In this study we leveraged the peak accumulation of pSyn inclusions in the SNpc, observed two months following intrastriatal PFF injections in rats and months prior to neurodegeneration27,28, to identify early differential gene expression in synucleinopathy.
Through the use of LCM guided by GFP under the hTH promoter, we were able to isolate and collect the SNpc from samples, providing enrichment of RNA from nigral dopamine neurons that possess pSyn inclusions as well as immediately proximal glia. Overall, transcripts downregulated in early synucleinopathy in both sexes were primarily associated with neurotransmission and the dopamine pathways, whereas those upregulated in early synucleinopathy were primarily associated with the neuroinflammatory response.
Neurotransmitter release is a choreographed process that requires the interaction of numerous proteins. Both RNASeq and the ddPCR results in early synucleinopathy show the downregulation of several transcripts encoding protein involved in vesicle organization and neurotransmitter release, such as Syn1 (synapsin 1), Syn2 (synapsin 2), Syn3 (synapsin 3), Vamp2 (synaptobrevin 2), Snap25 (SNAP25), Stx1b (syntaxin 1b), Stxbp1 (Stxbp1 also known as MUNC18), Pclo (piccolo), Bsn (bassoon), Erc2 (CAST1), Rims1 (Rims1), Rab3a (RAB3a), Rab3c (RAB3c), Rab27b (RAB27b), Syt1 (synaptotagmin 1), Syt2 (synaptotagmin 2), Syt3 (synaptotagmin 3), Cplx1 (complexin 1), Cplx2 (complexin 2), and Nsf (N-ethylmaleimide sensitive factor, vesicle fusing ATPase). Synaptic vesicles in the axon terminals have been hypothesized to be organized and trafficked into three pools prior to neurotransmitter release: the resting or reserve pool, the recycling pool, and the readily releasable pool. The reserve pool is organized by synapsin proteins interacting with α-syn and has been suggested to be the largest pool of SVs that can feed into the recycling pool if needed45–47. The recycling pool is comprised of SVs that can move into the readily releasable pool when needed, and the readily releasable pool is comprised of synaptic vesicles that are found in the active zone45–52. Decreases in Syn1 have been reported in α-syn overexpression in culture53, rodent models54,55, cultured mouse hippocampal neurons treated with PFFs31, as well as in post-mortem tissue from patients with dementia with Lewy bodies (DLB)53. Similarly, Syn2 decreases in α-syn overexpressing mice55, cultured mouse hippocampal neurons treated with PFFs38, and post-mortem tissue from PD patients13. Reported changes in Syn3 are variable, with increases reported in post-mortem PD tissue55, and decreases reported in iPSCs harboring the A53T SNCA mutation56. Downregulation of synapsins in early synucleinopathy suggests disorganization of synaptic vesicles, likely delaying vesicle movement to the active zone and impairing dopamine release in the nigrostriatal pathway.
During the neurotransmitter release process, synaptic vesicles in the active zone dock and prime for release via interactions between vesicular-SNARE protein synaptobrevin, and the target-SNARE proteins SNAP-25, syntaxin, and Munc18 on the synaptic membrane48–52. The complexin proteins bind to the formed SNARE complex to regulate fusion between the vesicle and cell membrane 48–52. Priming of synaptic vesicles is also modulated by the interaction of Rab proteins and the presynaptic cytoskeletal matrix of the active zone, which is comprised of scaffolding proteins such as piccolo, bassoon, CAST, ELKS, Munc13 and the RIMS proteins48–52. Together, these are predicted to aid in maintaining proximity to calcium channels48–52. Stimulation of the neuron to fire results in the binding of calcium with synaptotagmin on the synaptic vesicle, which displaces complexin and results in membrane fusion and release of the neurotransmitters48–52. Regarding components of the SNARE complex, syntaxin 1 is decreased in post-mortem tissue from DLB patients57, as well as synuclein overexpression models in mice55 and rats58. Changes in synaptobrevin 2 vary by method, with decreases observed in hippocampal cultures overexpressing α-syn or with intracellular PFF induced inclusions38,53, and increases in α-syn overexpressing mice55. SNAP-25 protein is also variable, with decreases reported in post-mortem DLB and PD with dementia patients59,60 and cultured mouse hippocampal neurons treated with PFFs38; and increases in α-syn overexpressing in mice55. Complexins 1 and 2 are decreased in α-syn overexpressing mice55, with complexin 2 also decreased in post-mortem tissue from PD patients55. Involving priming, decreases in Rab3A have been reported in post-mortem tissue from PDD and DLB patients59,60, as well as synuclein overexpression models in mice55 and rats58. Decreases in the cytomatrix scaffold protein piccolo in cultured mouse hippocampal neurons overexpressing a-syn have also been reported53. We observed decreases in synapsins and synaptobrevins, and the complexins and piccolo transcripts generally mirror previous findings associated with protein changes; however, this does not hold true for synaptotagmin. Where we see a decrease in Syt transcripts, synaptotagmin 1 is increased in α-syn overexpressing mice55, and synaptotagmin 2 is increased in post-mortem and priming in the active zone based on deficits in the SNARE complex and cytomatrix proteins, and impaired membrane fusion and fusion pore opening due to decrease synaptotagmin. These biological changes would be expected to decrease dopamine release in the nigrostriatal pathway in early synucleinopathy and potentially produce stress in the axon terminals that could contribute to axonopathy.
Previous studies revealing dysregulation of neurotransmission transcripts and proteins associated with PFF-induced pSyn inclusions used hippocampal cultures31,38. However, nigral dopamine neurons have unique firing patterns61 and use proteins involved in dopamine synthesis and handling that hippocampal cultured cells do not express. In DaNs, dopamine is synthesized in a stepwise process where L-tyrosine is converted to L-DOPA by tyrosine hydroxylase (TH), and DOPA decarboxylase (DDC) converts L-DOPA to dopamine. Models employing α-syn overexpression show decreased Th and Ddc transcripts62 and parallel reduction in TH protein/activity63–67 and DDC protein67. Similarly, we observed decreases in Th and Ddc transcript levels in early synucleinopathy. However, previous findings show that TH immunoreactivity and dopamine synthesis and storage in the nigrostriatal system remains relatively unchanged until 4 months after PFF injection to rats28,68, suggesting a delay between Th transcript and protein decreases.
After synthesis, dopamine is packaged into synaptic vesicles for release via vesicular monoamine transporter 2 (VMAT2)69,70. Dopamine released in the synapse can interact with post-synaptic dopamine receptors on medium spiny neurons, presynaptic dopamine receptors on the axon terminal, or can be transported back into the presynaptic terminal via the dopamine transporter (DAT). Presynaptic dopamine neurons express the inhibitory subtype D2 dopamine receptor subtype (D2R) on the presynaptic terminal. D2R is a G-protein coupled receptor which ultimately decreases TH activity to reduce dopamine synthesis71. Activity of G-protein coupled receptors are modulated by the regulator of G-protein signaling (RGS) proteins, with RGS8 being a known regulator of the D2R activity72,73. In the present analysis, both Drd2 and Rgs8 transcripts decrease in early synucleinopathy. As D2R inhibits TH activity, it is possible that downregulation of Drd2 is a compensatory mechanism to promote dopamine synthesis. Decreased Rgs8 transcript however is contradictory to the compensatory dopamine synthesis hypothesis, as the decrease in RGS8 protein would likely promote D2R related inhibition of TH activity, since RGS8 normally functions by increasing the rate of D2R inactivation.
Our RNAseq analysis identified decreased transcription of the monoamine transporters Slc18a2 (VMAT2) and Slc6a3 (DAT), in early synucleinopathy. At the 2-month post-PFF time point, previous work shows that decreased DAT function in the striatum is already present68, suggesting transcriptional changes translate to functional losses. Similar to our findings, overexpression of α-syn in culture results in inhibition of VMAT2 activity and decreased DAT protein and function74–80. As the primary function of the nigrostriatal dopamine neurons is to synthesize and release dopamine – thereby maintaining balance in the basal ganglia – downregulation of these genes can negatively affect the overall function of these neurons. Decreases in TH and DDC protein would reduce dopamine synthesis, where decreases in VMAT2 and DAT protein would negatively affect dopamine packaging into vesicles for release and reuptake/reuse of dopamine, respectively. Impairment in the dopamine pathway along with deficits in synaptic vesicle organization and general neurotransmission machinery likely produce synaptic dysfunction and thus, may contribute to the nigrostriatal axonopathy observed in early PD81,82.
In addition to dopamine release and reuptake, dopamine metabolism also has the potential to influence the health of the neurons. Dopamine can be metabolized by monoamine oxidase to produce DOPAL, which can be converted to DOPAC by ALDH1A183,84. In our results, Aldh1a1 transcript decreased early in synucleinopathy. Inhibition or dysfunction of ALDHs lead to the accumulation of DOPAL, a toxic metabolite of dopamine. DOPAL can act by inhibiting TH activity85, producing toxic quinones resulting in ROS86, and modifying and promoting α-syn oligomerization which can impair synaptic vesicle function87–95. The resulting synaptic distress can decrease dopamine release, leading to axonopathy, and ultimately, neurodegeneration. Deficiencies in ALDH1A1 have previously been associated with PD, and the subtype of dopamine neurons which produce ALDH1A1 show increased vulnerability in PD43. Downregulation of both ALDH1A1 mRNA17,96 and ALDH1A1 protein97,98 have been observed in post-mortem PD tissue from patients with both familial and idiopathic PD97,98. Related to the loss of ALDH1A1, DOPAL accumulation in the putamen has been found in post-mortem PD tissue99. Given the downregulation of Aldh1a1 that we observe in early synucleinopathy and potential loss of a protective enzyme, DOPAL accumulation and toxicity are mechanisms which could contribute to synucleinopathy progression and should be examined in future studies.
Along with differential expression of several genes related to neurotransmission and the dopamine pathway, we showed that synuclein transcripts Snca and Sncg decrease in early synucleinopathy induced via PFFs. This is in stark contrast to what would be expected in α-syn overexpression models and familial forms of PD, such as SNCA duplication and triplication100. However, previous work has shown decreases in SNCA transcript in LCM-dissected post-mortem PD vs control nigral neurons23, as well as in primary mouse hippocampal neurons with PFF-induced α-syn inclusions31. The mechanism whereby Snca and Sncg are downregulated in inclusion bearing neurons remains unknown; however, it is possible that formation of inclusions affects transcriptional regulators controlling Snca and Sncg.
Nigral samples in our study included multiple cell types immediately adjacent to α-syn inclusion bearing nigral dopamine neurons, including microglia and astrocytes. We found that upregulated transcripts were primarily associated with the neuroinflammatory response, suggesting that the immune system may be triggered by early α-syn pathology. The immune pathways upregulated (Dot plot GO enriched pathway census; (Fig. 2e)) included immune effector processes, cytokine production, B cell activation, and leukocyte activation. Cross-referencing via DropViz (Supplementary File 7) to sort known glial transcripts further identified upregulation of pathways involving the complement system, astrocyte function, antigen presentation, and phagocytosis. Many of these upregulated inflammatory pathways mirror findings from the SN of early Braak stages 1 and 2 PD subjects13, including upregulation of MHC-II and increased presence of B cells. Furthermore, elevation of complement system components has been detected in postmortem PD tissue101,102. Increased microglial MHC-II, astrocytic glial fibrillary acidic protein (GFAP) and complement expression have previously been detected in the pSyn-bearing SN following PFF injection using immunohistochemical methods27,30,35,103,104. While these neuroinflammatory results are intriguing, additional experiments are needed to determine the cellular source(s) (e.g. microglia vs. astrocytes vs. other cell types) of the upregulated immune transcripts in our dataset. Further, the neuroinflammatory transcriptome we observe is likely PFF model stage specific105, as has been reported in other disease models106. Future studies will be required to determine whether the neuroinflammatory transcriptome in the SN during established pSyn aggregation is distinctive from the transcriptome observed during initial aggregate accumulation or during/following nigral degeneration. Overall, our findings illustrate the complex heterogeneity of the immune system response to accumulation of α-syn aggregation in the SN and suggest that identification of critical drivers of neuroinflammation will be required to develop efficacious anti-inflammatory therapeutic strategies.
Another point to consider are the sex differences that we observed in our results. There is a greater incidence of PD, approximately two-fold higher, in males than females107,108; however, females with PD progress faster and have a higher mortality rate108,109. We unexpectedly observed a significant difference between male and female rats in the number of pSyn immunoreactive neurons in the SNpc at the 2-month time point: ~32% more pSyn inclusion containing neurons were observed in the SNpc in males compared to females. We also observed over twice as many differentially expressed transcripts in PFF males than PFF females. These data suggest that the differences in pSyn pathology may underlie transcriptomic differences, but it is also possible that sex-specific differences in the transcriptomic response to PFFs influence pathology. Gene ontology using sex-specific differential transcript expression identified additional pathways that did not emerge when consensus differentially expressed genes (both sexes) were examined. The top pathway identified in males was “Parkinson’s disease”. Enriched pathways identified in males which would be expected to be related to PD are pathways involved in bioenergetics and the mitochondria such as “carbon metabolism”, “glycolysis/gluconeogenesis”, “citrate cycle (TCA cycle)”, and “pyruvate metabolism”; pathways involved in response to protein aggregates such as “lysosome” and “ubiquitin mediated proteolysis”; and pathways involved in sleep disturbances “circadian rhythm” “circadian entrainment”. An example of a pathway implicated in females but not males was the “Wnt signaling pathway”, which interestingly shows sex-specific alterations in female post-mortem PD tissue12. Taken together, our results identified sex-specific changes in transcription during early synucleinopathy, but future studies are required to delineate true sex-specific differential transcript expression associated with α-syn inclusions in the SN in which equivalent pSyn deposition is achieved.
It is important to note the limitations of our approach. First, our refined bulk RNAseq approach does not allow for disambiguation of gene expression changes between neuronal or glial subtypes. However, by cross-referencing our transcriptomic data with the DropViz SN neuron and glial databases, as well as direct cell-specific confirmation of mRNA expression using FISH, we have begun to identify markers that can be used in future single-cell RNASeq studies to specifically identify inclusion-bearing nigral neurons in animal models and PD SNpc. Second, our study is limited to understanding the effects of synucleinopathy on the nigral transcriptome, meaning the impact of synucleinopathy on protein expression in the SNpc or striatum remains to be determined. A third limitation is the poor annotation of the rat genome and overall lack of genetic tools in rats. For example, ~ 50% of the differentially expressed transcripts lacked annotation in males and ~ 45% in females. Even after manual annotation, ~ 14% of differentially expressed transcripts in males and 9% in females could not be annotated to a known gene in the rat genome. This incomplete annotation likely resulted in unmatched gene names and gaps in the data. Poor genome annotation could also contribute to misalignment/misidentification of transcripts. For example, Eno3, an enzyme in glycolysis/gluconeogenesis, was one of the DTEs identified via RNASeq; however, Eno3 is primarily expressed in skeletal muscle and was below detectable limits in the SN with ddPCR (data not shown), whereas Eno2 is expressed in the brain110–112. This emphasizes the need to validate RNASeq data, especially in rats.
In conclusion, using our approach to enrich collection of α-syn inclusion bearing nigral tissue, we have identified large-scale changes in the transcriptome and multiple pathways altered in early synucleinopathy. Given that the differential transcript expression that we observe occurs months prior to neurodegeneration, the dataset that we have generated can provide insight into the underlying mechanisms of synucleinopathy progression. If validated in PD SN, targets initially identified in the rat PFF model could guide development of future disease-modifying therapies.