Abstract
Intraneuronal accumulation of α-synuclein (αSyn) is key in the pathogenesis of Parkinson’s disease (PD). Published studies suggest that this process begins in the enteric nervous system (ENS) and propagates into the brain decades before clinical diagnosis of PD. The triggers and mechanisms underlying the accumulation of αSyn remain unknown but evidence is growing that immune pathways and in particular colitis may play a critical role. Here we demonstrate that patients with inflammatory bowel disease (IBD) exhibit αSyn accumulation in their colon. We then confirmed in an experimental model of IBD that intestinal inflammation can trigger αSyn accumulation in the ENS of wildtype and αSyn transgenic mice. We discovered that the type and degree of inflammation modulates the extent of αSyn accumulation in the colon and that macrophage-related signaling limits this process. Remarkably, experimental colitis at three months of age exacerbated the accumulation of aggregated phospho-Serine 129 αSyn in the midbrain, including the substantia nigra, in 21-month but not 9-month-old αSyn transgenic mice. This was accompanied by loss of nigral tyrosine hydroxylase-immunoreactive neurons, another neuropathological hallmark of PD. Together, our data suggest a critical role for intestinal inflammation in the initiation and progression of PD.
Introduction
Parkinson’s disease (PD) is a progressively debilitating neurodegenerative disease affecting 1% of the population above 60 years (1). Typical symptoms are motor impairments including muscle rigidity, tremor, and bradykinesia. Neuropathologically, PD is hallmarked by dopaminergic cell loss in the substantia nigra (SN), a resultant loss of striatal dopaminergic signaling (2, 3), and the presence of intraneuronal inclusions called Lewy bodies and neurites (4, 5). Lewy pathology is enriched in α-synuclein (αSyn), a presynaptic protein that tends to misfold, aggregate, and become phosphorylated under pathological conditions (2, 3). Alpha-synuclein is also genetically linked to rare familial forms of PD and certain SNPs close to the αSyn gene (SNCA) locus are associated with an increased risk for developing PD (6). These findings in humans make αSyn a focal point of biomarker and drug development programs for PD.
Several years before the first appearance of motor features, many patients exhibit a variety of non-motor symptoms including constipation, sleep disorder, depression, and hyposmia (7–9). Co-occurrence of some of these non-motor symptoms increases the risk of progressing to PD even in otherwise healthy individuals in population-based and cohort studies (10–13). Constipation may be a particularly important non-motor feature of preclinical PD, with 28-61% of patients often exhibiting gastrointestinal dysfunction years before progressing to motor symptoms (9, 12, 14). This suggests an early connection between the status of the intestinal environment and the progression to PD (15). In line with this connection and due to the discovery of αSyn immunoreactive inclusions in neurons of the submucosal plexus in people with PD, Braak and colleagues have suggested an early involvement of the enteric nervous system (ENS) in the pathogenesis of PD (4, 5, 16, 17). It is currently unclear whether occurrence of αSyn aggregates in the peripheral nervous system is specific for PD since individuals with dementia with Lewy bodies and normal people can also show such inclusions in various tissues (18–21). Yet based on their findings, Braak and colleagues hypothesized that αSyn immunoreactive inclusions first appear in the ENS and then progress over time to the brainstem regions, including the dorsal motor nucleus of the vagus nerve and midbrain areas (4, 5, 16, 17). Several studies in preclinical models have recently demonstrated a potential propagation of αSyn pathology in the gut and the brain (22–26). Intriguingly, a clinical study reported that vagotomy in a Danish population decreases the risk of PD (27). Determining the critical factors for αSyn pathology in the ENS could provide insights into disease etiology, enabling the development of novel treatment paradigms to slow disease progression. A key factor that could trigger ENS αSyn accumulation and pathology spread is inflammation.
Interestingly, a recent finding in children with gastrointestinal inflammation suggests an immune regulatory function of αSyn (28). Immune pathways are indeed activated in the brain and colon of PD cases (29, 30). Also, several genes associated with an increased risk of developing PD have an immune system-related function (31), and a recent genome-wide association study identified common genetic pathways linking PD and autoimmune disorders (32). Most prominently, LRRK2, a major genetic risk factor for PD also confers increased risk for developing inflammatory bowel disease (IBD) (33) and is known to modulate the function of monocytes, macrophages and other immune cells (34, 35). Intriguingly, IBD is now confirmed to be associated with an increased risk for developing PD and specifically blocking the TNF pathway reduces this risk (36–38).
Here we tested the hypothesis that intestinal inflammation (e.g. colitis) triggers αSyn aggregation in the ENS and the subsequent development of αSyn pathology in the brain. We discovered that patients with IBD exhibited increased αSyn accumulation in the submucosal region of the colon. Experimental colitis in wild type and αSyn transgenic mice demonstrated that the type and degree of inflammation modulates the extent of αSyn accumulation in the colon and that macrophage-related signaling limits this process. Remarkably, experimental colitis at 3 months of age exacerbated the accumulation of aggregated αSyn in midbrain, including the SN, in 21-month but not 9-month-old αSyn transgenic mice. This was accompanied by loss of nigral tyrosine hydroxylase-immunoreactive neurons. Together, our data suggest a critical role for intestinal inflammation in the initiation and progression of PD.
Results
IBD patients show αSyn accumulation in the ENS and local macrophages
Recent epidemiological data links inflammatory bowel disease (IBD) to an increased risk of developing PD (36–38). In order to explore if IBD is associated with enteric αSyn accumulation we performed immunohistochemistry for αSyn in cryo-sections from colonic biopsies of patients with ulcerative colitis (UC, n = 11, mean age 31 years), Crohn’s disease (CD, n = 11; mean age 35 years), and from healthy subjects (HS, n = 8; mean age 51 years). We observed in eight UC cases various degrees of αSyn accumulation, mostly in neuritic structures (Figure 1). Interestingly, the eight UC cases, and four patients with CD (images not shown) also showed marked intracellular αSyn staining in many infiltrating monocytic cells. In contrast, only one HS showed a few immunoreactive cells (images not shown). This finding in human tissue suggests a potential role of local inflammation in the development of enteric αSyn accumulation.
Experimental IBD exacerbates αSyn load in submucosal plexus of αSyn transgenic and wildtype mice
During the process of further characterizing a (Thy1)-h[A30P]αSyn transgenic mouse line (39) we detected human αSyn accumulation in all innervated organs that were analyzed (Supplemental Figure 1). This included the myenteric and submucosal plexuses of the ENS, where it co-localized with peripherin, a specific marker for peripheral nerves (Figure 2A). We observed an age-dependent increase of baseline αSyn inclusion in both plexuses between the age of three and twelve months (Figure 2B). Since this transgenic model has the capacity to increase the αSyn load in the ENS over several months we wanted to test the hypothesis whether IBD-related inflammation in the colon may trigger or exacerbate local αSyn accumulation more acutely, e.g. within a few days or weeks.
Administration of dextran sulfate sodium (DSS) in the drinking water in acute or chronic paradigms is a well-established mouse model of IBD colitis, resulting in infiltration of leukocytes into the submucosa with various degrees of destruction of the colonic tissue and the mucosa (40). We administered DSS at different concentrations and durations to (Thy1)-h[A30P]αSyn transgenic mice (Figure 2C), resulting in leukocyte infiltration in a dose-dependent manner (Figure 2D and 3A). Intriguingly, in the acute paradigm, 2.5% but not 1% DSS triggered intracellular accumulation of αSyn in the submucosal plexus (Figure 3A, B). In the chronic DSS paradigm, a dose-dependent increase of αSyn load in the submucosal plexus was observed (Figure 3A). To confirm that this finding was independent of transgenic expression of human αSyn, acute and chronic (consistent dose) DSS paradigms were respectively applied also in wildtype mice. In both paradigms, we observed aggregation of endogenous murine αSyn in the submucosal plexus (Figure 3C, D). A separate experiment also confirmed that the observed effects of DSS could not be attributed to increased gene expression of murine or the transgenic human αSyn (Supplemental Figure 2). Together, these results confirmed the validity of this experimental IBD paradigm to test the effect of inflammation on αSyn accumulation in the ENS.
Colitis induced by peroral DSS but not by peritoneal administration of LPS aggravates αSyn accumulation in colonic submucosal plexus of αSyn transgenic mice
In order to explore effects of different approaches to induce inflammation in or nearby the gut in (Thy1)-h[A30P]αSyn transgenic mice, we compared the outcome of an acute 5% DSS in drinking water with acute intraperitoneal LPS administration (Figure 2C and 4). At d7, both agents had induced variable degrees of leukocyte infiltration in the submucosa of the colon with a marked destruction of the mucosa induced by only DSS (Figure 2D). As before, the DSS-exposed mice presented with increased mean accumulation of αSyn in the ganglia of the submucosal plexus (Figure 4A). In contrast, no change in αSyn load was detected in the myenteric plexus, consistent with lack of leukocyte infiltration in this anatomical region of the colon (Figure 4B). Interestingly, LPS-induced inflammation did not increase αSyn accumulation in the colonic nervous plexuses (Figure 4C, D). The two inflammatory insults resulted, however, in a differential expression of cytokines, and consistent with leukocyte recruitment, CCL2 was elevated in both (Figure 4F, G). Interestingly, in the LPS paradigm, mRNA for IL-10 was markedly elevated, whereas DSS strongly increased IL-6 and also IL-1β but not IL-10. Together these results indicate that certain types of local inflammation can increase the intracellular accumulation of aggregated αSyn in the periphery (i.e. in the colon).
Lack of Cx3cr1 signaling during DSS colitis aggravates αSyn load in the submucosal plexus of αSyn transgenic mice
In both the IBD patients as well as the (Thy1)-h[A30P]αSyn transgenic mice that experienced acute DSS colitis, we observed by their morphology αSyn positive infiltrating cells (Figure 5). In the mice these infiltrating cells were positive for the macrophage marker Iba-1 (Figure 5C-D). In order to explore further the role of monocytes/macrophages in the accumulation of αSyn in our DSS model, we added an experimental arm with (Thy1)-h[A30P]αSyn transgenic mice crossed with mice that have a deletion of Cx3cr1 induced by an insertion of GFP (Cx3cr1-GFP knock-in mice) (Figure 4A, B). The CX3CR1-CX3CL1 axis plays an important role in maintaining the function of the lamina propria macrophage population of the gastrointestinal wall and lack of this signaling pathway in experimental colitis models may either aggravate or ameliorate the induced pathology (41–43). In our experiment exposure to DSS resulted in leukocyte infiltration in the Cx3cr1-deficient αSyn transgenic mice at a minimally higher mean percent leukocyte area covered than for the Cx3cr1-competent group (Supplemental Figure 2A). We observed, however, a significantly higher mean level of αSyn accumulated in the submucosal plexus in αSyn transgenic mice lacking Cx3cr1 compared to αSyn transgenic mice expressing Cx3cr1 (p = 0.001, two-way ANOVA with Tukey HSD post-hoc analysis; Figure 4A). Consistent with the Cx3cr1-sufficient αSyn transgenic mice, no overt increased accumulation of αSyn was observed in the myenteric plexus (Figure 4B). These results indicate a potential link between monocyte/macrophage signaling and αSyn accumulation in ENS in this experimental IBD model.
Systemic IL-10 ameliorates DSS-induced colitis and associated enteric αSyn accumulation in αSyn transgenic mice
Interestingly, we observed above that LPS-induced colonic leukocyte infiltration did not result in increased accumulation of αSyn in the ENS of the colon and that the main difference in cytokine expression between the DSS and LPS paradigms was increased IL-10 expression in the LPS group (Figure 4). Interleukin-10 is an important regulator of monocytes/macrophages, and genetic ablation of IL-10 signaling or blocking IL-10 with specific antibodies was shown to enhance DSS colitis (44, 45). To mimic the effect of higher levels of IL-10 in an acute model of DSS colitis, we administered intravenously murine IL-10 (mIL10), which was recombinantly engineered onto two different murine IgG variants to extend the half-life of mIL-10 in circulation (mIgG1(v1)-mIL10 and mIgG1(v2)-mIL10, respectively). As shown previously, DSS induced a marked increase in leukocyte infiltration and αSyn accumulation, which was similar in the untreated and control IgG treated group (Figure 6A, B). Both mIgG1(v1)-mIL10 and mIgG1(v2)-mIL10 significantly reduced leukocyte infiltration despite administration of DSS (p<0.0001, one-way ANOVA with Tukey HSD post-hoc analysis; Figure 6A, B). Also, both mIgG1-mIL10 variants had reduced mean levels of human αSyn in the submucosal plexus whereas mIgG1(v2)-mIL10 reached significance (p=0.02, one-way ANOVA with Tukey HSD post-hoc analysis; Figure 6B). Interestingly, the significantly reduced αSyn accumulation was associated with detectable serum exposure of mIgG1(v2)-mIL10 whereas mIgG1(v1)-mIL10 was no longer detectable at the end of the in vivo phase, after 7d (Figure 6C). These results underline further an important role for the IL-10 pathway in keeping αSyn accumulation at a reduced level throughout the course of experimental IBD. Together, the genetic and pharmacological modulation of DSS colitis corroborates a relevant role for monocyte/macrophage pathways in the development of αSyn accumulations in the ENS of the colon.
Experimental colitis-induced enteric αSyn accumulation at young age persists for months and exacerbates brain pathology and dopaminergic cell loss in old αSyn transgenic mice
The previously highlighted hypothesis by Braak and colleagues associates αSyn brain pathology with previous development of αSyn pathology in the ENS (5, 46). However, it is currently unknown whether previous inflammatory conditions propel this ENS αSyn accumulation that may further worsen αSyn aggregation in the brain. To test this, we first explored αSyn accumulation in the submucosal plexus of (Thy1)-h[A30P]αSyn transgenic mice that had experienced DSS colitis previously but had fully recovered two months later. As expected, leukocyte infiltration in the submucosal plexus had returned to normal levels following the two-month recovery period (Supplemental Figure 3A). Remarkably however, αSyn accumulation in the ganglia of the submucosal plexus was still almost two-fold higher when compared to αSyn transgenic mice that were not exposed to DSS, and this was exacerbated in αSyn transgenic mice deficient for Cx3cr1 (Supplemental Figure 3B).
To assess development of brain αSyn pathology, we exposed 3-month old hemizygous (Thy1)-h[A30P]αSyn transgenic mice to DSS or normal drinking water (chronic increasing dose paradigm) and then returned all to normal drinking water (Figure 2C). After aging up to 9 or 21 months of age (i.e. mice aged for an additional 6 or 18 months after the chronic DSS paradigm), various brain regions were analyzed for pathologically accumulated αSyn (proteinase K resistant, pSer129-αSyn immunoreactive inclusions). Twenty-one-month old mice that only received water during their lifetime showed low levels of pathological αSyn inclusions in the brain (Figure 7 and Supplemental Figure 4), which is consistent with previous observations in the brain of the hemizygous (Thy1)-h[A30P]αSyn transgenic mice (47). In marked contrast, the majority of 21-month-old and previously DSS-exposed hemizygous (Thy1)-h[A30P]αSyn transgenic mice presented with higher degrees of pathological αSyn in the same brain areas where mice on water displayed much lower pSer129-positive immunoreactivity, reminiscent of the homozygous (Thy1)-h[A30P]αSyn transgenic mice (47). Intriguingly, αSyn transgenic mice that were aged to 9 months presented with extremely low levels of pathological αSyn inclusions in the brain (Figure 7 and Supplemental Figure 4). For the αSyn transgenic mice that aged up to 21 months, the accumulated αSyn pathology recorded in the SN was accompanied by a significant loss of tyrosine hydroxylase (TH) and Nissl positive cells (p≤0.05, Student’s T-test; Figure 8). Together, DSS colitis at a young age caused an age-dependent exacerbation of αSyn inclusion pathology and a loss of nigral dopaminergic cells in the brain of αSyn transgenic mice, similar to the human neuropathological hallmarks of PD.
Discussion
Currently, there is no therapy for PD available to slow or stop its progression and an obstacle in the quest to develop one is that we do not understand the triggers of PD (48). Intraneuronal accumulation of aggregated αSyn (i.e. Lewy bodies and neurites) is a key neuropathological hallmark, thus the distribution of Lewy pathology in postmortem brain tissues has been used for disease staging in PD (2, 4, 46). Abnormal accumulation of αSyn has also been observed in the peripheral nervous system in PD and in some individuals at risk of developing the disease (49, 50). Similar to this finding in people, αSyn immunoreactive inclusions have also been detected in the ENS of a transgenic mouse model prior to changes in the CNS (51). It is also suggested, based on preclinical models and postmortem pathology in various organs including the brain, that αSyn pathology propagates in a temporospatial and prion-like manner (2, 4, 5, 52, 53). However, the initial factors triggering αSyn aggregation are yet to be established (48). Moreover, the involvement of peripheral stimuli in the aggregation and pathogenic spread of αSyn is only beginning to unravel.
In this study, we provide evidence that patients with IBD have increased αSyn accumulation in their ENS (Figure 1) and that an experimental IBD-like inflammation (DSS colitis) triggers αSyn accumulation in the ENS of wildtype mice and in a transgenic mouse model of PD (Figure 3). Interestingly, in IBD patients and in the mouse model of IBD, macrophages filled with αSyn could be observed in the inflamed area (Figure 5). Aggravation of enteric αSyn accumulation in αSyn transgenic mice lacking Cx3cr1 signaling and amelioration of inflammation and enteric αSyn load by systemic IL-10 suggest a modulatory role of monocytes/macrophages in this process (Figure 4 and 6). Remarkably, we further observed that the aggravation of αSyn accumulation in the ENS persisted even after two months of recovery from DSS colitis, indicating that the observed effect is not a transient phenomenon (Supplemental Figure 3). Most importantly, 18 months but not 6 months post DSS colitis, αSyn transgenic mice had developed Parkinson-like brain pathology (Figures 7 and 8 and Supplemental Figure 4). This included elevated proteinase K resistant pSer129-αSyn pathology in the midbrain, including the SN, and other brain regions and an average decrease of 30-50% of TH- and Nissl positive cells in the SN. Together, this colitis model recapitulated the proposed accumulation of αSyn in the ENS in wildtype and αSyn transgenic mice long before clinical diagnosis of PD in humans. Additionally, the subsequent age-related development of αSyn pathology together with the loss of nigral dopaminergic neurons in the brain of αSyn transgenic mice mimicked a progression of the disease similar to what is proposed in humans.
We established that a mechanism by which peripheral inflammation promote αSyn accumulation in the colon potentially involves monocytes and macrophages. Interestingly, both peroral DSS and peritoneal LPS administration provoked strong local immune reactions resulting in leukocyte infiltration into the submucosa of the colon. This inflamed region is anatomically right next to the submucosal plexus and is separated by a muscular layer from the myenteric plexus (Figure 2). This might explain why the effect of αSyn accumulation was only observed in the nerves of the submucosal plexus but not in the ganglia of the myenteric plexus in our (Thy1)-h[A30P]αSyn transgenic mouse model of colitis. The mechanism underlying how intraperitoneally administered LPS leads to submucosal leukocyte infiltration probably involves the monocyte attractant chemokine CCL2 (Figure 4), but the specifics remain unclear. CCL2 was also upregulated in the colon of our DSS model. However, in contrast to peritoneal LPS, where infiltration of macrophages followed a patchy pattern, DSS-related macrophage infiltration was distributed across the whole submucosa. Also, the primary mechanism of peroral administered DSS is to destruct the mucosa of the colon, similar to some forms of ulcerative colitis, resulting in the transient disintegration of the intestinal epithelial barrier. The subsequent immune response to the infiltration of commensal bacteria evoked also in our (Thy1)-h[A30P]αSyn transgenic mouse model the expression of cytokines such as IL-1β and IL-6. This upregulation was absent in the LPS paradigm that had an intact mucosa. By acting on tight junctions, IL-1β and IL-6 can increase intestinal barrier permeability (gut leakiness), facilitating the recruitment of additional immune cells to the site of the inflammatory signal eventually culminating in widespread immune activation (54, 55). Consistent with the breach of barrier permeability in our mouse model, some PD patients express increased colonic cytokines such as IL-1β, IL-6 and TNF, occurring together with increased intestinal permeability (29, 56). In this context, it is striking that IBD patients on anti-TNF therapy have a reduced risk of developing PD (37). Here we demonstrate that patients with IBD also present with αSyn accumulations in the ENS as well as in infiltrating leukocytes nearby. In this regard, it is interesting to note that mucosal macrophages with intralysosomal αSyn content were previously described in the intact human appendix (57). These macrophages were in close proximity to the axonal varicosities of the vermiform appendix which showed an enriched staining for αSyn in the mucosal plexus. Indeed, we recently demonstrated that the vermiform appendix contains aggregated and truncated αSyn that has the propensity to aggregate recombinant αSyn in vitro (18). Also, monomeric and oligomeric αSyn species reportedly act as chemoattractants for neutrophils and monocytes, enhancing the maturation of dendritic cells in the ENS (28, 58). With such a role in intestinal immunity therefore, it is possible that the strong tissue destruction induced by DSS in the present study released αSyn, which perhaps served as a chemoattractant for monocytes. It is likely that increased abundance of extracellular αSyn and altered intestinal permeability, along with the DSS-evoked inflammatory response may have provided an enabling milieu allowing further αSyn accumulation in the ENS of the colon (59). However, the pathways leading to αSyn accumulation in macrophages remain to be elucidated. Mechanistically, it has been shown that excess endogenous αSyn compromises phagocytic activity in human iPSC-derived macrophages (60). Macrophages and other immune cells are also regulated by LRRK2, an established risk gene for PD and IBD. It will be interesting to explore in our colitis paradigm the effect of LRRK2 (e.g. its effect on autophagy in macrophages (34, 35, 61)) and other pathways in macrophages on αSyn accumulation.
Most profoundly, besides the accumulation of αSyn in the submucosal plexus, we found that a single chronic DSS paradigm at young age exacerbated αSyn pathology in the CNS of (Thy1)-h[A30P]αSyn transgenic mice much later in life (Figure 7). But how does this severe αSyn inclusion pathology develop in the brain of these mice? One hypothesis is that the brain αSyn pathology observed in this study could be due to direct effects of peripheral immune activation on the brain and brain vasculature. Hypothetically, colitis induced by DSS or LPS could compromise both active and passive components of the blood-brain barrier, which may render the CNS vulnerable to immune cells from the periphery (62, 63). Alternatively, certain peripheral triggers can directly affect microglial activity in the brain. For instance, short-chain fatty acids derived from gut microbiota appear to regulate function and maturation of microglia in the mouse CNS (64) and inflammatory mediators from gut microbiota could be carried in the bloodstream and induce pathological and behavioral changes in an αSyn transgenic mouse model (65). Moreover, rats and nematodes develop αSyn inclusions after exposure to the bacterial amyloid protein curli, a protein which stimulates microgliosis, astrogliosis, and secretion of IL-6 and TNF (66). Intriguingly, a recent study reported that peripherally applied inflammatory stimuli induce acute immune training (that exacerbates β-amyloid pathology) and immune tolerance in the brain that reprograms microglia, an effect which can persist for at least six months (67).
Another hypothesis is that the brain αSyn pathology observed may have accumulated following the transfer of pathogenic αSyn seeds from the gut via the vagal nerve. It is already established in rodents and non-human primates that αSyn pathology in the CNS can propagate from cell to cell, perhaps in a prion-like fashion (52, 53, 68, 69) and that certain pathogenic forms of αSyn may serve as templates in recipient cells, seeding and inducing further aggregation of endogenous αSyn in these cells (68, 70– 72). Interestingly, several studies in preclinical in vivo models have demonstrated that pathogenic αSyn seeds can be transferred from the peripheral nervous system to the CNS. Indeed, aggregated recombinant αSyn injected intraperitoneally, intramuscularly or into the gastric wall of certain mouse models of PD results in αSyn inclusions in the CNS (24, 73), with the latter being transmitted via the vagal nerve (74). Another study showed that vagotomy in mice prevents propagation of pathology into the CNS (74), a finding that is parallel to a peripheral rotenone-related αSyn rat model in which vagotomy reduced propagation of αSyn pathology (75), and to a human study that reported that vagotomy in a Danish population also decreases the risk of PD (27). Similarly, after injection of PD brain lysate and recombinant αSyn into the intestinal wall αSyn pathology trafficked through the vagal nerve to the dorsal motor nucleus of the vagus (23). Others have shown that virus-based expression of human αSyn in the medulla oblongata resulted in a transfer of αSyn towards rostral brain regions in a rat model (76). The findings in these previous studies, along with the present findings recall the Braak hypothesis and highlight the critical connection of the vagal nerve. Indeed, in the present study, αSyn pathology was much more prominent in the reticular nucleus (including the vagal area) and midbrain areas (compared to the rostral areas) 18 months post DSS colitis, suggesting the potential involvement of this nerve in the accumulation of αSyn in the CNS.
In summary, here we report that individuals with IBD exhibit αSyn accumulation in the colon concomitant with infiltrating monocytes/macrophages positive for αSyn. Employing a experimental model of IBD, we provide proof-of-principle demonstrating that αSyn accumulated in the colon of (Thy1)-h[A30P]αSyn transgenic and wildtype mice upon DSS colitis and that this process can be modulated by monocyte/macrophage-related signaling. We further demonstrate that chronic DSS colitis in young (Thy1)-h[A30P]αSyn transgenic mice robustly exacerbated αSyn accumulation in the aged brain which was accompanied by reduced numbers of TH- and Nissl positive neurons in the nigra. The present findings are in consonance with retrospective and cohort studies demonstrating a link between IBD and PD in various populations including those in Taiwan, Denmark and the US (36, 37, 77), although this may not be the case in all populations (78). The exact mechanisms connecting the gut pathology to the pathology in the brain including the role of monocytes/macrophages, remain to be established. In conclusion, this novel mouse model of a potential preclinical stage of PD recapitulated the accumulation of αSyn in the ENS and the subsequent age-related development of brain pathology. Together, our data suggest a critical role for intestinal inflammation in the initiation and progression of PD.
Methods
Mice
Male C57BL/6 wild type mice (Jackson Laboratories, Bar Harbor, USA), hemizygous Tg(Thy1-SNCA*A30P)18Pjk ((Thy1)-h[A30P]αSyn) (39) and Tg(Thy1-SNCA*A30P)18Pjk crossed with Cx3cr1tm1Litt ((Thy1)-h[A30P]αSyn /CX3CR1-def; homozygous for Cx3cr1-GFP knock-in allele; (79) transgenic mice were used for the study. (Thy1)-h[A30P]αSyn transgenic mice express mutant human αSyn under the neuron selective Thy1 promoter. (Thy1)-h[A30P]αSyn transgenic mice were crossed to Cx3cr1-def transgenic mice which express eGFP replacing fractalkine gene expression. All mice were maintained on a C57BL/6 background for more than 10 generations and under specific pathogen-free conditions. To the extent possible, littermates were used in the experiments. Health status was monitored daily during experiments.
Human Subjects
Samples from patients with Crohn’s disease (CD), ulcerative colitis (UC) or healthy subjects (HS) were provided by the tissue bank, Institute of Pathology, University of Bern. Briefly, specimens were obtained from patients who underwent surgical procedures at the University Hospital (Inselspital) in Bern, Switzerland between 2004 and 2011. Three selected male patients previously clinically diagnosed with UC with a reported disease duration > 6 year (n=3) and undergoing steroid therapy combined with either metronidazole or mesalazine. CD patients were of mixed gender and aged 22-56 years ranging from 2 months to 11 years post disease diagnosis undergoing treatment with either infliximab or mesalazine in combination with steroids. Healthy subjects were of mixed gender with no report of inflammatory bowel disease, aged 40-59. All samples contained the mucosa and submucosa regions including minor parts of the circular muscle layer. Following surgical removal, tissue samples were immediately immersed in O.C.T. compound (VWR International GmbH, Dietikon, Switzerland), frozen in liquid nitrogen and stored at −80°C. Diagnosis of disease status was made according to established criteria for histopathological analysis.
Experimental IBD in mice with DSS and LPS
Paradigms for the induction of inflammation were either 1 week (acute) or 3-4 weeks (chronic) with or without an incubation phase of 2, 6 or 18 months post application (Figure 2). Acute systemic inflammation was induced by intraperitoneal Lipopolysaccharide (LPS) application (80) of 0.5 mg/kg in 100 µl injection volume on day 1 and 4 (Sigma-Aldrich Chemie GmbH, Steinheim, Germany, LPS 055:B5). Acute colitis was induced by application of 36-50kDa Dextran Sulfate Sodium (DSS) (81) (160110, MP Biomedicals, LLC, Illkirch, France) at 0%, 1%, 2.5% or 5% in autoclaved drinking water for 5 continuous days respectively, followed by 2 days of water (1 DSS application cycle). Chronic colitis was achieved by 4 repeating DSS application cycles. The DSS concentration during 4 weeks of chronic colitis was either 1% or 2.5% for 4 weeks or 2.5%-4% raised 0.5% every week for 4 weeks. Mice from same littermate group were randomized per cage into vehicle and inflammation inducing agent.
IL-10 treatment and exposure measurement
Two different forms of mouse IgG bound murine IL-10 (mIgG(v1)-mIL10 and mIgG(v2)-mIL10) were diluted in pre-prepared sterile formulation buffer comprised of 0.5% mouse serum supplemented with 25mM citrate, 300mM arginine to a final concentration of 0.75mg/ml and the pH adjusted to 6.7 on the day of application. Each mouse was treated once with 150 µg i.p concurrently with the initiation of the acute colitis paradigm with 5% DSS. The concentrations of mIgG-mIL10 fusion proteins in murine serum samples were determined by enzyme-linked immunosorbent assays (ELISA) specific for the Fab moiety of the administered mIgG-mIL10 fusion protein. Biotinylated mIgG-mIL10-specific target molecules were used for capturing, goat anti-mIg IgG-HRP conjugate and peroxidase substrate ABTS was used for quantitative detection of mIgG-mIL10 fusion proteins.
Immunohistochemistry
Animals were injected with a lethal dose of pentobarbital (150 mg/kg). Upon full anesthesia, mice received transcardial perfusion with room temperature phosphate buffered saline (PBS). For biochemical and immunohistochemical analysis, one section of either the proximal colon was fresh frozen and stored at −80°C or post-fixed in 4% paraformaldehyde (PFA) solution for 24 h. Following post-fixation, organs were incubated in 30% sucrose/PBS at 4°C for at least 48 h before further processing. Subsequently, enteric tissue was cryotome-sectioned to 35 µm thick longitudinal sections (approx. 1 cm length). The brain was collected and post-fixed for 24 h in 4% PFA followed by 30% sucrose in phosphate buffer until cryo-sectioning of floating sections at 40 μm. Histological analysis of the colon was performed using standard hematoxylin staining. Immunohistochemical staining was accomplished using the Vectastain Elite ABC Kits and Peroxidase Substrate Kit SK-4100 (Vector Laboratories, Burlingame, CA, USA) or fluorescently labelled secondary antibodies (Alexa coupled to dye 488, 555 or 647, Life Technologies, Zug, Switzerland). The following primary antibodies have been used for overnight incubation at a dilution of 1:1000: monoclonal antibody to human α-synuclein (sc-12767, Santa Cruz Biotechnology, Heidelberg, Germany), polyclonal antibody to murine α-synuclein (Syn1, BD Transduction Laboratories, Allschwil, Switzerland), polyclonal antibody to the peripheral neuronal marker Peripherin (Millipore Corporation, Billerica, MA, USA), and polyclonal antibody to macrophage marker Iba1 (Wako Chemical GmbH, Neuss, Germany). To detect phosphorylated αSyn (pSer129 pathology) in the free-floating brain sections, monoclonal antibody (ab51253, Abcam, Cambridge, USA) to human αSyn was used at a dilution of 1:10000. Prior to the pSer129 staining, the free-floating brain sections were incubated for 10 min at room temperature in a phosphate buffered saline solution containing 10 μg/mL proteinase K (Cat # 25530015; Invitrogen, California, USA). TH-immunoreactive cells were detected using a polyclonal antibody (657012, Millipore Sigma) at a dilution of 1:1000. To measure the density of Nissl-positive cells, the TH-stained cells were counter-stained with Cresyl violet. The slides were incubated in 0.1% Cresyl violet solution for 9 min and then dehydrated in 95% and 100% ethanol and then xylene prior to coverslipping with Cytoseal 60 mounting media (Thermo Fisher Scientific). Quantifications of the blind-coded TH/Nissl stained slides were done using Stereoinvestigator (version 2017.01.1; MBF Bioscience, Williams, VT, USA) on Imager M2 microscope (ZEISS) coupled to a computer. We analyzed 5-7 nigral sections per animal, and a total of 7-8 animals per treatment group. We outlined the substantia nigra pars compacta and counted every TH-immunoreactive and Nissl-positive cell in that area and computed the number of cells per section, generating the mean cell density per animal. We then calculated the mean density of cells per treatment group and analyzed the data using unpaired Student’s T-test after confirming normality and homoscedasticity in Prism 7.0 (GraphPad Software).
Imaging and stereological quantification of αSyn deposits in enteric nervous system
Imaging and stereological quantification was performed on a Zeiss Axio Imager Z2 fluorescence microscope (Carl Zeiss AG, Jena, Germany). Leica TCS SP5 confocal system using an HCX PL APO CS 40x 1.3 oil UV or an HCX PL APO LB 63x 1.4 oil UV objective was utilized for image recording. Accumulation of αSyn in the ENS was assessed on a random set of 3 adjacent 35 µm thick, αSyn-immunostained sections comprising the myenteric and submucosal neuronal plexuses. Analysis was performed with the aid of Stereologer software (Stereo Investigator 10, MBF Bioscience, Williams, VT, USA) as described previously (82). In the myenteric plexus ganglion volume was defined by multiple outlined plexuses containing a range of 5-20 neuronal cells and quantified by the optical fraction fractionator technique. In contrast to the myenteric plexus, the submucosa consists of compact plexuses with 1-5 cells including interconnecting neurites. Therefore, the entire submucosa was set as region of interest, analyzed with the area fraction fractionator technique. Results of the submucosal plexus are displayed by percent area containing αSyn deposits. For the IL-10 experiment, αSyn positive granules from immunofluorescence images were counted for each image. Granules were filtered based on having a size between 12 and 50000 pixels and a minimal intensity value greater than 300. The filtering step was included to exclude small background particles and macrophages (very large spots). The counts were then aggregated to the animal level by summing the granule counts of all images per animal and then normalizing for (i.e. dividing by) the number of images for a given animal. Upon exploratory data analysis two animals were excluded: one mouse because it only had one image and another due it being an outlier, based on its infiltration score and image data.
Blinding
For analyses of colon and brain tissue on slides, a second individual assigned unique codes to stained slides. Therefore, the experimenter conducted the analyses blinded to the identity of the mice. For randomization of treatment groups see above.
Quantification of leukocytes infiltration
To determine the leukocyte covered area in the colon after LPS or DSS application, three adjacent hematoxylin stained sections were quantified. Total area of colon sections and localizations of leukocyte assemblies within the tissue architecture were identified and outlined utilizing Stereologer Software (Stereo Investigator 6, MBF Bioscience, Williams, VT, USA). Percentage of leukocyte covered area has been set in proportion to total area of the analyzed colon section. For the IL-10 experiment, hematoxilin stained colon slices were examined by an expert pathologist blinded to treatment conditions. A score of 0-3 was assigned to each section for each of the 3 layers lamina propria, submucosa and muscularis based on the degree of inflammatory infiltration. A score of 0 denoted no inflammation and a score of 3 indicated extensive infiltration. The mean of the values for all 3 layers was taken as the final measure of leukocyte infiltration per mouse.
Quantification of αSyn/Iba1 double positive macrophages
The number of α-syn+/Iba1+ positive cells was evaluated by quantification of 10 random regions in 2 adjacent sections of the proximal colon. The region of interest was set to contain the myenteric plexus/circular muscle layer and the submucosal plexus. Cells were assessed for positive αSyn staining and concomitant co-localization with the macrophage marker Iba1 was quantified.
Scoring of pSer129 pathology and brain heatmap
We evaluated pSer129 pathology on a full series of immunostained coronal sections from 10 mice per treatment group (i.e. water vs. DSS-treated groups) on blind-coded slides using a previously described method (83). We visualized pathology from one hemisphere of all brain sections (apart from the olfactory area) using NIKON Eclipse Ni-U microscope and assigned scores ranging from 0 to 4 to each brain area based on the relative abundance of PK-resistant pSer129-positive inclusions (i.e. cell bodies and neurites). In this case, 0 = no aggregates, 1 = sparse, 2 = mild, 3= dense, 4 = very dense. For the heatmap, we obtained the average score values of each brain area for each treatment group. The average data for each treatment group (n=10/ group) was then represented as a heatmap in a sagittal mouse brain background (http://atlas.brain-map.org/atlas?atlas=2#atlas=2&structure=771&resolution=16.75&x=7755.7470703125&y=3899.625&zoom=-3&plate=100883867&z=5).
Densitometry of pSer129 αSyn brain pathology
The density of pSer129 pathology in 12 major brain areas (reticular nucleus, pontine reticular nucleus, periaqueductal gray, gray and white layer, reticular formation, substantia nigra, ventral tegmental area, thalamus, hypothalamus, central amygdala, pallidum and striatum) was determined in the water and DSS-treated animals. A NIKON Eclipse Ni-U microscope was used to acquire 20x magnification images (without condenser lens) from all the indicated brain areas, using the same exposure time for all images. In all cases, images were acquired on three sections separated by 420 μm intervals (localized between Bregma). We then processed the acquired images using Image J64 (84), created a mask (to exclude background) that redirects to the original image for analysis, measured the total area and the mean grey value of the area that had inclusions. For brain areas such as periaqueductal gray that do not fill the entirety of the field to be analyzed, we drew a contour of the area and the analysis was performed only within that contoured area. We subsequently calculated the grey value of the area per square pixels for each image (i.e. A.U./px2 = mean grey value x area stained/total area assessed). Based on this, we calculated the average grey value per square pixels for each brain area for each animal (n = 6 mice/group), and then extended this calculation to determine the average grey value per square pixels for each treatment group and each of the twelve brain areas of interest.
mRNA expression
To assess mRNA expression levels from the proximal colon, RNA was extracted from fresh frozen tissue with MagnaLyser green beads (Roche Diagnostics, Mannheim, Germany) and Qiazol Lysis (Reagent cat.no.79306, Hilden, Germany) purified on MagnaPure LC (HP Kit no.03542394001, F. Hoffmann - La Roche AG, Rotkreuz, Switzerland) and amplified via real-time PCR (4ng RNA/reaction; Lightcycler 480, Roche Diagnostics Corporation, Indianapolis, USA). Amplification of mRNA was performed by using TaqMan probes for human or murine specific α-synuclein and for selected cytokines/chemokines (Applied Biosystems Europe B.V., Zug, Switzerland). Target mRNA was normalized to tissue-specific murine GAPDH levels and displayed as relative expression after 30 amplification cycles.
Statistics
Statistical analysis of gut pathology and inflammation was performed using GraphPad Prism 6.04 or 7.0 software (GraphPad Software, Inc. La Jolla, CA, USA). The results are expressed as mean values ± standard errors of the mean (SEM). Student’s T-test (or Welch’s T-test for unequal variances) was used to compare two groups and ANOVA was used for multi-comparison of groups followed by Tukey HSD post-hoc analysis. For the statistical analysis of the pSer129 αSyn brain pathology, negative-binomial mixed-effects models with a random intercept for each sample were used to analyze the dataset via the ‘lme4’ (http://lme4.r-forge.r-project.org/) package in R v 3.4.4. To analyze the pSer129 αSyn cell count dataset, an offset for the total area examined was included to model the densities. Linear contrasts with false discovery rate (FDR) adjustments were then used to test our hypotheses and account for multiple testing (for brain area and experimental group). Like the pSer129 dataset, the Iba-1/αSyn-double positive dataset were analyzed using negative-binomial regression and Tukey HSD adjusted contrasts to test our hypotheses.
For the mRNA expression analysis, data quality was assessed by inspecting the distribution of Cp values of reference endogenous genes across samples, by inspecting the level of Cp variation between technical replicates and by exploring the samples multivariate signal distribution as in a principal component analysis. Relative gene expression levels were expressed as 2-(Cpgene – CpRef). Statistical analyses to assess the effect of the experimental conditions on the log2 gene expression levels were done with linear models using the limma package (Bioconductor/R, Smyth, 2005). These analyses were implemented in R v3.1.1.
For the statistical modelling of the effects of the IL-10 treatment on αSyn counts, as well as infiltration scores, the levels for IgG1(v1)-IL10 and IgG1(v2)-IL10 treatment were compared to the positive (vehicle/DSS) control. Additionally, since levels of the control antibody treatment (IgG1(v1)) were very similar to the positive control, the two groups were pooled in further contrasts in which effects of individual antibodies or control IgG was assessed. For αSyn counts, a linear model on the treatment groups with one-degree freedom contrasts was applied. For the infiltration score a Kruskal-Wallis test, with the same contrasts, was used.
Study approval
The human subjects’ study was conducted with the approval of the local Ethical Committee in Bern No. 47/04. Written informed consent was obtained from each patient. The animal experiments were approved by a Roche internal review board and the local authorities.
Author contributions
S.G., N.M. and L.S. planned and performed the in vivo experiments, colon immunostaining, analysis, and quantification; S.G. and N.M. drafted a first version of the manuscript; E.Q. performed, imaged, quantitated pSer129, TH and Nissl staining in the brain sections, and drafted a more advanced version of the manuscript with J.A.S., who also provided helpful discussion. F.B. and K.O.S. supported the image acquisition and image analysis for the colon samples; M.St. performed imaging and data analysis of experiments with wildtype mice; G.D.P. and J.S.P. performed statistical analysis of the DSS experiments; H.R. and M.H. performed mRNA analyses; M.Se. trained S.G. and L.S. on mouse necropsy and supported their work; P.M. performed expert pathology staging on leukocyte infiltration; T.E. and A.W. provided mIgG-mIL-10 fusion proteins and measured serum exposure; Z.M. performed statistical analysis for the pSer129 αSyn immunohistochemistry data. M.L.E.G. provided helpful discussion and project planning. A.H. co-mentored S.G. and N.M., performed expert pathology staging on leukocyte infiltration and contributed to experimental planning. C.M. trained S.G. on the colitis model, provided human tissue and expert input on the experimental IBD model. M.B. and P.B. co-mentored Roche Postdoctoral Fellows S.G. and N.M., conceived and oversaw the study, and performed experimental planning; M.B., P.B. and E.Q. wrote the final version of the manuscript.
Conflicts of Interest
At the time of the study S.G. and N.M. were Roche Postdoctoral Fellows employed by Roche and L.S., F.B., G.D.P., J.S.P., K.O.S., H.R., M.H., M.Se. M.St., P.M., A.W., T.E., A.H. and M.B. are or were fulltime employees or trainees at Roche and they may additionally hold Roche stock/stock options. S.G. and L.S. are currently employees of Neurimmune AG, Schlieren, Switzerland. P.B. has received commercial support as a consultant from Renovo Neural, Roche, Teva, Lundbeck A/S, AbbVie, NeuroDerm, Fujifilm Cellular Dynamics, Living Cell Technologies, IOS Press Partners, and Axial Biotherapeutics. Additionally, P.B. has received commercial support as a consultant from Renovo Neural, Roche, Teva, Lundbeck A/S, AbbVie, NeuroDerm, Fujifilm Cellular Dynamics, Living Cell Technologies, IOS Press Partners, Axial Biotherapeutics and CuraSen. Additionally, P.B. has received commercial support for grants/research from Renovo, Roche, Teva, and Lundbeck. P.B. has ownership interests in AcouSort AB. The other authors do not have conflicts of interest with regard to this research.
Acknowledgments
We acknowledge the human donors for providing tissue used in this study. We thank Drs. L. Ozmen, A. Bergadano, and A. Su for their tremendous support in maintaining the mouse colony and establishing of relevant animal experiment licenses, and we are grateful to the animal care takers, veterinarians and many unnamed staff at Roche for their valuable work with the mice in this study. In addition, at Roche we thank Dr. K.G. Lassen for critical input to the paper, Dr. C. Ullmer for co-mentoring S.G. and providing scientific input, Dr. L. Collin for helping with confocal imaging and we are grateful to Dr. T. Kremer, N. Haenggi, D. Mona, A. Girardeau, and J. Messer for providing support in tissue dissections and G. Walker and R. Lauria for technical support. Ms. E. Schulz from VARI assisted with immunostaining of the brain tissue. We thank the Contract Research Organization Frimorfo for carefully sectioning the brains for this study. We acknowledge Drs. L. Gaudimier (née Chicha) and F. Pan-Montojo for scientific discussions early in the project and Dr. W. Zago from Prothena for valuable scientific input throughout the project. P.B. reports relevant grants from NIH (R01DC016519-01, 1R21NS106078-01A1 and 5R21NS093993-02), Department of Defense (W81XWH-17-1-0534), The Michael J. Fox Foundation for Parkinson’s Research, and Cure Parkinson’s Trust. Finally, we thank the Roche Postdoctoral Fellowship Program for supporting S.G. and N.M.
References
- 1.↵
- 2.↵
- 3.↵
- 4.↵
- 5.↵
- 6.↵
- 7.↵
- 8.
- 9.↵
- 10.↵
- 11.
- 12.↵
- 13.↵
- 14.↵
- 15.↵
- 16.↵
- 17.↵
- 18.↵
- 19.
- 20.
- 21.↵
- 22.↵
- 23.↵
- 24.↵
- 25.
- 26.↵
- 27.↵
- 28.↵
- 29.↵
- 30.↵
- 31.↵
- 32.↵
- 33.↵
- 34.↵
- 35.↵
- 36.↵
- 37.↵
- 38.↵
- 39.↵
- 40.↵
- 41.↵
- 42.
- 43.↵
- 44.↵
- 45.↵
- 46.↵
- 47.↵
- 48.↵
- 49.↵
- 50.↵
- 51.↵
- 52.↵
- 53.↵
- 54.↵
- 55.↵
- 56.↵
- 57.↵
- 58.↵
- 59.↵
- 60.↵
- 61.↵
- 62.↵
- 63.↵
- 64.↵
- 65.↵
- 66.↵
- 67.↵
- 68.↵
- 69.↵
- 70.
- 71.
- 72.↵
- 73.↵
- 74.↵
- 75.↵
- 76.↵
- 77.↵
- 78.↵
- 79.↵
- 80.↵
- 81.↵
- 82.↵
- 83.↵
- 84.↵