Introduction

Parkinson’s Disease (PD) is a frequent neurodegenerative disorder characterized by the progressive loss of dopaminergic neurons in the nigrostriatal pathway, leading to bradykinesia, tremor, rigidity, and postural instability [1, 2]. These cardinal motor symptoms are typically addressed by administration of dopaminergic drugs or by deep brain stimulation. PD patients also experience non-motor symptoms including sleep, affective, and cognitive dysfunctions often preceding the motor disabilities [3, 4]. These comorbidities are in large part refractory to current PD treatments and are thought to be caused by neurodegenerative processes occurring in concomitance to the loss of midbrain dopaminergic neurons. However, the pathology underlying non-motor symptoms remains poorly understood.

Post-mortem studies in PD patients provided first insights into the brain areas which might be involved in the etiology of non-motor dysfunctions in PD. Besides the profound degeneration of the substantia nigra pars compacta (SNc), these studies found cell loss and reduced neurotransmitter release in other monoaminergic brain regions, including the dorsal and median raphe nuclei (DRN and MRN, respectively), and the locus coeruleus (LC) [1, 58]. The DRN constitutes the main source of serotonin in the brain with serotonergic cells (DRN5-HT) accounting for 30-50% of its neurons [9]. DRN5-HT neurons have been implicated in numerous neuropsychiatric diseases, rendering them a potential neural substrate for non-motor symptoms in PD. In fact, several studies have shown that serotonergic markers and transmitter levels are altered in Parkinson patients as well as in non-human primate and rodent models of PD [1017]. Notably, alterations in the serotonergic system have also been related to non-motor comorbidities in PD [18, 19]. Yet, functional investigations of DRN5-HT in rodent models of PD have led to conflicting results showing both increased and decreased activity in DRN5-HT neurons themselves as well as in their downstream targets [2022]. Besides the serotonergic neurons, the DRN comprises other neuronal populations, including a group of dopamine neurons (DRNDA), in which Lewy bodies have been found in PD patients [6]. DRNDA neurons have been linked to the regulation of pain, motivational processes, incentive memory, wakefulness and sleep-wake transitions [2327], but their ultimate behavioral significance is yet to be elucidated [2832]. In addition, the physiology and pathophysiology of DRNDA neurons in PD remains elusive. The sparsity of research on DRNDA neurons is likely due to the technical challenges associated with targeting this population among the diverse cell types in the DRN and adjacent structures (e.g., retrorubral field, periaqueductal grey and LC), which often co-express signature genes, hampering their molecular identification and region-specific manipulations with cre driver lines [9, 3336].

Recently, this issue has been addressed by Pinto and colleagues who showed that DRNDA neurons are most faithfully labelled in transgenic mice in which the expression of cre is linked to the dopamine transporter (DAT-cre) [34]. Previously, the membrane properties of DRNDA neurons have only been addressed in mice in which DRNDA neurons were identified based on the expression of the transcription factor Pitx3 or the enzyme tyrosine hydroxylase (TH) [33]. In Pitx3-GFP mice, about 70% of GFP-positive neurons are TH- positive as shown by immunohistochemistry. Moreover, 40% of TH-positive neurons in the DRN are not labelled in these mice, suggesting that this line targets a subpopulation of DRNDA neurons [33]. The widely used TH-cre reporter line has been found to show ectopic expression of cre in non-dopaminergic neurons, probably caused by a transient developmental expression of TH [37, 38]. In addition, the TH-cre line also labels noradrenergic neurons in the neighboring LC [37], which produces most of the noradrenaline (NA) in the brain and is involved in mood control, cognition, and sleep regulation [39]. The large overlap of functions ascribed to the LC and DRN is thought to result from the complex reciprocal synaptic connections between these two brain areas: notably, the LC provides noradrenergic input to the DRN [40, 41] and also receives input from DRN5-HT neurons [4244].

Here, we used ex vivo whole-cell patch clamp recordings and morphological reconstructions to characterize the electrophysiological and morphological properties of DRNDA and DRN5-HT neurons in wild type and DAT- tdTomato mice. Moreover, we studied the impact of catecholamine depletion on DRNDA and DRN5-HT populations in the 6-OHDA toxin model of PD.

Results

DRNDA and DRN5-HT neurons are electrophysiologically distinct cell-types

To investigate the electrophysiological and morphological profile of DRNDA neurons and to compare it to DRN5-HT neurons, we performed whole-cell patch clamp recordings in coronal slices of adult wild type and DAT-cre mice crossed with tdTomato reporter mice (Figure 1A). All neurons were filled with neurobiotin and Alexa488 while recording. Alexa488 allowed us to take snapshots of recorded neurons at different time points, thus facilitating the topographical registration of recorded neurons to the post-hoc stained slices (Suppl. Figure 1). Using this approach, we obtained complete sets of electrophysiological and morphological data from 75 neurons in the DRN. Cells were identified as DRN5-HT or DRNDA neurons based on tryptophan hydroxylase (TPH) or TH immunoreactivity, respectively (Figure 1A-D). None of the recorded neurons was positive for both TPH and TH (n = 0/412). During the recordings, we used a series of depolarizing and hyperpolarizing current steps and ramps that allowed us to characterize active and passive membrane properties in detail (Figure 1E – G). Based on the electrophysiological data, we first tested possible differences between TH-positive neurons recorded in wild-type mice and tdTomato-positive neurons recorded in DAT-tdTomato mice. We found no differences between these two groups (n = 13 TH-positive vs n = 30 tdTomato-positive neurons, Suppl. Figure 2A) and neither within the subset of tdTomato-positive neurons when comparing TH-positive to TH-negative neurons (n = 25 TH-positive vs n = 5 TH-negative neurons, Suppl. Figure 2A–E). Out of 114 tdTomato-positive neurons only one cell displayed a different electrophysiological profile than all other DRNDA neurons, suggesting a false-positive rate of 0.8%. That neuron was TH-negative and displayed profoundly distinct intrinsic properties, and was therefore excluded (Suppl. Figure 2F, G). Taken together, the electrophysiological results support the use of the DAT-tdTomato mouse line when studying DRNDA neurons and data from both mouse lines were pooled. Recordings of DRNDA neurons showed that they share several properties characteristic of other dopaminergic populations located in the SNc and the ventral tegmental area, such as a slowly ramping membrane potential during constant current injections giving rise to delayed spiking and postinhibitory hypoexcitability (Figure 1E; [45, 46]). Moreover, like other dopaminergic populations in the midbrain, most DRNDA neurons displayed rebound oscillations and the vast majority of cells expressed sag currents (Figure 1E, F). When comparing the electrophysiological properties of DRNDA to DRN5-HT neurons, we observed numerous differences between these two cell types, but here we focus on the five most significant ones. While DRN5-HT neurons spike with short delays in response to current steps and maintain a relatively constant action potential (AP) amplitude, DRNDA neurons display a longer delay to the first spike and the amplitude of subsequent action potentials drops (Figure 1F-I). Additionally, the action potentials of DRN5-HT neurons rise faster, while their afterhyperpolarization (AHP) is longer compared to DRNDA neurons (Figure 1H, I). Lastly, the capacitance of DRN5-HT neurons is significantly larger than in DRNDA neurons (Figure 1I).

DRNDA and DRN5-HT are electrophysiologically distinct cell-types.

(A) Scheme of the location of the DRN (pink) in a coronal section (top) and at higher magnification together with two patch pipettes (center). Bottom: A representative slice stained post recording for TPH, TH and neurobiotin revealing serotonergic neurons (arrows). The ventricle is indicated with a dashed line. (B) Top: Differential interference contrast (DIC) microscopy image (left) of neurons that were filled with Alexa488 (right) and neurobiotin. Center, bottom: Staining of the same neurons revealing a TPH-positive (DRN5-HT) neuron and a TPH- and TH-negative cell. (C) Top: DIC image of recorded neurons that were filled with Alexa488 and neurobiotin. Center, bottom: Staining of the same neurons revealing tdTomato- and TH-positive (DRNDA) neurons. (D) Representative fluorescent (top), DIC (center) image and overlay (bottom) of a tdTomato-positive neuron in a DAT-tdTomato mouse. (E) Representative recordings depicting postinhibitory hypoexcitability, slowly ramping currents and rebound oscillations in DRNDA neurons. (F) Representative voltage responses to current injections in a TPH-positive (DRN5-HT) and TH-positive (DRNDA) neuron. (G) Ramping current injections reveal action potential (AP) amplitude accommodation. Grey circles indicate the onset and peak of APs. (H) Amplitude and duration of the AP and action potential afterhyperpolarization (AHP) in a DRN5-HT and DRNDA neurons. Grey circles indicate onset, peak, and end of the AP and AHP. (I) Quantification of electrophysiological properties distinguishing DRN5-HT from DRNDA neurons (n= 28-43 cells per group, N = 9; Wilcoxon Rank Sum Test). (J) Principal component analysis (PCA) of five electrophysiological parameters (insert) and hierarchical cluster analysis based on PCA1 and PCA2 (Ward’s method, Euclidean distance). Intrinsic properties were sufficient to separate TPH-positive cells (blue dash) from TH-positive (red dash) cells. Bottom dashes indicate wild type (black) and DAT-tdTomato (gray) mice. Data are shown as mean ± SEM, ***p < 0.001. Scale bars: A, 100 μm; B-D, 10 μm.

Next, we tested if DRNDA neurons can be distinguished from DRN5-HT neurons based on these five electrophysiological parameters. To this end, we standardized the data and ran a principal component analysis (PCA) including all DRN5-HT neurons (i. e. all TPH-positive), all TH-positive neurons recorded in wild- type mice and all tdTomato-positive cells recorded in DAT-tdTomato mice (except for one outlier shown in Suppl. Figure 2F, G). Plotting the first two principal components (PC) showed two separate clusters (Figure 1J, insert). Unsupervised hierarchical cluster analysis based on PC1 and PC2 revealed the same two major clusters and potential subclusters (Figure 1J). Mapping the molecular identity of the cells onto the dendrogram revealed the separation of DRN5-HT and DRNDA neurons, while there was no branching according to mouse line (wild type vs. DAT-tdTomato), further corroborating the validity of DAT-tdTomato mice as a marker for DRNDA neurons. Overall, these data suggest that electrophysiological parameters themselves are sufficient to distinguish between DRN5-HT and DRNDA neurons.

In addition to DRNDA and DRN5-HT the DRN contains an unknown number of cell types and 47 out of 120 recorded neurons were neither TH-positive, nor TPH-positive and did not express tdTomato. To test whether DRNDA can also be distinguished from those populations based on their electrophysiological profile, we ran a PCA on 20 standardized parameters and used the first three PCAs for unsupervised hierarchical clustering (Suppl. Figure 3). Our analysis suggests that there might be four major electrophysiological cell types in the DRN. In contrast to DRNDA and DRN5-HT neurons, a large proportion of the remaining cells showed rebound spiking and biphasic AHPs, resembling the profiles of local interneurons in other brain areas (Suppl. Figure 3D, E). Interestingly, the clustering also indicated that three TH- and tdTomato-negative neurons belonged to the DRNDA neurons and further analysis showed that they were indistinguishable from molecularly identified DRNDA neurons (Suppl. Figure 3C, F). These findings indicate that clustering can be used to identify neurons that otherwise would have been excluded due to a lack of post-hoc staining data or genetic driver lines.

Overall, our data show that DRNDA neurons constitute an electrophysiologically distinct class of neurons in the DRN expressing several hallmark properties, which are sufficient to identify them within the local DRN circuitry.

DRNDA and DRN5-HT neurons have different morphological profiles

Next, we characterized the morphological profile of DRN5-HT and DRNDA neurons. We focused on the analysis of somatic and dendritic properties since a complete reconstruction of the axonal arborization could not be retrieved from the slices. The analysis of the somatic properties showed that DRN5-HT neurons had larger cell bodies than DRNDA neurons (Figure 2A-E), as measured in their area, perimeter, length, and width (Figure 2B-E). Cell bodies also differed in shape, with DRNDA neurons having more circular somata than DRN5-HT neurons, as indicated by the circularity index (Figure 2F). Analyzing the dendritic properties, we found that DRN5-HT neurons had four to five primary dendrites, compared to only two to three in DRNDA neurons (Figure 2G). Moreover, dendrites of DRNDA neurons were frequently bipolar with the main primary dendrites starting from opposite extremes of the soma. Both populations had relatively few bifurcations (Figure 2H) but the DRN5-HT neurons had significantly more terminations (Figure 2I). The overall dendritic length did not differ between the DRN5-HT and DRNDA neurons: both populations had a mix of short and long dendrites (Figure 2L). These data suggest that DRN5-HT neurons have denser dendritic arborization than DRNDA neurons, mostly due to larger numbers of primary dendrites.

DRNDA and DRN5-HT have distinct morphological profiles.

(A) Top: representative digital reconstruction of a DRN5-HT from a wild-type mouse. Bottom: representative digital reconstruction of a DRNDA from a wild-type mouse. Scale bar: 10 μm. (B-F) Morphological parameters describing the soma size and shape of DRN5-HT and DRNDA neurons in Sham-lesion mice (DRN5-HT: n = 20, N = 3; DRNDA: n = 27, N = 3; unpaired t-test). (G-J) Morphological parameters describing the dendritic tree of DRN5-HT and DRNDA neurons in Sham-lesion mice (DRN5-HT: n = 8, N = 3; DRNDA: n = 7, N = 3; unpaired t-test). Data are shown as mean ± SEM, ***p < 0.001.

Altogether, our results show that DRN5-HT and DRNDA neurons have distinct morphological properties. DRN5- HT neurons are mostly multipolar neurons, with a big and complex soma and multiple primary dendrites, while DRNDA neurons have smaller and more circular cell bodies with bipolar dendrites.

DA and NA depletion distinctly affect the membrane properties of DRN5-HT neurons

To elucidate how DRN5-HT and DRNDA neurons might be affected in PD, we characterized these populations in a mouse model of PD based on bilateral injection of the neurotoxin 6-OHDA in the dorsal striatum. This approach leads to a partial lesion of catecholamine neurons, reproducing an early stage of parkinsonism in which particularly non-motor symptoms are manifested [47]. In line with previous studies, we observed a 60-70% reduction of TH levels in the striatum (Suppl. Figure 4, [48]). Only mice meeting this criterion were included in the study. Immunostaining showed that the striatal 6-OHDA injection did not cause degeneration of DRN5-HT or DRNDA neurons (Suppl. Figure 5).

Striatal injection of 6-OHDA has also been found to produce a partial loss of NA neurons in the LC [48]. In the present study, we determined the specific impact of NA dysfunction on the physiology of DRN5-HT and DRNDA neurons by pre-treating a group of mice with desipramine (DMI), a selective inhibitor of noradrenaline reuptake, before injecting 6-OHDA (DMI+6-OHDA mice). We then assessed the intrinsic properties of DRN5-HT and DRNDA neurons in Sham-lesion (Sham), 6-OHDA- and DMI+6-OHDA-treated mice. Whole-cell recordings obtained from DRN5-HT neurons in control mice revealed that 36% of DRN5-HT neurons were spontaneously active in slices and the proportion of intrinsically active neurons was similar in mice injected with 6-OHDA (control: n = 12/33 DRN5-HT neurons, 6-OHDA: n = 7/18 DRN5-HT neurons, Figure 3A). However, DRN5-HT neurons recorded in DMI+6-OHDA mice showed an increased excitability: in this condition 73% of DRN5-HT neurons were spontaneously active and DRN5-HT neurons displayed lower rheobase currents than in control and 6-OHDA mice (Figure 3A, B). Since the noradrenergic system is protected by DMI in these mice, these findings suggest that the 6-OHDA induced lesion of midbrain dopaminergic neurons evoked the increased firing of DRN5-HT neurons.

Lesions targeting primarily nigrostriatal dopamine increase the excitability of DRN5-HT neurons whereas loss of NA affects their action potentials.

(A) Top: Pie charts showing the number of spontaneously active (dark) and silent (pale) DRN5-HT neurons in three conditions: Sham (left), 6-OHDA-injected mice (center) and 6-OHDA-injected mice pre- treated with desipramine (DMI + 6-OHDA, right). Bottom: Representative recordings of spontaneously active DRN5-HT neurons (I = 0 pA). (B) Quantification of the rheobase (left), the firing frequency of spontaneously active cells (center), and the resting membrane potential of silent DRN5-HT neurons (right). (C) Representative action potentials (APs) of DRN5-HT at low (left) and high (right) temporal. Grey circles indicate onset, offset, and peak of the APs as well as the end of the afterhyperpolarization (AHP). (D) Quantification of the amplitude (left) and duration (right) of the APs of DRN5-HT neurons. (E) Same as in (D) for the AHP. (F) Representative responses of DRN5-HT neurons to current steps (I = +75 pA). (G) Quantification of firing frequency / injected current. (H) Quantification of the delay to the first AP when injected with current eliciting 2Hz firing. (I) Quantification of the membrane time constant (tau) of DRN5-HT neurons (Sham: n = 22 - 32, N = 6 - 7; 6-OHDA: n = 11 - 16, N = 7; DMI + 6-OHDA: n = 18 – 21, N = 4; unpaired t-test). Data are shown as mean ± SEM, * p < 0.05, ** p < 0.01.

While the excitability of DRN5-HT neurons was not affected in 6-OHDA mice, we observed that their firing properties were profoundly altered: DRN5-HT neurons recorded in 6-OHDA mice displayed significantly smaller and shorter APs and AHPs compared to Sham mice (Figure 3C-E). Moreover, DRN5-HT neurons fired at higher frequencies and with shorter delays in response to current injections (Figure 3F-H). Finally, the membrane time constant of DRN5-HT neurons was shorter in 6-OHDA injected mice than in Sham mice (Figure 3I). Interestingly, these changes were not observed in 6-OHDA injected mice pre-treated with DMI, suggesting that the noradrenergic lesion – and not the dopaminergic lesion – underlies the changes in the firing properties of DRN5-HT neurons in 6-OHDA mice. Taken together, these results indicate that DRN5-HT neurons are affected in the 6-OHDA mouse model of PD. Specifically, lesions of the dopaminergic system increase the excitability of DRN5-HT neurons whereas lesions of the noradrenergic systems change the firing properties of DRN5-HT neurons.

6-OHDA lesion induced hypotrophy of DRN5-HT neurons

Morphological analysis revealed a reduced soma size of the DRN5-HT neurons in 6-OHDA mice, which was manifested in a decrease in area, perimeter, and major axes in comparison to control mice (Figure 4A-C). Moreover, the increase in the circularity of the 6-OHDA group indicated that the shape of the soma of DRN5- HT neurons was also altered by the lesion (Figure 4C). These modifications were not observed in DMI + 6- OHDA mice, suggesting that preserving the noradrenaline system protected the DRN5-HT neurons (Figure4A- C). Finally, the injection of 6-OHDA also caused a significant reduction in the number of primary dendrites of DRN5-HT neurons in the 6-OHDA group and a tendency to fewer termination points (Figure 4D). The number of bifurcations and the dendritic length were not affected by the lesion (Figure 4D). Globally, these results suggest that the lesion produced by 6-OHDA induces a hypotrophic phenotype in DRN5-HT neurons characterized by a shrinkage of the soma and a decrease in dendritic branching and that these alterations are partly NA-dependent.

Striatal injection of 6-OHDA induced a hypotrophic phenotype in the DRN5-HT, which is prevented by pre- treatment with DMI.

(A) Representative digital reconstructions of a DRN5-HT neuron in three different conditions: Sham (left), 6-OHDA-injected mice (center) and 6-OHDA-injected mice pre-treated with desipramine (right). Scale bar: 10 μm (B) Representative confocal pictures of soma from DRN5-HT neurons in Sham (top), 6-OHDA-injected mice (center), and 6-OHDA-injected mice pre-treated with desipramine (bottom). Scale bar: 10 μm. (C-G) Morphological descriptors of the soma size and shape in DRN5-HT neurons (Sham: n = 20, N = 4; 6-OHDA: n = 19, N = 4; DMI + 6-OHDA: n = 17, N = 3; unpaired t-test).(H-K) Morphological descriptors of the dendritic tree in DRN5-HT neurons (Sham: n = 8, N = 3, 6-OHDA: n = 6, N = 3: DMI + 6-OHDA: n = 6, N = 2; unpaired t-test). Data are shown as mean ± SEM, ***p<0.001, **p<0.01. Scale bar: 10 µm.

6-OHDA lesion affects the firing of DRNDA neurons independent of NA loss

Finally, we assessed whether the striatal 6-OHDA lesion affects the physiology of DRNDA neurons. Whole- cell patch-clamp recordings revealed that 58% of DRNDA neurons are spontaneously active in slices of Sham- lesion mice (Figure 5A). In contrast, the proportion of intrinsically active neurons increased to 77% and 78% of DRNDA neurons in 6-OHDA injected mice with and without pre-treatment with DMI, respectively. In line with this, silent DRNDA neurons recorded in DMI + 6-OHDA mice were also found to have more depolarized resting membrane potentials and hence to be closer to action potential threshold (Figure 5B). Yet, active DRNDA neurons tended to fire at lower frequencies in all 6-OHDA mice compared to control mice.

Lesions targeting primarily SN dopamine depolarize DRNDA neurons whereas concomitant loss of NA does not affect their action potentials.

(A) Top: Pie charts showing the proportion of spontaneously active (dark) and silent (pale) DRNDA neurons in three conditions: Sham (left), 6OHDA-injected mice (center) and 6-OHDA-injected mice pre- treated desipramine (DMI, right). Bottom: Representative recordings of spontaneously active DRNDA (I = 0pA). (B) Quantification of the rheobase (left), the firing frequency of spontaneously active (center), and the resting membrane potential of silent DRNDA neurons (right). (C) Representative action potentials (APs) of DRNDA at low (left) and high (right) temporal resolution. Grey circles indicate onset, offset, and peak of APs and the end of the afterhyperpolarization (AHP). (D) Quantification of the amplitude (left) and duration (right) of the APs of DRNDA neurons. (E) Same as in (D) for the AHP. (F) Representative responses of DRNDA neurons to current steps (I = 75pA). Grey circles indicate the delay to the first AP. (G-I) Quantification of firing frequency / injected current (G), the delay to the first AP when injected with current eliciting 2 Hz firing (H), and the membrane time constant (tau, I) of DRNDA neurons recorded (Sham: n = 43, N = 8; 6-OHDA: n = 29, N = 6; DMI + 6-OHDA: n = 40, N = 6; unpaired t-test). Data are shown as mean ± SEM, * p < 0.05.

In stark contrast to DRN5-HT neurons, the APs and AHPs of DRNDA neurons were not affected in any 6-OHDA mice (Figure 5C-E). In fact, we did not observe any change in the firing properties of DRNDA neurons that were dependent on the protection of the NA system with DMI (Figure 5). Like DRN5-HT neurons, DRNDA neurons displayed a reduction in spike latency in 6-OHDA mice, but the shortening of the delay was also present in DMI+6-OHDA mice. Together, these results suggest that the electrophysiological properties of DRNDA neurons are affected in the 6-OHDA mouse model of PD and that these changes are primarily due to the lesion of the nigrostriatal dopaminergic pathway.

Striatal 6-OHDA injections increased soma complexity and dendritic branching in DRNDA neurons

The morphological analysis of DRNDA neurons revealed that the striatal 6-OHDA injection caused a significant reduction of the circularity from 0.82 (Sham) to 0.77 (6-OHDA) (Figure 6A-C), indicating increased complexity of the DRNDA soma. We also observed that this effect was not present in the mice pre-treated with DMI (Figure 6C). The soma size was not affected by the lesion and no changes in the area, perimeter, length, and width were found (Figure 6C). However, dendritic branching, as shown by the increase in termination points and bifurcations in DRNDA neurons was increased in 6-OHDA-injected mice, but not in mice treated with a combination of DMI and 6-OHDA (Figure 6D). The number of primary dendrites and the overall dendritic length were not affected by the 6-OHDA lesion (Figure 6D). Overall, these data suggest that impairment of noradrenergic function caused by the striatal 6-OHDA injection altered the soma shape and dendritic branching of DRNDA neurons.

Striatal injection of 6-OHDA increased soma complexity and branching in DRNDA.

(A) Representative digital reconstructions of a DRNDA neuron in three different conditions: Sham (left), 6-OHDA-injected mice (center) and 6- OHDA-injected mice pre-treated with desipramine (right). (B) Representative confocal pictures of soma from DRNDA neurons in Sham (top), 6-OHDA-injected mice (center), and 6-OHDA-injected mice pre-treated with DMI (bottom). (C- G) Morphological descriptors of the soma size and shape in DRNDA neurons (Sham: n = 27, N = 7, 6-OHDA: n = 16, N = 4; DMI + 6-OHDA: n = 31, N = 5; unpaired t-test). (H-K) Morphological descriptors of the dendritic tree in DRNDA neurons. (Sham: n = 7, N = 3; 6-OHDA: n = 11, N = 4; DMI + 6-OHDA: n = 7, N = 3; unpaired t-test). Data are shown as mean ± SEM, *p<0.05. Scale bar: 10 µm.

Discussion

In the present study we combine ex vivo whole-cell patch clamp recordings with morphological reconstructions and immunohistochemistry, to show that DRNDA neurons have a distinct electrophysiological profile, which is sufficient to distinguish them from DRN5-HT neurons as well as other neuron classes in the DRN. Utilizing this approach, we also reveal that, in a 6-OHDA mouse model of PD, DRN5-HT neurons display distinct pathophysiological changes depending on the loss of dopamine and noradrenaline. Notably, degeneration of noradrenergic neurons affects not only the electrical properties of DRN5-HT neurons but also evokes hypotrophy of their cell bodies. In contrast, the loss of nigrostriatal dopamine mainly affects the electrophysiological properties of DRNDA neurons while concomitant loss of noradrenaline alters their morphology.

We used an extensive electrophysiological characterization protocol to quantify the differences between the DRNDA and DRN5-HT populations. The electrophysiological properties agree with previous studies, such as the spontaneous firing pattern seen in DRNDA neurons [33], and the slow AHP of DRN5-HT neurons [21]. Standard electrophysiological parameters were used to create a classification tool, which efficiently identifies DRN5-HT and DRNDA cells, including dopamine neurons confirmed by TH staining and / or by fluorescent expression in DAT-tomato mice (Figure 1). Importantly, the DRNDA neurons recorded from wild- type and DAT-tdTomato mice did not differ in their electrical properties, indicating that the transgene does not interfere with the membrane properties of this population.

We showed that DRNDA neurons share electrophysiological properties with other dopaminergic populations in the midbrain [45, 46] but have electrophysiological profiles distinct from DRN5-HT neurons as well as other neuronal populations in the DRN. Most of the parameters extracted in our characterization rely on intracellular recordings of the membrane potential. However, some properties such as the spontaneous firing and action potential kinetics could be useful for in vivo characterization, even in extracellular recordings [4951]. In addition to the DRN5-HT and DRNDA neuronal populations, a large fraction of neurons displayed electrophysiological properties that were distinct from these two groups (Suppl. Figure 3), suggesting that there are other neuronal subtypes in the DRN network, such as previously reported GABAergic, glutamatergic, and peptidergic neurons [9, 5257].

In line with previous studies, the majority of DRN5-HT neurons were large multipolar or fusiform neurons with four to five primary dendrites, very distinct from the DRNDA neurons [5860]. Very little is known about the morphology of the DRNDA neurons, but previous studies identified small ovoid cells in the DRN which are likely to correspond to the DRNDA cells [33, 61]. Out of 25 reconstructed DRN5-HT neurons, only one displayed dendritic spines. Previous studies in rats, described the presence of dendritic spines in most DRN5- HT neurons [62]. However, the study was performed in thicker slices and the dendritic spines were scarce in the primary and secondary dendrites, while they became dense in the distal dendrites, thus it is possible that in our study those dendrites were not present [62].

In the present study we assessed the impact on DRN cells of a striatal bilateral 6-OHDA lesion performed with or without DMI pre-treatment, which has been shown to protect the NA neurons from the 6-OHDA- induced degeneration [48, 63, 64]. We found that both DRN5-HT and DRNDA populations were affected in a cell-type specific manner by the combined action of 6-OHDA on DA and NA, with DRN5-HT neurons being particularly sensitive to changes in the noradrenergic system. Loss of SNc dopaminergic neurons alone (6- OHDA + DMI) – which are known to target DRN5-HT neurons directly – increased the excitability and spontaneous activity in DRN5-HT neurons [53]. This is in line with previous ex vivo and in vivo studies showing that the DRN5-HT neurons display increased firing rates in rodents pre-treated with DMI and injected with 6-OHDA [21, 65] As hypothesized by Prinz et al. [21], the selective loss of midbrain DA may induce a homeostatic increase in the excitability of DRN5-HT neurons. Our data contrast with a previous in vivo study showing decreased firing activity in DRN5-HT neurons where injection of 6-OHDA was preceded by treatment with DMI and fluoxetine [22]. This dissimilarity may be related to species-specific (rat vs mouse) and technical (intracerebroventricular vs striatal injections, recordings performed at 10 days vs 3 weeks after the 6-OHDA injection). Importantly, in the same study identification of DRN5-HT neurons was not molecularly confirmed and the data may include other spontaneously active DRN neurons. In fact, our recordings show that there are non-serotonergic neurons in the DRN, which are spontaneously active and display a regular, slow firing frequency similar to DRN5-HT neurons, highlighting the importance of unequivocal identification of DRN cell types.

The present study shows that combined DA and NA lesioning affects DRN5-HT neurons more profoundly than selective loss of DA (Figure 3, 4). In mice treated with 6-OHDA only, several electrophysiological and morphological properties were altered (Figure 3, 4). The time constant and AHP of DRN5-HT neurons were shorter and the neurons responded with higher firing frequencies to current injections than in sham. This finding suggests that the pronounced AHP and long tau of these neurons may act as a “brake” limiting their maximum firing frequency in control conditions and that this brake is reduced when the NA system is lesioned. Future studies are needed to assess if DRN5-HT in fact fire at higher rates in vivo in mice treated with 6-OHDA. In contrast, such changes in DRN5-HT neurons were prevented when the NA system was protected by pre-treatment with DMI. These findings indicate an important role for NA as mediator of changes in the activity and properties of DRN5-HT neurons. The changes produced by the 6-OHDA lesion on the DRNDA population were less pronounced than and different from those in DRN5-HT neurons. In terms of electrophysiological properties, the observed changes were primarily in the DA only lesion (6-OHDA + DMI), suggesting that unlike DRN5-HT, DRNDA neurons are affected by the loss of midbrain DA rather than the accompanying changes in NA (Figure 5). Conversely, the morphological changes produced by 6-OHDA in DRNDA neurons appeared to require a concomitant DA and NA lesion, suggesting a stronger impact of the NA modulation (Figure 6).

Our results show that DRN neurons are affected by depletion of both DA and NA, thus raising the possibility that non-motor symptoms in PD are a result of the intricate organization of DA and NA neuromodulation as well as the interactions between the different DRN neuronal populations. Moreover, our results highlight the complex interplay between different neuromodulator systems [66], in this case NA, DA, and 5-HT in the DRN. The NA system is profoundly affected in PD [1, 8], but the precise pathophysiological processes resulting from loss of NA, and specifically the impact on DRN, are yet to be elucidated.

In conclusion, our study provides a quantitative description and classification scheme for two major neuronal populations in the dorsal raphe nucleus, DRN5-HT and DRNDA neurons. We identified novel electrophysiological and morphological changes in these populations in response to DA and NA depletion in the basal ganglia. Considering the involvement of DRN and LC in the development of non-motor comorbidities, this study provides useful insights to understand better how these areas are affected in the parkinsonian condition. Moreover, our data pave the way for future experiments to characterize these subpopulations in terms of receptor expression and synaptic connectivity to shed light on their functional roles particularly with regard to the wide variety of non-motor symptoms observed in PD.

Methods

Experimental model details

All animal procedures were performed in accordance with the national guidelines and approved by the local ethics committee of Stockholm, Stockholms Norra djurförsöksetiska nämnd, under ethical permits to G. F. (N12148/17) and G. S. (N2020/2022). All mice (N=30) were group-housed under a 12 hr light / dark schedule and given ad libitum access to food and water. Wild-type mice (‘C57BL/6J’, #000664, the Jackson laboratory) and DAT-cre (Stock #006660 the Jackson laboratory) mice crossed with homozygous tdTomato reporter mice (‘Ai9’, stock #007909, the Jackson laboratory) were used.

6-OHDA model

Three-month old, male and female C57BL/6J or DAT-tdTomato were deeply anesthetized with Isoflurane and mounted on a stereotaxic frame (Stoelting Europe, Dublin, Ireland). To achieve a partial striatal lesion, each mouse received a bilateral injection of 1.25 μl of 6-hydroxydopamine hydrochloride (6-OHDA, (Sigma- Aldrich, 4μg/μl) or vehicle (0.9 % NaCl + Ascorbic Acid 0.02%) in the dorsolateral striatum, according to the following coordinates: anteroposterior +0.6 mm, mediolateral ± 2.2, dorsoventral -3.2 from Bregma, as previously described [47, 67]. One group of mice (referred to as DMI + 6-OHDA) was pre-treated with desipramine hydrochloride (DMI, Sigma-Aldrich, 25mg/kg i.p.) 30 minutes before the 6-OHDA infusion in order to protect the noradrenergic system [48].

Slice preparation and electrophysiology

Three weeks after the 6-OHDA injection, mice were deeply anaesthetized with isoflurane and decapitated. The brain was quickly removed and immersed in ice-cold cutting solution containing 205 mM sucrose, 10 mM glucose, 25 mM NaHCO3, 2.5 mM KCl, 1.25 mM NaH2PO4, 0.5 mM CaCl2 and 7.5 mM MgCl2. In all experiments, the brain was divided into two parts: the striatum was dissected from the anterior section for Western Blot and the posterior part was used to prepare coronal brain slices (250 um) with a Leica VT 1000S vibratome. Slices were incubated for 30-60 min at 34°C in a submerged chamber filled with artificial cerebrospinal fluid (ACSF) saturated with 95% oxygen and 5% carbon dioxide. ACSF was composed of 125 mM NaCl, 25 mM glucose, 25 mM NaHCO3, 2.5 mM KCl, 2 mM CaCl2, 1.25 mM NaH2PO4, and 1 mM MgCl2. Subsequently, slices were kept for at least 60 min at room temperature before recording.

Whole-cell patch clamp recordings were obtained in oxygenated ACSF at 35°C. Neurons were visualized using infrared differential interference contrast (IR-DIC) microscopy (Zeiss FS Axioskop, Oberkochen, Germany). DAT-tomato positive cells were identified by switching to epifluorescence using a mercury lamp (X-cite, 120Q, Lumen Dynamics). Up to three cells were patched simultaneously. Borosilicate glass pipettes (Hilgenberg) of 6 - 8 MOhm resistance were pulled with a Flaming / Brown micropipette puller P-1000 (Sutter Instruments). The intracellular solution contained 130 mM K-gluconate, 5 mM KCl, 10 mM HEPES, 4 mM Mg-ATP, 0.3 mM GTP, 10 mM Na2-phosphocreatine (pH 7.25, osmolarity 285 mOsm), 0.2% neurobiotin (Vector laboratories, CA) and Alexa-488 (75 µM) was added to the intracellular solution (Invitrogen). Spontaneous firing was recorded in cell-attached mode before break-in and all other whole- cell recordings were made in current-clamp mode. The intrinsic properties of the neurons were determined by a series of hyperpolarizing and depolarizing current steps and ramps, enabling the extraction of sub- and suprathreshold properties. Recordings were amplified using MultiClamp 700B amplifiers (Molecular Devices, CA, USA), filtered at 2kHz, digitized at 10-20kHz using ITC-18 (HEKA Elektronik, Instrutech, NY, USA), and acquired using custom-made routines running on IgorPro (Wavemetrics, OR, USA). Throughout all recordings pipette capacitance and access resistance were compensated for and data were discarded when access resistance increased beyond 30 MOhm. Liquid junction potential was not corrected for.

Immunofluorescence

Following the recordings, slices were fixated overnight at 4°C in 4% paraformaldehyde solution. Slices were then washed with PBS 1X. For the immunofluorescence, slices were treated with PBS 1X + Triton 0.3% and then incubated with a blocking solution of Normal Serum 10% and Bovine Serum Albumin 1% for 1h at room temperature. Afterward, slices were incubated overnight at 4°C with the following primary antibodies: rabbit anti-TH (Millipore, 1:1000), mouse anti-TPH (Sigma Aldrich, 1:600) and Streptavidin (Jackson Immunoresearch, 1:500). The following day, primary antibodies were washed out and slices were incubated with the appropriate fluorochrome-conjugated secondary antibodies.

For the immunostainings in the striatum and SNc and cell counting in DRN, mice were deeply anesthetized and transcardially perfused with PFA 4%. The brains were extracted and post-fixed in PFA 4% for 24 hours. 40 um coronal slices were prepared with a vibratome (Leica VT1000 S) and processed as described above.

Confocal Microscopy Analysis

The slices were imaged using Confocal (ZEISS LSM 800) at 10X and 40 X and z-stacks were retrieved. For cell identification, colocalization between neurobiotin and TH or TPH was evaluated.

Morphological analysis

For morphological analysis of dendrites, the confocal z-stacks were used in a semi-manual reconstruction using neuTube [68] and custom code, as previously described [69]. Soma morphology was analyzed by tracing manually the cell body profile, excluding dendritic trunks, in order to measure area (µm²), perimeter, major and minor axis length (µm) and circularity values. Circularity, calculated as the ratio between the squared perimeter and the area (i.e. perimeter²/4π area), can be a value between 0 and 1 (1 for circular shapes and values < 1 for more complex shapes.

The morphological analysis was performed on the neurobiotin stacks.

Western Blot

The striata were sonicated in 1% sodium dodecyl sulfate and boiled for 10 min. Equal amounts of protein (25 μg) for each sample were loaded onto 10% polyacrylamide gels and separated by electrophoresis and transferred overnight to nitrocellulose membranes (Thermo Fisher, Stockholm, Sweden). The membranes were immunoblotted with primary antibodies against actin (1:30,000, Sigma Aldrich, Stockholm, Sweden) and TH (1:2000, Millipore, Darmstadt, Germany). Detection was based on fluorescent secondary antibody binding (IR Dye 800CW and 680RD, Li-Cor, Lincoln, NE, USA) and quantified using a Li-Cor Odyssey infrared fluorescent detection system (Li-Cor, Lincoln, NE, USA). The TH protein levels were normalized for the amount of the corresponding actin detected in the sample and then expressed as a percentage of the control (Sham-lesion).

Statistical Analysis

Statistical analysis of morphology data was performed using GraphPad Prism 9.2.0. Data are reported as average as ± SEM and analyzed by unpaired t-test. N indicates the number of mice, while n indicates the number of cells. Significance was set at p<0.05.

Acknowledgements

We thank Elin Dahlberg for technical assistance and Kristoffer Tenebro Berglund for taking care of the mice. We also thank the members of the Silberberg and Fisone labs and the AND-PD consortium members for comments and discussions.

Funding

This work was supported by a Wallenberg Fellowship from the Knut & Alice Wallenberg Foundation (KAW 2017.0273), the Swedish Brain Foundation (Hjärnfonden, FO2021-0333), and the Swedish Medical Research Council (VR-M, 2019-0 1254) to GS, and a Swedish Research Council International Postdoc Grant (2020-06365) to YJ. RT, RA, GF, and GS are supported by an EU grant (H2020 848002 AND-PD).

Author contributions

Conceptualization: LB, YJ, RT, RM, GF, GS.

Investigation: YJ (electrophysiology, clustering analysis), LB (6-OHDA lesions, morphological analysis, cell counting, western blot).

Visualization: YJ (electrophysiological and clustering results), LB (6-OHDA model, morphological reconstructions).

Supervision: GS and GF. Writing—original draft: YJ and LB. Writing—review & editing: LB, YJ, GF, GS.

Competing Interests

The authors declare that they have no competing interests.

Data availability

All raw data is available upon reasonable request.