Characterization of the mouse model with h-α-Syn overexpression in raphe 5-HT neurons
We first examined the h-a-Syn transgene expression by injecting AAV5 construct with chicken-β-actin (CBA)-promoter that encodes h-α-Syn (referred to as AAV5) into mouse raphe nuclei. The AAV5 construct was validated in previous studies [44,46]. To investigate the time-course of h-α-Syn expression, mice were sacrificed at 1, 4, and 8-weeks post-injection. We detected a progressive increase of h-α-Syn mRNA expression in the raphe nuclei compared to control group (p<0.0001), without changes of murine α-Syn mRNA levels (Fig. 1a,b). The maximal increase of h-α-Syn mRNA expression was found at 8-weeks post-injection and reached 290% compared to murine α-Syn mRNA levels (Fig. 1b), reproducing the levels of α-Syn accumulation in patients with duplication or triplication of the gene [13,14].
A more exhaustive histological analysis was performed in order to evaluate whether the h-α-Syn mRNA co-localized with tryptophan hydroxylase (TPH)-positive cells, specific 5-HT neuronal marker. We found progressively significant increases in the number of TPH-positive cells expressing h-α-Syn mRNA, as well as increases in the intracellular h-α-Syn density in both dorsal raphe (DR) and median raphe (MnR) nuclei (Fig. 1c,d). At 8-weeks post-infusion, ~84% and 63% of TPH-positive cells expressed the transgene in DR and MnR, respectively, versus ~60% and 30% detected at 4-weeks post-infusion (p<0.0001). In addition, a variable number of non-TPH-positive cells targeted by the vector were also observed in the raphe nuclei as well as in adjacent brain regions, such as ventrolateral, lateral and dorsolateral periaqueductual gray (Supplementary Fig. 1). Although we did not identify the non-TPH-positive cells, most of which maybe GABAergic, our data indicate that the brain volume transduced by the AAV5 serotype is greater compared to other serotypes, such as 9 or 10 [47], which leaves different populations of cells inside and outside the raphe nuclei expressing the transgene. Next, we did not detect any significant loss of TPH-positive neurons distributed throughout the complete rostro-caudal extent of the raphe nuclei for at least 8-weeks post-infusion (Fig. 1e,f).
In parallel, we found that AAV5 infusion induced a marked time-dependent increase of h-α-Syn protein density in the raphe nuclei compared to control group, as assessed by immunohistochemistry (p<0.0001; Fig. 2a,b). The maximum level of h-α-Syn protein reached in raphe nuclei was found at 8-weeks post-injection (~197% compared to vehicle-infused mice), overlapping with the maximum number of TPH-positive neurons expressing h-α-Syn transgene (Fig. 1c,d).
Phosphorylation of α-Syn at amino-acid serine-129 (p-α-Syn) is a post-transcriptional modification found in ~90% of α-Syn inclusions in human PD brain tissue and animal models overexpressing α-Syn, frequently used as an indicator for α-Syn aggregation [44,48,49]. We, therefore, measured p-α-Syn signal intensity in the raphe nuclei. Our data revealed that AAV5-induced h-α-Syn overexpression leads to strong and progressive phosphorylation of α-Syn in raphe nuclei (Fig. 2c,d). The accumulation of p-α-Syn started early with positive cells being observable already 1-week post-injection, reaching the highest levels at 8-weeks, in close parallelism with h-α-Syn protein accumulation (p<0.0001). Similarly, using ELISA to identify α-Syn aggregation [50,51], we detected significant increases in filamentous α-Syn levels in raphe nuclei 4-weeks later compared to vehicle-infused mice (p<0.0001; Fig. 2e). We confirmed the α-Syn aggregates in raphe nuclei using the h-α-Syn proximity ligation assay (PLA) which allows detecting proteins in closely interaction [52]. While raphe tissue sections from control mice showed no signal, an abundant h-α-Syn puncta signal was found in AAV5-injected mice beginning 4-weeks post-infusion (Fig. 2f).
Extensive accumulation of h-α-Syn to efferent brain regions
Next, we addressed the question of whether raphe α-Syn pathology could spread to anatomically connected brain regions [53-57]. To investigate the accumulation of h-α-Syn to efferent brain areas, we stained brain sections from vehicle- and AAV5-injected mice, including medial prefrontal cortex (mPFC), cingulate cortex (Cg), caudate-putamen (CPu), and hippocampus (HPC), using antibodies against h-α-Syn and serotonin transporter (SERT). We examined the co-localization of h-α-Syn-positive and SERT-positive fibers at three time points, 1, 4 and 8-weeks later (Fig. 3a,b and Supplementary Fig. 2). Following injection of AAV5 into raphe nuclei, we found an abundant and progressive presence of h-α-Syn-positive fibers from 1-week post-injection (p<0.0001), and subsequently a significant loss of SERT-positive axons was observed in various brain regions innervated by 5-HT neurons 8-weeks later (p<0.05; Supplementary Fig. 2). As previously reported [44], the h-α-Syn signal was solely axonal and no h-α-Syn cell bodies were detected. The highest density of fibers co-localizing h-α-Syn and SERT was seen at 8-weeks post-injection, reaching values of 35.2±2.4%, 31.3±5.7%, 22.9±4.5%, and 19.8±1.7% in CPu > Cg > mPFC > HPC, indicating that the h-α-Syn was anterogradely transported along the axons towards the synaptic terminals. Moreover, at 4 and 8-weeks post-injection, h-α-Syn-positive and SERT-positive fibers had developed a striking distorted appearance with swellings (Fig. 3a,b). Confocal analysis showed an increased accumulation of h-α-Syn protein density in the axonal swellings in all brain areas examined including mPFC, Cg, CPu, and HPC (p<0.001). Notably, the axonal changes were accompanied by the presence of PLA-identified h-α-Syn aggregates, suggesting accumulation/aggregation of h-α-Syn as a result of impaired axonal transport (Supplementary Fig. 3) [58].
To determine whether the h-α-Syn accumulation in innervation networks can alter the density of synaptic proteins, such as the synaptic vesicle (SV)-associated proteins, which are ubiquitously expressed in the brain and modulate vesicular function [59], we performed an exploratory analysis of the SV2A protein levels in Cg and CPu from mice overexpressing h-α-Syn in raphe 5-HT neurons. Immunofluorescence analysis showed a progressive increase in punctate SV2A staining in these regions compared to vehicle-treated mice (p<0.01; Fig. 3c,e). Moreover, we identified that this punctate SV2A staining co-localized with h-α-Syn protein in the axonal swellings throughout the Cg and CPu (Fig. 3d), which could lead to altered presynaptic 5-HT function.
Overexpression of h-α-Syn in 5-HT neurons triggers deficiencies in forebrain 5-HT neurotransmission and BDNF expression and elicits a depressive phenotype
We next performed microdialysis experiments in CPu and mPFC of freely moving mice at 4-weeks post-injection in order to examine whether the axonal h-α-Syn accumulation affects forebrain 5-HT neurotransmission. No differences in baseline 5-HT concentration was found in both CPu and mPFC between the different groups (Table 1). However, the infusion of the depolarizing agent veratridine (50 μM) by reverse dialysis significantly increased the extracellular 5-HT levels in CPu (p<0.0001) and mPFC (p<0.01) of control mice, but not in mice overexpressing h-α-Syn (Fig. 4a,b), suggesting a marked deficiency in 5-HT reserve pools of AAV5-injected mice.
We previously showed that changes in endogenous α-Syn levels modify SERT function in cortical and striatal 5-HT terminals of wild-type mice [9]. Local infusion of SERT inhibitor, citalopram (1-10-50 μM), dose-dependently increased the extracellular 5-HT concentration in CPu and mPFC, but, this effect was more pronounced in vehicle- than in AAV5-injected mice (p<0.001; Fig. 4c,d). Furthermore, the systemic administration of 5-HT1A receptor agonist 8-OH-DPAT (1 mg/kg, intraperitoneal) comparably reduced the 5-HT release in mPFC of both phenotypes, but not in CPu. In this region, 8-OH-DPAT effect on 5-HT release was lower in mice injected with AAV5 than with vehicle (p<0.01), suggesting that the inhibitory feedback mechanism mediated by 5-HT1A receptor activation is attenuated by the h-α-Syn overexpression (Fig. 4e,f).
Functional deficits in monoamine neurotransmission, particularly 5-HT, have been extensively associated with anxiety and depression [60,61], main neuropsychiatric symptoms with a high prevalence in PD patients [29-31]. Previous data showing altered 5-HT neurotransmission provided justification for further analysis, so we conducted several behavioural paradigms routinely used to assess anxiety- and depressive-like behaviours in rodents. AAV5-injected mice showed an increased immobility time in the tail suspension and forced swimming tests compared to vehicle-injected mice (vehicle: 193.5±4.9 s, AAV5: 226.9±7.0 s, p<0.0004 and vehicle: 100.0±5.1 s, AAV5: 160.3±6.9 s, p<0.0004, respectively; Fig, 4g,h). Moreover, AAV5-injected mice showed an increased latency in the novelty suppressed feeding paradigm (p<0.01; Fig. 4i). None of these behavioural changes were driven by changes in the locomotor activity as assessed by the open field test, and comparable anxiety-like behaviour was observed in the dark-light box between both groups (Supplementary Fig. 4).
We next examined if the depressive-like phenotype is related to changes in brain-derived neurotrophic factor (BDNF) expression, since several studies support that BDNF dysfunction leads to depression and decreased BDNF levels were observed in the blood and post-mortem brain samples in patients with depression suffering from PD [62-64]. In intra-raphe AAV5 mouse model, we found significant decreases of BDNF mRNA in different HPC regions (p<0.01; Fig. 4j,k), suggesting that α-Syn-induced axonal pathology may be partially caused by defects in synaptic plasticity associated with the insufficient neuronal supply of BDNF, among other neurotrophic factors.
IND-ASO therapy reduces the h-α-Syn accumulation and alleviates the depressive-like phenotype
Recently, we designed indatraline-conjugated ASO sequences (1233- and 1337-IND-ASO) that were successfully delivered to monoaminergic neurons in vivo [44,45]. Both IND-ASOs selectively reduced the accumulation of α-Syn mRNA and protein in aged monkeys and in mouse models overexpressing wild-type or mutant human α-Syn forms in tyrosine-hydroxylase-positive neurons of substantia nigra compacta and locus coeruleus [44,45]. Therefore, we extended these previous observations and examined the IND-ASO effects (1337 sequence) on the raphe α-synucleinopathy model. For this purpose, AAV5-injected mice into raphe nuclei were treated with vehicle or IND-ASO (100 μg/day, intracerebroventricular) for 4-weeks and euthanized at 7 days after completion of treatment (Fig. 5a). Mice injected with AAV5 and treated with IND-ASO showed a ~35% reduction in h-α-Syn mRNA expression in raphe 5-HT neurons compared to AAV5-injected mice and treated with vehicle (p<0.001; Fig. 5b,c). Remarkably, IND-ASO treatment did not alter the murine α-Syn mRNA expression or the number of TPH-positive cells (AAV5/IND-ASO: 2158±209; AAV5/Vehicle: 2426±198), which supports the specificity and safety of the employed ASO sequence.
Furthermore, the mice treated with IND-ASO also showed reduced levels of h-α-Syn (~31%) and p-α-Syn (~65%) proteins in the raphe nuclei compared to vehicle-treated mice assessed by immunohistochemistry (p<0.001; Fig. 5b-e). Notably, we detected a reduced h-α-Syn density across different brain regions analysed (mPFC, Cg, CPu, and HPC) after IND-ASO treatment (p<0.05; Fig. 5f,g and Supplementary Fig. 5a). Likewise, a slight reduction was observed in the accumulation of SV2A protein in Cg and CPu of mice injected with AAV5 and treated with IND-ASO compared to AAV5-injected mice treated with vehicle (p<0.05; Fig. 5h,i).
Finally, we found that AAV5-injected/IND-ASO-treated mice showed improved 5-HT neurotransmission in CPu compared to AAV5-injected/vehicle-treated mice, reaching extracellular 5-HT levels comparable to control mice assessed by microdialysis procedure (p<0.001; Fig. 4a,c and 5j). In parallel, IND-ASO-treated mice showed increased BDNF mRNA expression in the hippocampus compared to vehicle-treated mice (Supplementary Fig. 5b) and also exhibited better performance in the tail suspension and forced swimming tests with a reduced immobility time (p<0.001; Fig 5k).
The present data indicate that the α-synucleinopathy in the raphe 5-HT neurons evoked a widespread alteration of 5-HT function and of BDNF expression in HPC. Both effects are likely associated to the depressive-like phenotype, given that the prevention of h-α-Syn overexpression by IND-ASO normalized 5-HT deficits and mouse behaviour. Overall, these observations support that α-Syn accumulation in raphe 5-HT neurons partly underlies early non-motor, neuropsychiatric symptoms in PD.