To determine if chronic chemogenetic activation of DA neurons for 2 weeks induces degeneration, we first quantified its effects on axonal integrity in the striatum. Strikingly, compared to vehicle-treated mice, hM3Dq-expressing animals treated with CNO lost ≈40% of their dopaminergic axons in the dorsal striatum, as evidenced by decreases in both TH immunoreactivity and reporter mCherry immunofluorescence (Figure 2A and B). The same perturbation had a lesser impact on nucleus accumbens and no impact on olfactory tubercle DA afferents. Notably, TH immunoreactivity and mCherry immunofluorescence also decreased in nonresponder mice following 2 weeks of treatment (Figure 2—figure supplement 1A), although the extent of decrease was somewhat less than in responders. After a shorter 1-week treatment period in a small cohort, we also observed a trend for decreased TH immunoreactivity in the CPu (Figure 2—figure supplement 1B).

We also assessed the impact of 2 weeks of activation on DA neuron survival in the midbrain. With stereological quantification, we found no difference in neuron number between CNO-treated animals and vehicle-treated controls in the SNc or VTA (Figure 2C). Therefore, at this time point, CNO-treated mice exhibit severe axonal degeneration without DA neuron death.

To ensure that the loss of DA terminals was not due to an off-target effect of CNO, we assessed TH expression in CNO alone animals. These mice did not show a decrease in dopaminergic axons in the CP (Figure 2D), supporting the conclusion that the axonal degeneration requires hM3Dq activation.

We next assessed the impact of more prolonged activation on degeneration of DA neuron somata. Mice treated with CNO for 4 weeks lost the majority of axons projecting to the CP (Figure 2E and F), while fibers in the nucleus accumbens and olfactory tubercle were again preferentially spared. Moreover, following 4 weeks of CNO treatment, there was a significant decrease in the overall number of TH+ midbrain (SNc and VTA) DA neurons, assessed by stereology (Figure 2G and H). Although the loss of TH reactivity can occur in the absence of neuronal death, there was a similar decrease in the number of mCherry+ DA neurons (labeled by DsRed) in the midbrain. The alignment of these findings supports true loss of DA neurons, although quantitation of mCherry+ DA neurons specifically informs the impact on transduced DA neurons, and this analysis does not account for the variability between injections. Together, these data suggest that, similar to PD progression in humans (Kordower et al., 2013), axonal impairment precedes cell body loss in this model. No preferential loss of SNc vs VTA cell bodies was detected at this time point.

hM3Dq activation may increase neural activity by increasing intracellular calcium levels (Roth, 2016), and increased neural activity should also increase intracellular calcium. To determine the impact of 2 weeks of hM3Dq activation on intracellular calcium concentrations in DA neurons, we bred DAT^IREScre mice to Ai148D mice that contain a Cre-dependent GCaMP6f calcium indicator and injected them bilaterally with the Cre-dependent hM3Dq AAV (Figure 3A). The fluorescent signal was used to guide implantation of an optical fiber in the midbrain of animals (Figure 3B and C), with confirmed hM3Dq expression based on increased wheel running in response to i.p. CNO (Figure 1—figure supplement 1A). Mice were administered CNO or vehicle water and then placed in arenas for 10 min fiber photometry recording sessions every 1–3 days over the 2-week treatment period. Mean fluorescence levels served as a proxy for population-level intracellular calcium concentrations. We observed a small trend for increased fluorescent signal the day after starting CNO that persisted for days 3 through 5 but did not reach significance (Figure 3D–F). Interestingly, this was followed by a second, much larger increase in calcium levels between 10 and 17 days (Figure 3D–F), that occurred in parallel with axon loss (Figure 2A, Figure 2—figure supplement 1B). To determine if the increase in neural activity was reversible, we performed two additional recording sessions after removing CNO from the drinking water. Although there was a small trend for decreased fluorescence, this did not reach significance.

Although GCaMP fluorescence is not quantitative, the consistent difference in baseline in animals tested concurrently on the same system provides confidence in our interpretation. The frequency and amplitude of calcium transients decreased with chronic CNO treatment, and during washout, the frequency remained decreased while the amplitude returned to vehicle levels (Figure 3—figure supplement 1A). Notably, the decrease in frequency or amplitude is unlikely to represent a ceiling effect (i.e. where a high baseline might occlude transient detection), because while the absolute baseline and transient amplitude returned to pre-treatment levels during the washout, the frequency of transients did not. Instead, the decreases in transient frequency and amplitude may reflect less bursting activity in the neurons. As sharp peaks in genetically encoded calcium indicator signals tend to reflect bursting activity, the sustained decrease in frequency during washout may reflect more permanent dysfunction in the burst firing patterns of these neurons.

Together, these data support massive increases in DA neuron intracellular calcium over time, which may be due to hM3Dq receptor activation, calcium dysregulation, and the onset of degeneration. We also assessed time spent moving and distance traveled during photometry sessions in a subset of animals (Figure 3—figure supplement 1B). Paralleling the small initial increase in baseline calcium, there was a strong trend for CNO to increase activity over time vs controls in both measures of gross locomotion during the first week of treatment (two-way ANOVA distance traveled p=0.07, percent time spent moving p=0.06). Interestingly, the increased open-field movement in the CNO group decreased back to baseline by day 11 (Figure 3—figure supplement 1B), just as calcium levels began to climb markedly (Figure 3F) and in parallel to axonal degeneration (Figure 2A, Figure 2—figure supplement 1B).

To gain additional insight into the mechanisms of degeneration in this model, we used spatial genomics. We performed Visium spatial transcriptomics on midbrain and striatal sections from DAT^IREScre mice that were bilaterally injected with AAV-DIO-hM3Dq-mCherry at 3 months and then received either CNO (GqCNO) or vehicle (GqVeh) drinking water for 1 week prior to harvest. Additional non-injected control mice also received CNO (CNO alone) (Figure 4A). Mice were harvested following 1 week of CNO treatment in order to quantify transcriptomic changes at a time point when some axons still remain and prior to somatic degeneration (Figure 2A, Figure 1—figure supplement 2A, Figure 4—figure supplement 1A), thereby focusing on early gene expression changes prior to neuronal death.

We next identified 55 μm diameter barcoded discs (Visium Spatial Gene Expression) overlaying regions enriched for DA neurons within the SN and VTA in the midbrain, and the caudate putamen (CP) (Figure 4B). In the midbrain, selected discs contained at least one entire TH-positive (and mCherry-positive for injected animals) soma and also expressed at least two of three characteristic DA genes (Slc6a3 which encodes the DA transporter DAT, Slc18a2 which encodes the vesicular monoamine transporter 2 [VMAT2], or Th) above a preset threshold level (Figure 4C). Interestingly, Slc6a3 and Slc18a2 were significantly decreased in the SN and VTA of GqCNO mice, perhaps indicating a loss of a dopaminergic phenotype. In agreement with this, we found that striatal DA levels were markedly decreased by HPLC (Figure 4—figure supplement 2). Principal components analyses of the SN and VTA in the midbrain demonstrate that GqCNO midbrain dopaminergic regions are transcriptionally similar to one another and distinct from control groups (Figure 4D). This suggests that chronic DREADD activation induces a robust transcriptomic response in DA neurons that is independent of viral transduction or CNO administration. Consistent with this, we employed the NEUROeSTIMator method (Bahl et al., 2024) to predict neural activity scores based on the transcriptomic assessment of the SN and VTA, finding that GqCNO samples had higher predicted activity scores than controls (GqVEH and CNO alone) (Figure 4E).

We hypothesized that chronic activation of DA neurons likely also leads to specific gene expression changes in striatal target neurons. In the striatum, discs were selected based on the expression of Ppp1r1b (which encodes DARPP32) (Figure 4C). Notably, principal components analysis between GqCNO and control groups revealed less separation in the striatum than the midbrain (Figure 4D), perhaps because of the multiple inputs on striatal neurons in addition to dopaminergic axons, although future investigation is required.

We next compared the transcriptomic profiles of discs within each brain region between GqCNO and GqVeh mice, as well as in GqCNO vs CNO alone mice. Only genes that had significant differential expression between GqCNO and both control groups were assessed for pathway changes (Figure 4F). Consistent with chronic DREADD activation, pathway analyses of GqCNO vs control groups in the SN and VTA showed significant enrichment of GO Biological processes, including ‘chemical synaptic transmission’, ‘synaptic vesicle exocytosis’, ‘positive regulation of transporter activity’, and ‘vesicle transport along microtubule’ (Table 1). Also consistent with DREADD activation, GO Molecular Functions in the VTA included ‘GTP binding’ and ‘syntaxin binding’ (Table 1). Interestingly, both CP and SN GO Molecular Functions included ‘calcium channel regulator activity’, suggesting a larger circuit-level change in calcium channel regulation as a result of chronic DREADD activation (Table 1).

Consistent with this, there were a number of differentially expressed genes (DEGs) in the SN and/or VTA that regulate neural transmission, calcium, and activity (Figure 4G, Figure 4—figure supplement 1B and C, Figure 4—figure supplement 3, Figure 4—figure supplement 4). These included increased expression of S100a6, a calcium sensor and modulator (Donato, 2003) and Sgk1, a serine/threonine kinase regulated by intracellular calcium involved in the regulation of ion channels, membrane transporters, cellular enzymes, and transcription factors (Lang et al., 2010). Interestingly, Sgk1 expression trended up in the CP, supporting circuit-level changes in calcium signaling. Increased S100a6 expression has been observed in AD and ALS (Filipek and Leśniak, 2020), while Sgk1 upregulation has been associated with DA neuron death in rodent toxin models of PD (Iwata et al., 2004; Stichel et al., 2005). Conversely, the expression of calmodulin-2 (Calm2), which mediates control over a large number of enzymes and ion channels as a result of calcium binding (Means et al., 1991), and Trpc6, which is thought to form a nonselective calcium-permeant ion channel (Hsu et al., 2007), were both decreased. Decreased Trpc6 may be a homeostatic response to chronically high intracellular calcium. These changes in calcium-regulating genes are consistent with the changes in baseline intracellular calcium observed in DA neurons by GCaMP photometry (Figure 3) and may reflect early processes that either cause or contribute to the onset of degeneration, or alternatively compensate to protect against degeneration.

Pathway-level changes in transmission were driven by decreases in several synaptic-vesicle genes, including synaptotagmin-1, a calcium sensor that triggers transmitter release including striatal DA release (Xu et al., 2009); Snap25, a t-SNARE that regulates transmitter release (Kádková et al., 2019); Ap2a1, a component of the adapter protein complex AP-2 (Jackson et al., 2010); and Syngr3, a synaptic vesicle-associated protein that may regulate DAT function (Ho et al., 2023). Interestingly, Snca was also significantly downregulated in the VTA. α-Synuclein is enriched in presynaptic terminals and has been shown to modulate DA release (Bendor et al., 2013). Together, these results suggest that chronic DREADD activation of DA neurons results in significant changes in synaptic transmission which may result in altered calcium regulation and DA release.

To begin to interrogate the role of calcium in hyperactivity-induced degeneration, we performed a pilot experiment in which mice injected with saline in the left midbrain and AAV-hM3Dq(DREADD)-mCherry in the right midbrain were then implanted subcutaneously with slow-releasing placebo or isradipine pellets (Figure 4—figure supplement 5A). Mice were not acutely tested for responsiveness to CNO, as wheel running data is less clear with unilateral manipulation. Instead, DREADD expression was validated post hoc by immunofluorescence, and animals with little or no DREADD expression were excluded from further analysis. Isradipine blocks Ca_v1.2 and Ca_v1.3 channels that are highly expressed in SNc neurons (Chan et al., 2007; Dragicevic et al., 2014). Isradipine treatment has been shown to return SNc DA neurons to a juvenile form of pacemaking, and chronic treatment has been shown to reduce calcium-dependent mitochondrial oxidant stress (Chan et al., 2007; Guzman et al., 2018). To confirm the presence of systemic isradipine, we measured isradipine levels in blood plasma via HPLC (Figure 4—figure supplement 5B). Following 2 weeks of vehicle or CNO administration with placebo or isradipine release, we assessed striatal TH immunoreactivity and mCherry immunofluorescence. Consistent with previous results, we observed decreases in TH and mCherry in CNO-treated AAV-Gq hemispheres, but chronic isradipine administration did not rescue this loss (Figure 4—figure supplement 5C). However, the absence of rescue may result from a lack of statistical power due to small sample size or due to the presence of nonresponders. Further investigation is required to fully test the contribution of calcium signaling in hyperactivity-induced degeneration.

We also identified DEGs in the striatum. Upregulated genes included: Mt2, a metallothionein protein that acts as an antioxidant to protect against free radicals (Carpenè et al., 2007); Acsl3, an acyl-CoA synthetase that stimulates fatty acid synthesis (Fernandez and Ellis, 2020) and could supply energy substrates; and the mitochondrial uncoupling protein Ucp2. Downregulated genes included: Unc13a, which is involved in neurotransmitter release via vesicle maturation and has been implicated in ALS (Dittman, 2019; Diekstra et al., 2014), and Vgf, a secreted growth factor whose expression is also decreased in PD (Quinn et al., 2021).

We also evaluated the SN DEGs (n=59) and additional genes of interest identified within our DREADD mouse model with gene expression changes within human control and early PD subjects (Figure 4—figure supplement 6A and B). We employed the Nanostring GeoMx assay to assess whole transcriptome changes of DA neurons (TH+ masked, see Methods) within the SNc ventral tier of controls and early PD subjects. Of the 59 DEGs, 51 were found within the human dataset, and these genes exhibited broadly congruent directional changes between mouse and human (chi-square p-value=0.05). This suggests that similar mechanisms may be at play in the PD midbrain and the DREADD mouse model. To investigate this further, we considered genes that emerged from the mouse dataset involved in calcium, transmission, and DA and evaluated them in the early PD samples (Figure 4—figure supplement 6B and C). We found that the expression of Syt1, Syngr3, and Calm2 was all decreased in both mouse and human datasets. Similarly, we found that expression of Slc18a2 (VMAT2) and Slc10a4 were also decreased in the mouse and human data. These findings may suggest common mechanisms of calcium dysregulation and altered DA metabolism in an early stage of degeneration in our mouse model and in human PD. Additionally, differential gene expression analysis revealed that Hspa4 and Eif3g were downregulated (FDR<0.02) within the SNc ventral tier DA neurons of early PD subjects compared to controls and in the mouse model (Figure 4—figure supplement 6A), highlighting other potential common pathogenic mechanisms. Additional investigation is required to determine the functional impact of these genetic changes on DA neuron and DA neuron subtype function and vulnerability.

Prior studies have activated midbrain DA neurons by both optogenetics and DREADDs to study the impact of DA neuron activity on sleep (Eban-Rothschild et al., 2016) and the spread of pathological α-synuclein (Helwig et al., 2022). However, in order to accurately model the impact of neural activity on neurodegeneration, we hypothesized that activity must be chronically increased. Unlike other studies which have used once or twice daily injections of CNO, we delivered CNO in the drinking water to ensure a more continuous, sustained impact on neural activity. We confirmed that chronic 1-, 2-, or 4-week hM3Dq activation results in sustained increases in intracellular calcium levels and functional behavioral readouts, suggesting that our model induces long-lasting changes in DA neuron activity. Moreover, ex vivo electrophysiology demonstrated that the basal level of DA neuron activity does indeed increase with CNO treatment. An important question, however, is whether our model accurately recapitulates how neural activity increases in physiological or pathological situations.

Compensatory increased or decreased rates of DA neuron pacemaking may help to maintain striatal DA as it is depleted in PD or to protect against energy failure by decreasing energy consumption. Meanwhile, changes in bursting may reflect circuit-level adaptations to PD in response to ongoing degeneration. Irregular or silenced pacemaking is commonly observed in toxin-based models of PD (Harden and Grace, 1995; Liss et al., 1999; Bilbao et al., 2006; Liss et al., 2005; Yee et al., 2014), while increased induced bursting is observed after massive loss of DA neurons or complete loss of complex I (Hollerman and Grace, 1990; Chen et al., 2009; Stachowiak et al., 1987; González-Rodríguez et al., 2021). These changes could further promote the degeneration of remaining DA neurons. As such, in future studies, it will be important to understand how changes in the pattern of activity (i.e. bursting vs pacing) impact degeneration. In other PD models, DA neuron activity is altered prior to degeneration (Bishop et al., 2010; Good et al., 2011; Lin et al., 2021). The energetic demands of neural activity position altered activity as a possible contributor to the degenerative process in energetically vulnerable DA neurons, either as an initiating force or as a driver of disease progression.

One possible outcome of chronic hyperactivation is depolarization block (DB), a process that occurs even in healthy neurons wherein prolonged excitation induces persistent depolarization and blockade of spike generation (Curtis et al., 1960). DB occurs in midbrain DA neurons following chronic treatment with antipsychotic drugs (Grace and Bunney, 1986), and DB in SNc neurons is specifically associated with extrapyramidal symptoms (Grace et al., 1997; White and Wang, 1983). Lesions of DA neurons with 6-OHDA can also increase susceptibility to DB (Hollerman and Grace, 1989). In our chronic hyperactivation model, there is a strong decrease in dark cycle wheel running behavior that may be consistent with DB (Figure 1C, Figure 1—figure supplement 1B). Prolonged excitation along with progressive degeneration in our model may put DA neurons at greater risk of entering DB. However, CNO-treated animals have increased rather than decreased open-field movement (Figure 3—figure supplement 1B). Our NEUROeSTIMator data also predicts higher activity scores in GqCNO samples compared to GqVEH and CNO alone controls (Figure 4E). Moreover, while DB can’t be assessed in slices due to its reliance on afferent input, we observed increased pacemaking in CNO-treated neurons, further arguing against prominent induction of DB. Given that VTA and SNc neurons can have differential susceptibilities to DB in response to antipsychotics, it is possible that a subpopulation of DA neurons in our model periodically enter DB without gross effects on locomotion. Additional data such as in vivo electrophysiology will be necessary to fully investigate the possibility of DB.

As expected, chronically increasing neural activity increased the movement of mice during the light cycle. However, it was notable that movement as measured by wheel use in the home cage did not increase as robustly as did movement in the open-field chamber (Figure 3—figure supplement 1B), perhaps indicating that the impact on movement is greater in novel environments. This may reflect the effects of midbrain DA neurons on motivation and exploration (Wang et al., 2013; Jing et al., 2019). The later return to control levels of locomotion midway through CNO treatment may indicate the onset of synaptic dysfunction or degeneration. In addition, our electrophysiologic analyses focused on cell body function, but the loss of terminals starting by ≈1 week (Figure 2—figure supplement 1B) raises the possibility that synaptic function might be disrupted even earlier, in parallel to behavioral changes. Meanwhile, chronically increasing activity unexpectedly and robustly decreased movement during the dark cycle (when mice should be most active) within 3–4 days of starting CNO (Figure 1C, Figure 1—figure supplement 1B). One possibility is decreased dark cycle movement results from a disruption in sleep, which might occur given the mild increased use of running wheels during the light cycle. Indeed, prior studies show that acute inhibition or stimulation of VTA DA neurons by optogenetics or DREADDs suppresses or promotes wakefulness, respectively (Eban-Rothschild et al., 2016; Taylor et al., 2016). These data support a critical role for VTA DA neurons in maintaining wakefulness and suggest that increased activity might promote wakefulness in our activity model. Future investigations assessing the impact of chronic SNc vs VTA DA neuron activation on circadian rhythms and sleep are needed to determine if these DA neurons similarly influence sleep in the setting of prolonged stimulation, and how potential sleep changes might contribute to the effects of chronically increased neural activity on motor function.

How does increasing the activity of SNc and VTA DA neurons impact these neurons, as well as their striatal targets? Analysis by spatial transcriptomics near the onset of axonal degeneration reveals a clear decrease in the expression of genes involved in DA synthesis (Th), uptake (Slc6a3), and storage (Slc18a2) in the SNc and VTA, perhaps reflecting neuronal stress that precedes degeneration or an attempt to decrease DA release. Further investigation will be needed to determine whether these gene expression changes are associated with functional deficits. Decreases in presynaptic transmission-related genes (Syt1, Snap25, Snca, Ap2a1) in midbrain DA neurons are consistent with axonal dysfunction and loss that are observed later in CNO treatment. Moreover, consistent with the role of hM3Dq DREADDs in increasing cytosolic calcium (Roth, 2016) on a systems level, we observed a significant change in expression of genes involved in the regulation of calcium channels (Table 1), as well as serum and glucocorticoid-inducible kinase 1 (Sgk1), a protein-serine/threonine kinase activated by calcium (Imai et al., 2003), in both the midbrain and striatum. Our photometry analysis also raises the possibility of a central role for calcium in activity-induced degeneration, as calcium levels in midbrain DA levels rose markedly following 7 days of CNO (Figure 3F), as axonal degeneration progresses (Figure 2—figure supplement 1B).

The mechanism of preferential degeneration remains to be determined but is likely triggered by increased cytosolic calcium levels associated with hM3Dq activation (Roth, 2016). SNc DA neurons are believed to be intrinsically vulnerable to increased calcium due to their pacemaking activity driven by Ca_v1.3 channels (Chan et al., 2007), combined with the high energetic requirements of removing calcium from neurons and lack of calbindin expression (Surmeier and Schumacker, 2013). Further investigation is required to determine if DREADDs differentially increase calcium levels in SNc vs VTA DA neurons, and/or if SNc DA neurons are more vulnerable to the same level of elevation. Further, it remains unknown whether increased calcium leads directly to degeneration, for instance by triggering energy failure or increasing ROS, or if it indirectly causes neurodegeneration by triggering potentially toxic downstream processes such as increased cytosolic or extracellular DA release.

Excessive glutamate leads to large influxes of calcium that can trigger excitotoxicity leading to neurodegeneration and cell death (Szydlowska and Tymianski, 2010). Chronic, mild calcium elevation can also dysregulate mitochondrial calcium buffering and increase sensitivity to cytosolic calcium toxicity (Verma et al., 2022), and the route of calcium entry can also impact toxicity (Sattler et al., 1998). In a pilot study, isradipine, a blocker of Ca_v1.2 and Ca_v1.3 channels, did not confer protection against hyperactivity-induced degeneration in our model (Figure 4—figure supplement 5). However, it is possible that other interventions against calcium-induced toxicity, such as inhibition of NMDA receptors or the mitochondrial calcium uniporter, may be protective. Alternatively, degeneration could be driven by the toxicity of excessive extracellular DA as a consequence of elevated activity. Future studies will explore these potential mechanisms.

If increased DA neuron activity contributes to degeneration in PD, then one might also expect gene expression changes in our model to mirror those seen in PD. Indeed, we found that mouse DEGs exhibited broadly congruent directional changes in the human dataset (Figure 4—figure supplement 6). This raises the possibility that similar mechanisms of degeneration and adaptation may be at play in both our mouse model and in PD. Interestingly, Hspa4 was significantly downregulated in the mouse and human datasets. Heat shock proteins broadly function to remove misfolded proteins, but the function of Hspa4 in neurons is not well described. However, in rat neural stem cells, Hspa4 is upregulated by selegiline, a type B monoamine oxidase inhibitor used to treat PD, and Hspa4 upregulation reduces ROS levels and mitochondrial DNA damage following hydrogen peroxide exposure (Abdanipour et al., 2018; Elkenani et al., 2022). In MPTP-treated mice, deacetylation of Hspa4 is linked to decreased microglial activation and neuroinflammation (Yang et al., 2022). In our activity model, decreased Hspa4 expression might indicate a failure to respond to inflammation, promoting neurodegeneration. Further studies will be required to investigate the role of Hspa4 and inflammation in chronic hyperactivation-induced toxicity.

Notably, we compared our mouse transcriptomic data at a stage when there is mild terminal loss and no somatic degeneration with early human PD patient tissues where there is limited DA neuron loss, raising confidence in the relevance of the comparison. However, it is important to consider that our Visium transcriptomic analysis has limited cell-type resolution. Therefore, important areas of future investigation include higher spatial resolution in the analysis of mouse tissue, and assessment of additional disease stages, brain areas, and cell types in human PD patient tissues.

In some cases, disease proteins or circuit changes may decrease neural activity. As such, it will be important to understand in which contexts activity increases or decreases, and also how decreasing DA activity influences degeneration. Indeed, if activity does promote degeneration, it will also be important to understand if decreasing DA neuron activity is protective, and if activity can be decreased or modulated without compromising motor function. Interestingly, chronic nicotine, which may mediate the protective association of smoking in PD (Quik, 2004; Nourse et al., 2021), inhibits SNc DA neurons through agonism of nicotinic acetylcholine receptors expressed on presynaptic GABAergic terminals (Xiao et al., 2009). In addition, a proposed mechanism of action for the beneficial effects of deep brain stimulation (DBS) in PD is the inhibition of the subthalamic nucleus. As such, it will also be important to know how DBS influences DA neuron activity and if it has the potential to influence neurodegeneration. Interestingly, recent clinical trial data raise the possibility that DBS might slow disease progression when administered to early-stage patients (Hacker et al., 2020).

In summary, our data show that chronically increasing the activity of DA neurons can produce toxicity, and that SNc neurons may be more susceptible to this effect than VTA neurons. Considering that DA neuron activity may increase to compensate for other dying DA neurons, or in response to certain disease proteins, our data support the hypothesis that increased neural activity contributes to the pathophysiology of at least a subset of PD. Proving this, and determining how this evolves with disease stage, will be important goals for future research.

A detailed protocol for collecting fiber photometry data can be found at dx.doi.org/10.17504/protocols.io.bp2l6xwxzlqe/v1. 3-month-old Ai148D mice (RRID:IMSR_JAX:030328) bred with DAT^IREScre were injected with rAAV8-hSyn-DIO-hM3Dq-mCherry as described above. After allowing 3 weeks for expression of the hM3Dq construct, locomotion was tested using IP injection of CNO to determine robust hM3Dq expression. Mice that exhibited increased locomotion following CNO injection were then implanted with optical fibers (400 µM, 0.48 NA). Mice were administered vehicle or CNO water for 14 days.

In vivo calcium data was acquired using a custom-built photometry system. An RZ5P fiber photometry processor (TDT, RRID:SCR_024878) and Synapse software (TDT, RRID:SCR_006307) were used to control LED output and acquire the photometry signal. Using this system, two LEDs were used to control GCaMP and isosbestic excitation (470 nm and 405 nm, respectively, Thorlabs). LEDs were sinusoidally modulated at 211 Hz (470 nm) and 531 Hz (405 nm) and entered a four-port fluorescence mini cube (Doric Lenses). The combined output was coupled to a fiber-optic patch cord (400 µm, 0.48 NA, Thorlabs), which then mated to the fiber-optic cannula in the mouse brain. The emitted light was collected onto a visible femtowatt photoreceiver module (AC low, Newport) and sampled at 60 Hz. Photometry data was then extracted via proprietary TDT software using MATLAB (MathWorks, RRID:SCR_001622). Code for the fiber photometry analysis can be found at https://doi.org/10.5281/zenodo.10819072.

A detailed protocol for open-field analysis can be found at dx.doi.org/10.17504/protocols.io.36wgq33rklk5/v1. Spontaneous locomotor activity was assessed while simultaneously recording fiber photometry data. Videos acquired during photometry sessions were analyzed using Ethovision software (Noldus, RRID:SCR_000441) to calculate total distance traveled and percent of time spent moving.

A detailed protocol for implantation of slow-release pellets can be found at dx.doi.org/10.17504/protocols.io.n92ldr428g5b/v1. DAT^IREScre mice were unilaterally injected in the right midbrain with AAV-hM3Dq as detailed above and in the left midbrain at the same coordinates with 1 μL of sterile saline at 200 nL per minute. Mice were allowed to heal for 2 weeks. Mice were then implanted with either placebo or isradipine slow-release pellets (Innovative Research of America). Pellets were designed to release 3 μg/g/day of isradipine. Briefly, mice were anesthetized with isoflurane, and a placebo or isradipine pellet was implanted subcutaneously, and the wound was closed with sutures. Administration of either vehicle or CNO water was initiated the same day as pellet implantation. Blood was drawn from anesthetized mice and incubated in tubes containing K3 EDTA solution on ice. Tubes were centrifuged at 4°C at 10,000×g for 10 min. Collected supernatant was stored at –80°C until analyzed. Samples were sent to Charles River Laboratories for LC-MS/MS bioanalysis.

A detailed protocol for collecting striatal punches can be found at dx.doi.org/10.17504/protocols.io.dm6gp97xdvzp/v1. Striatal tissues flash-frozen for Visium Spatial Transcriptomics (described below) were used for monoamine quantification. Briefly, flash-frozen brains were embedded in optimal cutting temperature (OCT) and mounted on a cryostat to collect dorsal striatal punches. Punches were stored at –80°C before shipment to the Vanderbilt Neurochemistry Core for catecholamine quantification by HPLC (Berthet et al., 2014).

Animals were anesthetized with 2,2,2-tribromoethanol and perfused with PBS followed by 4% paraformaldehyde (PFA) in PBS. Intact brains were removed, post-fixed in 4% PFA overnight at 4°C, and cryoprotected in 30% sucrose. 40-μm-thick coronal sections were cut on a sliding microtome (Leica) and stored in cryoprotectant (30% ethylene glycol [Sigma], 30% glycerol [Fisher Scientific] in PBS).

A detailed protocol for immunofluorescence can be found at dx.doi.org/10.17504/protocols.io.kxygx38owg8j/v1. Brain sections were rinsed with PBS followed by 0.2% Triton X-100 in PBS. The sections were then transferred to a blocking solution containing 4% fetal bovine serum (JR Scientific) and 0.2% Triton X-100 in PBS for 1 hr. Following overnight incubation in primary antibody, sections were rinsed in 0.2% Triton X-100 in PBS and incubated for 2 hr with a suitable secondary antibody. Sections were rinsed again in 0.2% Triton X-100 before mounting and coverslipping with antifade mounting medium (Vector Laboratories H1400, H1500).

A detailed protocol for peroxidase staining can be found at dx.doi.org/10.17504/protocols.io.n92ldm127l5b/v1. Sections were quenched with 3% H₂O₂ and 10% methanol in PBS and blocked in 10% fetal bovine calf serum, 3% BSA, and 1% glycine in PBS with 0.5% Triton X-100. They were incubated with primary antibody followed by biotinylated secondary and subsequently streptavidin-conjugated horseradish peroxidase (HRP) (1:500; Vectastain ABC kit, Vector Laboratories). Immunostaining was visualized with hydrogen peroxide and 1 3,3′-diaminobenzidine (DAB, Sigma).

The following primary antibodies were used: rabbit anti-DsRed (1:1000; Takara, RRID:AB_3083500), rabbit anti-tyrosine hydroxylase (1:1000; AB152, Millipore, RRID:AB_390204), mouse anti-NeuN (Millipore MAB 377, RRID:AB_2298772), and chicken anti-mCherry (Abcam ab205402, RRID:AB_2722769). Secondary antibodies Alexa Fluor goat anti-mouse 647 (Thermo Fisher Scientific Cat# A-21235, RRID:AB_2535804), anti-rabbit 488 (Thermo Fisher Scientific Cat# A11034, RRID:AB_2576217), anti-rabbit 594 (Thermo Fisher Scientific Cat# A11037, RRID:AB_2534095), or anti-chicken 594 (Thermo Fisher Scientific Cat# A-11042, RRID:AB_2534099) IgG were used (1:500). A biotinylated goat anti-rabbit IgG (1:300, Vector Laboratories, BA-1000, RRID:AB_2313606) was used for peroxidase staining.

Stained samples were visualized using an automated fluorescence microscope (Keyence BZ7000) and a laser-scanning confocal microscope (Zeiss LSM880).

Total numbers of TH-positive and mCherry-positive neurons were quantified by an experimenter blinded to groups. Region selection of SN and VTA was done under ×5 magnification following the definition by Fu et al., 2012. Imaging was done under ×63 magnification by a Zeiss Imager microscope (Carl Zeiss Axio Imager A1) equipped with an XYZ computer-controlled motorized stage and an EOS rebel T5i Digital Camera (Canon), and counting was done using MBF Bioscience’s Stereo Investigator (RRID:SCR_024705). Counting frame size was 60×60 μm², and systematic random sampling grid was 130×130 μm², with a section interval of 6.

A detailed protocol for optical density analysis can be found at dx.doi.org/10.17504/protocols.io.81wgbxo2nlpk/v1. Images of DAB-stained striatal sections were taken with an automated light microscope (Keyence) by a blinded experimenter. ImageJ (RRID:SCR_002285) was calibrated for optical density and subsequently used to draw regions of interest (ROIs) around striatal areas and to measure mean optical density. The Allen Brain Atlas was used as a reference brain atlas (Allen Mouse Brain Atlas, mouse.brain-map.org and atlas.brain-map.org). Optical density values were background-subtracted using an adjacent brain region with low TH expression levels, the piriform cortex.

All statistical analyses including the n, what n represents, description of error bars, statistical tests used, and level of significance are found in the figure legends or corresponding results. A minimum sample size of n=3 mice was used, except for spatial transcriptomics where capture area space was limited. Two-way repeated-measures ANOVA followed by Holm-Sidak multiple comparisons was used for comparing vehicle vs CNO groups. One-way ANOVA followed by Holm-Sidak multiple comparisons was used for comparing multiple groups in mouse spatial transcriptomic data. Three-way ANOVA followed by Holm-Sidak multiple comparisons was used for assessing chronic isradipine histology. t-Test or permutation (nonparametric) analysis was used for ex vivo electrophysiology data. Mouse DEGs were ranked according to the defined SNS and FCS metrics. Congruence between directional changes of mouse and human was quantified using a chi-square test. ‘Limma voom’ methodology was used to assess differential gene expression between control and PD samples, incorporating covariates (see Human Spatial Gen Expression). All analyses were performed using GraphPad Prism version 9 (RRID:SCR_002798), Microsoft Excel, and R version 4.2.2 (RRID:SCR_000432).