of age18. Immunohistochemistry (IHC) on mechanically separated duodenal ME demonstrated significantly increased s129p+ αS surrounding HuC/D+ myenteric ganglia, a specialized group of enteric neurons within the ME, of 3KL versus wild-type (WT) mice (Fig. 1a–c) at 3 months. We focused on the duodenum, a region highly innervated by the vagus nerve. We also observed s129p+ αS in the ME of postmortem PD tissues (Extended Data Fig. 1a,b). Further, increased total gut transit time was observed in 3-month 3KL versus WT, suggesting impaired ENS function and constipation (Fig. 1d). However, we found no changes in murine myenteric ganglia and glial volume, measured by HUC/D and glial fibrillary acidic protein, indicating the absence of neuronal death despite αS and enteric pathology (Extended Data Fig. 1c,d). To investigate underlying molecular changes, we performed digital spatial profiling (DSP) on duodenal myenteric plexus in 3-month 3KL versus WT mice using a multiplexed antibody panel consisting of PD-associated targets (Fig. 1e). We observed increased expression of proteins related to autophagy and lysosomal biology in myenteric ganglia of 3KL versus WT, including PINK1, LRRK2 and VPS35 that are associated with familial PD (Fig. 1e). Of note, DSP confirmed upregulation of s129p+ αS in myenteric plexus and elevated total αS levels in both the lamina propria and myenteric plexus (Extended Data Fig. 1e). In the brain, we observed increased punctate αS s129p+ staining in cholinergic neurons of the dorsal motor nucleus of the vagus of 3KL animals at 6 months compared with age-matched WT controls (Extended Data Fig. 1f–k).

Fig. 1: ME-Macs engulf and modify αS pathology in 3KL αS transgenic mice.

a, Schematic of gut cross-section. b, Confocal image of duodenal myenteric plexus of 3-month WT versus 3KL. c, Quantification of s129p+ αS pathology: n = 6–8 mice; 1 datapoint, average of 1 mouse; 3–10 ROIs per mouse. Three experiments, data analysed using an unpaired t-test. d, Whole gastrointestinal transit time in 3-month WT versus 3KL, n = 9–10. Two experiments, unpaired t-test. e, Hierarchically clustered heatmap of ROI-specific nCounter digital counts across PD-relevant protein targets in duodenal myenteric plexus of 3-month WT versus 3KL, n = 2, 2–3 ROIs per mouse. f–j, Confocal images of engulfed αS by murine MHCII+ duodenal (f) and human CD209+ jejunal (g) ME-Macs, quantified murine LAMP1 lysosomal volume (h) and engulfment of s129p in murine (i) and αS (2F12) in human (j) LAMP1+ lysosomes. n = 8 mice per genotype (h,i) and n = 6 NHC and 8 PD postmortem samples (j). Two experiments, Mann–Whitney test (h–j). k, Volcano plot showing differentially expressed proteins in sorted ME-Macs of 3-month WT versus 3KL. n = 2 per genotype, with 3 mice pooled per biological unit. l, Biological processes enriched in 3-month 3KL ME-Macs, one-sided hypergeometric test. m, Schematic of SAA on sorted duodenal ME-Macs versus enteric neurons. n,o, αS aggregation kinetics through SAA in duodenal enteric neurons (n) and ME-Macs (o). p, SAA-positive versus negative count in different cell lysates. A sample was counted positive if aggregation onset (lag time) was at least 2 h shorter than negative control (PBS). n = 11 (enteric neurons), n = 12 (ME-Macs). Six experiments, Fisher’s exact test. Data are mean ± s.e.m. (error bars). FC, fold change; GI, gastrointestinal; NO, nitric oxide; ROS, reactive oxygen species; ThT, thioflavin T. Scale bars, 50 μm (a); 10 μm (b); 5 μm (f–j), insets 2 μm (f–j).

Source data

Inflammation has been linked to the initiation and progression of αS pathology, and many PD risk genes are enriched in immune cells7,19. Macrophages are the most abundant innate immune cells in the intestinal layers. ME-Macs have a crucial role in maintaining ENS integrity, with their loss leading to constipation8,9,14. We observed no significant differences in ME-Mac distribution across the duodenum, jejunum and ileum of 3-month WT versus 3KL animals (Extended Data Fig. 2a,b). Fluorescence-activated cell sorting (FACS) analysis of ME and lamina propria revealed unaltered differentiation of intestinal macrophages from monocyte precursors at 3 months, 4 months and 6 months of age13 (Extended Data Fig. 2c–f). By contrast, ME-Macs demonstrated increased expression levels of CSF1R and major histocompatibility complex class II (MHCII) (Extended Data Fig. 2g). We questioned whether ME-Macs, as professional phagocytes of the ENS, engage in the clearance of accumulating αS in surrounding myenteric neurons (Fig. 1f–j and Extended Data Fig. 2h,i). We assessed levels of lysosomal LAMP1 in ME-Macs and found an almost twofold increase in LAMP1 expression in ME-Macs from 3KL versus WT (Fig. 1h). We observed a roughly 14-fold increase in s129p+ αS engulfment by ME-Macs (that is, in LAMP1+ lysosomes) in 3-month 3KL mice compared with age-matched WT controls and a similar 14-fold increase in ME-Macs in postmortem jejunum of PD versus neurologically healthy control (NHC) participants (Fig. 1i,j and Supplementary Table 1a). Further, we found significant (P ≤ 0.05) differential expression of 474 proteins using mass spectrometry in sorted CX3CR1hiCD11clo ME-Macs of 3-month 3KL versus WT mice including lysosomal GRN, CTSB and CD68 (Fig. 1k and Extended Data Fig. 2k). Observed downregulated proteins included NDUFV3, NDUFB10, NDUFA4 and NDUFA7, part of the mitochondrial NADH-ubiquinone oxidoreductase complex that has been implicated in PD20 (Fig. 1k). Notably, 3-month 3KL ME-Macs acquire signatures related to phagosome maturation and lysosome pathways (collectively referred to as the coordinated lysosomal expression and regulation network)21, suggesting continuing clearance response in ME-Macs exposed to s129p+ αS and confirming our DSP data (Fig. 1l and Extended Data Fig. 2j).

We then asked whether s129p+ αS engulfed by 3KL ME-Macs shows altered pathological activity compared with WT ME-Macs and enteric neurons. We adapted the αS seed amplification assay (SAA) to isolated ME-Macs and enteric neurons to detect prion-like misfolded αS by means of templated aggregation of monomeric αS, which results in amplified binding to thioflavin T22 (Fig. 1m). Notably, lysates from isolated ME-Macs of 3-month 3KL demonstrated accelerated aggregation response within 10 hours in contrast to WT ME-Macs, suggesting increased aggregation activity in 3KL ME-Macs (Fig. 1n,o). No activity was observed in isolated enteric neurons from either 3-month 3KL or WT mice, despite the expression of total αS in enteric neurons but absence in ME-Macs (Fig. 1n–p and Extended Data Fig. 2l). The presence of αS aggregates in 3KL ME-Macs was further confirmed using the amyloid-binding dye Amytracker in isolated ME-Macs by flow cytometry (Extended Data Fig. 2m). In line with SAA data, we observed increased Amytracker signal in ME-Macs of 3-month 3KL versus WT mice but did not find positive signal in lamina propria Macs (Extended Data Fig. 2m,n). Together, these results indicate ME-Macs as potential modulators of αS amyloid aggregates in the ENS.

To further explore the gut–brain axis in synucleinopathy, we directly injected human PD brain-extracted αS fibrils (PD) versus identically extracted human neurological healthy control (NHC) brain preparations into ME of SncaWT/GFP knock-in mice3,4,23 (Extended Data Fig. 3a–e and Supplementary Table 1b). We used a purification protocol specifically designed to select for amyloid and other highly insoluble proteins24. The brain-extracted αS fibrils are hereafter designated as PD-αS and NHC-αS, respectively. Duodenal injection of PD-αS but not NHC-αS in ME increased endogenous αS expression in myenteric ganglia (Fig. 2a,b). We observed significant upregulation of s129p+ αS immunoreactivity at 1 month and 3 months postinjection, predominantly around neuronal cell bodies and to a lesser extent, in extraganglionic neurons, similar to 3-month 3KL (Fig. 2a–d). Prolonged total gut transit time was found post-PD-αS injection (Fig. 2e). Significant increase of s129p+ αS inclusions was observed at 3 months post-PD-αS injection in the brainstem (Fig. 2f,g). Further, an increased fraction of s129p+ αS in dopaminergic neurons of the substantia nigra pars compacta (SNpc) was observed, suggesting a potential route for pathological spread from the ENS to the CNS (Fig. 2h,i). We observed a selective loss of dopaminergic neurons in the SNpc at 3 months post-PD-αS versus NHC-αS injections (Fig. 2j,k and Extended Data Fig. 3f). By contrast, no dopaminergic neuronal loss was observed in the ventral tegmental area (VTA), consistent with the region-specific dopaminergic neuronal vulnerability in patients with PD25 (Fig. 2k). We then investigated whether ME-Macs engulfed s129p+ αS in response to PD-αS injections and observed a roughly 13-fold increase in lysosomal s129p+ αS in ME-Macs in PD-αS-injected animals compared with NHC-αS-injected controls at 1 month postinjection (Extended Data Fig. 3g–i). However, we found no overt production of inflammatory cytokines in the ME, suggesting the absence of local inflammation in the ME induced by αS injections (Extended Data Fig. 3j). These data indicate that body-first αS pathology, triggered by PD-αS injection into the ME, propagates from the ENS to the CNS, leading to region-specific dopaminergic neuron loss.

Fig. 2: ME injection of PD-extracts triggers αS pathology along the gut–brain axis.

a,b, αS s129p+ pathology in duodenal myenteric plexus at 1 month (a) and 3 months (b) post-NHC-αS versus PD-αS injection. c, Quantification of αS-GFPhigh cells per mm2 at 1 month and 3 months postinjection. n = 6 for αS-NHC, 1 month; n = 8 for αS-PD, 1 month; n = 5 for αS-NHC and αS-PD, 3 months. Two experiments. d, Quantification of αS s129p+ pathology at 1 month and 3 months postinjection. One datapoint, average of 1 mouse, 5–9 ROIs per mouse. n = 6 mice for αS-NHC, 1 month; n = 8 for αS-PD, 1 month; n = 5 mice for αS-NHC and n = 6 mice for αS-PD, 3 months. Two experiments. e, Gastrointestinal transit time at 1 month and 3 months postinjection. n = 7 mice for αS-NHC, n = 6 mice for αS-PD, 1 month; n = 7 mice for αS-NHC, n = 5 mice for αS-PD, 3 months. Two experiments, data analysed using two-way ANOVA (c–e) with Bonferroni’s multiple-comparison test (d). f, Widefield images of s129p+ αS pathology in brainstem at 3 months postinjection, blue arrows highlight αS inclusions. g, Quantification of s129p+ αS pathology in brainstem at 1 month and 3 months postinjection, n = 4 mice for αS-NHC, 1 month and n = 5 mice for αS-NHC, 3 months; αS-PD, 1 month and 3 months. Two experiments, two-way ANOVA with Bonferroni’s multiple-comparison test. h, αS s129p+ pathology in SNpc at 3 months post-NHC versus PD-extract injection. i, CTCF of s129p αS normalized to total αS, *n *= 5 mice per group. Two experiments, unpaired t-test. j, TH+ immunostainings in SNpc and VTA using DAB, 3 months post-treatment. brightfield micrograph of dopaminergic neurons in the SNpc. k, Unbiased stereological quantification of TH+ cells in SNpc and VTA, n = 7 mice for αS-NHC, SNpc and VTA, n = 8 mice for αS-PD, SNpc and VTA. Three experiments, mixed-effect model with multivariate t-distribution posthoc test. Data are mean ± s.e.m. (error bars). Scale bars, 100 μm (a,b), insets 20 μm (a,b,j); 20 μm (f,j); 50 μm (h); 500 µm (j). DA, dopaminergic.

Source data

A gut-to-brain T cell axis in PD

Recent studies reported increased CD3+ T cell numbers in intestinal biopsies of constipated patients with PD versus healthy control participants and in PD mouse models that show prodromal constipation26,27. Further, increased numbers of T cells are found in the plasma and brain of patients with PD, but with unknown origin5,6,28. In addition to endolysosomal pathways, proteomic signature analysis on isolated ME-Macs from 3KL mice revealed upregulation of antigen presentation and leukocyte migration pathways (Fig. 1l and Extended Data Fig. 2j). MHCII, which enables antigen presentation to T cell receptors (TCRs), was found increased on ME-Macs (Extended Data Fig. 2g). These data suggest a potential engagement of adaptive immune cells after ME-Mac uptake of pathological αS. Hence, we asked whether CD3+ T cells are expanded in the ENS in response to αS pathology. IHC on myenteric plexus of 3-month 3KL versus WT mice showed increased numbers of CD3+ T cells in 3-month 3KL mice (Fig. 3a–c). Of note, T cells were primarily found adjacent to myenteric ganglia (Fig. 3a,b). Flow cytometry confirmed increased proportions of CD3+ T cells among total CD45+ cells isolated from 3-month ME, with an increased CD4+ to CD8+ ratio in 3-month 3KL mice (Fig. 3d–e and Extended Data Fig. 4a). This is in line with previous studies demonstrating a predominant role for CD4+ T cell responses in αS pathology in mice and human PD29. Of note, T cell expansion was less obvious in the lamina propria (Extended Data Fig. 4b). Further, the increased proportions of CD3+CD4+ T cells in myenteric plexus were also observed in the ME PD-αS versus NHC-αS injection model, as demonstrated by IHC and flow cytometry (Fig. 3f–j). We assessed T cells in the ME of human patients with PD by performing IHC on postmortem jejunum and observed similar expansion of CD4+ T cells in myenteric plexus of PD versus NHC patients (Fig. 3k,l and Supplementary Table 1a). Together, these results show that CD4+CD3+ T cells regionally infiltrate the ME of PD mouse models and patient tissues.

Fig. 3: T cells expand in the ENS and travel to the CNS in synucleinopathy.

a,b, CD3+ T cells in duodenal myenteric plexus of 3-month WT (a) versus 3KL (b) mice. Inset, ROI for counting. c,d, CD3+ T cells in 3KL myenteric plexus per ROI, quantified by IHC (c) and FACS (percentage of ME CD45+ cells) (d). n = 6 mice (WT), n = 7 mice (3KL) (c) or n = 10 mice (WT), n = 12 mice (3KL) (d). Two (c) and four (d) experiments, data analysed using an unpaired t-test with Welch’s correction (c,d). e, FACS plots of CD4+CD8+ ME T cells in 3-month WT versus 3KL. f,g, CD3+ T cells in duodenal myenteric plexus post-NHC-αS (f) versus PD-αS (g). Inset, ROI used for counting. h,i, CD3+ T cells in 3KL myenteric plexus per ROI, quantified by IHC (h) and FACS (percentage of ME CD45+ cells) (i). n = 6–7 (h) or n = 4 (i) mice per genotype. Two experiments, unpaired i-test (h) with Welch’s correction (i). j, FACS plots of CD4+CD8+ ME T cells post-NHC-αS versus PD-αS. k,l, DAB staining of ME CD4+ T cells in jejunum of PD versus NHC patients (k) and quantification (l). n = 7–8 per disease status. Two experiments, unpaired t-test. m,n, CD3+ T cells in dura mater of 3-month 3KL versus WT (m) or 1 month post-NHC-αS versus PD-αS (n) quantified by FACS (percentage of total dura mater CD45+ cells), n = 4 mice (WT), n = 6 mice (3KL) (m) or n = 4 mice per treatment (n). Two experiments, unpaired t-test. o, Schematic of T cell photoconversion in VHD mice, 5 days post-NHC-αS versus PD-αS. p, Dendra-red CD3+ T cells in dura mater at 1 month post-NHC-αS versus PD-αS injection and UV versus non-UV photoconversion, n = 5–6 per genotype. Note minor leakage in the absence of UV illumination. Four experiments, two-way ANOVA. q, Dendra-red versus dendra-green CD3+ T cells in dura mater, 1 month post-PD-αS injection and UV illumination in ME. Representative of five mice (overview). r, Boxplot quantifying clonotype overlap between ME and dura mater through the Jaccard index (Methods, equation (1)) of NHC-αS- and PD-αS-injected groups for α (A) and β (B) TCR chains. Bonferroni-adjusted unpaired t-test with Welch’s correction. Data are shown as boxplots: the median is the central line; hinges indicate the first and third quartiles; whiskers extend to the most extreme values within the 1.5× interquartile range and points beyond the whiskers are plotted as individual outliers. s, Alluvial plots of the top ten most frequent T cell clone β chains in ME and dura mater of one representative mouse from PD-αS. t, Boxplot quantifying the clonotype overlap between the ME and dura mater through the expanded index (Methods, equation ( 3)) of NHC-αS- and PD-αS-injected groups for α (A) and β (B) TCR chains. Bonferroni-adjusted unpaired t-test with Welch’s correction. Data are mean ± s.e.m. (error bars). Scale bars, 500 μm (a,b,f,g); 50 μm (k); 1 mm (q); insets 50 μm (a,b,f,g); 25 μm (q).

Source data

Next, we asked whether T cells could distribute from the ENS to CNS according to the spatiotemporal distribution of αS pathology. Many studies have demonstrated infiltration of CD3+ T cells into the CNS in preclinical models of PD and postmortem tissue, but the peripheral site of origin remained elusive29. We first evaluated CD3+ T cell numbers in the dura mater, a peripheral-brain interface in the leptomeninges that allows access of T cells into the brain parenchyma30. We found that CD3+ T cells were significantly elevated in the dura mater of PD-αS-injected and 3KL mice (Fig. 3m,n and Extended Data Fig. 4c). To test whether dural CD3+ T cells could originate from the intestine, we injected PD-αS versus NHC-αS in Vav-H2B-Dendra2 (VHD) mice expressing H2B-dendra2 protein under control of haematopoietic Vav, allowing tracking of dendra2 photoconverted immune cells31 (Fig. 3o). Ultraviolet (UV) exposure of ME from VHD mice caused irreversible photoconversion in ME CD3+ T cells, Ly6G+ neutrophils and CD11b+ myeloid cells at 24 hours post-UV illumination (Extended Data Fig. 4d–f). UV exposure did not label circulating CD3+ T cells in the blood at 1 hour post-UV (Extended Data Fig. 4g). We characterized photoconversion kinetics in T cells at 3 days, 7 days and 31 days post-UV illumination and PD-αS injection and detected circulating dendra-red+ T cells in the blood between 3 days and 7 days postillumination but declining at 31 days, whereas photoconverted T cells were increasingly observed in the dura mater between 7 days and 31 days postillumination (Fig. 3p and Extended Data Fig. 4h,i). Notably, roughly 2% of lymphocytes are dendra-negative, probably reflecting incomplete Vav promoter efficiency. A small percentage of T cell photoconversion was detected in the lamina propria, highlighting their potential contribution to the dura mater T cell pool (Extended Data Fig. 4h). Minor dendra-red cell presence was observed in dura mater of control illuminated αS-treated mice, indicating slight leakiness of the model (Fig. 3p and Extended Data Fig. 4j). In addition, photoconverted CD64+ cells were detected in the dura mater of PD-αS-injected mice, suggesting migration of myeloid cells from the duodenum (Extended Data Fig. 4k). Whole-mount imaging of the dura mater of PD-αS-injected mice confirmed presence of dendra-red CD3+ T cells (Fig. 3q and Extended Data Fig. 4l,m). Together, these data indicate expansion of ME T cells in the context of αS pathology, and that a proportion of intestinal T cells migrates to the dura mater.

To confirm that T cells in the ME and brain were clonally related, and to measure clonal expansion, we performed TCR sequencing on ME and dura mater, striatum and hippocampus samples collected 10 days post-NHC-αS and PD-αS injection. The number of TCRs recovered from striatum and hippocampus at this early time point was very small, so we focused analysis on the dura mater. The diversity (inverse Simpson) of the repertoires showed a trending decrease following PD-αS injection, suggestive of antigen-driven clonal expansion (Extended Data Fig. 5a). In agreement with the increased dendra-red CD3+ T cells observed in PD-αS-injected mice, there was an increase in TCR sequence sharing between ME T cells and dura mater T cells following injection of PD-αS (Fig. 3r). The degree of overlap was greater for TCRα than for TCRβ, reflecting the higher diversity of the TCRβ32. Many of the TCRs shared between ME and dura mater showed significant clonal expansion, consistent with previous antigen recognition (Fig. 3s and Extended Data Fig. 5b,c). A trend for increased sharing of expanded ME T cells to the dura mater in PD-αS-injected versus NHC-αS-injected mice was observed (Fig. 3t). Taken together, these data indicate expansion of ME T cells in the context of αS pathology, and that a proportion of expanded intestinal T cells migrates to the dura mater, potentially mediated by PD-αS.

ME-Mac–T cell crosstalk through TGFβ1

We next explored whether the observed T cell expansion involved crosstalk with ME-Macs that are reactive to αS pathology. IHC and 3D reconstruction revealed close proximity between ME-Macs and T cells in the myenteric plexus of 3-month 3KL mice and patients with PD (Fig. 4a). To elucidate transcriptional changes in ME-Macs and T cells on αS pathology and cellular interactions among them, we performed single-cell RNA sequencing (scRNA-seq) using the 10X Genomics platform on fluorescence-activated cell-sorted duodenal ME CD3+ cells and ME-Macs of 4-month 3KL and WT mice, a time point chosen to capture profiles after initial T cell expansion (Fig. 4b and Extended Data Fig. 6a,b). After quality control, we ran unsupervised clustering on 2,748 cells and identified clusters of ME-Macs (Adgre1 and Csf1r) and T cells (Cd3 and Cd8a) (Fig. 4b and Extended Data Fig. 6b–d). We found three transcriptionally distinct ME-Mac subclusters based on Cd163 and Ccr2 expression, resembling previously described tissue-resident macrophage subtypes associated with blood vessels (Lyz2, Mrc1, Pf4) and neurons (H2-Dma,* H2-Aa*, Cd74)33 (Fig. 4b, Extended Data Fig. 6d and Supplementary Table 2). Cd163*+* ME-Macs expressed transcripts involved in leukocyte migration and chemotaxis, whereas Ccr2*+* ME-Macs were enriched in transcripts related to antigen processing (Extended Data Fig. 6e). The third cluster, hereafter referred to as Cd163*−Ccr2−* ME-Macs, demonstrated upregulation of genes related to leukocyte migration and cytokine signalling. We validated the presence of CD163+, CCR2+ and CD163−CCR2− IBA1+ ME-Macs using flow cytometry and IHC (Fig. 4c,d and Extended Data Fig. 6f–l). Flow cytometry revealed no differences in absolute subcluster numbers between WT and 3KL mice at 3–4 months (Extended Data Fig. 6h). However, CCR2+ ME-Macs were generally significantly fewer in 3KL mice, reaching significance at 6 months. IHC showed altered proportions of CD163+ and CD163−CCR2− ME-Macs in 3-month-old 3KL mice but no differences in PD-αS- versus NHC-αS-injected animals (Extended Data Fig. 6i–l).

Fig. 4: ME-Macs and T cells interact by means of TGFβ1 in body-first PD models.

a, Confocal images of ME-Macs and ME T cells in 3-month 3KL and PD postmortem ME. Representative of more than four experiments. b, Uniform manifold approximation and projection (UMAP) of unsupervised clustering of ME-Macs and T cells from ME assigned into colour-coded subclusters. scRNA-seq data obtained from fluorescence-activated cell-sorted ME-Macs and ME CD3+ cells of 4-month 3KL and WT subjected to 10X Genomics scRNA-seq (n = 2,748 cells). Four biologically independent samples were used, and samples were sequenced in n = 2 batches from WT versus 3KL. One sample represents ME pooled from 4 mice. c,d, FACS plots (c) and confocal images (d) demonstrating CD163+, CCR2+ and CD163−CCR2− duodenal ME-Macs in 3-month WT versus 3KL (c) or 3KL (d). Data representative of three experiments. e,f, Confocal images (e) of s129p engulfment by CD163+ and CCR2+ duodenal ME-Macs in 3-month WT versus 3KL and quantification (f), n = 4 per genotype. One experiment, two-way repeated measures ANOVA. g, Heatmap showing top ligands expressed in ME-Macs ranked on the basis of area under the precision–recall curve (AUPRC) with 3KL T cells as receivers. h, Circos plot showing ligand–receptor pairs between ME-Macs and T cells in 4-month 3KL. Bottom, top eight ligands expressed by 3KL ME-Macs. Top, differentially expressed receptors in 3KL versus WT T cells. i,j, Confocal images (i) of duodenal ME T cells in Cx3cr1*+/+.Tgfb1LoxP* versus Cx3cr1CreERT2.Tgfb1**LoxP at 10 days post-NHC-αS versus PD-αS injection and quantification (j), n = 4 mice (CTRL, NHC-αS and PD-αS; FLOX, NHC-αS) and n = 5 mice (FLOX, PD-αS). Two experiments, two-way ANOVA, Bonferroni’s multiple-comparison test. Data are mean ± s.e.m. (error bars). Scale bars, 30 μm (3KL, a); 50 μm (PD, a); 100 μm (d); 50 μm (i). AUPRC, area under the precision–recall curve.

Source data

We next examined whether CD163+ and CCR2+ ME-Macs showed increased capacity to engulf s129p+ αS in PD models. In 3-month 3KL mice, both populations showed increased αS engulfment compared with WT (Fig. 4e,f). Similarly, in the injection model, αS engulfment was increased in both ME-Mac subsets 1 month post-PD-αS injection (Extended Data Fig. 6m,n). These findings indicate that ME-Macs show an enhanced ability to internalize αS in the context of PD-associated pathology.

To evaluate how ME-Macs could modulate T cells transcriptionally, we used NicheNet on our scRNA-seq data to predict ligand–receptor interactions between ME-Macs and T cells34 (Fig. 4g,h and Extended Data Fig. 7a,b). We focused on disease-specific interactions under 3KL conditions, selecting receptor transcripts that were differentially expressed in T cells and corresponding ligands expressed by 3KL ME-Macs. The ligand with the highest regulatory potential was Tgfb1, which was predominantly expressed by CCR2+ and CD163+ ME-Macs, whereas its canonical receptor Tgbr2 was expressed by ME T cells35 (Fig. 4g,h and Extended Data Fig. 7c). We next investigated the role of TGFβ1 in ME-Mac–T cell crosstalk in the context of αS pathology. Tamoxifen-treated Cx3cr1CreERT2.Tgfb1**LoxP mice were injected with NHC-αS or PD-αS injections 4 months later, when Tgfb1 deletion is retained only in self-maintaining macrophages8 (Extended Data Fig. 7d–f). Notably, PD-αS failed to induce T cell expansion in Cx3cr1CreERT2.Tgfb1**LoxP mice, unlike in Cx3cr1*+/+.Tgfb1LoxP* and NHC-αS controls (Fig. 4i,j). Whereas absolute IBA1+ ME-Mac counts, CD163+ or CCR2+ proportions, and lysosomal numbers remained unchanged, a proportional increase in CD163−CCR2− IBA1+ ME-Macs was observed in both genotypes (Extended Data Fig. 7g–j). T cell numbers in the lamina propria were unaffected12 (Extended Data Fig. 7k,l).

ME-Mac targeting reduces PD pathology

We next asked whether ME-Macs could directly modulate the expansion of T cells to the brain in response to αS pathology. We aimed to deplete ME-Macs by injecting anti-CSF1R antibody (AFS98) into duodenal ME at 24 hours before αS treatment9 (Fig. 5a and Extended Data Fig. 8a). We combined anti-CSF1R with anti-CCR2 antibody (MC21) treatment to circumvent immediate monocyte replenishment and to extend the depletion of ME-Macs in the myenteric plexus8. Of note, this approach resulted in minor depletion of other immune cells, including reduced numbers of ME monocytes, ME eosinophils and lamina propria monocytes at 5 days post-treatment, as well as decreased circulating neutrophils at 5 days and circulating monocytes at 24 hours and 5 days post-treatment (Extended Data Fig. 8b,c). Microglia and border-associated macrophages were not affected by this treatment (Extended Data Fig. 8d). T cell expansion on PD-αS versus NHC-αS injection was abolished in the myenteric plexus at 1 month post-anti-CSF1R/anti-CCR2 versus IgG treatment, suggesting that ME-Macs are potentially involved in the T cell expansion in αS models (Fig. 5b,c). Further, we observed amelioration of s129p+ αS pathology at 1 month post-PD-αS injection in mice pretreated with anti-CSF1R/anti-CCR2 compared with IgG, suggesting the contribution of ME-Macs in the progression of αS pathology along the ENS (Fig. 5d). Given that fractions of T cells could travel from the ME to the