USA: +1 508 290 1604 
Sign in
Neural circuit dynamics of timing
Timingis an integral part of an adaptive behavior to predict and prepare for an upcoming eventor action. I am particularly interested in solving how the brain networksprocess timing.Timing relies on the internal dynamics of the brain unlike other sensory inputs coming from the external world. Organisms moveas a prompt response to sensory inputs, but they also act by internally driveninformationsuch as timing. Thus, solving the neural basis of timing will help us to understand the brain better as a place where the internal and external worlds meet.
Multiple theories are existing on where the timing is processed, and among them, the cortico-basal ganglia circuitsare known to play a critical role in timing behaviors, especially for the cognitively controlled interval timing1. The abnormal timing behaviors have been reported in many neurological and developmental disorders such as Parkinson’s disease (PD), schizophrenia, and autism spectrum disorders2,3. Intervention strategies using rhythmic entrainment or music therapy are also known to be beneficial for those patient groups4. However, the neural mechanism of timing processes in the basal ganglia circuits is still not clear.Revealing the neural basis oftiming will improve theunderstanding of circuit dysfunctions in basal ganglia disorders andprovide better framework for the intervention strategies such as rhythm and music treatment or brain stimulations.
- Past research achievement: timing and behavioral inhibition
I had researched timing behaviors during my Ph. D. in rodents and humans using electrophysiology and computational modeling. During my postdoc training, I have extensively researched a behavioral inhibition in the basal ganglia circuits using electrophysiology, optogenetics and a novel population analysis. Thesecombined research backgrounds uniquely position me to make significant contributions tothe understandings of basal ganglia circuit dynamics of timing.
- a) Timing inbasal ganglia disorders and its relation to neural oscillations
Abnormal timing behaviors are commonly observed in basal ganglia related disorders, and my research contributed to a better understanding of specific relationships between the compromised basal ganglia functions and resulting behavioral alterations. Specifically, Ihave shownrepetitive temporal behaviors in a rodent model of obsessive-compulsive disorder5 and distorted timing behaviors in patients with PD,which was explained usingBayesian modeling3.
As an underlying neural mechanism of timing, the role of neural oscillations has beenemphasized.To explain the involvement of internal dynamics and neural oscillations, I developed an oscillatory computational model of timing and working memory. This modelpredicts that both types of information can bemultiplexed in the same brain area with different frequencies of neural oscillations6. Consistent with this idea, I have demonstrated that neural oscillations of cortico-striatal circuits increase during both encoding and comparison of tonedurations using rodent electrophysiology7.
- b) Prepared to inhibit a specific action: GPe population dynamics of internal representation
Behavioral inhibition can be prepared in advance, which we call ‘proactive inhibition’. Proactive inhibition requires internal preparation so that the system can facilitate ‘reactive inhibition’, which refers to apromptbehavioral inhibition as a response to a stop cue.The cortico-basal ganglia circuit, especially the indirect pathway has been suggested tomediate proactive inhibition.
To parse the specific roles of the indirect pathway during proactive inhibition, I recorded the neural activities of rat globus pallidus pars externa (GPe,a core structure of the indirect pathway)while the animals were performinga stop-signal task that was modified to probe proactive inhibition8. Individual neurons of GPe showed diverse activity patterns which made it difficult to interpret the result. Thus, I developed a novel population analysis method of examining the neural dynamics in functionally defined axes. Using the two axes defined for initiation and selection of actions, I was able to dissociate the neural representationsof ‘when’ and ‘what’ to do.
As a result, I revealed that the population neural dynamics ofGPe representinternal preparation and biases. When prepared to inhibit certain actions, GPe population activity positioned farther away from the neural population dynamics ofthe action initiation (Figure 1A). Moreover, the population states during preparation were predictive of distinct types of errors: failures to inhibit, failure to initiate an action and choosing wrong actions (Figure 1B).
Figure 1. A. Trajectories of GPe neural population. Comparison between Maybe-Stop and No-Stop conditions shows that proactive inhibition moves GPe population activity farther from initiation of movement. B. Distinct state-space positions predict distinct types of errors. Compared to the correct trials, failure to go within a limited time and wrong choice started from biased state-space locations at the time of the Go cue.
- c) Proactive inhibition controls the sensorimotor output in the SNr
The internal representation of proactive inhibition should interact with sensory inputs in the brain to produce an adaptive behavioral output. The possible mechanism of this interaction is the crosstalk between multiple cortico-basal ganglia loops. Cortico-basal ganglia loops are thought to be organized in anatomically segregated multiple parallel modules of limbic, associative, and sensorimotor areas9.Associative loops include dorsomedial striatum andmiddle substantia nigra reticulata (SNr)while sensorimotor loops include dorsolateral striatum and lateral SNr (Figure 2). Where and how the parallel loops interact with each other is an essential question to understand how cognition controls behavior.
We hypothesized that these loops can crosstalk in the SNr where the cells are highly interconnected with each other through collateral inhibition. Using a proactive inhibition task in rats, I uncovered that the proactive inhibition is mainly represented in the middle SNr (associative loop) while the lateral SNr (sensorimotor loop) does not show the clear effect of proactive inhibition10. To show direct relation between SNr cells, I used a cutting-edge technique of optogenetic tagging of parvalbumin (PV) negative cells. Optogenetic stimulation of PV negative SNr cellsthat are mainly located in the medial/middle SNr, the activity of typical sensorimotor SNr cells (decrease with action initiation) was inhibited. This finding shows that the proactive inhibition inputs from the associative SNr can control the sensorimotor SNr outputs through collateral inhibitions, and this is a critical finding that shows how cognitive control modulates the basal ganglia outputs. Figure 2. Parallel loops of basal ganglia
- Proposed research:
My lab will focus on the role of cortico-basal ganglia loops in timing behaviors. However, it has been known difficult to dissociate timing components from other correlating factors such as movement. Using my expertise in timing behaviors, I will develop a novel task paradigm that can precisely extract timing components from other behavioral components. Specifically, I will focus on disentangling the circuit dynamics that are related to ‘timed initiation of an action’ in contrast to ‘cued initiation of an action’.This will reveal the neural process of timing by contrasting the actions initiated mainly by internal dynamics of timingand the actions initiatedas a prompt response to a sensory input.
The dissociation of circuit mechanisms involved in timed versus cued action initiation is critically relevant to the pathophysiology of PD. Patients with PDshow difficulties in timing and self-initiating movement while showing less impaired function with cue evoked movement initiation11. Moreover, external auditory cueing is widely established as an effective tool to normalize gait disturbance in PD patients.
To reveal the involved circuits, I will develop a ‘timed or cued initiation task’ for rodents that has two conditions contrasting movement initiated by timing versus cue. To show the circuit dynamics, my lab will use simultaneous multi-area recording, opto-tagging of specific cell types, optogenetic stimulation, and a new population analysis.Knowledge in population analysis will be essential to reveal the circuit interactions due to the complexity and heterogeneity of neuronal responses. My skillset in those methodologies, with expertise in neural responsesin basal ganglia and timing behaviors, will be critically important to find the neural circuit dynamics of timing.The first project will focus on GPe-SNr interaction during timed action initiation in healthy and PD model rats. The second project will reveal the circuit interactions of the SNr-thalamus-motor cortex during timing (Figure3).
Figure 3. Proposed research
Project1. GPe-SNr interactions duringtimed initiation in healthy and PD model rats.
Information from direct, indirect, and hyperdirect pathways converges into SNr cells, and the converged information results in a decrease of SNr cell activities to release an action. Among the inputs, the indirect pathway received significant attention as the ‘no-go’ pathway and suggested being dysfunctional in multiple neurological and mental disorders. Also, there is evidence suggesting that rhythmic timing is represented in GPe. However, surprisingly little is known about the effect of GPe inputs on SNr and its functional role.
Typical SNr cells show a decrease of activation to release a behavior; however, previous literature and my research showed that there are other functional subtypes that show an increase with action initiation. The functional role and circuit mechanisms of this type of cells are not well understood. I hypothesize that those increasing types of SNr cells are strongly modulated by GPe inputs, and they play a critical role in timed action initiation. Furthermore, I hypothesize that those SNr subtypes are the core of impaired action initiation in PD. To support this project, I will apply for a grant to Brain and Behavior Young Investigator Grant, Stanley Fahn Junior Faculty Award and/or Aligning Science Across Parkinson’s collaborative research network.
Aim1-1: SNr cells that receive strong GPe inputs are involved in timed action initiation. To show the GPe and SNr interaction, electrophysiological recordings of SNr will be usedwith optogenetic stimulation of the axonal target of GPe to SNr. I hypothesize that GPe input to SNr plays a more critical role in timed initiation compared to cued initiation. Stimulating GPe input to SNr will prove this if,1) SNr cells that are significantly modulated by GPe input stimulation will show increased activity during timed action initiation unlike typical SNr cells, 2) theseincreasing SNr subtypes will encode timed action initiation more significantly compared to cued initiation, and 3) optogenetic stimulation of GPe input to SNr will facilitate timed movement initiation with stronger effect compared to cued action initiation.
Aim1-2: Change of SNr subtypes and population dynamics in the rodent model of PD. Circuit changes of direct and indirect pathways with dopamine cell loss have been known as the core of the pathophysiology of PD. These circuit input changes will cause changes infiring patterns of individual SNr cells as well as the population dynamics of SNr cellsthroughabundant collateral inhibition. Using 6-OHDA injection rats (PD model) with or without L-DOPA treatment, the activity patterns of a specific SNr subtype as well as the SNr population states will be examined. First, SNr population activity will be recorded during timed and cuedinitiation task with L-DOPA administration. After selecting the functional cell subgroup that are more involved in the timed initiation, their activity patterns without L-DOPA will be examined. Also, it will be tested if the optogenetic stimulation of the GPe input to SNr willrestore the activity patterns of the selected subtype like the L-DOPA treatment state. Population analysis will also be applied to examine the role of the functional subtype in the timed action initiation as well as in the generation of oscillatory patterns.
Project2:Investigate how SNr output modulatesthe cortex through the thalamusduringtiming
SNr integrates multiple inputs and sends outputs to multiple targets, and one of the major targets is the thalamus. Thalamus also sends inputs to the cortex, so the circuits through the cortico-basal ganglia-thalamus-cortex make a closed loop thatcan send feedback information to its origin.Previous studies have examined the neural correlates of timing within a few areas of theseloops; however,timing information processes between areas are rarely known.
To revealthe timing information processes ata circuit level, my lab will focus on revealing the SNr-thalamus-cortex circuits withtiming of action initiation. Examining this specific behavior will show the communication between cortex and basal ganglia through dynamic circuit changes over time.However, each brain area receives multiple circuit inputs, thus entangling the circuit information flow at the neural population level is a complicated process. To solve this problem, I will use simultaneous multi-area recordings with axonal target stimulation and population analyses. To support this project, I will apply for an R01 NIH grant.
Aim2-1. Timing information to motor thalamus mediated by associative or motor SNr? The model of parallel closed loopssuggests separate circuits of middle SNr-associative thalamus-medial prefrontal cortex loops and lateral SNr-motor thalamus-motor cortex loops. However, recent anatomical evidence supports that the motor thalamus receives axonal inputs not only from lateral SNr but also from medial/middle SNr, as a potential locus of crosstalk between parallel closed loops. To understand the communication between SNr and motor thalamus during timing, the middle and lateral SNr axonal inputs to motor thalamus will be stimulated with optogenetics while recording motor thalamuscell activities. With this method, the cells that are only sensitive to the lateral SNr inputs can be separated from the cells that are sensitive to both inputs. These two types of thalamic neurons will be compared to prove which type represents the timing information more strongly.The fining will reveal the specific SNr-thalamus connection delivering timing information.
Aim2-2. Timing information to motor cortex mediated by premotor or motor thalamus?Representation of timing has been reported in the frontal, premotor and motor cortex as in the form of cortical dynamics.The motor cortex receives direct inputs both from the premotor and thalamus, so it is not clear if the motor cortex receives timing information directly from the premotor cortex or through the cortico-basal ganglia-thalamus loops. Using a similar method described in Aim 2-1, the motor cortex cells will be categorized by the inputs from the premotor cortex and thalamus, and the categorized subgroup cells will be compared during timing. Given the known complexity of functions of motor cortex cells, population analysis will also be applied to show the population dynamics of timing in each subgroup of cells.
