Therapeutic deep brain stimulation disrupts movement-related subthalamic nucleus activity in parkinsonian mice

Curation statements for this article:
  • Curated by eLife

    eLife logo

    Evaluation Summary:

    The paper would be of interest to neuroscientists and clinician scientists interested in better understanding the mechanism of deep brain stimulation (DBS) in the treatment of Parkinson's disease. Using a combination of electrical artifact-free calcium imaging and electrical stimulation, it probes the effects of stimulation on the neural dynamics of basal ganglia structures that correlate with motor improvement. The key claims are well supported with a convincing discussion of the caveats of the methods used.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #3 agreed to share their name with the authors.)

This article has been Reviewed by the following groups

Read the full article See related articles


Subthalamic nucleus deep brain stimulation (STN DBS) relieves many motor symptoms of Parkinson’s disease (PD), but its underlying therapeutic mechanisms remain unclear. Since its advent, three major theories have been proposed: (1) DBS inhibits the STN and basal ganglia output; (2) DBS antidromically activates motor cortex; and (3) DBS disrupts firing dynamics within the STN. Previously, stimulation-related electrical artifacts limited mechanistic investigations using electrophysiology. We used electrical artifact-free GCaMP fiber photometry to investigate activity in basal ganglia nuclei during STN DBS in parkinsonian mice. To test whether the observed changes in activity were sufficient to relieve motor symptoms, we then combined electrophysiological recording with targeted optical DBS protocols. Our findings suggest that STN DBS exerts its therapeutic effect through the disruption of movement-related STN activity, rather than inhibition or antidromic activation. These results provide insight into optimizing PD treatments and establish an approach for investigating DBS in other neuropsychiatric conditions.

Article activity feed

  1. Evaluation Summary:

    The paper would be of interest to neuroscientists and clinician scientists interested in better understanding the mechanism of deep brain stimulation (DBS) in the treatment of Parkinson's disease. Using a combination of electrical artifact-free calcium imaging and electrical stimulation, it probes the effects of stimulation on the neural dynamics of basal ganglia structures that correlate with motor improvement. The key claims are well supported with a convincing discussion of the caveats of the methods used.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #3 agreed to share their name with the authors.)

  2. Reviewer #1 (Public Review):

    In this study, the authors aimed to address the important question of the mechanism of deep brain stimulation (DBS) in treating Parkinson's disease, based on a mouse model that the authors established previously.

    The strength of the study lies on 1) avoiding the interference of stimulation artefacts of using electrophysiological recording technique, and 2) examining effects on cell-type or projection-specific targets.

    However, there are several critical problems in this study. First, the low temporal resolution and the averaged population signal (rather than from individual neurons) of the fibre photometry data prevents in-depth enough analysis of the effects of DBS on the target areas to draw useful conclusion. Thus, all interpretations were based on an average rise in GCaMP-reported calcium signals with pretty low temporal resolution. As a result, important readouts that were analysed in many previous studies such as the firing patterns (e.g. rhythmic) or synchrony among neurons were missed by this approach. Take one example. The conclusion that antidromic activation is excluded as a possible mechanism is based partly on the lack of good correlation of the averaged calcium signal with the behavioral improvement. However, such a lack of correlation is also evident in the averaged calcium signal and the improvement in movement behavior under 60Hz and 100 Hz stimulation (Figure 2). While a higher average in calcium signal is observed under 60Hz DBS than 100Hz, the improvement in motor behavior is lower than that induced by 100 Hz DBS. This highlights the severe limitation of the fibre photometry data in revealing the therapeutic mechanism of DBS.

    Second, there is no clear elucidation of the pathological changes revealed by the fibre photometry in PD mice to illustrate what is normal and what is abnormal, and how the DBS rectifies the abnormal changes. For example, when we need to interpret the effect of the DBS on calcium activities in the subthalamic nucleus (STN), the substantia nigra pars reticulala (SNr) and the primary motor cortex (M1), what abnormal GCaMP signal did the authors find, compared with healthy control mice? Without such information, it is difficult to get a sense of what an increase in GCaMP signal in STN, SNr and M1 mean with respect to motor control, and therefore what it means with respect to the effect of DBS. With the specific context of a peak (actually a biphasic waveform) of the calcium activity in the PD anima, it is puzzling that a surge of STN is correlated with movement onset, while in principle it should result in movement termination. Therefore, it is critical to know if there is there such a correlation in healthy animals. If yes, this may not indicate a pathological change that needs to rectified by DBS. If no, how the pathological appearance of such change leads to parkinsonian motor symptoms (akinesia, bradykinesia etc) must be established.

    Third, it is well-known that clinical DBS employed at least 120 Hz stimulation. In fact, the authors had also demonstrated in their previous report that the optimal stimulation frequency in the mouse model is around 180Hz. But the present study utilised clearly suboptimal frequencies (60 and 100Hz only) to address the mechanism. It is possible that different mechanisms or combinations of mechanism may take place under different stimulation frequencies. As such, any conclusion drawn from this study may not represent the whole picture.

    Given the above consideration, I do not think that the authors have achieved the aim of their study, as the results cannot convincingly support their conclusions.

  3. Reviewer #2 (Public Review):

    Schor et al. investigated three theories underlying the mechanism of electrical DBS by studying whether therapeutic DBS inhibits STN or SNr activity, antidromically activates the hyperdirect M1 neurons, or disrupts the movement-related neural activity in the STN. Contrary to the former two theories, they found that therapeutic electrical STN DBS excites STN and SNr neurons as shown by their average GCaMP calcium signals, and the activation of M1 hyperdirect neurons is not required for the behavioral rescue in parkinsonian mice as demonstrated by the continued improvements with removal of M1. They show evidence in support of the 3rd theory that the primary effector of STN DBS is the alteration of the pattern and not the rate of activity in the STN. They show that the attenuation of the neural activity in the STN around movement onset is not only correlated with the best behavioral responses, but that mimicking this activity using optical stimulation is sufficient to improve movement, providing a causal link between STN dynamics and DBS-induced therapeutic benefit.

    • By using calcium-imaging instead of in vivo electrophysiology, the authors are able to identify the acute effects of DBS on the neural dynamics of adjacent structures without having to worry about electrical artifacts. This method provides a technical advancement to the field of pre-clinical DBS studies.
    • Additionally, they provide a rigorous and robust correlation between the STN calcium signals and electrophysiology using both in vivo and ex vivo simultaneous recordings. The correlations in the photometry-based average and single-unit firing activity of basal ganglia nuclei shown for both short and long-time scale interventions like DBS and L-DOPA provide some foundation for the interpretation of the effects of DBS on neural dynamics across experiments.
    • The peak-trough distance in Fig. 6 highlighting the effect of therapeutic STN DBS on the fiber photometry signal, a proxy for the movement-related activity in the STN was an effective and impactful result demonstrating the primary therapeutic mechanism of STN DBS.
    • The discussion addresses key issues of alternate explanations generated by their data as well as caveats of their methods. Specifically, the acknowledgement that the STN and SNr signals could be a cause, or an effect of the movement changes seems like an important acknowledgement for future studies. Additionally, the clarification of basal-ganglia cell-types and limitations with respect to temporal resolution demonstrates an acknowledgement of the complexity of this circuit.

    Minor weaknesses:
    The conclusions of this paper are mostly well supported by data, but some more controls to validate the photometry method, as well as establish baseline firing characteristics could be performed:
    Calcium-imaging controls:
    • The loss of correspondence between spiking and calcium at highest stimulation frequencies produces concern regarding the interpretation of the calcium signals with the "high effect stim" which includes multiple stim protocols with frequency > 100Hz. Added controls with higher frequency stimulation could improve interpretability of the GCaMP signals for DBS at higher frequencies (Fig 1A)
    • In addition to establishing baseline firing dynamics in the STN in parkinsonian mice, identification of the movement-related calcium activity in a healthy control mouse would establish whether therapeutic DBS shifts the activity in the direction of healthy STN firing activity (Fig. 1B). This could enhance the relevance of the finding that the movement-related neural dynamics in the STN are abolished with DBS.
    Variability in firing between STN and other sites:
    • Comparison from WT mice with synapsin-GCaMP in the SNr vs. Cre-dependent GCaMP was unclear
    • Validation of correspondence between calcium imaging and in vivo firing activity only performed in the STN. Similar control experiments in the SNr could be valuable since it has a different baseline firing rate.
    Evidence against theory suggesting STN inhibition as a mechanism of DBS:
    • To further establish that the inhibition of STN activity was not correlated with the behavioral improvement, a correlation between the movement velocity and the change in STN activity could be shown like shown for the M1 analysis (SFig. 5E).
    • Interpretation of Fig 7G and K suggesting that the attenuation of movement-related dynamics with the 50 Hz protocol is causal to the velocity increase is confounded by the inhibition of the firing rate (Fig. 7E).
    Optical Stimulation Setup:
    • The discrepancy between the in vivo and ex vivo firing rate response to 50 Hz optical stimulation could be further explained or experimentally explored in order to make the findings consistent with previously used in vivo validation (Fig 1). Using different frequencies of optical stimulation could identify whether there was frequency-dependence to the effect.

  4. Reviewer #3 (Public Review):

    In this study the authors use GCaMP6s fiber photometry to determine neural activity in the subthalamic nucleus (STN), substantia nigra pars reticulata (SNr) and STN projecting neurons in motor cortex (M1-STN) while therapeutic electric deep brain stimulation (DBS) is delivered via electric stimulation pulses to the STN of mice rendered parkinsonian with unilateral 6-OHDA lesions. This technique avoids electrical stimulation artifacts obscuring the effects of DBS on neural activity that makes electrical recordings during DBS problematic. The authors find that activity levels in STN and SNr are increased during therapeutic DBS. They also find that equally therapeutic L-DOPA treatment as measured by locomotor speed improvement leads to a reduction in STN and SNr activity. Overall, these findings contradict a rate coding model as underlying therapeutic DBS or indeed parkinsonian motor deficits. Another candidate mechanism for DBS given by antidromic activation of STN projecting neurons in motor cortex is partially refuted by finding inconsistent activity changes in M1-STN neurons during DBS and importantly the maintenance of therapeutic DBS effects on locomotor speed after ablating motor cortex unilaterally. In contrast a consistent mechanism correlated with therapeutic effects correlated across all conditions and treatments was the suppression of locomotor related activity increases in STN. The authors then causally assessed this mechanism by optically stimulating glutamatergic STN neurons with 50 Hz pulse trains which also improved locomotor speed and suppressed locomotor activity increases in STN as assessed by electrical recordings in this case.


    This study directly addresses an important discussion on the mechanism(s) by which DBS acts. It assesses activity levels during DBS in 3 key areas (STN, SNr and M1-STN) and provides some of the clearest evidence yet that the traditional rate model of parkinsonism is not suitable to explain the effect of STN DBS.

    By directly comparing levodopa treatment and STN DBS under matching conditions, the opposite rate changes seen with these treatments while giving a similar therapeutic effect provide convincing key evidence.

    A key insight is also given by the maintenance of locomotor speedup with DBS after ipsilateral M1 ablation. While it is known that spontaneous mouse locomotion does not depend on cortex, this finding provides clear evidence for direct STN DBS effects on descending pathways.

    The mechanism of DBS action most consistent in the recordings across conditions that provide therapeutic effect is a suppression of locomotor related activity increases in STN. This mechanism is directly tested by optical stimulation of STN neurons which indeed is sufficient to exert a therapeutic effect.


    The assessment of therapeutic efficacy remains limited to locomotor speed changes in mice. It is well known that Parkinson's Disease (PD) patients have multiple symptoms that are differentially responsive to different DBS locations or in some cases not responsive to DBS. The current findings strictly only apply to effects of unilateral 6-OHDA lesions on spontaneous locomotor speed. As this therapeutic effect is not dependent on cortex, it may or may not transfer to cortically dependent parkinsonian deficits in humans.

    The assessment of antidromic activation is limited to M1, while STN projections also emanate from more frontal premotor and more posterior sensory areas of cortex.

    As acknowledged by the authors, therapeutic effects were only validated for 1 min periods. Long term effects could differ.

    While fiber photometry overcomes the problem of electrical stimulation artifacts, it loses cellular resolution and has low temporal resolution. Important aspects of previously proposed pathological dynamics of STN activity such as bursting and beta oscillations could not be assessed. Therefore a potential primary mechanism of DBS given suppressing such dynamics could be missed. Endoscopic imaging with single cell resolution would provide a stronger approach.