Neurodegeneration in the centrally-projecting Edinger–Westphal nucleus contributes to the non-motor symptoms of Parkinson’s disease in the rat

Background The neuropathological background of major depression and anxiety as non-motor symptoms of Parkinson’s disease is much less understood than classical motor symptoms. Although, neurodegeneration of the Edinger–Westphal nucleus in human Parkinson’s disease is a known phenomenon, its possible significance in mood status has never been elucidated. In this work we aimed at investigating whether neuron loss and alpha-synuclein accumulation in the urocortin 1 containing (UCN1) cells of the centrally-projecting Edinger–Westphal (EWcp) nucleus is associated with anxiety and depression-like state in the rat. Methods Systemic chronic rotenone administration as well as targeted leptin–saporin-induced lesions of EWcp/UCN1 neurons were conducted. Rotarod, open field and sucrose preference tests were performed to assess motor performance and mood status. Multiple immunofluorescence combined with RNAscope were used to reveal the functional–morphological changes. Two-sample Student’s t test, Spearman’s rank correlation analysis and Mann–Whitney U tests were used for statistics. Results In the rotenone model, besides motor deficit, an anxious and depression-like phenotype was detected. Well-comparable neuron loss, cytoplasmic alpha-synuclein accumulation as well as astro- and microglial activation were observed both in the substantia nigra pars compacta and EWcp. Occasionally, UCN1-immunoreactive neuronal debris was observed in phagocytotic microglia. UCN1 peptide content of viable EWcp cells correlated with dopaminergic substantia nigra cell count. Importantly, other mood status-related dopaminergic (ventral tegmental area), serotonergic (dorsal and median raphe) and noradrenergic (locus ceruleus and A5 area) brainstem centers did not show remarkable morphological changes. Targeted partial selective EWcp/UCN1 neuron ablation induced similar mood status without motor symptoms. Conclusions Our findings collectively suggest that neurodegeneration of urocortinergic EWcp contributes to the mood-related non-motor symptoms in toxic models of Parkinson’s disease in the rat. Supplementary Information The online version contains supplementary material available at 10.1186/s12974-022-02399-w.


Background
Parkinson's disease (PD) is characterized by classic motor symptoms including rigor, tremor and bradykinesia attributed to neurodegeneration in the substantia nigra pars compacta (SNpc) and consequent loss of striatal dopaminergic afferentation [1]. According to the selfassessment of patients, non-motor symptoms deteriorate their quality-of-life more dramatically than motor symptoms themselves [2]. Therefore, the attention has been recently turning to PD-associated non-motor symptoms, such as anxiety and depression. While the main neuropathological hallmarks of motor symptoms such as Lewy body (LB) and Lewy neurite accumulation as well as nigral neurodegeneration are relatively well-investigated [1], much less knowledge has collected explaining the non-motor psychiatric symptoms. Anxiety disorders as well as major depression have been linked with altered noradrenergic, serotonergic and dopaminergic neurotransmission [3], but the underlying neuropathology of PD-associated mood disorders is poorly understood. Not surprisingly, the currently available pharmacotherapeutic strategies sharing the goal to increase serotonin (5-HT) and/or norepinephrine levels in the brain, provide limited efficacy [4].
Some neuropathological observations may explain the association between PD and mood disorders. For instance, LBs accumulate in the serotoninergic raphe nuclei [3], dopaminergic ventral tegmental area (VTA) as well as in the norepinephrinergic locus ceruleus (LC) [5]. Besides these, neurodegeneration and LB accumulation have been observed in PD in the dorsal vagal nuclear complex and in the Edinger-Westphal nucleus (EW) [6,7].
Classically, cholinergic parasympathetic preganglionic neurons of the EW project to the ciliary ganglion controlling the sphincter pupillae and ciliary muscles. Based on this, patients with PD would be expected to suffer from compromised pupillary light reflex and lens accommodation [8]. However, this is usually not the case, even though it is known that the "rostral part of the EW suffers 50% neural loss and 3% of the cells contain Lewy bodies" [6]. This paradox suggested us that the recently defined peptidergic centrally-projecting (EWcp) division [9] of the EW may be affected.
In this work our goal was to investigate the putative underlying mechanisms of mood disorders in PD. Our main hypothesis was that the specific loss or damage of EWcp/UCN1 neurons directly contributes to anxiety and depression as non-motor symptoms of PD.
To test this, a systemic chronic rotenone treatment model for PD was applied in the rat [26]. As EWcp/ UCN1 neurons carry leptin receptors [10], a leptin-conjugated saporin-induced selective ablation of UCN1 neurons was also performed to validate the depression-like phenotype in rats. Behavioral, functional-morphological, biochemical and histopathological tools were used to test the motor coordination, mood status as well as morphological changes in the brain.

Animals
In-house bred male Wistar rats (breeding pairs purchased from Animalab Kft., Vác, Hungary) were housed in three-per-cage (40 × 25 × 20 cm) groups in temperature and humidity-controlled environment. Animals were supplied with standard rodent chow and tap water ad libitum. Experiments were approved by the Animal Welfare Committee of Pécs University, the National Scientific Ethical Committee on Animal Experimentation in Hungary and the National Food Chain Safety Office in Hungary (license No: BA02/2000-49/2017).

Sucrose preference test
The anhedonia level was determined by sucrose preference test (SPT). Seventy-two hours before testing 1 m/v% sucrose solution and tap water were offered. After 24 h, the place of the bottles was reversed. On the test day, water deprivation was applied between 6 a.m. and 4 p.m. Then, the two bottles were offered for individually caged animals for 3 h. Sucrose preference was calculated according to the widely used formula [28].

Open field test
The open field test (OFT) was performed in a brightly lit box [50 × 50 × 38 cm] with black background. Video recordings of 5 min were analyzed with SMART Junior Tracking software (PanLab, Barcelona, Spain) to assess the time spent in the periphery of the box, and the traveled distance [29].

Rotarod test
Rotarod performance test (RPT) was used to assess motor coordination. Animals were first pre-trained to run on the accelerating rotating axis three times. For testing, rats were placed on the axis of the device (47750, Ugo Basile, Gemonio, Italy) and the time interval rats spent on the rotating axis was registered [30].

Targeted toxin-induced lesion of EWcp/UCN1 neurons
The EWcp contains neuron populations that do not express UCN1 [14,31]. Because in this project, our aim was to study the significance of UCN1 neuron loss, we applied a selective neuron ablation by saporin. This neurotoxin enters neurons only if it is conjugated to a substance that is internalized by receptor-mediated endocytosis and irreversibly inhibits the cells' protein synthesis [32]. As EWcp/UCN1 neurons carry leptin receptors [10], leptin-conjugated saporin injection provides a reliable tool to perform selective UCN1 neuron ablation [33].
Thirty-two rats were intraperitoneally anesthetized by ketamine (78 mg/kg, Richter, Budapest, Hungary) and xylazine (13 mg/kg, Eurovet, Nagyatád, Hungary). Using a stereotaxic apparatus, 0.08 μl leptin-conjugated saporin (n = 16) or saporin (n = 16) solution (#KIT-47, ATS INC, Carlsbad, CA, USA) was injected into the EWcp area. From a point 1.5 mm left and 4.8 mm caudal to the Bregma [34] an oblique injection path closing 19° angle with the vertical axis was set and the Hamilton needle was introduced to a depth of 5.8 mm [33,35]. Compounds were injected in 30 s. Thirty seconds later, the needle was retracted by 0.5 mm, and 1 min later, it was removed.
If the magnitude of UCN1 neuron loss was less than 20% upon leptin-saporin treatment, the animal was excluded (8 rats) from the experiment. Six rats were excluded, because the injection path deviated from the planned route, or the needle caused physical damage in the EWcp. Finally, nine saporin (control) and leptinsaporin-injected rats were included into the statistics. Behavioral testing was performed after 10 days recovery as described above. Behavioral data were converted into Z-scores using the formula Z = (X − µ)/σ, where (X) represents the raw data of the individual animal, (µ) and (σ) the mean and standard deviation of the control group, respectively [36]. Tissue sampling was performed 2 weeks after surgery.

Multiple immunofluorescence labeling
Antisera were characterized earlier as referenced in Table 1. Omission and non-immune serum replacement of primary or secondary antibodies prevented all immunolabelings (images not shown).

RNAscope in situ hybridization combined with immunofluorescence
In the rotenone model, four EWcp sections per animal were pretreated for RNAscope according to our recently developed protocol [37] optimized for 30 µm-thick PFA fixed sections. Subsequent steps of RNAscope protocol was performed according to the supplier's suggestions. Ucn1 mRNA was visualized by Cy5 (1:3000). A mouse Ucn1 mRNA probe (Cat No: 466261, Advanced Cell Diagnostics, Newark, CA, USA) was used as mouse and rat UCN1 show full peptide sequence identity [38]. Some randomly selected sections were hybridized with triplex positive control probes for the rat (320891) or with triplex negative control probes (320871). After channel development positive control gave well-recognizable signal, while no fluorescence was seen in negative controls (images not shown). Subsequently, slides were further processed for doubleimmunofluorescence using goat anti-UCN1 (1:175) and rabbit IBA1 (1:1000) for 48 h at 4 °C. After washes, Alexa 488-conjugated donkey anti-goat and Cy3-conjugated donkey anti-rabbit sera (1:500, 3 h) were used. Sections were counterstained with 4′,6-diamidino-2-phenylindole (DAPI) and covered with antifade medium.
Morphometry including counting of cells, nerve terminals and signal dots of the RNA labeling, densitometry and cell size measurement was performed on non-edited digital images. Every quantitation was carried out on four to six images per animal. The intensity of fluorescent signal was semi-quantified by measurement of the cytoplasmic signal corrected for the background, yielding the specific signal density (SSD) [18]. The measurement and unbiased stereological cell counting [39] was performed by ImageJ software (version 1.52a, NIH). Glial cell morphology was conducted according to the scoring system by Harrison et al. [40]. For publication purposes, representative images were cropped, contrasted and edited using Adobe Photoshop software.

Statistics
Data were presented as mean of the group ± standard error of the mean. Statistics were performed by two-sample Student's t test after assessment of data distribution. Outlier data beyond the 2-sigma range were excluded. Spearman's rank test was also applied to search for correlations. Mann-Whitney U test was used to assess results of glial activity scores. Alpha was set to 5% in all cases.
When testing our main hypothesis, we found that the number of UCN1 neurons decreased upon rotenone treatment (Fig. 3A-C, t(22) = 2.52; p = 0.019) and they showed pyknotic morphology (Fig. 3A, B, D, t(18) = -3.62; p = 0.0019). Occasionally, UCN1 neurons were found to disintegrate, where IBA1-ir microglial processes surrounded the UCN1-ir debris of the neurons (Fig. 3G, H' , Additional file 1: video 1A, Additional file 2: video 1B). Microglia interacted about 10-times more frequently with UCN1 neurons in rotenone-treated rats (Fig. 3I, t (18) = − 4.28; p = 0.0004) than in controls. Alpha-synuclein-ir cytoplasmic inclusions were observed in 11.94 ± 1.28% of the UCN1 neurons in rotenonetreated rats (Fig. 3B, E), unlike in controls (Fig. 3A, E; 0.62 ± 0.33%; t(20) = − 8.52; p < 10 -6 ). GFAP labeling in the EWcp (Fig. 3J, K) revealed that rotenone induced also astrocyte activation ( Fig. 3L; Mann-Whitney test; U(22) = − 14.50; p = 0.0009). To test whether astrogliosis was restricted to EWcp within the periaqueductal gray matter (PAG), astrocyte morphology was assessed in the ventrolateral PAG (Fig. 3J, K) and no remarkable astrogliosis was seen here (Fig. 3M, U (22 Rotenone treatment resulted in 126%-elevation of UCN1 peptide immunosignal (compare Fig. 3A with 3B or 3 N" with 3O") ( Fig. 3Q, t(22) = − 11.11; p < 10 -6 ). We successfully combined immunofluorescence with RNAscope in situ hybridization for Ucn1 mRNA. The EWcp perikarya contained high amount of cytoplasmic Ucn1 mRNA that was often inhomogeneously distributed in clusters of confluent signal dots, especially in rotenone-treated animals. This uneven distribution made the quantitation of cytoplasmic mRNA signal difficult. In contrast, well-countable Ucn1 mRNA signal puncta were dispersed in the karyoplasm of neurons (Fig. 3N' , O'). Quantitation of intranuclear Ucn1 mRNA transcripts revealed decreased Ucn1 mRNA expression by 57.8% in rotenone-treated rats ( Fig. 3P; t(18) = − 4.08; p = 0.0008). A strong negative correlation was found between cellular (See figure on next page.) Fig. 2 Neuroinflammatory markers support the validity of the rotenone model. Increased tumor necrosis factor alpha (TNF-alpha, green in A and B) specific signal density (SSD) in glial fibrillary acidic protein (GFAP; red in A and B)-expressing astrocytes was detected in the substantia nigra pars compacta (SNpc) as shown in graph C also. No TNF-alpha immunoreactivity was observed in ionized calcium binding adaptor molecule 1 (IBA1)-containing microglia (white in A and B). Inducible nitric oxide synthase (iNOS, red in D and E) immunoreactivity was detected only in very few IBA1 (white in A and B)-containing SNpc microglia in control rats (boxed area and red arrowhead in D). Rotenone treatment increased the number of iNOS immunoreactive (-ir) IBA-immunopositive microglia (inset in E) that occasionally formed active cell clusters (yellow arrowheads in E). Some faintly iNOS-ir nerve cell bodies (blue arrowheads in E and graph G) were also found and occasionally, TH-ir neurons also appeared to show weak iNOS positivity (see inset in E). IBA1 (green in H and I) and cluster of differentiation 68 (CD68, red in H and I) double-labeling revealed that upon rotenone treatment (I), the microglial cells in the centrally-projecting Edinger-Westphal nucleus (EWcp) co-express CD68 (J) suggesting their reactivity. Black bars: vehicle (oil) injected rats, red columns: rotenone-treated group. *p < 0.05, **p < 0.01, ***p < 0.001 in Student's t test, n = 6. Bars: 50 µm. Scatter plot K demonstrates the negative correlation between the TH immunoreactivity in the SNpc and EWcp microglial activity scores. Scatter plot L illustrates the inverse relationship between SNpc/TH immunoreactivity and astroglial activity scores

Selective EWcp/UCN1 neuron ablation replicates mood status-related symptoms observed in parkinsonian rats
EWcp/UCN1 neurons were selectively ablated by targeted leptin-saporin toxin injection to test whether the administration causes similar behavioral anomalies to those observed in the rotenone model.

Discussion
In the systemic toxic model of Parkinson's disease, rotenone was chosen to assess whether the PD-like state can be associated with morphological changes also within the peptidergic EWcp and whether it is accompanied with depression-like phenotype and anxiety. Next, a local neuron ablation was performed to examine whether the selective EWcp-damage causes comparable deterioration in the mood status.

Anxiety and depression as non-motor symptoms of PD are detected in the rotenone model
Although not without limitations [41,42], the rotenone model of PD reproduces many features of the human disease, including systemic mitochondrial impairment, selective dopaminergic nigrostriatal damage, microglia activation, alpha-synuclein accumulation with formation of LBs [43,44]. The RPT, a widely used method [45,46] unambiguously revealed severe motor coordination deficit in rotenone-treated rats. Reduced OFT locomotor activity further supported the deterioration of motor skills in full agreement with earlier studies [47,48]. OFT also suggested increased anxiety in rotenone-treated rats [49]. In SPT, reduced sucrose consumption in parkinsonian rats suggested diminished reward-seeking behavior [50]. Although these in vivo results confirmed the efficacy of the PD model, one has to consider the limitations of these tests as motor skills of rats might directly affect their performance. Nevertheless, after pre-testing other mood-status assessing paradigms (e.g., forced swim test, light-dark box test) in our pilot studies, we decided for OFT and SPT as parkinsonian animals were also capable of performing these, and the within and between subject error was acceptable. The number of SNpc/TH cells was reduced by 32.4% in rotenone-treated rats. Although this is considerably less than the 70% neuron loss in humans [1], rats with 20% loss of SNpc/TH neurons showed significantly impaired motor skills. Furthermore, in line with others, we found neurons with cytoplasmic alpha-synuclein inclusions in SNpc/TH [51] that correspond to LB-like structures. These two observations strongly suggest the validity of our model. Increased SNpc/astroglia TNF-alpha immunoreactivity and elevated microglial and neuronal iNOS signal in the SNpc strongly suggest that the rotenone treatment-related oxidative stress and consequent neuroinflammation contributed to the neuron loss [52,53]. Furthermore, in the rotenone model a diffuse striatal axonal loss and also drastic focal TH-ir fiber loss may be observed [54]. Indeed, we also measured reduced CPu/ TH-ir fiber content in rotenone-treated rats, and found some circumscribed striatal neuropil areas with almost undetectable TH-ir (Fig. 1S).

EWcp/UCN1 neurons suffer neurodegeneration and surviving cells show functional damage in the rotenone model
The magnitude of neuronal loss, the presence of cytoplasmic alpha-synuclein-ir inclusions, corresponding to LB-like structures and the increased microglial activity in the SNpc was in line and in statistical correlation with micro-and astroglial activity as well as with UCN1-neuron loss of the EWcp suggesting their similar susceptibility to the toxic agent. This is not surprising as EWcp neurons are developmentally also closely related with SNpc/TH neurons [55]. Although obvious cytoplasmic alpha-synuclein immunoreactive inclusions suggested the presence of LB-like structures in the SNpc and EWcp, we have to state an important limitation here is that our antibody, similarly to numerous other commercially available sera [56], was not specific for the high molecular weight isoforms found in LBs in the rat.
Beyond the obvious interaction between UCN1-ir cell fragments (Fig. 3H') and CD68-ir reactive microglia (Fig. 2I) [57] that unambiguously proves the loss of EWcp/UCN1 cells, we also provide evidence that surviving UCN1 cells suffered functional damage. We assume that the mitochondrial complex I inhibition in UCN1 cells caused energy deficit [58] with consequent UCN1 peptide accumulation due to impaired axonal transport. The increased SSD of UCN1 suggests peptide accumulation in the perikarya of cells, which was also observed in stressed mice [19,59]. The reduction of Ucn1 mRNA message is in line with findings in EWcp samples of chronically stressed rats [23] and depressed suicide victims [25]. Adrenal-, thymus-and bodyweight data suggest increased HPA axis activity that is commonly associated with depression-like states in animal models [19,60] and human depression [17]. Although PD-related neurodegeneration was shown to affect the DR [3], VTA and LC [5], we did not observe remarkable damage of these serotonergic, dopaminergic and noradrenergic cell populations in our rats. We cannot exclude that a higher rotenone dose [30] or a longer treatment period would have also affected these or other centers [53], that were not examined in this project but may affect mood status. Nevertheless, our rat model ultimately supports further the significance of the EWcp/UCN1 in PD-associated mood disorder.

Local targeted toxic UCN1-neuron ablation results in altered mood status
To further corroborate that EWcp/UCN1 neuron loss contributes to deterioration in the mood status, a local targeted UCN1 cell ablation was conducted. The leptinconjugated saporin is internalized into leptin receptor positive cells leading to disturbed protein synthesis [61] and apoptotic neuron loss [62]. As in this area the leptin receptor expression is restricted to a part of EWcp/ UCN1 cells [10,63], a selective partial UCN1 neuron death was expected. After pre-tests with multiple leptin-saporin doses we achieved a cell loss that was comparable with that observed in the rotenone model. Earlier works [33] and pre-tests suggested that the cell death requires 2 weeks in the leptin-saporin model. Nevertheless, the occasional caspase 3 immunoreactivity in UCN1 neurons suggested that in some cells the leptin-saporin-induced apoptotic process [64] was still in progress at the time of tissue sampling. Some activated micro-and astroglial cells were seen also in saporininjected controls most likely due to the direct physical effect of the surgery. In leptin-saporin-treated rats both glial cell types were more reactive and microglia tended to interact more frequently with the UCN1 neurons. Nevertheless, the selectivity of ablation was confirmed by the unchanged number of UCN1 immunonegative, but NeuN positive EWcp cells upon leptin-saporin treatment. In leptin-saporin-injected animals, increased OFT-anxiety and SPT-anhedonia suggested that the loss of EWcp/UCN1 cells provoked mood changes but normal motor coordination and locomotor activity excluded parkinsonism. In line with this, dopaminergic neuron loss, apoptotic activity or reactive gliosis was not detected in the SNpc.

Conclusion
In this study we successfully applied the rotenone model of PD and detected the anxious and depressionlike phenotype, considered as non-motor symptoms of PD in the rat. Selective local UCN1 neuron ablation evoked similar mood status without motor symptoms. With respect to the limitations, our findings collectively suggest that the impairment of the UCN1 neurons in the EWcp contribute to the non-motor symptoms of PD. Human studies have to determine the potential diagnostic and therapeutic significance of these observations. Selective partial UCN1 neuron ablation in the centrally-projecting Edinger-Wespthal nucleus (EWcp) causes increased anhedonia and anxiety level without affecting motor skills. Leptin-saporin-injected rats (red bars) show reduced preference to the sweetened water in the sucrose preference test (A). Open field test (OFT) revealed longer period of time spent in the peripheral part of the arena (B). Neither the total traveled distance in OFT (C) nor the rotarod test (D) revealed compromised motor performance of leptin-saporin-treated rats in comparison to saporin-injected controls (black bars). UCN1 (red)-NeuN (green) double labeling (E, F) revealed reduced UCN1/NeuN cell count upon leptin-saporin treatment (F, G), while the number of UCN1 immunonegative NeuN neurons remained unchanged in the EWcp (H). Ionized calcium binding adaptor molecule 1 (IBA1)-immunoreactive (ir) microglia (green in I and J) showed increased activity upon leptin-saporin treatment (J, K) and tended to interact with UCN1-ir cells more frequently (see the blue arrowheads in J and graph P). Astrocytes showing glial fibrillary acidic protein (GFAP, red in I and J) were significantly (L) more active upon leptin-saporin injection (J), although some active cells were also seen in saporin-treated controls (I) representing reactive gliosis due to the effect of surgical manipulation. UCN1-ir (green in M and N) neurons were seen to contain the apoptotic marker caspase 3 (see the boxed area and blue arrowhead in N) in leptin-saporin injected rats (O), while in saporin-injected control animals, only some glial nuclei were positive for caspase 3 (M). Saporin (Q) and leptin-saporin injection (R) into the EWcp did not affect the number of tyrosine-hydroxylase (TH, green in Q and R)-ir cells in the substantia nigra pars compacta (SNpc) as revealed by unbiased stereological cell counting (T). Caspase 3 (red in Q and R) revealed no remarkable apoptotic activity of SNpc/TH (S). Both microglia (IBA, green in U and V) and astroglial (GFAP) activity scores remained basal (W and X, respectively) in both groups (n = 6-9) t: tendency (p = 0.051), *p < 0.05, **p < 0.01, ***p < 0.001 in Student's t test or in Mann-Whitney U test in K, L, W, X. ns not significant. Bars: 50 µm