1. haSyn1-120 mice show anxiety and depression before motor impairment.
The study of anxiety and depression in PD animal models has been hindered by the challenge of identifying and replicating premotor disease stages (Schapira et al., 2017). To address this issue, we utilized haSyn1-120 mice, which exhibit progressive anatomical degeneration prior to the onset of motor impairment, offering a broad time window for investigation (Tofaris et al., 2006; García-Reitböck et al., 2010). To identify the neuronal circuits involved in the anxiety and depression signs comorbid with PD, we evaluated these mental signs, along with motor abilities, in haSyn1-120 mice at 3, 8, and 18 months.
We assessed the presence of anxiety-like traits with the open field arena (OF), the elevated plus maze (EPM), and the novelty-suppressed feeding (NSFT) behavioral tests. In the OF test, anxiety signs became evident at 8 months and worsened at 18 months, evidenced by reduced time in the center (63.9% ± 8.2 at 8 months and 78.2% ± 8.2 at 18 months; p < 0.0001 vs. 3-month-old transgenic mice and control mice) and increased time in the corners (40.6% ± 9.0 at 8 months and 58.2% ± 9.0 at 18 months; p < 0.0001 vs. 3-month-old transgenic mice and control mice; 112.5% ± 8.2 at 18 months; p < 0.01 vs 8-month-old transgenic mice) (Fig. 1A). Control mice at 3, 8 and 18 months of age did not show any significant differences and were grouped together for simplicity. In the EPM, anxiety-like behavior was present at 8 and 18 months, indicated by reduced time (62.8% ± 7.7 at 8 months and 56.0% ± 7.7 at 18 months; p < 0.0001) and the number of entries (27.9 ± 1.5 at 8 months and 57.9 ± 1.5 at 18 months; p < 0.0001) in the open arms vs. 3-month transgenic mice and control mice (Fig. 1B). Consistent with this, the latency to first bite in the NSFT was significantly increased (65.4% ± 21.4 at 8 months and 121.4% ± 21.4 at 18 months; p < 0.001 vs. 3-month-old mice and control mice and 133.8% ± 23.9 at 18 months; p < 0.01 vs 8-month-old mice) (Fig. 1C). In line with these results, depressive signs were present at 8 months and worsened at 18 months of age when mice underwent the social interaction (SI), sucrose preference and tail suspension tests. Compared to control mice, there was a significant reduction in time spent in the social chamber (27.8% ± 4.04 at 8 months and 26.3 ± 3.3 at 18 months; p < 0.0001) (Fig. 1D) and a significant reduction of sucrose at 8 months (22.8% ± 4.6; p < 0.01) vs 3-month-old transgenic mice and control mice), a reduction that nearly doubled at 18 months (40.% ± 4.6; p < 0.0001 vs 3-month-old mice and control mice and 77% ± 4.8; p < 0.05 vs 8-month-old mice). There were no changes in water intake, thus ruling out drinking impairments (Fig. 1E). In addition, there was a progressive increase in immobility time (71% ± 7.4 at 8 months and 110.6% ± 7.4 at 18 months; p < 0.0001 vs 3-month-old transgenic mice and control mice and 123.1% ± 7.4 at 18 months; p < 0.01 vs 8-month-old mice) during tail suspension (Fig. 1F).
The appearance of motor impairments marks the transition from premotor to motor PD stages (Schapira et al., 2017). To determine the time window of the premotor period in the haSyn1-120 model, we studied the progression of motor impairment. In haSyn1-120 mice, fine motor deficits were not evident until 18 months of age, as measured by a decrease in distance covered (46.4% ± 4.1; p < 0.001 vs 3-month-old transgenic mice or control mice and 42.8%± 4.3; p < 0.001 vs 8-month-old mice) in the OF (Fig. 1G) and increased number of falls (3 ± 0.6; p < 0.0001 vs 8-month-old mice) in the rotarod (Fig. 1H). Indicating the presence of hypokinesia and motor coordination impairment at the 18th month, as previously described (Tofaris et al., 2006). Taken together, these results show that in haSyn1-120 mice, the onset of anxiety/depression signs precedes motor impairment, reproducing the sequential onset of the pre-motor and motor symptoms in PD.
2. Age-dependent accumulation of aSyn pathology in DRNDA neurons of the haSyn1-120 mice.
To explore the possible mechanisms of the onset of anxiety/depression signs we focus on the pathological changes in DRNDA and DRN5HT neurons. DRNDA and DRN5HT neurons are two separate populations (Dougalis et al., 2012; Huang et al., 2019) that express the biomarkers TH or TPH2 (Fig. S1A), respectively. Anatomically, these two types of neurons are intermingled at mid-rostrocaudal areas of the DRN, although DRNDA neurons appear more rostral than DRN5HT neurons, which are also present in more caudal areas (Fig. S1B). These DRNDA neurons co-express L-amino acid decarboxylase (AADC) and vesicular monoamine transporter type 2 (VMAT2) together with TH, indicating they can synthesize, store, and release dopamine. However, immunofluorescence in TH + cell somas or fibers indicates that dopamine transporter (DAT) expression levels are almost undetectable, suggesting that dopamine re-uptake capability is low (Fig. S2). The great majority of DRNDA neurons also express the calcium chelator protein calbindin (Fig. S2).
In post-mortem PD samples, aSyn deposits in the raphe are in dopaminergic neurons (Halliday et al., 1990). In haSyn1-120 mice, where haSyn expression is under the control of the TH promoter, progressive haSyn deposits occur exclusively in DRNDA (TH+/aSyn+) cells, not in DRN5HT (TPH2+/aSyn-) neurons (Fig. S3), indicated by a significant time-dependent increase of aSyn + cells at 18 months (113.2% ± 8.6; p < 0.01 vs 3-month-old mice and 115.1% ± 8.9; p < 0.01 vs 8-month-old mice) (Fig. 2A).
Although stereological counts revealed no changes in the number of DRNDA (TH+) neurons (Fig. 2B), we observed a time-dependent shrinkage of their soma size that becomes statistically different at 8 months (19% ± 6; p < 0.0001 vs. control), a time point that overlaps with anxiety and depression signs. This shrinkage is even greater at 18 months (30% ± 5; p < 0.0001 vs. control and 13.7% ± 6.3; p < 0.01 vs. 8-month-old transgenic mice) when anxiety and depression worsen (Fig. 2B). Consistent with neuronal atrophy, pathological changes in DRNDA (TH+) neurons of 18-month-old haSyn1-120 mice include cytoplasm reduction and vacuolization, as well as dystrophic and swollen processes (Fig. S4). This phenotype is similar to what has been described in these mice for the substantia nigra pars compacta (SNc) (Tofaris et al., 2006). These results strongly demonstrate that aSyn pathology triggers neurodegenerative processes in DRNDA neurons. In contrast, we observed no changes in the number or soma size of DRN5HT neurons (Fig. 2B), further confirming that in this mouse line, this neuronal population is spared from aSyn pathology.
Since synaptic terminal loss is one of the main features of neuronal dysfunction and degeneration (Goedert et al., 2013; Brooks et al., 2014), we quantified the density of dopaminergic terminals in subcortical DRN projection fields, such as the BNST and CeA (Hasue and Shammah-Lagnado, 2002; Dougalis et al., 2012; Zhou et al., 2022; Poulin et al., 2018). In 8-month-old transgenic mice, there was a significant reduction in the proportional TH-stained area in the BNST (34.7% ± 5.2; p < 0.0001 vs control) and in the CeA (20.5% ± 5.2; p < 0.0001 vs control), which occurs in parallel with DRNDA atrophy and anxiety/depressive signs. Notably, this reduction in TH + fibers also coincides with progressive haSyn accumulation in the BNST and the CeA (Fig. 2C). In contrast, there were no changes in the density of serotonergic (SERT+) terminals in the BNST or in the CeA.
Finally, we tested DRNDA functional decline by measuring dopamine release in the BNST and CeA in behaving animals. In control and symptomatic haSyn1-120 mice (8 months old), we transduced the viral vector AAV9-hSyn-dLight 1.1-eGFP, encoding the dopamine sensor d-Light, in the BNST and in the contralateral CeA (Fig. 3A-B). Using fiber photometry during a sucrose-preference behavioral test, we detected a fluorescence transient in the BNST at the time of licking; this signal was significantly lower in haSyn1-120 mice than in control animals (Fig. 3C: 0.006 ± 0.05 average z-scored ΔF/F vs 0.46 ± 0.11 control average z-scored ΔF/F; p < 0.01 within the time window of half second, corrected by Bonferroni (see Methods)). The CeA showed similar results during the uplifting test (Fig. 3D) (0.42 ± 0.13 average z-scored ΔF/F vs 1.17 ± 0.16 control average z-scored ΔF/F; p < 0.01).
Collectively, these results suggest that pathological aSyn accumulation specifically and progressively disrupts DRNDA neuronal integrity (evidenced by a reduction in the cell soma size, synaptic terminal density, and dopamine neurotransmission in BNST and CeA) through processes that temporally overlap with anxiety and depression signs.
3. Selective lesions of DRNDA neurons accelerate anxiety and depressive signs in asymptomatic haSyn1-120 mice.
If DRNDA neuronal dysfunction is associated with anxiety/depression signs, lesioning this neuronal cell population in asymptomatic haSyn1-120 mice should accelerate the appearance of affective traits. For this purpose, we injected the 6-OHDA toxin in the DRN of 3-month-old haSyn1-120 mice (Fig. 4A), when anxiety/depression signs were not detectable, nor were alterations of DRNDA morphology or dopaminergic innervation in BNST/CeA.
As predicted, DRNDA 6-OHDA lesion accelerated the onset of anxiety signs (Fig. 4B), indicated by a significant increase in time spent in the corners of the OF (54% ± 22.0; p < 0.05 vs naïve) and by reduced time spent in the open arms during the EPM test (58.8% ± 14.4; p < 0.001 vs naïve). Upon lesion, time spent in the social chamber was reduced (43.9% ± 4.8; p < 0.0001 vs naïve) while immobility time during tail suspension was increased (188.9% ± 5.1; p < 0.0001 vs naïve) (Fig. 4C), indicative of depression-like behavior. We did not find any change in motor performance when tested on a rotarod or in the distance traveled in the open field, suggesting that DRNDA 6-OHDA lesions do not affect motor abilities (Fig. 4D).
We confirmed the efficiency of the 6-OHDA lesion in the DRN by stereological counts. TH + cells in the DRN were reduced (50.6% ± 7.1; p < 0.05 vs naïve) (Fig. 4E), accompanied by fewer dopaminergic (TH+) terminals in the BNST (29.1% ± 5.0; p < 0.01) and in the CeA (30.4% ± 4.9; p < 0.0001 vs naïve) (Fig. 4F). In contrast, the number of DRN5HT neurons (TPH2) and terminal density (SERT+) in BNST/CeA did not change (Fig. 4E-F). These results demonstrate that in asymptomatic haSyn1-120 mice, specifically damaging DRNDA neurons accelerates the onset of anxiety and depression phenotypes.
4. Selective chemogenetic activation of DRNDA neurons rescues anxiety and depressive states in symptomatic haSyn1-120 mice.
To conclusively establish a direct link between DRNDA functional decline and the appearance of anxiety/depression traits, we designed rescue experiments. We chemogenetically activated DRNDA neurons in symptomatic 8-month-old haSyn1-120 mice, which were coinjected within the DRN with the AAV9-Thp-Cre and AAV9-DIO-hM3DGq-mCherry viral constructs to achieve region and cell-type specificity (Fig. 5A). The efficiency of viral transduction was verified by double-staining for the markers TH and mCherry; 90.2% of TH + neurons in the DRN of injected mice were also mCherry+ (Fig. 5B and S5). One month after viral injection, in symptomatic hM3DGq/DRNDA-haSyn1-120 mice, systemic CNO administration alleviated anxiety signs. That is, compared with naive mice of the same age, the time spent in the center increased (95% ± 9.9; p < 0.01 vs 8 months) and corners of the OF area decreased (29.6% ± 10.5; p < 0.01 vs 8 months). Time spent (67.3% ± 8.3; p < 0.001 vs 8 months) and the number of entries (112.9% ± 3.8; p < 0.001 vs 8 months) in the open arms of the EPM also increased, reaching values similar to those of asymptomatic, 3-month-old haSyn1-120 mice (Fig. 5C).
In a similar pattern to the anxiety phenotype, in symptomatic haSyn1-120 mice, chemogenetic activation of DRNDA neurons ameliorated depression signs. Compared with symptomatic saline-haSyn1-120 mice, CNO significantly increased the time spent in the social chamber (122.2% ± 4.4; p < 0.0001 vs 8 months) and reduced the immobility time during the tail suspension test (47.5% ± 10.8; p < 0.01 vs 8 months) (Fig. 5D). CNO administration alone did not rescue anxiety/depression traits in naive (sham-operated) symptomatic 8-month-old haSyn1-120 mice (Fig. S6) nor did a saline injection in symptomatic hM3DGq/DRNDA -haSyn1-120 mice.
Taken together, these results robustly demonstrate that in symptomatic 8 months old haSyn1-120 mice, selective activation of DRNDA neurons is sufficient to rescue anxiety/depression signs. This further supports the key role of dysfunctional DRNDA neurons in the onset of emotional comorbidities in PD mice.
5. Dorsal raphe nucleus activity decays in symptomatic haSyn1-120 mice while its variability remains unaffected.
Although in haSyn1-120 mice, DRN5HT neurons are not directly affected by aSyn pathology (Fig. 2), we cannot rule out the possibility of impaired DRN functional integrity.
To assess DRN neuronal activity, we recorded and analyzed the firing rate (FR) and coefficient of variation (CV) of the FR (see Methods). We compared these measurements between 3- and 8-month-old haSyn1-120 mice and control mice. Our study included a total of 242 DRN neurons from 19 mice, including 6 control mice, 7 transgenic mice at 3 months, and 6 at 8 months of age. From these, we analyzed 109 neurons from control mice, 89 neurons from 3-month-old mice, and 44 neurons from 8-month-old haSyn1-120 mice. The recorded neurons exhibited four distinct firing patterns. Approximately 43% of the neurons displayed non-regular spiking behavior, as explained in the Methods section. The remaining 56% consisted of pacemaker-firing cells, which could be further categorized into bursty neurons (7%) and non-bursty neurons (the remaining percentage, as explained in the Methods section). All three groups of neurons showed changes in their mean firing rate during the progression of the haSyn1-120 mouse model. Specifically, the non-regular neurons with a distinct Poisson firing type exhibited a one-fold decrease in the FR, from 3.6 ± 0.6 Hz in control mice to 1.7 ± 0.2 Hz in 3-month-old haSyn1-120 mice. This reduction was maintained at 8 months. Based on the firing pattern, these neurons are likely dopaminergic cells in the DRN (Brown et al 2009, Aghajanian et al 1978). The pacemaker neurons displayed a progressive decrease in FR, which was evident in the 3-month-old mice and became significant at 8 months of age. Based on previous descriptions (Aghajanian et al. 1978, Hajós et al., 2007, Mlinar et al., 2016), it is possible that these pacemaker neurons correspond to the DRN serotonergic neurons. Additionally, we identified a small percentage (4%) of inhibitory neurons whose firing activity did not show significant changes. Overall, there were no statistically significant differences in FR variability over the recording session (represented by the CV) induced by the development of the aSyn pathology over time, in any of the cell populations. These results indicate that although DRN5HT neurons are not morphologically altered, their FR is reduced, possibly secondary to DRNDA neuronal dysfunction.
6. LCNA neuronal pathology is related to anxiety/depression in the haSyn1-120 mice.
The progressive accumulation of aSyn in haSyn1-120 mice could affect, in addition to DRNDA neurons, noradrenergic neurons of the locus coeruleus (LCNA), which similarly express the catecholaminergic synthesizing enzyme TH. The contribution of LCNA neuronal decline to anxiety/depression is not well understood, although noradrenaline dysfunction has been widely described in the early stages of PD when emotional alterations appear (Carey et al., 2021; Remy et al., 2005; Paredes-Rodríguez et al., 2020; Benarroch, 2018).
As expected, double immunofluorescence analysis in the LC of the haSyn1-120 mouse line showed that pathological haSyn is expressed in LCNA (DbH+/aSyn+) neurons (Fig. S8 and Fig. 7A). Stereological counts of TH + neurons revealed a progressive decrease in the number of LCNA neurons, which is significant from 8 months of age onwards (Fig. 7A), thus overlapping with the presence of anxiety/depression signs. Moreover, progressive shrinkage in the soma size (Fig. 7A) of LCNA neurons was already present in 3-month-old mice, suggesting that in the model LCNA pathology precedes DRNDA affectation (Fig. 2B) and highlighting the higher vulnerability of LCNA neurons to aSyn pathology, as described in post-morten brain samples of PD patients (Braak et al., 2003; Benarroch, 2018).
Noradrenaline modulates the activity of the BNST/CeA (Daniel and Rainnie, 2015) and DRN5HT neurons (Brown et al., 2002), exerting excitatory input through alpha-1 noradrenergic receptors (Khamma et al., 2022). Therefore, we studied if LCNA degeneration in the haSyn1-120 model reduces noradrenergic innervation to the BNST/CeA and the DRN, and whether a reduction is relevant for the manifestation of anxiety/depression-like behavior. In haSyn1-120 mice, the specific NA marker, DbH, was indeed reduced in the BNST, CeA, and DRN, a reduction that reached statistical significance at 8 months of age (Fig. 7B), thus temporally overlapping with LCNA neuronal loss and with the appearance of anxiety/depressive signs described in Fig. 1.
To test the specific contribution of LCNA pathology to anxiety/depression, we lesioned this neuronal population with bilateral injections of 6-OHDA in asymptomatic 3-month-old haSyn1-120 mice (Fig. 7C), when anxiety/depression signs and DRNDA pathology were not yet evident. Similar to lesioning DRNDA neurons, LC 6-OHDA lesions also accelerate the appearance of anxiety/depression signs (Fig. 7D-E): anxiety indicated by increased time spent in the corners in the OF and reduced time spent in the open arms during the EPM (Fig. 7D), and depression indicated by reduced time spent in the social chamber and increased immobility time during the tail suspension test (Fig. 7E). After 6-OHDA lesions, stereological counts confirmed a significant reduction of TH + cells in the LC and noradrenaline (DbH+) fibers in the DRN, CeA, and BNST (Fig. 7F). These results demonstrate that the loss of LCNA neurons as well as disruption of noradrenergic innervation in limbic areas accelerates anxiety and depression traits in asymptomatic haSyn1-120 mice.