C1q-dependent clearance of alpha-synuclein allows macrophages to transiently limit enteric synucleinopathy in male mice

Abstract

Deposition of misfolded α-synuclein (αsyn) in the enteric nervous system (ENS) is found in multiple neurodegenerative diseases. It is hypothesized that ENS synucleinopathy contributes to both the pathogenesis and non-motor morbidity in Parkinson’s Disease (PD), but the cellular and molecular mechanisms that shape enteric histopathology and dysfunction are poorly understood. Here, we employ a fibrillar injection model of enteric synucleinopathy in male mice and demonstrate that ENS-resident macrophages, which play a critical role in maintaining ENS homeostasis, initially respond to enteric neuronal αsyn pathology by upregulating machinery for complement-mediated engulfment. Pharmacologic depletion of ENS-macrophages or genetic deletion of C1q enhanced enteric neuropathology. Conversely, C1q deletion ameliorated gut dysfunction, indicating that complement partially mediates αsyn-induced gut dysfunction. However, this C1q-dependent clearance mechanism diminished over time and its failure temporally correlated with the further increase in ENS pathology. These findings highlight the importance of enteric neuron-macrophage interactions in removing toxic protein aggregates that putatively shape the gastrointestinal manifestations of PD.

Misfolded α-synuclein in the gut contributes to Parkinsons’s disease pathogenesis and gut symptoms. Gut macrophages clear α-synuclein via C1q–complement pathways; loss of this response worsens neuropathology and improves gut function, implicating neuroimmune crosstalk.

Introduction

Aggregates of α-synuclein (αsyn) are found throughout the enteric nervous system (ENS) in both Parkinson’s Disease (PD) and Lewy Body Dementia (LBD)^1–5. It is widely hypothesized thatENS αsyn pathology may precede central nervous system (CNS) pathology in a subset of PD patients, and via prion-like propagation, spreads to vagal fibers and subsequently to the brain^6–10. Moreover, ENS Lewy pathology is thought to contribute to decreased intestinal motility and, by extension, to symptoms such as severe constipation and delayed gastric emptying which can incur high morbidity in PD and LBD^11–16. Thus, enteric synucleinopathy has important implications for pathogenesis and pathophysiology of CNS synucleinopathies.

Several reports have recently shown that gut-seeded αsyn pathology can spread to the brain in animal models^17–20. These studies have focused on the kinetics, anatomical routes, and modulators of the progression from gut-to-brain. However, relatively little is known about the factors that initially shape the local histopathology within the ENS and associated functional deficits of enteric synucleinopathy.

The ENS contains specialized macrophages that play important roles in homeostasis, development, and infectious states^21–26. Given the role for CNS macrophages in the progression of PD pathology^27–29, it has been hypothesized that these ENS macrophages may play a role in the spread of enteric αsyn aggregates to the CNS³⁰. CNS-resident macrophages have been shown to engage in both beneficial and detrimental functions during neurodegenerative diseases and represent promising targets for new therapies^31–33. Prominent examples include activating the adaptive immune response, which can lead to neuronal death^34–36, and microglia-mediated complement-dependent synaptic pruning^37–40, which mechanistically links protein aggregates to neuronal dysfunction. Given the range of states macrophages can adopt, and the disease-modifying potential of macrophage phenotype, it is important to elucidate the precise response of ENS macrophages to synucleinopathy and how this response reciprocally contributes to αsyn pathology in enteric neurons.

Here we show induction of enteric synucleinopathy disrupts enteric neuron-macrophage interactions and elicits a macrophage response characterized by upregulated engulfment machinery, specifically complement-related genes. Macrophages utilize complement to clear αsyn from nearby neurons and constrain the spread of αsyn pathology. However, this mechanism facilitates intestinal dysmotility and eventually fails as synucleinopathy spreads locally within the ENS, although pathology does not ascend to the brainstem. Overall, our results reveal a mechanism by which ENS-resident macrophages sculpt the histopathological and functional manifestations of enteric synucleinopathy and shed light on potential mechanisms underlying the gastrointestinal aspects of Parkinson’s Disease.

Results

PFF-induced enteric synucleinopathy alters ENS neuron-macrophage interactions

To assess the potential role for the gut-resident immune system in modulating local or distal synucleinopathy, we employed a recently described gut-first model of α-synuclein (α-syn) pathology¹⁷ by injecting pre-formed fibrils (PFF) of mouse recombinant α-syn directly into the stomach and duodenal wall of wild-type (WT) mice (Fig. 1a, b). The PFF had a mean length of 39.37 nm (standard deviation 17.69, Fig. 1a). We first assessed the kinetics for local seeding of α-syn pathology within the enteric nervous system using whole-mount immunofluorescent microscopy at 4, 14, 30, and 60 days post injection (dpi), and found that enteric synucleinopathy, defined as phospho-Serine 129 α-syn+ enteric soma, was first observed within the duodenal myenteric plexus of PFF-injected mice at 14dpi, and further increased through 30dpi and 60dpi (Fig. 1c, d). We focused on duodenal myenteric plexus and did not assess for pathology further along the GI tract. The establishment and initial development of enteric synucleinopathy was observed without changes to total neuronal area (Supplementary Fig. 1a, b). These data confirmed that duodenal PFF injections induce local progressive α-syn pathology within enteric neurons over a 2-month time period.

Fig. 1. PFF-induced enteric synucleinopathy alters local neuroimmune interactions.

a Representative transmission electron microscope images of mouse recombinant α-synuclein PFFs (left) and quantification of fibril length (right) across 3 independent samples. b Cartoon depiction of intestinal PFF injections. Created in BioRender. Mackie, P. (2026) https://BioRender.com/d3wzc1k. c Representative confocal images of the myenteric plexus (Tuj1, pseudo colored magenta) and pSer129 α-synuclein (pseudo colored yellow) at 4-, 14-, 30-, and 60-days post injection (dpi) with either saline or PFFs; scale bar: 50 μm. d Counts of the average number of pSer129+ neuronal soma/ganglia within the myenteric plexus shows PFF but not saline injection induces neuronal pSer129 pathology starting at 14dpi, with a time-dependent increase to 30dpi and 60dpi. Two-way ANOVA with Tukey’s multiple comparison test: interaction p < 0.0001, injection p < 0.0001, time p < 0.0001; n = 3–5 animals/group/time point. e Thresholded and binarized confocal images depicting the contact sites (white) between the myenteric plexus (magenta mask) and myenteric macrophages (cyan mask) at 14dpi and 30dpi; scale bar: 50 μm. f Quantification of the macrophage-neuron contact expressed as the percent of ganglia area covered. Two-way ANOVA with Tukey’s multiple comparison test; condition p = 0.0007, n = 4–5 animals/group/time point. g Representative orthogonal image of a myenteric macrophage with an internalized punctae of pSer129; scale bar: 10 μm. h Quantification of the average pSer129+ punctae size in myenteric macrophages at 30dpi showing a statistically significant increase in PFF-injected mice. Two-sided unpaired t-test, p < 0.05; n = 5 animals/group. i Whole gastrointestinal transit time measured via Carmine Red assay showing slowed whole gut motility in PFF-injected mice at 30dpi. Two-sided unpaired t-test p < 0.05, n = 8–10 animals/group. j PFF-injected mice also had decreased stool output at 30dpi as measured by pellet number and weight. Two-sided Mann–Whitney test p < 0.01 (weight and count), n = 5–7 animals/group. Graphs in (d), (f), (h), (i), (j) represent mean +/− SEM.

We then examined how development of PFF-induced neuronal α-syn pathology altered local neuro-immune interactions between enteric neurons and resident macrophages. ENS-resident macrophages made contact with enteric ganglia, as previously described²⁴ (Fig. 1e). At 14dpi, the amount of macrophage-neuron contact was the same between both groups. However, by 30dpi, enteric macrophages in PFF-injected mice displayed significantly increased contacts with enteric ganglia compared to 30dpi saline controls (Fig. 1e, f). Importantly there was no difference in total macrophage coverage area (Supplementary Fig. 1c, d). Moreover, at 30dpi, punctae of pSer129 α-syn within enteric macrophages were clearly detectable, and these punctae were significantly increased in the PFF-injected group compared to control (Fig. 1g, h). We did not observe any changes to enteric macrophage morphology at the 30dpi timepoint (Supplementary Fig. 1f–h). Thus, local neuronal α-syn pathology is associated with altered macrophage-neuron interactions by 30dpi.

To further investigate the functional significance of enteric neuronal α-syn pathology, we assessed gastrointestinal transit times (GTT). Similar to previously published results¹⁸, PFF-injected animals had significantly slower gastrointestinal transit (Fig. 1i). This was also correlated with decreased fecal output (Fig. 1j). These data are consistent with the interpretation that α-syn pathology disrupts ENS function and that both GTT and fecal output assays have sufficient sensitivity to detect this dysfunction.

We next sought to characterize the gut-to-brain spread of α-syn pathology, which has been reported to varying degrees in this model^17–19. We did not observe overt pSer129 pathology within the Dorsal Motor Nucleus of the Vagus (DMV) or midbrain dopaminergic nuclei at 30dpi or at 60dpi (Supplemental Fig. 1e). Thus, gastrointestinal PFF injections induce α-syn pathology confined to the ENS, altering both local neuroimmune interactions and ENS function in the absence of CNS pathology.

Macrophage depletion exacerbates early enteric α-syn pathology

Given the effects of PFF injection on ENS macrophages, we next asked if macrophages had a functional role in shaping the development of enteric neuronal α-syn pathology. To do this, we employed a macrophage depletion model²⁴ in which we injected mice with a monoclonal anti-CSF1R antibody at low doses. Antibody injection depleted ENS macrophages by 7 days post injection, and macrophages remained depleted for 21days post-injection (Supplementary Fig. 2a,b). Interestingly, we still observed macrophages in the lamina propria 7 days after depletion, but not in the muscularis (Supplementary Fig. 2c). Also as expected, anti-CSF1R antibody injection did not deplete CNS microglia (Supplementary Fig. 2d).

After characterizing depletion kinetics, we first injected mice with anti-CSF1R antibody then performed gastrointestinal PFF injections 7 days later (Fig. 2a). At 30days post-PFF injection, we quantified pSer129 α-syn+ neurons. As expected, PFF injection alone significantly increased the number of pSer129 α-syn+ neurons compared to saline controls; however, PFF injection+ macrophage depletion resulted in a further increase in neuronal α-syn pathology compared to PFF injection alone (Fig. 2b, c).

Fig. 2. Macrophage depletion enhances enteric α-synuclein pathology.

a Schematic of design for single macrophage depletion experiments. Created in BioRender. Mackie, P. (2026) https://BioRender.com/zfb68ch. b Representative confocal images of myenteric ganglia (Peripherin, pseudo colored magenta) with pSer129 α-synuclein pathology (pseudo colored yellow) at 30dpi; scale bar: 20 μm. c Counts of average number of pSer129+ neuronal soma per myenteric ganglia replicating that PFF injection induces neuronal pSer129+ α-synuclein pathology at 30dpi and additionally showing macrophage depletion with PFF injection further increases neuronal pSer129 α-synuclein burden. One-way ANOVA (p < 0.0001) with Sidak’s test for multiple comparisons; n = 3–5 animals/group. d Cartoon of design for double macrophage depletion experiments. Created in BioRender. Mackie, P. (2026) https://BioRender.com/h0rzp0u. e Representative confocal images of myenteric ganglia (Peripherin, magenta) with pSer129 α-synuclein pathology within ganglia at 30dpi; scale bar: 20 μm. f Quantification of pSer129 α-synuclein pathology in myenteric soma showing that PFFs alone induces an increase in pathology compared to saline and that PFF injection with double macrophage depletion further increases pathology. One-way ANOVA with Tukey’s test for multiple comparisons p < 0.0001, n = 4–5 animals/group. Graphs in (c) and (f) represent mean +/− SEM.

However, we also observed repopulation of ENS macrophages by 30 days post-PFF injection (Supplementary Fig. 2a, b) in our single depletion study. Therefore, we next performed a double depletion study in which we first injected anti-CSF1R antibody, performed our PFF injections 7 days later, then re-depleted macrophages with an additional anti-CSF1R antibody injection at 14 days post PFF injection (21 days after original macrophage depletion) (Fig. 2d). This resulted in sustained depletion at 30 days post PFF injection (Supplementary Fig. 2e). Sustained macrophage depletion yielded similar results as single depletion—PFF injection alone increased enteric α-syn pathology, and PFFs+ depletion resulted in a further significant increase in neuronal pSer129 α-syn+ neurons compared to PFFs alone (Fig. 2e, f). Thus, macrophage depletion significantly enhances ENS α-syn pathology in early stages, suggesting a potential protective effect of ENS macrophages at initial time points.

Single-cell transcriptomic identification of muscularis macrophages

The potential role of ENS macrophages in shaping enteric α-syn pathology prompted us to further investigate the molecular landscape of intestinal macrophage populations during development of synucleinopathy. To this end, we employed both FACS enrichment of live CD45+ cells and magnetic isolation of live CD11b+ cells from digested mouse duodenum at 30dpi and performed single-cell RNA sequencing using the 10x Genomics platform (Supplementary Fig. 3). Following quality-control filtering, we identified 27,648 high-quality cells which separated into 10 transcriptionally distinct clusters (Supplementary Fig. 4a–c, Supplementary Data 1). To conduct a more thorough analysis on the macrophage transcriptional response to synucleinopathy, we sub-selected cells enriched for macrophage/myeloid markers Cx3cr1, Aif1, and Lyz2 and re-clustered them at a higher resolution which yielded 9 transcriptionally-distinct macrophage clusters (Supplementary Fig. 4d, Fig. 3a, b, Supplementary Data 2). We confirmed macrophage identity by validating that these clustered expressed at least one of the structural macrophage markers Itgam, Cx3cr1, Aif1, or Lyz2 (Supplementary Fig. 5a).

Fig. 3. scRNA-seq identifies muscularis macrophage subsets.

a Uniform manifold approximation and projection (UMAP) plots of subsetted gut macrophages from saline- and PFF-injected mice at 30dpi showing 9 transcriptionally distinct clusters of macrophages. b Dot plot of dendritic cell and macrophage clusters showing the top 2–3 marker genes identifying each cluster. c Violin plots of genes that were differentially expressed between Mp2/Mp6 and Mp1/Mp3/Mp4/Mp8. d 60× confocal image with 2× digital zoom of muscularis propria in duodenal cross section showing macrophages (IBA1, green) expressing C1a (red). Experiments repeated in 3 separate animals, 2 sections/animal; scale bar: 5 μm. e 60× confocal image with 2× digital zoom of muscularis propria in duodenal cross section showing macrophages (IBA1) lack expression of S100a11 (red). Experiments repeated in 3 separate animals, 2 sections/animal; scale bar: 5 μm. f UMAP of 3 integrated scRNA-seq datasets: this paper (Saline and PFF), Chiaranunt et al. 2023, and De Schepper et al. 2019 after subsetting macrophages identifying 9 distinct clusters. g Dot plot of clusters from integrated dataset showing expression top 2–3 marker genes/cluster. h Composition of each cluster broken down by original study. i Feature plot of composite Mp2 gene signature C1qa, C1qb, C1qc, Apoe, Cd81, Wfdc17, Npl, Lyz2, Csf1r, and Msr1 projected onto the integrated dataset (left) and feature plot of composite Mp6 gene signature C1qa, C1qb, C1qc, Mmp13, Pf4, Wnt4, Ptgs1, Cd63, C5ar1, and C3ar1 projected onto the integrated dataset (right). Both gene signatures were strongly expressed by clusters 2 and 4. j Feature plots of the genes differentially expressed between our putative muscularis and mucosal macrophages projected onto the integrated dataset.

Our first challenge was to determine which macrophage clusters represented muscularis macrophages. We examined the expression levels of genes previously reported to be differentially expressed between muscularis vs mucosal macrophages as well as genes differentially expressed between our two larger neighborhoods—Mp1, Mp3, Mp4, and Mp8 vs Mp2 and Mp6 (Supplementary Fig. 5b, c, Fig. 3c, Supplementary Data 3). Some genes such as C1qa, C1qb, and C1qc, which were recently reported to be enriched in muscularis macrophages, were significantly upregulated in Mp2, Mp6 and Mp9. Other genes enriched in these clusters included transcription factor Ms4a7; F11r, which was recently shown to be specifically increased in ENS-resident macrophages²³; and Adamdec1, which was reported to be upregulated in blood vessel-associated macrophages²² (Fig. 3c, Supplementary Fig. 5c). Furthermore, Ctss, a canonical microglia-associated gene that is increased in other peripheral nerve-associated macrophages, was relatively enriched in Mp2 and Mp6 compared to other clusters (Supplementary Fig. 5c). Csf1r, which is specifically required for muscularis macrophage survival²⁴, was also expressed highly in Mp2 and Mp6 (Supplementary Fig. 5c). From these data and supporting literature, we hypothesized that Mp2 and Mp6 likely represented muscularis macrophages. To validate this notion, we performed IHC on cross-sections of mouse duodenum. Based on our scRNA-seq data, we defined Mp2 and Mp6 as C1q+ and S100a11− (Fig. 3c). Thus, we predicted muscularis macrophages to be C1q+ but S100a11−. Via double immunolabelling with either IBA1 and C1q or IBA1 and S100a11. We observed IBA1+/C1q+ but not IBA1+/S100a11+ cells in the muscularis layer (Fig. 3d, e) Conversely IBA1+/S100a11+ cells were found throughout the mucosal layer (Supplementary Fig. 5e). Of note, we did observe IBA1+/C1q+ cells throughout the mucosa as well (Supplementary Fig. 5d), indicating this is not a specific marker for muscularis macrophages contrary to recent reports⁴¹. Based on these spatial data, we concluded that Mp2 and Mp6 likely represented muscularis macrophages.

To further analyze these macrophage subsets, we pooled our data with data from two other published scRNA-seq studies of intestinal macrophages (Supplementary Fig. 5f)^22,42. Chiaranunt et al. enriched their dataset for colonic lamina propria macrophages, while De Schepper et al.’s dataset is mostly long-lived ileal macrophages. After integrating the data and sub-setting out macrophages based on Aif1, Cx3cr1, and Lyz2 expression (Supplementary Fig. 5g), we identified 9 clusters that still displayed substantial separation based on the original study (Fig. 3f–h, Supplemental Data 4). Cluster 0 represented Lyve1+ macrophages and was composed of cells from the Chiaranunt et al. study and was highly similar to the cluster 0 identified in that study. Similarly, clusters 1 and 3 were marked by high Ccr2 expression and were also mostly comprised of cells from Chiaranunt et al. Cluster 2, on the other hand, was mostly comprised of cells from the De Schepper et al. and thus represented self-maintaining macrophages. Cluster 5 was mostly comprised of cells from this study. Clusters 4 and 7 contained cells from multiple studies. Notably, Cluster 4 was defined by high expression of genes such as Adamdec1 and Ctss (Supplementary Fig. 5h), suggesting these cells may represent vasculature and nerve-associated macrophages. Overall, these analyses likely reflect both technical variation and the role that anatomic location along the GI tract and ontogeny play in the transcriptional state of macrophages while highlighting that some gene signatures are conserved across multiple studies.

We also examined the expression of genes that we had identified as distinguishing muscularis from mucosal macrophages (Fig. 3c and Supplementary Fig. 4a, b). Genes that we identified as marking mucosal macrophages such as S100a11, Tmsb10, Napsa, and Lsp1, were highly expressed in clusters 1, 3, 5, and 7 and were absent from clusters 2 and 4 (Fig. 3j, Supplementary Fig. 5h). Notably clusters 1 and 3 were from a study that enriched for lamina propria macrophages, supporting the notion that genes like S100a11 mark mucosal/lamina propria macrophages.

Conversely, genes enriched in our muscularis macrophages such as Ms4a7, Selenop, Adamdec1, and Ctss were preferentially or exclusively expressed in cluster 4 as well as clusters 0 and 2 (Fig. 3j, Supplementary Fig. 5h). Additionally, we computed scores based on the core gene signature of our Mp2 cluster: C1qa, C1qb, C1qc, Apoe, Cd81, Wfdc17, Npl, Lyz2, Csf1r, and Msr1. We projected these scores onto our integrated UMAP and found that clusters 2 and 4 strongly exhibited this signature (Fig. 3i). Similarly, clusters 2 and 4 were also enriched in the core gene signature of our Mp6 cluster: C1qa, C1qb, C1qc, Mmp13, Pf4, Wnt4, Ptgs1, Cd63, C5ar1, and C3ar1 (Fig. 3i). These data indicated that cluster 4 in the integrated dataset likely contained our Mp2 and Mp6 macrophages and therefore potentially represented muscularis macrophages. Thus, our Mp2 and Mp6 clusters, which we identified as muscularis macrophages, were transcriptionally similar to but still distinct from the self-maintaining macrophages identified by De Schepper et al. Overall, these comparative analyses supported our identification of Mp2 and Mp6 as muscularis macrophages.

Enteric synucleinopathy is associated with upregulation of engulfment and clearance machinery in muscularis macrophages

After determining that Mp2 and Mp6 represented muscularis macrophages, we next investigated the potential effects of PFF injection on macrophage transcriptome. Analyzing DEGs between saline- and PFF-injected mice across all Mp clusters revealed modest perturbations in macrophage transcripts such as upregulation of iron-related genes like Ftl1, chemotactic genes Ccl5, and lysosomal gene Cd63 in macrophages from PFF-injected mice. Conversely, inflammasome-related genes such as Nlrp1b and Il1r2 were downregulated (Fig. 4a, Supplementary Data 5). Interestingly, Cd63 was preferentially expressed in Mp6 and, to a lesser extent, Mp2 compared to other Mp clusters (Fig. 4b). Additionally, Ftl1, while widely expressed, was enriched in Mp2 and Mp6 whereas Nlrp1b, which was downregulated in PFF-injected mice, was selectively depleted in Mp2 and Mp6 (Fig. 4b). Consistent with these observations, when we analyzed the composition of our saline and PFF groups by cluster, we found that Mp6 was significantly enriched in our PFF group (Fig. 4c). Thus, based on these data, PFF injections preferentially affected the transcriptomes of Mp2 and Mp6—our muscularis macrophage clusters—with Mp6-like muscularis macrophages being enriched in PFF-injected mice.

Fig. 4. Enteric synucleinopathy is associated with upregulation of clearance pathways in muscularis macrophages.

a Volcano plot of differentially expressed gene (DEG) analysis between all macrophage clusters from PFF-injected mice vs all macrophage clusters from saline-injected mice reveals 83 significantly upregulated genes including Ftl1, Cd63, and Ccl5, and 21 significantly downregulated genes including Nlrp1b, Il1r2, and Gab2. Statistical analysis performed with Wilcoxon-rank sum test with Bonferroni correction. b Feature plots of top DEGs between PFF and saline groups projected on our dataset. Cd63 and Ftl1, which were upregulated in the PFF group, showed highest expression in Mp6 and, to a lesser extent, Mp2. Nlrp1b, which was downregulated in the PFF-group, had the lowest expression in Mp2 and Mp6. c Composition of the clusters by treatment group indicating that Mp6 was significantly enriched in the PFF group relative to the saline group. Fisher’s exact test with Bonferroni correction, p < 0.0001. Gene Ontology Biological Process (GO BP) analysis on the significantly expressed marker genes of Mp2 (d) and Mp6 (e) indicating that Mp2 was mainly characterized by enrichment of terms relating to the multi-vesicular body system including processing by the lysosome while Mp6 was mainly characterized by enrichment for antigen-presentation terms and terms relating to synapse elimination including pruning and negative regulation of long-term potentiation. These data were compiled from n = 4 to 6 mice per group.

We then functionally annotated Mp2 and Mp6 using Gene Ontology. Our Mp2 cluster was highly enriched in terms related to the multivesicular body-lysosome pathway, which has been implicated in α-syn aggregation⁴³, whereas Mp6 was highly enriched in antigen presentation terms, synaptic pruning, and negative regulation of long-term potentiation (Fig. 4d, e). The synaptic pruning term was corroborated by high expression of C1q genes, C3, and early complement component receptors in Mp6 macrophages (Fig. 3b, Supplementary Data 2). Additionally, the transcriptomic signature of Mp6 had similar features when compared to the Disease Associated Microglia (DAM) signature⁴⁴, particularly genes related to engulfment and lysosomal degradation such as Ctsb, Fcrg1, Cd68, and Lyz2 (Supplementary Fig. 5i). These findings support the role of macrophage-mediated engulfment and degradation in neurodegenerative diseases. Overall, these data suggest that enteric αsyn pathology modestly alters the macrophage transcriptome and lead to an upregulation of genes related to clearance, preferentially in muscularis macrophages.

Complement-dependent engulfment of α-syn by macrophages couples enteric pathology and dysfunction

Thus far, our data indicate that myenteric macrophage may have a role in clearing α-syn pathology in the ENS. Specifically, Mp2 and Mp6 had high expression of complement system genes. Furthermore, Mp6, which was preferentially enriched in PFF-injected mice, displayed upregulation of genes related to synapse elimination—which can be a complement-dependent process⁴⁵. Therefore, we then validated the C1q expression indicated by our scRNA-seq on a protein and spatial level, as C1q genes were most highly expressed by Mp6 and, to a lesser extent, Mp2. Using whole-mount immunofluorescence, we confirmed that ENS macrophages expressed C1q protein, and that expression was increased in PFF-injected mice (Fig. 5a, b). Interestingly, we observed C1q+/pSer129 α-syn+ punctae within ENS macrophages, and these were also increased in PFF-injected mice (Fig. 5a, c) suggesting complement-mediated internalization of α-syn by ENS macrophages. These findings were associated with increased deposition of C1q on myenteric neurons (Fig. 5d, e), which, taken together, suggest that myenteric macrophages utilize complement to engulf αsyn from enteric neurons and limit the spread of α-syn pathology within the ENS.

Fig. 5. Myenteric macrophages increase complement-associated internalization in enteric synucleinopathy.

a Representative confocal images of myenteric macrophages (MHCII, blue) at 30dpi saline or PFF showing expression of C1q (magenta) and co-localized punctae of C1q with pSer129 α-syn (magenta, white arrowheads) in PFF mice; scale bar: 20 μm. b Quantification of C1q+ area and intensity within myenteric macrophages showing upregulation in PFF-injected mice two-sided Mann–Whitney test, p < 0.05, n = 3–5 animals/group). c Quantification of the number of C1q+/pSer129 α-syn+ punctae within myenteric macrophages showing an increase in PFF mice (two-sided unpaired t-test, p < 0.01, n = 3–5 animals/group). d Representative confocal images of myenteric ganglia (peripherin, cyan) at 30dpi with either saline or PFF showing pSer129 α-syn+ soma (yellow), and deposition of C1q+ punctae (magenta) on neuronal soma; scale bar: 20 μm. e Quantification of the C1q+ area (left) and intensity (right) on myenteric neurons in saline and PFF mice showing an increase with PFF injection (unpaired t-test, p < 0.05, n = 3–4 animals/group). Graphs in (b), (c), and (e) represent mean +/− SEM.

Complement-dependent engulfment has frequently been implicated in excessive synaptic elimination leading to functional deficits^37,38,40. In the CNS, α-syn is known to be enriched at presynapses⁴⁶, but the subcellular localization of α-syn within enteric neurons is unknown. As predicted, we found that enteric α-syn highly co-localized with pre-synaptic but not post-synaptic markers (Supplementary Fig. 6). These data indicate that enteric neuronal α-syn is enriched at pre-synaptic spaces and suggests that myenteric macrophages may use C1q-dependent complement to engulf α-syn from the pre-synapse.

To validate that complement dependent engulfment of α-syn was a protective mechanism against enteric synucleinopathy, we repeated our gut PFF injections in C1qa^−/− mice and WT controls (Fig. 6a). Loss of C1q resulted in a further increase in PFF-induced enteric neuronal pSer129 α-syn+ staining compared to WT controls (Fig. 6b, c). The exacerbated neuronal pathology was accompanied with a decrease in the number of internalized pSer129 α-syn+ punctae within macrophages (Fig. 6d, e), consistent with the finding that macrophages in C1qa^−/− mice were less efficient at clearing α-syn from neurons, leading to spread of pSer129+ pathology. Consistent with our previous analysis (Fig. 1e, f), there was a positive correlation between the neuronal pSer129 α-syn+ pathology load and the amount of neuron-macrophage contact that was unaffected by loss of C1q (Fig. 6f), suggesting that myenteric macrophage chemotactic signaling remained intact.

Fig. 6. Myenteric macrophages require complement to engulf α-syn and limit neuronal pathology.

a Schematic depicting experimental design for C1qa^−/− injection experiments. Created in BioRender. Mackie, P. (2026) https://BioRender.com/tgvzfvz. b Representative confocal images of myenteric ganglia (peripherin, green) showing qualitatively increased pSer129 α-syn+ (magenta) neurons in C1q^−/−-PFF animals; scale bar: 20 μm. c Quantification of the number of pSer129 α-syn+ neuronal soma expressed as a percentage of total peripherin+ soma within the myenteric plexus of saline- and PFF-injected mice at 30dpi in both C1q^−/− and WT controls showing a PFF-induced increase in pathology that was further enhanced in the C1q^−/− mice (two-way ANOVA p_condition < 0.0001, p_genotype < 0.01, p_interaction < 0.05. Tukey’s test for multiple comparisons WT-saline vs WT-PFF p < 0.05, WT-saline vs C1q^−/−-saline p = 0.9987, C1q^−/−-saline vs C1q^−/−-PFF p < 0.0001, WT-PFF vs C1q^−/−-PFF p < 0.001). N = 3–7 animals/group. d Representative orthogonal confocal image of myenteric macrophage (MHCII, cyan) with internalized pSer129 α-syn+ punctae (magenta) in PFF injected WT mice at 30dpi; scale bar: 10 μm. e Quantification of the number of internalized pSer129 α-syn+ within MHCII+ myenteric macrophages showing decreased punctae in C1q^−/− macrophages (two-sided unpaired t-test, p < 0.01). N = 5–7 animals/group. f Linear regression analysis of 139 myenteric ganglia comparing the percentage of pSer129+ neurons against the neuron-macrophage contact showing a statistically significant positive correlation that did not differ between genotypes (p = 0.5609, left). Average contact size between macrophages and myenteric ganglia in PFF-injected WT and C1q^−/− mice revealing no significant difference between groups (two-sided unpaired t-test, right). g Fecal output measured by the number of pellets or the total weight of pellets after a 30-minute period showing a decrease in the PFF group that was partially rescued by loss of C1q. Pellet count (left): two-way ANOVA p_condition < 0.001, p_genotype < 0.05 with Tukey’s test for multiple comparisons WT-saline vs WT-PFF p < 0.01, WT-saline vs C1q^−/−-saline p = 0.9819, C1q^−/−-saline vs C1q^−/−-PFF p = 0.100, WT-PFF vs C1q^−/−-PFF p < 0.05. Pellet weight (right): two-way ANOVA p_condition < 0.0001, p_genotype < 0.001 with Sidak’s test for multiple comparisons WT-saline vs WT-PFF p < 0.01, WT-saline vs C1q^−/−-saline p = 0.2319, C1q^−/−-saline vs C1q^−/−-PFF p < 0.05, WT-PFF vs C1q^−/−-PFF p < 0.01. N = 3–7 animals/group. Graphs in (c), (e), and (g) represent mean +/− SEM.

Thus far, our data indicate that myenteric macrophages utilize complement to engulf α-syn from enteric neurons, possibly from pre-synaptic compartments. Because our data also showed that PFF-induced mice had impaired GI motility (Fig. 1h, i), we asked if loss of C1q could rescue this phenotype. Indeed, we found decreased fecal output in our PFF-injected mice relative to saline controls that was partially rescued by loss of C1q (Fig. 6g). Thus, loss of C1q decoupled enteric neuron histopathology and functional deficits, indicating that pSer129 pathology induces enteric dysfunction in part via a complement C1q-dependent mechanism.

Macrophage clearance of α-syn wanes over time

Previous studies have indicated that α-syn can activate various intracellular stress pathways and induce an inflammatory response^47–49. Thus, we asked if Mp6 or Mp2 macrophages (i.e., muscularis macrophages) from PFF-injected mice exhibited genetic signatures related to protein processing and related stress pathways. Indeed, Cd93, which encodes the complement C1q receptor, was differentially expressed in Mp6 macrophages from PFF mice (Fig. 7a), corroborating our data implicating C1q in macrophage-mediated αsyn engulfment. Additional upregulated genes included Psap, Ctsb, and Ctsl, all of which have been implicated in the degradation of αsyn⁵⁰ (Fig. 7a). Furthermore, Gene Ontology analysis of the DEGs between muscularis macrophages from PFF-injected compared to saline-injected mice revealed that muscularis macrophages from the PFF group had upregulation of genes related to protein processing, amyloid fibril formation, response to unfolded protein, and cell death (Fig. 7b). These analyses indicate that while muscularis macrophages may help clear αsyn, the internal processing may induce cellular stress pathways.

Fig. 7. Macrophage clearance of α-syn within the ENS fails over time as synucleinopathy progresses.

a Violin plots of select differentially expressed genes between PFF and Saline Mp2 and Mp6 macrophages. b Gene ontology biological pathways analysis of DEGs between PFF and Saline muscularis macrophages (Mp2 and Mp6) showing enrichment in genes related to the responses to unfolded protein, protein processing and presentation, and programmed cell death. c Representative confocal images of ENS-resident macrophages (MHCII, blue), pSer129 α-syn (magenta), and C1q (yellow). Arrow heads indicate co-localized pSer129 + /C1q+ punctae within the macrophage; scale bar: 20 μm. d Quantification of the number of pSer129 + /C1q+ punctae per macrophage showing an initial increase at 30dpi that then decreases by 60dpi. Ordinary one-way ANOVA (p < 0.01) with Sidak’s multiple comparisons test (**p < 0.01). e Representative confocal images of myenteric neuronal soma (Peripherin, cyan) and C1q (yellow) showing complement deposition at 30dpi PFF; scale bar: 10 μm. f Quantification of C1q deposition (yellow) on peripherin+ area (cyan) shown as % of saline control. C1q deposition increased at 30dpi PFF and then decreased to near-control levels at 60dpi. Ordinary one-way ANOVA (p < 0.01) with Sidak’s test for multiple comparisons (*p < 0.05, **p < 0.01). N = 3–5 animals/group/time point in (d) and (f). Graphs in (d) and (f) represent mean +/− SEM.

Given that our data demonstrated further increase of pSer129 α-syn+ pathology from 30 to 60dpi (Fig. 1b, c), we asked whether this macrophage response was sustained or if it eventually failed. We repeated gut PFF injections and waited up to 60dpi then re-examined the enteric nervous system for αsyn pathology and C1q. While macrophages at 60dpi maintained a comparable level of C1q expression relative to 30dpi, they contained fewer internalized C1q+/pSer129+ punctae (Fig. 7c, d), indicating lack of a sustained engulfment response. There was also decreased C1q deposition on enteric neurons at 60dpi compared to 30dpi (Fig. 7e, f). Collectively, these data suggest that ENS macrophages lose their C1q-dependent engulfment capacity in enteric synucleinopathy.

Discussion

The current understanding of PD has evolved to recognize the importance of ENS involvement both in the pathogenesis of the disease and in the non-motor symptoms that significantly compromise quality of life for the patient. A major challenge has been elucidating the cell-to-cell neuroimmune interactions and molecular processes that shape the course of enteric synucleinopathy within the context of PD-like pathology. Here, we found that ENS-resident macrophages initially adopt a protective phenotype characterized by complement-dependent engulfment of α-syn from enteric neurons. Furthermore, our data suggest that this mechanism may contribute to the synucleinopathy-associated impairments in GI motility. Finally, our results also suggest that this clearance mechanism is not sustainable in this model as it fails by 60 days, although without progression of pathology to the CNS. Collectively, our data highlight complement-dependent engulfment as a dynamic mechanism by which ENS-resident macrophages modify both the functional and histopathological aspects of enteric synucleinopathy.

The complex and multi-dimensional macrophage response to PFFs and synucleinopathy

Macrophages such as microglia can adopt a multitude of states and phenotypes in different neurodegenerative diseases⁵¹. In PD specifically, several groups have shown that microglia or other CNS-associated macrophages engage in T-cell recruitment^35,52 and inflammasome activation^53–55 to drive neurodegeneration. A previous study reported upregulation of multiple pro-inflammatory cytokines in the gut 7 days following PFF injection¹⁸, suggesting that perhaps ENS macrophages engage a similar pro-inflammatory program to microglia in response to α-syn. In contrast to these findings, we found downregulation of inflammasome-related genes in PFF macrophages. A potential explanation is that the macrophage response during fibril clearance is distinct from that associated with the developing neuronal pathology. This possibility is supported by our own data which indicate macrophages decrease their engulfment of α-syn over time, suggesting that their response to enteric synucleinopathy is highly dynamic. Characterizing the full trajectory of the myenteric macrophage response will be critical for the design and timing of successful therapeutic interventions to delay or prevent synucleinopathy.

Importantly, our findings indicate that synucleinopathy-associated macrophages are characterized by upregulation of synaptic pruning-related genes and antigen processing/presentation genes. While we focused on the former for this study, it is crucial to acknowledge the role the adaptive immune system may play in PD progression. Indeed, our sequencing data suggested shifts in the intestinal lymphocyte compartment of PFF-injected mice. Multiple studies indicate that α-syn reactive T-cells are found in the circulation of PD patients^56–58. Lymphocytes can be educated in the gut before traveling to the CNS^59–61. Thus, it is possible that the C1q-dependent mechanism of α-syn engulfment we describe herein is the first step in antigen processing for downstream T-cell activation. More work is required to de-couple these pathways and balance the beneficial and detrimental effects of ENS macrophage activation.

Endolysosomal stress is thought be a convergence point for multiple pathophysiologic processes across both idiopathic and familial PD⁶² and is thus the subject of intense research. Our data suggest that continued engulfment of toxic α-syn upregulates pathways related to endolysosomal stress. It is possible that activation of these stress pathways may be associated with the decrease in C1q-mediated engulfment we observed at later timepoints as synuclein continues to spread. While it is tempting to approach this challenge with repopulation techniques or other pharmacologic stimulations such as DREADDs or M-CSF injections, it is likely that continued engulfment may induce deleterious responses associated with increased endolysosomal stress, such as inflammasome activation or pyroptosis^63–65. Therefore, further investigation is needed to identify mechanisms to enhance protective macrophage-dependent breakdown of pathologic α-syn species.

Mechanisms of ENS dysfunction in synucleinopathy

Our results indicate that loss of C1q increases enteric α-syn pathology but paradoxically improves dysmotility. A potential explanation for this could be that ENS macrophages preferentially engulf α-syn from enteric synapses, thereby limiting the trans-synaptic propagation of misfolded α-syn while simultaneously decreasing neurotransmission efficiency. Consistent with this idea, our findings indicate that enteric α-syn preferentially localizes to the pre-synaptic space. Moreover, although excessive synaptic pruning due to α-syn has not been shown, it is a well-studied mechanism for cognitive dysfunction in Alzheimer’s amyloid-related pathology^37–39. Additionally, ENS macrophages engage in synaptic refinement during development²³. Therefore, to our knowledge, our findings are the first to suggest a role for macrophage-mediated synaptic pruning in response to neuronal α-syn pathology. Notably, increased complement deposition has also been reported in the substantia nigra of PD patients post mortem⁶⁶. Future studies may investigate whether or not microglia may also engage in synaptic pruning in the context of α-syn pathology and elucidate how macrophages determine which synapses to target for engulfment. Finally, the divergence of pathology and function in our studies is similar to that reported in the literature with Aβ³¹. It is worth noting that, although myenteric macrophages decrease their clearance of α-syn in the ENS by 60dpi, we still did not observe spread of pathology to CNS nuclei. While there are multiple factors that may contribute to the lack of spread, it is possible that future interventions for enteric synucleinopathy should be targeting GI motility as an endpoint rather than local α-syn pathology load. Overall, this finding underscores the importance of examining both pathologic and functional outcomes to assess the outcomes of new interventions.

Limitations of the model and study design

Here we employed multiple orthogonal approaches including sequencing, pharmacologic depletion, genetic knockouts, and imaging, to describe a role for myenteric macrophages in response to synucleinopathy. However, there are still outstanding limitations that future work should address. First, because we utilized injections of PFFs to induce synucleinopathy and studied time points focused on early spread of pathology, our model did address how synucleinopathy begins. Complementary models using infection, toxins, or gut inflammation should be employed to determine the dynamics of the macrophage response to α-syn pathology as it progresses across time. Furthermore, fibrils are a finite nidus and may be taken up by macrophages upon injection. To ensure that the macrophage response characterized here is truly due to neuronally-derived α-syn, a complementary model using viral overexpression of α-syn to compare with our outcomes would be informative. Ideally, the analysis of PD patient ENS macrophages would provide the strongest support for the translatability of our findings; however, duodenal muscularis tissue with identical disease stage is extremely difficult to obtain. More concerted efforts at collecting this tissue post-mortem will be crucial to advancing the field.

Two of our experiments utilized pharmacologic depletion of macrophages via a monoclonal antibody against the mCSF1R receptor. One experiment resulted in transient depletion that allowed for macrophage repopulation by 30dpi whereas the other resulted in more sustained depletion at least up to 30dpi. While our data suggest that antibody-mediated depletion may preferentially affect muscularis macrophages (Supplementary Fig. 2b), the interplay of macrophage subtypes and kinetics of macrophage repopulation add potential confounders. Future studies with refined and selective depletion methods to conclusively determine the roles of specific macrophage populations will also be informative.

Additionally, our C1q data was based on the use of a global knockout mouse. While we and others have shown that macrophages are the predominant source of C1q within the gut⁴¹, it is still possible that global knockout of C1q may have other unknown compensatory effects. Furthermore, this model knocks out C1q from all macrophages. Subsequent studies examining the role complement-mediated engulfment by muscularis macrophages should be performed using conditional inducible knockouts. This approach would allow one to leverage the unique turnover kinetics of the muscularis macrophages compared to mucosal macrophages to assess their individual contributions. Future investigations may use this platform to further elucidate the role of muscularis macrophages in synapse removal and associated gastrointestinal dysmotility. Thus, while our data collectively indicate that muscularis macrophages affect ENS pathology, we cannot rule out the additional contribution of other cell types.

Another important limitation is the generalizability of our scRNA-seq findings. Our dataset was relatively small and from one portion of the gastrointestinal tract. We took several steps to compare our data to other published datasets and found that some of the core gene signatures we identified are consistent with published reports. For example, based on our analysis of integrated datasets from multiple studies, Ms4a7 may be able to grant genetic access to muscularis macrophages, particularly if used in combination with an orthogonal tool as has been done for dural macrophage subsets⁶⁷. Additionally, while the overall effect of PFF injection on the muscularis macrophage transcriptome was mild, our findings highlight the importance of endolysosomal pathways. These processes, particularly in macrophages, have been repeatedly implicated in multiple neurodegenerative diseases⁶⁸, such as the DAM phenotype in Alzheimer’s disease⁴⁴. Thus, our results support continued investigation into endolysosomal-targeted therapeutics for synucleinopathy. Nevertheless, more work must be done to rigorously characterize the transcriptomic profiles of muscularis macrophage subsets and how they compare to mucosal macrophage subsets, including additional studies in synucleinopathy models.

Finally, our study focused on the enteric phase of the gut-to-brain hypothesis of PD. We did not observe any α-syn pathology in the brain at the time points we investigated. Notably, various groups have reported differing time scales and extents of gut-to-brain transmission with some studies only showing transient pathology confined to the brainstem. Furthermore, while some of these reports have used wild-type mice, others have used transgenic mice that overexpress α-syn. Additionally, while the original hypothesis by Braak and colleagues proposed that pathology could spread in a retrograde fashion up the efferent arm of the Vagus nerve, recent studies suggest that transmission may occur by both efferent and afferent arms of the Vagus nerve and perhaps even the circulatory system. Therefore, more work is required to standardize this model and rigorously characterize the kinetics of α-syn pathology along the gut brain axis.

In conclusion, using orthogonal approaches, we found a role for complement-dependent, ENS-resident macrophage engulfment of α-syn which sculpts the development and initial spread of enteric synucleinopathy. This macrophage response may represent a targetable mechanism that helps link the histopathology and dysfunction observed in the ENS of Parkinson’s Disease.

Methods

Mice

All mice were used in accordance with national and institutional guidelines and under protocols approved by the University of Florida IACUC. C57B6/J mice were either purchased from Jackson Labs (animals for surgery) or bred in-house (validation experiments). C1qa knockout mice were also purchased from Jackson Labs (strain #031675, RRID IMSR_JAX:031675) at 2–3 months of age. Knockout of C1q was additionally validated in-house by immunofluorescent microscopy. Unless otherwise stated, mice were group-housed and kept in 12-h light/dark cycles with temperature and humidity controlled. Unless otherwise stated, mice were allowed ad libitum access to food and water. Following surgery, mice were singly-housed for 7–10 days.

Macrophage depletion

Macrophage depletion was performed based on the protocol established by Muller et al.²⁴. Mice were intraperitoneally injected with and a monoclonal antibody against CSF1R (InVivoMab BioXCell, see Supplementary Table 1 for clone) at a dose of 37.5 μg/kg. Surgery was performed 6 days after injection. For double depletion experiments, intraperitoneal injections were performed 6 days before surgery and 14 days after surgery (21 days total between injections).

Preformed α-synuclein fibrils (PFFs) generation

Recombinant mouse full length α-synuclein (αSyn) protein was expressed in BL21 (DE3)/RIL Escherichia coli (E. coli; New England BioLabs Inc) using the pRK172 bacterial expression vector containing the murine αSyn cDNA. Bacterial were harvested by centrifugation and re-suspended in high-salt buffer (0.75 M NaCl, 50 mM Tris, pH 7.4, 1 mM EDTA) containing a cocktail of protease inhibitors and heated to 100 °C for 10 min. Debris were removed by sedimentation at 20,000 × g for 30 min. Supernatants were dialyzed into 100 mM NaCl, 10 mM Tris, pH 7.5 and applied onto a Superdex 200 gel filtration column (Amersham Pharmacia Biotech, Inc., Piscataway, NJ) and separated by size exclusion. The fractions were assayed for the presence of the α-Syn proteins by SDS-polyacrylamide gel electrophoresis (PAGE) followed by Coomassie Blue R-250 staining. Proteins were concentrated using Amicon Ultra-15 units (Millipore Corp., Bedford, MA), dialyzed against 10 mM Tris, pH 7.5, applied to a Resource Q column (GE Healthcare) and eluted with a 0–0.5 M NaCl gradient. Protein concentrations were determined using the bicinchoninic acid (BCA) protein assay (Pierce, Rockford, IL) and bovine serum albumin (BSA) as a standard. The purified αSyn monomers were then assembled into PFFs by incubation at 5 mg/mL in sterile PBS at 37 °C with continuous shaking at 1050 rpm for at least 48 h using an Eppendorf Thermomixer. αSyn fibril formation was monitored with K114 fluorometry as previously described (PMID: 12950445). PFFs were diluted in sterile PBS to a final concentration of 2 mg/ml and fragmented into an array of shortened fibrils using a Branson 2800 water bath sonicator at room temperature for 1 h prior to freeze-down. Once fragmented, PFFs were aliquoted and stored at −80 °C until use. At the time of surgery one aliquot was thawed and used then discarded. All surgeries for this study were conducted with PFFs from the same preparation to limit batch effects.

Electron microscopy

EM images were obtained using a ThermoFisher G2 Talos L120C TEM (ThermoFisher Corp., Wilham, MA) operated at 120 kV and digital images were acquired with a Ceta CMOS 4 K × 4 K camera and TIA software ThermoFisher G2 Talos L120C TEM (ThermoFisher Corp., Wilham, MA).

Gut injection surgery

Mice ~3 months old were used for surgery. The gut injections were performed as previously described^17,18. Mouse anesthesia induction and maintenance was achieved with isofluorane at 1.5–3%. The abdomen was shaved and sterilized using chlorhexidine/ethanol swabs and a midline incision was made from the xiphoid process to the lower abdomen. Once the peritoneum was opened, the stomach and duodenum were teased out using wetted cotton tip applicators to minimize trauma. Gut injections were performed with a Nanofil syringe fitted with a 36 G beveled tip (World Precision Instruments). 4 injections in total were administered to the gut wall: 1 in the stomach antrum, 1 near the pylorus, 1 in the first segment of the duodenum and 1 in the second segment of the duodenum. Each injection was 2.5 μL. Fast Green FCF dye (Sigma Aldrich) was added to either the PFFs or sterile saline (control) to visually confirm injection into the gut wall rather than the gut lumen. Following confirmation of injection, the mice were closed and allowed to recover. Mice were housed independently for the first 7–10 days following surgery to allow for the abdominal wound to heal before being recombined with their littermates.

Duodenal whole mount immunofluorescence

At designated time points, mice were euthanized via isofluorane overdose followed by transcardial perfusion with cold PBS. The duodenum was extracted and for the remainder of the microdissection was kept submerged cold PBS on ice. The duodenum was flushed intra-luminally with PBS, and the mesentery was gently removed using fine dissection. The duodenum was opened along the mesenteric line and then pinned to a cured Sylgard disc with the mucosa side facing up. The mucosa was then gently dissected off using fine dissection technique under a microscope. Following dissection, the samples were drop-fixed in 4% paraformaldehyde for 20 min and then hand-washed in PBS 5 times. Dissected samples were then photobleached overnight and were then ready for immunostaining.

Duodenal samples were first blocked and permeabilized in 0.5% Triton-X, 3% BSA (Sigma), 3% normal goat serum (NGS) for 3 h at 37 °C. Primary antibodies were diluted in 0.5% Triton-X, 2% BSA, 2% NGS. For specific antibodies and dilutions, see Supplementary Table 1. Primary labeling incubated overnight with rocking at 4 °C. Following primary labeling, samples were washed in PBS 5 times at room temperature for 15 min each with rocking. Secondary antibodies were diluted in the same solution as primary antibodies and allowed to incubate with samples at room temperature for 2 h with rocking. Samples were then washed again 4 times for 15 min each, counterstained with 4’,6-diamino-2-phenylindole (DAPI) at a final dilution of 1:2000 for 20 min at room temperature and briefly given one final PBS wash. Samples were then mounted using Fluoromount Gold, sealed, and imaged.

Duodenal cryosections

For indicated experiments, mice were euthanized via transcranial perfusion with ice-cold PBS followed by ice-cold 4% PFA (Sigma). The duodenum was extracted and post-fixed in 4% PFA for 2–6 h and then transferred to 30% sucrose and stored at 4 °C overnight. The following day, the tissues were transferred into cryomolds filled with OCT solution and stored at −20 °C until ready for sectioning. Tissue was then cut in 20 μm sections on a cryostat, slide mounted, and stored at −20 °C until ready for staining.

Mouse brain immunostaining

FFPE brains

Mice were euthanized and transcardially perfused with cold PBS as above and the brains were removed and fixed in 70% ethanol, 150 mM NaCl for several days. Brains were sectioned into 8 segments and dehydrated through a series of 70–100% ethanol, followed by xylenes and then infiltrated with paraffin. Paraffinized brains were then embedded in blocks and cut in 6 μm sections mounted on glass slides.

Confocal imaging

Mouse brain and gut tissue was imaged using a Nikon A1 laser scanning confocal microscope using NIS Elements software (V5.0 Nikon). Samples were illuminated using 405, 488, 560, and/or 640 nm diode lasers (Obis) through either a ×20 (0.75NA Plan Fluor) mixed immersion, ×40 (1.3NA Plan Apo) oil-immersion, or ×60 (1.4NA Plan Apo) oil-immersion objective with emission detection set to 595/50, and 700/75 nm, respectively. To avoid bleed-through, all images were acquired using channel series mode.

For imaging of mouse duodenal whole mounts, the myenteric plexus was first visualized and inspected. Any tissues with incomplete staining of the myenteric plexus were discarded. For remaining samples, 4–6 z-stack images were randomly taken across the sample with the total width of the myenteric plexus captured for each image. For pSer129 neuronal counts, images were taken 1.0 microns apart, for all other images, images were taken at 0.5-micron increments.

Image processing was either performed in Fiji/ImageJ (see below) or, for representative images, in Nikon NIS Elements software. Representative images were background subtracted using the rolling ball algorithm and then denoised using the Elements Advanced Denoising feature set to a power of 2 for each channel.

scRNA-sequencing

Sample preparation

Single-cell RNA sequencing was carried out using the 10× Single-cell 3’ Dual Index kit (v3.1). For each library, 2 animals were pooled. Animals were terminally anesthetized and then perfused with ice-cold PBS with 2% fetal bovine serum (FBS). The duodenum from each animal was dissected and any remaining mesentery was carefully removed on ice. The duodenum was then flayed open, cleaned, and finely cut into small pieces using a fresh razor blade. The duodenal pieces were then added to 5 mL of Accumax dissociating agent and allowed to digest with mild agitation for 5 min at 34–37 °C. Following digestion, the mix was further mechanically dissociated using a wide-bore P1000 pipette tip with gentle trituration. Fresh PBS + FBS was then added to the mix and the tissue was spun down at 340 × g for 3 min. After the first spin, the sample was then filtered through a 40 µm tube-top filter and rinsed thoroughly with ice cold PBS + FBS. The sample was then spun down again under the same settings and the supernatant was aspirated. The remaining sample was re-suspended and counted under a microscope using a hemocytometer and Trypan Blue.

Once counted, cells were allocated to one of two separate isolation and enrichment strategies and then pooled together for scRNA-seq (schematic in Supplementary Fig. 3a). The rationale for this design was to broadly capture all immune cells in an unbiased fashion while also ensuring we sequenced enough macrophages. Therefore our two strategies were (1) FACS-based CD45+/DAPI− sorting for live immune cells and (2) magnetic-based enrichment of CD11b+ cells followed by FACS-based sorting of DAPI- (live) cells.

The first strategy was myeloid enrichment, which was accomplished using a CD11b magnetic bead MojoSort kit (Biolegend, see Supplementary Table 1). 10 million cells were allocated to CD11b positive selection according to the manufacturer protocol. Briefly, cells were first incubated with the biotin-antibody cocktail for 15 min on ice. Cells were then washed by hand and spun down at 300 × g for 3 min. Next, labeled cells were allowed to incubate in 100 µL of MojoSort Buffer with the streptavidin nanobeads on ice for 15 min. Cells were re-washed and spun again at 300 × g for 5 min and then resuspended in 1 mL and placed in a magnetic separator column. 3 rounds of separation, salvage, and washing were performed to enrich for CD11b+ cells. Following magnetic enrichment, cells were then incubated in a 1:2000 DAPI mix to distinguish live/dead cells. Since dead cells can decrease the quality of scRNA-seq datasets, we further enriched the proportion of live cells in our sample using DAPI and FACS. Briefly, after magnetic enrichment cells were incubated in a 1:2000 DAPI:PBS + 2% FBS dilution for 10 min and then sorted for live cells via FACS (DAPI-negative gate) on a BD FACSAria III and collected in microcentrifuge tubes with PBS + 2%BSA.

The second isolation and enrichment strategy was performed in parallel and sorted live CD45+ immune cells via FACS. 4–6 million cells were allocated for CD45 labeling. Briefly, cells were allowed to incubate with CD45-FITC at 0.5 µL per million cells in FACS buffer for 30 min, covered and on ice. Cells were then washed with PBS and a 1:2000 DAPI mix was added to allow for Live/dead labeling. CD45-labeled cell samples were then subjected to FACS using a BD FACSAria III. CD45-labeled cells were gated on DAPI-negative/CD45-FITC positive (live immune cells). Importantly, cells were sorted on the slowest speed using a 100 µm nozzle and sorted into pre-wetted microcentrifuge tubes filled with 100 µl of PBS + FBS. For each sort, 50–80,000 cells were sorted.

Following FACS, both the CD11b- and the CD45-enriched samples were washed 3 times to discard any debris/dying cells and filtered once more using a P1000 filter tip pipette. The CD11b+ magnetically sorted fraction and the CD45+ FACS fraction were then pooled for loading onto the 10X Genomics chip according to manufacturer protocol.

Library construction and sequencing

Only samples with at least 80% final viability by Trypan Blue were selected for 10X Genomics chip-loading and further library construction. For each library, the maximum number of cells was loaded into each well of the chip. GEM recovery and cDNA amplification was performed according to the 10X Genomics protocol. All libraries were subjected to QC using a bioanalyzer high-sensitivity chip before proceeding with library construction. A final QC was also conducted using a Bioanalyzer High Sensitivity chip prior to freeze-down or sequencing. Libraries were kept at −20 °C until they were sent for sequencing using an Illumina NovaSeq 6000 at the University of Florida’s Interdisciplinary Center for Biotechnology Research NextGen DNA Sequencing Core. All libraries were sequenced to a depth of 100,000 read pairs per cell, and all libraries were sequenced on a single flow cell to help minimize any potential batch effects.

Whole gut transit time assay

Mice were habituated in individually-housed cages in the experiment room for 6 h the day prior to the experiment. Mice were then fasted overnight. The next morning, mice were brought to the experiment room and allowed to acclimate for 1–2 h in their individual cages before assay start. To measure whole gut transit time, mice were oral gavaged with 300 μL of 2% carmine red in 0.4% methylcellulose in water. Cages were then checked starting at 30 minutes and checked every 10 min thereafter for a red fecal pellet. The first red fecal pellet was marked as the end time for the assay.

Fecal output

Fecal output assays were carried out in a flow hood. Mice were transferred from their home cage into individual, covered to-go cups for 30 min. At the end of the 30 min, the fecal pellets were collected and counted for each animal. Pellets were blotted dry to remove any urine and then weighed together.

Image analysis

All image analysis was performed using Fiji/ImageJ software and associated plugins. Details for each type of analysis are described below.

Neuronal pSer129 αSyn counts

Images were first background subtracted using the rolling ball method available in ImageJ then denoised using the Pure Denoise plugin with default settings selected. The appropriate channels were then subjected to thresholding and masking using the Triangle filter in ImageJ with the Autodetect set for threshold selection. Composite images were then made my combining the masked Tuj1 channel with the masked pSer129 channel. Throughout the pre-processing steps, images were quality checked. Any errors at any point in the process resulted in the image or slice being discarded from analysis (whichever was appropriate). A maximum intensity projection image was created and the total number of ganglia was counted. Ganglia were defined as areas where multiple fiber tracts came together and there were visible soma at the junction. To identify pSer129+ soma, the z-stack was scrolled through manually and positive soma were visibly identified, marked, and counted. In order to count, soma must be positive within the z-area of the myenteric plexus (i.e., positive signal above or below was not counted) and approximately half or more of the soma has to be positive for pSer129. Additionally, the shape of the signal had to approximate a cell body (i.e., scattered punctae were not counted). The total number of pSer129+ neurons was then added up and divided by the total number of ganglia to yield the average number of pSer129+ soma per ganglia.

Macrophage pSer129 punctae

For macrophage pSer129 punctae analysis, the same pre-processing workflow was followed as for the neuronal counting analysis. Following background subtraction, denoising, and thresholding and masking, the Image Calculator function in ImageJ was used to select pixels that were positive for both MHCII and pSer129. A sum slices projection was taken of the resultant image and the Analyze Particles Function was used to count the number of punctae as well as the average size of punctae within the macrophages. The average size of punctae was the metric reported.

Macrophage-neuron contact

Following the same pre-processing steps, the image calculator was again used to identify the pixels that were positive for both Tuj1 and MHCII. Regions of interest were then drawn around each ganglia in the Tuj1 channel. The double-positive area for MHCII and Tuj1 in each ROI was then recorded and taken as a percentage of the total Tuj1+ ganglia area. For PFF-injected mice, ganglia with pSer129+ neuronal soma were noted. The proportion contact area for each ganglion for every image was averaged to give the reported values per mouse.

C1q analysis

Images were subject to similar pre-processing as described above with the exception that a more stringent rolling ball radius (10) was selected for the background subtraction. To analyze the C1q deposition on neurons, the total peripherin+ area for each image was recorded. The image calculator was used to identify pixels that were positive for both C1q and peripherin. The total C1q+/peripherin+ area in square pixels was then divided by the total peripherin+ area and multiplied by 100 to yield the reported measure.

A similar process was followed to measure the average C1q+ area per macrophage. Following the same pre-processing steps, the image calculator was used to identify pixels positive for both MHCII and C1q after threshold and masking. Separately, a maximum intensity projection of the MHCII image was used to count the total number of macrophages in the field of view. The count, macrophages had to have a defined morphology, be entirely within the field of view, and not overlap with one another. The total C1q+/MHCII+ area was divided by the total number of macrophages counted to give the reported metric.

To identify C1q+/pSer129+ punctae within macrophages, the image calculator was again used with two iterations. The first was used to identify pixels that were double positive for both C1q and pSer129, the second was used to identify pixels positive for these two markers and MHCII. A sum slices projection was then made and the analyze particles function in ImageJ was used to calculate the number of total particles and their average size. The total number of particles was divided by the total number of macrophages counted (as above) to yield the average number of punctae per macrophage.

Sholl analysis

Z-stack images were acquired using 60× oil immersion objective on a confocal laser microscope equipped with a Nikon A1 system. The DMV region was defined by the presence of positive ChAT neurons. Morphometric analysis was performed using Image J software. Maximum intensity projections were subjected to thresholding to create a binary image. Individual microglia were then isolated in the image and used to perform Sholl analysis. For Sholl analysis, a line was drawn from the center of the soma to the end of the longest branch. Deprecated Sholl analysis on the Neuroanatomy plug-in then drew concentric circles from the start to the end of the line at 1 μm increments and counted the intersections at each radii.

Co-localization analysis

Co-localization analysis was performed using Nikon Elements NIS analysis software. Images were pre-processed by rolling ball background subtraction and advanced denoising. The co-localization analysis option was performed through each slice of the z-stack and the Pearson’s R for each slice was averaged to give a co-localization value for the image.

scRNA-sequencing analysis

Following sequencing, fastq files were passed through the CellRanger (v6.1.1) pipeline for alignment, mapping, barcode assignment and UMI counting. The default mouse genome in the CellRanger pipeline was used to map reads. Libraries were visually inspected individually for quality control. Any library where <70% of reads could be confidently mapped to cells was not used for downstream analysis. Libraries from the same condition were then aggregated using the CellRanger aggregation function available on their cloud platform. The resulting barcode and count matrices were then passed through to Seurat v4.0⁶⁹ for downstream analysis.

Files were read into Seurat and underwent further quality control based off of the feature and UMI counts, and the mitochondrial contamination. The minimum and maximum feature counts were set to 200 and 6500, respectively while the percent mitochondrial-derived RNA was capped at 15%. Only cells that met these requirements proceeded to the next step of analysis. The data were then normalized, and the top 2500 highly variable features were identified. The saline and PFF libraries were then integrated using the ‘FindIntegrationAnchors’ and ‘IntegrateData’ functions in Seurat. The resultant combined dataset was then used for the remaining analyses. For an initial clustering analysis, the data were first scaled, then PCA (npcs = 10) followed by dimensionality reduction via uniform manifold approximation and projection (10 dimensions) were performed. The FindNeighbors function (10 dimensions) and the FindClusters function (with resolution of 0.1) were then carried out. The default settings for all of the above functions were used except where specified. For DEG identification, the FindMarkers function was used to compare the gene counts in the Mp clusters or the T-cell clusters from the PFF vs the Saline cells.

To perform a high-resolution clustering analysis specifically on the macrophages, the “Mps” cluster was extracted from the larger dataset. These data were then renormalized and a new set of 2500 highly variable features were found. The Mp dataset was then re-scaled and a new PCA was performed (10 components). The number of components was determined by visual inspection of an elbow plot. Finally, the data were reclustered with a higher resolution of 0.7. To analyze proportions, the total number of cells from each category in each cluster were tabulated and then either a z-test or Fisher exact test were performed to confirm statistical significance. For comparison of signatures from clusters of interest with published datasets, all significant (adjusted p-value < 0.05) marker genes for that cluster were pulled into a list and compared against the published signature for disease associated microglia (DAM) or disease inflammatory macrophages (DIM) from the supplemental table of Silvin et al. 2022. For functional annotation, the significant marker genes were entered into the Gene Ontology Panther Database and the resultant terms were ranked on the Fold Enrichment. To identify DEGs between Mp6 vs Mp2, the FindMarkers function was used to directly compare the two clusters. Again, the significant DEGs following p-value adjustment were input into the Gene Ontology Panther Database for functional annotation. A similar process was carried out for the T-cell re-clustering and subsequent analysis.

Integrated dataset analysis

Datasets from De Schepper et al. Cell 2018 and Chiaranunt et al. Sci Immunol 2023 were downloaded from their respective online databases. For the De Schepper dataset, gene IDs were assigned using Ensembl m. musculus gene data base. The resultant Seurat objects were then integrated with the macrophage Seurat object from this study. This integrated object was filtered, normalized, and scaled as above and the standard Seurat workflow was performed. Select clusters expressing macrophage genes at high levels were isolated using the subset() function and the data were re-normalized and scaled. For heatmaps projecting the gene signatures of Mp2 and Mp6 on the integrated dataset, we used the Addmodulescore() function where the features argument contained the respective genes.

Statistics

Statistical analysis for scRNA-seq data was all performed in R Studio using the default settings of the functions described above. All other statistical analysis was performed in GraphPad Prism v9.0. Details for individual statistical tests can be found in the legend of each figure. For all datasets, normality was checked by visual inspection of a Q-Q plot. Sample sizes were based off existing literature. Randomization was performed at the time of surgery. Imaging experiments were performed and analyzed in a blinded fashion.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

Conceptualization: P.M.M., M.G.T., B.I.G., and H.K. Experimental Design: P.M.M., M.G.T., B.I.G., and H.K. Data Collection: P.M.M., J.M.K., M.H.B., T.H., W.H., G.M.L., G.P., A.M., and M.L.B. Data Analysis: P.M.M., J.M.K., M.H.B., W.H., G.M.L., G.P., and M.L.B. Writing manuscript (original draft): P.M.M. Revising and editing: P.M.M., M.G.T., B.I.G., and H.K.

Peer review

Peer review information

Nature Communications thanks Wouter Peelaerts and the other, anonymous, reviewers for their contribution to the peer review of this work. A peer review file is available.

Data availability

Source data are provided with this paper. Single-cell RNA sequencing data has been uploaded onto the NCBI Gene Expression Omnibus database (GSE314156). Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-68641-8.