Alternative α-synuclein transcript usage as a convergent mechanism in Parkinson's disease pathology

Abstract

α-Synuclein is implicated both in physiological functions at neuronal synaptic terminals as well as in pathological processes in the context of Parkinson's disease. However, the molecular mechanisms for these apparently diverse roles are unclear. Here we show that specific RNA transcript isoforms of α-synuclein with an extended 3' untranslated region, termed aSynL, appear selectively linked to pathological processes, relative to shorter α-synuclein transcripts. Common variants in the aSynL 3' untranslated region associated with Parkinson's disease risk promote the accumulation and translation of aSynL transcripts. The presence of intracellular dopamine can further enhance the relative abundance of aSynL transcripts through alternative polyadenylation site selection. We demonstrate that the presence of the extended aSynL transcript 3' untranslated region impacts accumulation of α-synuclein protein, which appears redirected away from synaptic terminals and towards mitochondria, reminiscent of Parkinson's disease pathology. Taken together, these findings identify a novel mechanism for aSyn regulation in the context of Parkinson's disease-associated genetic and environmental variations.

Figure 1. Altered aSyn transcript co-expression correlation networks in PD brain tissue

a–c, aSyn transcripts are globally rewired in PD brain tissue. a, Plot of the normalized DW score (y-axis) and DE (x-axis, plotted in log2) between PD and unaffected control brain tissue cohorts. Each circle represents an Affymetrix probeset. An aSyn probeset specific for longer 3′UTR isoforms (aSynL, 204467_s_at; highlighted in red) is most differentially co-expressed out of 22283 total in meta-analysis across all datasets. b, Schematic of aSynL network rewiring in PD. aSyn transcripts recognized by the aSynL probeset (aSynL) are presented in green. c, Transcript co-expression correlation tables for unaffected control (left) and PD brain tissue (right) cohorts. An aSynL probeset specific for longer isoforms (3′UTR1/204467_s_at; Fig. S1) is highlighted in green. A second aSyn probeset that targets the aSyn coding sequences regardless of 3′UTR structure (CDS1/211546_x_at) is adjacent. High correlations (r=1) are denoted in red, high anti-correlation (r=−1) in blue and weak correlation in white (r=0). n=10 for unaffected, n=15 for PD. d–f, A loss of correlation in expression levels of aSynL with other aSyn transcript isoforms is specifically associated with PD. Correlation tables of aSyn isoform expression are presented in laser-microdissected SN dopamine neurons from PD patient tissue or unaffected controls (d; n=10 and 18 per group) or in striatum samples from PD (e, upper panel, n=19 and 16 per group) or HD patients (e, lower panel, n=32 and 38 per group) along with corresponding unaffected controls. High correlation (r=1) is depicted in red, weak correlation (r=0) in yellow. In PD samples, longer aSynL transcripts (as detected by probeset 3′UTR1) are relatively unwired from shorter transcripts (as detected by CDS1, CDS2 [207827_x_at], or 3′UTR2 [204466_s_at]; the latter probeset recognizes the most proximal region of the aSyn 3′UTR; see Fig. S1). f, aSyn transcript co-expression was quantified in cortical tissue from 183 unaffected control individuals grouped according to their genotype for the PD-associated SNP (rs356168), using Illumina aSyn probes CDS3 and 3′UTR3 (Fig. S1b). Individuals homozygous for the PD-risk allele (“CC”, left) demonstrated significantly decreased aSyn transcript co-expression correlation relatively to those that harbor 1 (“CT”, middle) or 2 (“TT”, right) protective alleles.

Figure 2. Characterization of aSyn mRNA 3′UTR isoforms in unaffected and PD brain tissue

a, Relative abundance of the different aSyn 3′UTR species, determined by pA-RNAseq in 17 cortical brain samples from unaffected individuals. The frequency of 3′UTR species is expressed as the percentage of total aSyn transcript, with short species (290, 480 or 560 nt) in shades of green; medium (1070 nt) in orange and long (2520 nt) in red. b, Northern blot analysis of RNA from human brain or SH-SY5Y cells, as indicated. Blots were hybridized with probes targeting the aSyn CDS (Left panel) or the 3′UTR (Right panel). Nucleotide length is presented on the right; the corresponding 3′UTR size (color coded as per a) is indicated on left. c, Ratio of long 3′UTR aSyn mRNA to short 3′UTR aSyn mRNA species counts, evaluated by pA-RNAseq of cortical samples from unaffected individuals (n=17, black diamonds) and from PD patients (n=17, red triangles). Ratio corresponds to the long 3′UTR species (1070nt and 2520nt) read count divided by the shorter 3′UTR species (290, 480 or 560 nt) read count. Horizontal bars represent the means. *: p < 0.05, two-tailed t-test. d, Ratio of aSynL:total transcripts, as quantified by RT-qPCR in cortical samples from PD (n=18), ALS (n=16) or unaffected individuals (n=8). The mean transcript ratio of the control group was arbitrarily set to 1 as a reference. *: p < 0.05, ANOVA followed by Bonferroni correction. e, aSynL:total transcript ratios in cortical tissue samples from unaffected non-PD individuals (n=365) are presented as a function of rs356168 PD-associated risk allele load. Individuals harbor either 0 PD-risk alleles (“CC”, left), 1 PD-risk allele (“CT”, middle) or 2 PD risk alleles (“TT”, right). Association between the allelic load of the T causative variant and the aSynL:total ratio was evaluated by linear regression. f, Manhattan plot representing the association of 380,157 SNPs with the aSynL:total ratio in the non-PD brain tissue cohort as in (e). X-axis represents chromosomal location, Y-axis represents −log10 of the unadjusted p-value of association for each SNP with elevated aSyn transcript ratio. The aSyn 3′ locus SNP rs356168 (arrow) exhibits the highest association. Error bars represent the standard error of the mean (SEM).

Figure 3. Regulation of aSyn transcript isoform ratio in vitro and in vivo

a, aSynL:total ratio RT-qPCR quantification in rat primary cortical neuron cultures exposed to extracellular dopamine as indicated. N=8 per group. b, l-Dopa treatment (20mg/kg/day, 5 days) of 2 month-old control (DAT-Cre/Dicer^flox/+) mice but not Dicer–deficient mice (DAT-Cre/Dicer^flox/flox, having lost >95% of mDNs ) led to an increased aSynL:total ratio in midbrain. N=6 per group. c, In situ hybridization (ISH) with probes for human aSyn CDS (red) or specific for the aSyn long 3′UTR (blue) of primary cortical neuron cultures from aSyn-PAC transgenic mice. Cells were treated with dopamine (100μM), picrotoxin (100μM), or vehicle for 24h and co-stained with antibodies to aSyn (green). Scale bar, 10μm. d, aSynL:total ratio was quantified in terms of particle count per neuron using ISH probes as per (c). n>10 neurons/group from 3 independent wells e, SH-SY5Y cells were cultured for 8 h in the presence of EU (to label newly transcribed RNA; ‘pulse’) and subsequently maintained without EU for the indicated period of time (‘chase’). aSynL:total ratio was then evaluated by RT-qPCR in EU-labeled nascent RNA, as well as in total RNA. Pulse-chase analyses were conducted in the absence of dopamine (‘vehicle’; blue), in the presence of 100 μM dopamine during the EU labeling period only (‘dopamine pulse’; red), or during the chase exclusively (‘dopamine chase’; yellow). n=5 per group f, Total endogenous aSyn protein levels in SH-SY5Y cells treated with dopamine (100μM), picrotoxin (100μM) or vehicle for 48h. n=5/group. g, l-Dopa treatment (20mg/kg IP daily for 5 days) of 2 month-old aSyn PAC transgenic mice led to significantly increased aSyn protein in midbrain tissue. n>5 mice/group. h, Human SH-SY5Y cells were transfected with a firefly luciferase reporter vector harboring a 1.1 kb human aSyn 3′UTR element (along with a Renilla luciferase control vector), or with this luciferase vector modified to encode the rs356165 (C>T) or the rs78991202 (T>G) minor alleles. Dopamine (100μM) or picrotoxin (100μM) were added to the culture medium for 24 hrs before luciferase activity quantification. n=6 for each group. For all graphs, error bars are SEM; *:p<0.05,**:p<0.01, ***: p<0.001, as determined by ANOVA followed by Bonferroni correction.

Figure 4. Longer aSyn transcript 3′UTRs promote protein accumulation and mitochondrial localization

a, Predicted local secondary structure of aSyn 3′UTR RNA near the rs356165 and rs78991202 SNPs using RNA fold . A putative miR-34-3p binding site lies in this region (as determined by Targetscan analysis). Insert panel shows the predicted global structure of the aSyn 3′UTR, with black box denoting the area of interest. b–c, Left panels: Luciferase activity in HEK293 cells, quantified 24 h after transfection with the luciferase-aSyn 3′UTR reporter vector, along with a miR-34b-mimic (b) or a miR-34b-inhibitor (c). n=6 for each group. Right panels: Total endogenous aSyn protein levels (normalized to total aSyn mRNA levels as measured by RT-qPCR) in SH-SY5Y cells transfected with a miR-34b-mimic (b) or with a miR-34b-inhibitor (c). n=5 for each group. d, In SH-SY5Y cells exposed to dopamine (100 μM) or picrotoxin (100 μM) for 48 h, aSyn protein content is preferentially increased inmitochondrial preparations relative to whole cell aSyn content (upper panel) and decreased in membrane preparations (lower panel, n=5 for each group) e–f, Rat primary cortical neurons cultures at 3 DIV were transfected with a vector encoding a GFP-aSyn fusion protein (green) with either a short (0.3Kb) or a long (1.1kb) aSyn 3′UTR and stained with Mitotracker (e, in red) followed by confocal microscopy. Increased colocalization was observed in the context of the longer 3′UTR, both within the axonal growth cone terminal fields (L1 or S1 arrows, magnified in upper inserts) as well as in axonal processes (arrows L2 or S2, magnified in lower inserts). Scale bar, 10μm in main panel, 5μm in insets. (f) Collocalization quantification of aSyn-GFP and Mitotracker as in (d–e; n=12 randomly chosen fields per well, with >3 wells per group. g, aSyn protein levels in mitochondrial protein fractions isolated from 19 human cortical brain samples of PD-free individuals, grouped by their rs356165 genotype (AA, n=3; AG,n=12; GG, n=4). For all graphs, error bars represent the SEM. For 3b and 3d *:p<0.05,**:p<0.01, ***: p < 0.001, as quantified by ANOVA followed by Bonferroni post hoc correction. for 3c and 3f, *:p<0.05, by two-tailed t-test. For 3g, *, p<0.05, as evaluated by linear regression against rs356168 allelic load.

Figure 5. Regulation and consequences of aSyn 3′UTR alternative usage

Nascent aSyn RNA transcripts are co-transcriptionally processed to generate mRNA species with alternative 3′UTR elements. Increased cytoplasmic dopamine promotes the generation of an mRNA isoform that harbors a longer 3′UTR, aSynL. The presence of a longer 3′UTR, particularly in the context of disease-associated SNP variants at the aSyn 3′UTR locus, enhances the accumulation of aSyn protein, as well as the preferential localization of aSyn protein to mitochondria.