Primary human hepatocytes take up oligomeric α-syn assemblies in vitro
Given the abundant expression of Cx32 within liver hepatocytes and its recently identified role in promoting oligomeric α-syn uptake, we assessed whether the liver is susceptible to α-syn accumulation in PD. Therefore, we first investigated whether human hepatocytes could take up α-syn protein assemblies associated with PD [42]. Thus, we generated recombinant human α-syn oligomers (oα-syn) and fibrillar assemblies tagged to ATTO-550 and then characterized their ultrastructure using transmission electron microscopy [42]. Consistent with previous studies [31, 42], we identified the characteristic donut-like and filament ultrastructure typical for oα-syn and fibrillar assemblies, respectively, confirming that ATTO-550 has no discernible effect on assembly formation as previously reported (Fig. 1A) [31, 42]. Following protein assembly validation, we incubated oα-syn and non-sonicated fibrillar assemblies (to avoid oligomer formation [50]) with primary human hepatocytes isolated from five healthy adult human livers for 24 h. Subsequently, we assessed protein uptake by Western blot analysis and confocal imaging techniques. Consistent with previous results obtained from human neurons and oligodendrocytes [31], primary human hepatocytes in vitro displayed a preferential uptake for oα-syn assemblies compared to its fibrillar counterpart (Fig. 1B).
To further assess oα-syn uptake in primary human hepatocytes, we immunolabeled oα-syn with a pan-specific human α-syn antibody (14H2L1 (14H)) targeting the C-terminal region of the α-syn molecule (amino acids 121–140) (Additional file 10: Table I). As expected, we observed hepatocytic α-syn uptake localized intracellularly and at the cellular membrane as visualized by the ATTO-550 tag likely within a Cx32 gap junction plaque (Fig. 1C–F, arrow). Moreover, a direct co-localization between ATTO-550 and the human specific α-syn antibody 14H was clearly observed (Fig. 1C–F). We next immunolabeled the cells with Cx32 antibody and observed a partial co-localization between ATTO-550 labeled oα-syn assemblies and Cx32. This feature was particularly evident at the cellular membrane, the predominant subcellular localization of Cx32 (Fig. 1G–J, Additional file 1: Fig. 1A–L), thus validating a direct interaction between oα-syn and Cx32 during cellular uptake as previously reported [31]. Notably, in the absence of oα-syn treatment, no ATTO-550 fluorescence or α-syn immunoreactivity was observed, indicating cellular protein uptake and, importantly, validating the specificity of the antibody to human α-syn (Additional file 2: Fig. 2A–D). To further demonstrate a Cx32/α-syn protein interaction, we stably overexpressed Cx32-mCherry in the widely used human hepatocyte cell line HuH-7, a cell type that normally lacks Cx32 expression [51]. Following stable Cx32 protein expression, we next incubated HuH-7-Cx32 cells with ATTO-488-oα-syn protein assemblies for 24 h and then assessed protein uptake by confocal image analyses. Consistent with the results above, we observed protein internalization and most notably, a direct co-localization between Cx32 and ATTO-488 labeled oα-syn assemblies in human HuH-Cx32 cells (Fig. 1K–N). As expected, untreated cells showed no oα-syn reactivity (Additional file 2: Fig. 2E–H). We next immunoprecipitated (IP) α-syn from HuH-Cx32 cells treated with oα-syn for 24 h. Confirming the results above, IP of α-syn successfully pulled down Cx32 from HuH-Cx32 cells whereas untreated cells showed no Cx32 pull down (Additional file 2: Fig. 2I). It is worth noting that in contrast to unmodified cells, Cx32 expression in HuH-7 cells changed the cellular phenotype resembling a hepatocytic-like morphology, a signaling mechanism likely enhanced by the gap junction protein Cx32 (Additional file 2: Fig. 2J).
To determine whether human hepatocytes could degrade oα-syn following treatment, we incubated HuH-7 wild type or Cx32-expressing human hepatocytes with oα-syn assemblies for up to 72 h followed by assessment of protein uptake at different time points using Western blot analysis. Consistent with previous results [31], Cx32 expression in HuH-7 cells increased oα-syn uptake thus suggesting that Cx32 at the cellular membrane interacts with oα-syn and promotes intracellular uptake (Additional file 3: Fig. 3A–C) [31]. Notably, and consistent with the role of hepatocytes in toxin clearance and detoxification, wild type as well as Cx32 expressing HuH-7 hepatocytes showed a clear reduction of oα-syn assemblies over time (compare 24 h vs 72 h, Additional file 3: Fig. 2A, B). These results validate the degradation capacity of human hepatocytes to toxic substances including PD-associated pathology.
Identification of human α-syn pathology within the liver of the A30P mouse model of PD
To assess whether the liver is susceptible to the accumulation of α-syn pathology in vivo, we turned to a widely used mouse model of PD overexpressing approximately 2 times the levels of human α-syn compared to mouse α-syn harboring the Ala30Pro mutation (A30P) under control of the neuronal Thy-1 promoter (Additional file 11: Table II) [52]. Thus, we performed immunohistochemistry on brain and liver tissue sections from young (3 months) and aged (18 months) transgenic (Tg) PD mice using human specific α-syn antibodies (Additional file 10: Table I). As expected, we observed a clear accumulation of human α-syn in the brain of this model between 3 and 18 months of age (Fig. 2A, G, inserts). Strikingly, we also identified the presence of human α-syn deposits which appeared throughout the young A30P liver, localizing to regions near the portal and central veins (Fig. 2A–C). In some instances, we also observed the appearance of rosette-like α-syn deposits within the liver parenchyma (Fig. 2D–F). Interestingly, we also detected the presence of human α-syn deposits within the capsule of Glisson, the connective tissue of the liver which contains the vessels (data not shown). In aged A30P livers (18 months), we observed a progressive deposition of α-syn within the portal and central veins (Fig. 2G–I). We could also detect a significant increase in α-syn deposition within the liver parenchyma compared to young livers (Fig. 2J–L, Additional file 3: Fig. 3C). No staining or cross reactivity with mouse α-syn could be detected in neither A30P nor WT mice, regardless of age, when the primary antibody (14H) was omitted as a control experiment (Additional file 4: Fig. 4A–F). Notably, the aggressive accumulation of α-syn within the A30P liver appeared to correlate with the accumulation of α-syn within the brain of this PD model [34]. To eliminate the possibility that α-syn accumulation in the liver could be due to endogenous expression of the SNCA gene driven by the Thy-1 promoter, we performed semi-quantitative gene expression analysis (qRT-PCR) on Tg brain and liver samples as well as non-Tg tissue sample controls (18 months). Using specific probes targeting human α-syn (Additional file 10: Table I), we observed the expression of human α-syn in the brain of the A30P model but absent in WT mice. Moreover, neither A30P nor WT livers displayed any human SNCA mRNA expression (Additional file 4: Fig. 4G). We also assessed for the presence of endogenous mouse α-syn and as expected, we found such endogenous expression in the brain of WT and A30P mice but not in the liver (Additional file 4: Fig. 4H). Taken together, these findings show that α-syn in the liver in this mouse model likely originates from the brain. This could potentially be a way to facilitate clearance and detoxification out of the brain. Thus, our results suggest a potential liver involvement in the clearance of α-syn pathology.
Characterization of human α-syn within the liver of A30P mice
To further characterize the identity and accumulation of human α-syn within the liver of the A30P model, we performed immunohistochemistry on aged liver tissue sections using the widely used human specific α-syn clone Syn-211 (211) antibody which recognizes α-syn pathology in PD brain [53]. Using confocal image analysis, we observed a clear co-localization between the 14H and 211 antibodies, validating the identity of human α-syn within the liver of A30P mice (Fig. 3A–H). We next investigated whether the presence of α-syn within the aged liver is phosphorylated at serine 129 (pS129), a pathological hallmark associated with PD [54]. Using confocal image analysis, we observed a partial co-localization between pS129 and 211 antibodies (Fig. 3I–P), suggesting that the accumulation of α-syn pathology within the liver may undergo a similar post-translational modification to that in the brain. Alternatively, it is possible that human α-syn is transported from the brain to the liver in an already phosphorylated state.
We next investigated the aggregation state of α-syn within the liver using Thioflavin T, a molecular probe known to stain fully mature amyloid structures in multiple neurological conditions. In our hands, however, Thioflavin T failed to label any of the α-syn deposits within aged A30P liver (data not shown), suggesting a limited sensitivity of the probe and/or the lack of fully mature α-syn amyloidogenic structures within the liver. To overcome this technical limitation, we next stained tissue sections with a highly sensitive class of amyloid dyes commonly known as luminescent conjugated oligothiophenes (LCOs), which are ligands identified to stain protein aggregates that are formed early in the aggregation process [38,39,40, 55]. Using two well-characterized LCOs (p-FTAA and HS-68), we revealed the presence of p-FTAA-positive deposits (Fig. 4A–D) as well as a select number of HS-68-positive structures which co-localized extensively with the human specific α-syn antibody, 14H (Fig. 4E–H). These findings provide evidence for the presence of human α-syn within the liver in a partially aggregated state.
Given that the emission spectra of LCOs provide clues into the amylogenic state of its target [55], we next measured the wavelength spectra of α-syn liver deposits using the HS-68 ligand. Consistent with the lack of Thioflavin T staining, the emission spectra observed using the HS-68 LCO revealed two specific peaks, one peak around 530 nm and another peak approximately at 570 nm wavelength (Fig. 4I, J). A similar spectral signature was previously reported for immature amyloid beta (Aß) and tau aggregates in brains of young transgenic AD mice and might therefore be indicative of less mature assemblies [55, 56]. Interestingly, the emission peaks of the HS-68-positive α-syn deposits within the A30P liver resembled the peaks previously obtained from human PD brains using the same ligand but were markedly different to the wavelengths observed for MSA brain deposits (compare Fig. 4J with K [37]). Collectively, our results indicate that the A30P liver is susceptible to α-syn accumulation, thus suggesting a potential liver involvement in the clearance and detoxification of pathological protein aggregates from the brain.
Striatal injection of α-syn assemblies localize to the liver in WT mice
To further explore whether α-syn is transported from the brain to the liver, we took advantage of a widely established mouse model of PD involving α-syn injections in the brain leading to α-syn self-assembly and accumulation that is detected within the brain approximately within 20–30 days post-protein injection [57, 58]. Thus, we performed a single striatal injection of mouse oα-syn assemblies into wild type mice and assessed their presence within the liver 30 days post injection. Consistent with the detection of human α-syn within the A30P liver, we observed the appearance of injected oα-syn assemblies within the portal tracts (Fig. 5B–D), liver parenchyma and sinusoidal regions, the specialized channels which allow blood flow from portal tracts to the hepatic venule (Fig. 5F–H). As expected, adjacent control sections not treated with α-syn primary antibody showed no α-syn staining (Fig. 5A, E). Similarly, and consistent with our qRT-PCR analysis, we observed no staining within the portal tracts or liver parenchyma in non-α-syn injected wild type mice (data not shown). We then assessed whether the presence of injected oα-syn assemblies within the liver were pS129 positive. In contrast to the aged liver tissue sections from the A30P mouse model however, we found no evidence of pS129 immunoreactivity within either the portal tracts (Fig. 5I–L) or hepatocytic structures 1-month post-protein injection (Fig. 5M–P). These results suggest that α-syn is transported from the brain to the liver, likely via the circulatory system. Alternatively, it is also possible that the vagal or splanchnic nerves which directly innervate the liver via the portal area may play role in this process.
Human α-syn accumulation within the liver promotes inflammation
Given the inflammatory response generated by α-syn deposition in the brain and other organs outside the CNS, we next investigated whether the A30P liver is susceptible to inflammation as a result of a progressive α-syn deposition. We therefore performed hematoxylin and eosin (H&E) staining on young (3 months) and aged (18 months) liver tissue sections from WT and A30P mice. In WT mice, we observed no clear signs of inflammation regardless of age (Fig. 6A, B, E, F). In contrast, the A30P liver showed a progressive inflammatory pattern that appeared already by 3 months of age (Fig. 6C, D). Indeed, we observed focal inflammation areas throughout the liver parenchyma (Fig. 6C, D, squares). At 18 months of age, we observed a much more severe inflammatory reaction with more extensive focal inflammation within the liver parenchyma as well as in the portal tracts (Fig. 6G, H). In these livers, however, we found no evidence of fibrosis or steatosis in either WT or A30P mice regardless of age (Additional file 12: Table III). These findings suggest that the progressive accumulation of α-syn deposits within the A30P liver may be directly responsible for the progressive inflammation observed, as no detectable signs of inflammation were observed in WT mice regardless of age. These findings are consistent with the α-syn aggregation-dependent inflammation observed in the brains of the A30P mice and in PD patients [34].
Human α-syn within the liver partially co-localizes with inflammatory markers
Given the progressive inflammatory state of the A30P liver compared to livers from wild type mice, we next assessed whether inflammatory cells co-localize with α-syn following its transport to the liver. Thus, we performed confocal image analyses on tissue sections from aged A30P mice and identified the presence of CD45-positve leukocytes (Fig. 7A–D), CD11b-positive monocytes (Fig. 7E–H), and F4/80-positive Kupffer cells (Fig. 7I–L) that were next to or in close proximity to human α-syn deposits. However, none of these cell types co-localized with α-syn. Similarly, immunohistochemical staining for CD3 or MPO (myeloperoxidase) revealed the presence of T-cells and granulocytes within the liver parenchyma respectively, however, these cell types were also devoid of any human α-syn as no co-localization was observed (data not shown). Interestingly, we did observe a partial co-localization with the GFAP-positive perisinusoidal cells commonly known as hepatic stellate cells (Fig. 7M–P), specialized liver cells involved in fibrosis following liver injury [59], thus suggesting that in addition to inflammation, α-syn may promote liver toxicity. Taken together, our observations suggest that human α-syn is likely transported from the brain to the liver and its progressive accumulation within the liver promotes cellular toxicity resulting in liver inflammation.
Accumulation of α-syn within the liver is a general phenomenon of synucleinopathies
To further determine whether the accumulation of α-syn in the liver is not specific to the A30P model but rather a general phenomenon of synucleinopathies, we investigated the presence of human α-syn within the liver in a model of synucleinopathy overexpressing normal WT α-syn also under control of the Thy-1 promoter (L61, Additional file 11: Table II) [35]. As expected, we observed a progressive accumulation of human α-syn in the brain of this model between 3 and 12 months of age (Additional file 5: Fig. 5A and H, inserts). Consistent with our results from the A30P model, we also observed the presence of human α-syn within the liver of young L61 mice (3-month old) as small puncta distributed focally in portal veins and liver parenchyma (Additional file 5: Fig. 5A–F). In aged L61 livers 12 months old, we could reveal a progressive accumulation of human α-syn deposits (Additional file 5: Fig. 5G–L). Notably, α-syn deposition within the L61 liver appeared to occur to a much lesser degree than for the A30P model at any given time point analyzed (Additional file 12: Table III), despite the expression of both proteins under control of the same neuronal promoter (Additional file 11: Table II). Thus, the accumulation of liver α-syn appears to correlate with the amount of α-syn deposition within the brain of these synucleinopathy models [60].
We next assessed for the presence of α-syn within the liver of transgenic mice modeling MSA. These animals express human α-syn within oligodendrocytes approximately 3-times the levels human α-syn (Additional file 11: Table II) under control of the myelin basic promoter (MBP, MBP29) [36, 45]. At 4 months of age and consistent with the PD models above, we observed the presence of human α-syn deposits throughout the MBP29 brain (Additional file 6: Fig. 6A, insert). Within the MBP29 liver, we observed occasional α-syn puncta that were randomly distributed throughout the liver parenchyma (Additional file 6: Fig. 6A–C). In some instances, we also observed the presence of human α-syn surrounding the inflammatory cells within the sinusoidal region (Additional file 6: Fig. 6D–F). In contrast to the A30P model, α-syn pathology within the MSA mouse liver at 4 months was negative for pS129 immunostaining (data not shown). As expected, tissue sections lacking primary antibody were devoid of any human α-syn staining (Additional file 6: Fig. 6G–I). To also eliminate the possibility that human α-syn accumulation within the MBP29 liver is due to expression of the human SNCA gene, we performed qRT-PCR on brain and liver tissue samples from MBP29 and WT control mice. Using specific probes targeting human α-syn (Additional file 10: Table I), we observed restricted expression of the human SNCA transgene to the brain of this mouse model and as expected, absent in both the MBP29 and WT liver samples (Additional file 7: Fig. 7). Importantly, the limited lifespan of this aggressive MSA model [36] prevented us from investigating the accumulation of human α-syn pathology in older mice.
Finally, we assessed whether other proteins associated with neurodegenerative disorders are also deposited within the liver. To this end we performed immunohistochemistry on brain tissue sections from the samples from mice expressing the human amyloid precursor protein (APP) harboring the Beyreuther/Iberian mutation (AppNL−F mice) thus modeling AD [61]. As expected, we observed a clear accumulation of human amyloid beta (Aβ) within the brain of this mouse model at 24 months of age (Additional file 8: Fig. 8A, insert). However, and in contrast to the progressive α-syn deposition observed within the synucleiopathy mouse models shown above (A30P and L61), the AppNL−F mice (20 months) lacked any detectable age-dependent Aβ accumulation within the liver (data not shown). In mice of advanced age, however (24 months), we observed the presence of rare Aβ inclusions within the liver positive for an array of widely used human specific Aβ antibodies (Additional file 8: Fig. 8A–F). As expected, the lack of primary antibodies or immunostaining on WT mice failed to detect any Aβ inclusions (Additional file 8: Fig. 8G–L). Moreover, H&E analysis revealed no signs of inflammation in the liver even in the overly aged animals, thus validating the lack of local Aβ accumulation in this mouse model of AD (data not shown). Taken together, our immunohistochemical observations revealed a progressive accumulation of α-syn within the liver of synucleinopathy models that is absent in a model of AD. Intriguingly, accumulation of α-syn was more extensive in the A30P PD model [4] followed by L61 [2] and lastly the MSA model [1] (Additional file 12: Table III). Collectively, these findings suggest that the propensity for α-syn aggregation within the brain and the aggregation state play a role in its accumulation in the periphery, supporting the idea that α-syn within the liver in these models may originate from the brain.
Identification and characterization of α-syn pathology within the liver in PD cases
To validate the accumulation of α-syn within the liver as a pathological phenomenon in PD we assessed Lewy Body Disease cases (LBD, n = 16) with confirmed α-syn deposition in the brain (Braak 5–6, Additional file 13: Table IV) as well as aged-matched controls with no evidence α-syn deposition in the brain (M = 79.8 vs 76.5, n = 14). In a double-blinded setting, we identified the presence of α-syn within the liver that was detected by human-specific α-syn antibodies in several neuropathologically confirmed cases as well as in age-matched control tissues (Additional file 13: Table IV). Similar to our mouse models of PD, the presence of α-syn within the human liver was located to the sinusoidal regions, portal tracts, as well as liver parenchyma (Fig. 8A–L). Moreover, α-syn accumulation within hepatocytes appeared random but most often as round puncta near the nuclei (Fig. 8G–L). Importantly, even within the same affected region, not all hepatocytes contained α-syn and therefore did not seem to be equally vulnerable to α-syn deposition (Fig. 8G–L). Overall, we identified α-syn pathology within the liver in 12 out of 16 neuropathologically confirmed cases with α-syn pathology in the brain (75%). Out of these 12 cases, 6 showed α-syn accumulation within hepatocellular structures (37.5%). Similarly, 8 out of 14 control cases were positive for liver α-syn pathology (57%), whereas 4 of them showed α-syn within hepatocytes (28%). Only occasional cases showed detectable evidence of fibrosis, steatosis or inflammation but this did not correlate with brain pathology or α-syn accumulation within the liver (Additional file 13: Table IV). However, cholestasis was detected in 14 out of 16 neuropathologically confirmed cases (87.5%), whereas 9 out of 14 controls with no α-syn deposition in the brain were positive for the same condition (64%) (Additional file 13: Table IV). We then assessed whether PD or control cases positive for α-syn within the liver show pathological phosphorylation at serine 129 (pS129). While α-syn deposition within the liver was validated with the human-specific 211 antibody, none of the cases analyzed contained pS129 immunoreactivity (Additional file 9: Fig. 9A–L). Collectively, our data demonstrates that α-syn accumulation occurs in aged human livers with a tendency for higher prevalence in neuropathologically confirmed cases with α-syn deposition in the brain (75%) relative to controls without α-syn accumulation (57%). Thus, the propensity for α-syn accumulation within liver tissue may be indicative of a liver’s role in pathological protein clearance derived from either the brain or peripheral tissues.