Assessment of the electrophysiological consequences of sustained dopaminergic stimulation over cortical and striatal projections

MSc Brain and Cognitive Sciences

Second research project

Assessment of the electrophysiological consequences of sustained

dopaminergic stimulation over cortical and striatal projections

Mariana Duque Quintero

11617519

Host laboratory:

Synchronization in Neural Systems Laboratory

Donders Institute for Brain, Cognition and Behavior

August 2019 37 ECTS

Supervisor and First examiner: dr. Paul Anderson Second examiner: dr. Conrado Bosman

Abstract

Dopamine is an important neuromodulator in the brain, acting through widespread connections from the midbrain to striatal, cortical, and limbic areas, with consequences for cognition and behavior. Dopaminergic signaling can occur in a time effective manner with the potential to exert rapid control over the oscillatory activity of the innervated areas, driving intra and inter-area synchronization. The present work aimed to do an assessment of the consequences of the sustained stimulation of midbrain dopaminergic neurons at rhythms ranging from 1 to 20 Hz, on the activity of the prefrontal cortex and dorsal striatum, in terms of the oscillatory activity elicited by the stimulation and the occurrence of cross-frequency coupling. The assessment showed that the stimulation results in multispectral changes in oscillatory activity, where the largest effects occurred in the theta and beta band, with little variance between conditions and areas. Due to the structure of the multispectral result, where power changes appear to occur at the harmonics of theta, it is possible that the main effect of the stimulation occurred in this band, with a signal that did not behave as a sinusoidal. Furthermore, the cross-frequency coupling analysis showed that intra-area phase-amplitude coupling was not a consequence of the stimulation for the channels and conditions that were selected for the test. Methodological considerations are given for further analyses, suggesting that it would be important to describe the shape of the signal to determine what type of activity is elicited by the stimulation. Another methodological consideration was to do an exploration of the activity in the time and spatial domain to reveal other effects of the stimulation with spatial specificity or transient characteristics.

Introduction

Dopamine (DA) is a neuromodulator well known for its multiple functions in cognition and behavior. The roles of dopaminergic signaling have been elucidated by studying the dysfunctions that surge from dopamine depletion. These dysfunctions affect movement, motivation and learning (Schultz, 2007).

One of the means by which dopaminergic signaling controls behavior is through the rapid activation of dopaminergic neurons in the midbrain in the form of bursts. These bursts, occurring in the ventral tegmental area (VTA) and substantia nigra pars compacta (SNc), are able to induce rapid behavioral reactions, by the release of dopamine at the sub-second scale (Berke, 2018; Schultz, 2007). An important expression of the “bursty” activation of dopaminergic neurons is the response to reward predictive stimuli. Bursts of this type are in the order of 10 to 50 pulses, occasionally exceeding 100 pulses. Bursts from dopaminergic neurons can also occur spontaneously with discharges of 3 to 4 pulses, separated by long intervals (Schultz, 2007).

The bursty activation of dopaminergic neurons, also called phasic firing, is thought to be key for the rapid signaling of relevant events in the environment, which is thought to afford learning. In contrast, firing can also be tonic with only a few spikes per second, which has traditionally been thought to afford sustained motivation (Berke, 2018). This differential dopamine-dependent modulation of behavior is thought to be achieved by the selective activation of D1 receptors, when bursts rapidly increase the concentration of dopamine, or D2 receptors when the tonic activation generates low “background” levels of dopamine (Berke, 2018; Schultz, 2007). According to this division between phasic and tonic activation of dopaminergic neurons, it has been concluded that dopamine is a neuromodulator working at different timescales (Schultz, 2007), although this idea has been recently challenged by the suggestion that dopamine always modulates the excitability of the receptor neurons in a rapid fashion (Berke, 2018).

Another mean by which dopaminergic signaling controls behavior is by the action of DA neurons with widespread connections to striatal, cortical and limbic regions. At these innervations, dopamine can act with high temporal precision. Dopaminergic projections run through three main pathways; the nigrostriatal, mesolimbic and mesocortical. The nigrostriatal connects the SNc and the striatum, while the mesolimbic and mesocortical connect the VTA to limbic and cortical areas, respectively (Björklund & Dunnett, 2007; Haber, 2016). Neurons from the VTA/SNc complex have also been shown to have intermixed projections, where striatal DA innervations come from the VTA, and limbic and cortical DA projections come from the SNc (Björklund & Dunnett, 2007). This means that midbrain dopaminergic projections are not strictly confined to the main pathways. Additionally, dopaminergic projections are highly organized, where dopamine neurons affect spatially restricted domains in the cortico-striatal circuit. This organization is thought to have a functional role (Haber, 2016).

Considering the widespread connectivity of dopaminergic projections, and considering that these neurons can exert rapid actions over the innervated areas through phasic firing, it can be thought that dopaminergic neurons have the potential to modulate neuronal function in a time effective manner at a mesoscopic level. This means, exerting control over the activity of a neural population, which expresses as neuronal oscillations.

Neuronal oscillations occur due to the rhythmic synchronization of a neuronal population, where neuronal excitation is followed by inhibition in a cyclical manner. This sequence of activity creates short temporal windows that focus spike output, and sensitivity to presynaptic input, to time windows that can be predicted from the oscillatory activity of the population (Fries, 2015). Information is successfully integrated when presynaptic inputs consistently arrive when the receiving population is excitable. This is thought to be facilitated by the entrainment of pre and postsynaptic rhythms, allowing neuronal communication (Draguhn & Buzsaki, 2004). While communication can occur when an oscillation in a sending group entrains to an oscillation that is intrinsically generated in the receiving group, a sending group can also drive the oscillatory activity of a receiving group (Fries, 2005). Thus, one way in which dopaminergic neurons could control function at a mesoscopic level is by modulating the oscillatory activity of the neuronal population that receives the dopaminergic input.

To ease cross-structure communication through the proposed mechanism, the action of dopaminergic neurons should be able to generate oscillatory rhythms in the area of influence, and at the same time, should be able to send its outputs in a synchronized manner. In vitro work has shown that local rhythms can be originated in the neocortex by the action of neurotransmitter signaling at an ample range of frequencies (Roopun, 2008). However, the frequencies of dopaminergic activation that the innervated areas are responsive to, are unknown.

Regarding the synchronization of dopaminergic output, a recent study showed that neurons in the VTA exhibit rhythmic firing at a rate between 3 to 6 Hz (Duvarci et al., 2018). This rhythmic firing which occurred during a task demanding working memory, was associated to dopaminergic activity, since working memory function has shown to be dependent on dopamine (Graham V. Williams & Patricia S. Goldman-Rakic, 1995). Rhythmic firing of dopaminergic neurons can also be artificially induced by optogenetic stimulation, with consequences for behavior. Using laser pulses at rates from 1 to 50 Hz targeting dopaminergic neurons in the VTA, researchers generated light evoked firing in these neurons, which was entrained to the laser trains (Tsai et al., 2009). Interestingly, when stimulation was delivered at a phasic rate of 50 Hz, animals developed place preference conditioned to the optogenetic stimulation. In line with this evidence it can be thought that synchronized activity of dopaminergic neurons may have a role in the control of behavior through dopaminergic transmission.

One way in which dopaminergic signaling appears to influence the population activity of the innervated areas is by inducing cross-frequency coupling, particularly between the phase of a low frequency oscillation and the amplitude of a higher frequency (Fujisawa & Buzsáki, 2011; Lohani, Martig, Deisseroth, Witten, & Moghaddam, 2018; López-Azcárate et al., 2013). This form of coupling is said to establish a regulatory structure that serves inter-area communication, where the low frequency phase carries information about local neuronal excitability, and the amplitude of the higher frequency carries information about a general increase in population synaptic activity (Canolty & Knight, 2010). Thus, influencing cross frequency coupling would mean influencing information processing at different spatiotemporal scales, by the coordination of oscillatory activity at different frequency bands.

Evidence about this form of coupling in the cortico-striatal circuit was given by a study showing that for a working memory task, there was a cross-frequency entrainment in the prefrontal cortex between a slow oscillation at 4 Hz, and a fast oscillation at gamma frequency (30-80 Hz) (Fujisawa & Buzsáki, 2011). In a more causal test, the VTA was optically stimulated for 5 seconds at a rate of 20 Hz, leading to phase-amplitude coupling between low theta (4.5 to 8 Hz) and high theta (8 to 13 Hz) with high gamma (55 to 100 Hz) (Lohani et al., 2018). Similarly, theta (~6 Hz) to high gamma (50 to 100 Hz)

phase-amplitude coupling was also observed across various cortico-striatal structures of the nigrostriatal pathway after the injection of the dopamine agonist apomorphine (López-Azcárate et al., 2013). These structures were the motor cortex, caudate putamen, subtalamic nucleus and substantia nigra pars reticulata. Interestingly, phase amplitude coupling was also observed for the saline condition and the condition treated with dopamine antagonist haloperidol. Yet the apomorphine injection shifted the modulating frequency from delta (~3 Hz) to theta (~6 Hz), with respect to the other two conditions. These results indicate that dopaminergic activity, either due to a task demand, the direct stimulation of the VTA or a pharmacological manipulation, is able to generate cross-frequency entrainment in cortico-striatal structures. Also, they indicate that the entrainment can express between delta or theta, and gamma. However, exactly what type of dopaminergic activity is able to generate these patterns is not yet clear.

Based on the previous studies, it has been hypothesized that a key function of phasic dopaminergic activation is to create temporal windows for synchronization in frontal and striatal areas. In line with this hypothesis, the present work aimed to do an initial assessment of the consequences of sustained stimulation of the VTA/SNc at rhythms ranging from 1 Hz to 20 Hz, over the population activity of two innervated areas: the prefrontal cortex and the dorsal striatum. The consequences of the stimulation were measured in terms of the oscillatory activity elicited by the stimulation, and the occurrence of cross-frequency coupling.

This work presents an analysis of the oscillatory activity that is generated in the dorsal striatum and the prefrontal cortex by the stimulation of the VTA/SNc at this range of frequencies, by giving a description of the power changes that arise after the stimulation. Additionally, this work presents a preliminary test of intra-area phase-amplitude coupling during the time of optical stimulation, for the stimulation conditions that showed the biggest modulatory effects.

Materials and methods

Animals:

Animals were male, tyrosine hydroxylase (TH) cre-recombinase (Cre) rats with a Long-Evans background. These rats express Cre recombinase under the control of TH, the enzyme necessary for dopamine synthesis (Witten et al., 2011) Rats were maintained on a 12/12 hour light/dark cycle (lights on at 07:00), with ad libitum access to food and water. Experiments began at an age of 8-9 weeks. The recording sessions were conducted during the light phase of the daily cycle. All experiments were performed according to the Dutch national guidelines on the conduct of animal experiments.

Surgical Procedures:

Viral infusions and electrode implants were carried out in two separate surgeries (a minimum of three weeks apart). Rats were anesthetized with 5% isoflurane (2-3% maintenance, 1L/min equal parts 02:

(0.5 mg/kg) and Xylocaine (no more than 0.4 ml) under the skin at the incision site. The top of the skull was exposed and holes were made for viral infusion needles or electrode implantation, as well as skull screws. At the end of all surgeries, rats were removed to a heating pad and monitored until they were ambulatory. Rats were monitored daily for one week following surgery and recordings began at least 4 weeks after viral injection surgery.

 Viral injections:

The virus AAV-EF1a-DIO- hChR2 (H134R)-EYFP-WPRE (titer > 1 e12 _{particles/ml, University of North}

Carolina Vector Core) was injected in TH:Cre rats with an age of 8 weeks. The virus was targeted to the ventral tegmental area (VTA) and the substantia nigra (SNc), in the right hemisphere. Viral injection co-ordinates were: + 0.8 & + 1.8 mm medial-lateral (ML), -5.2 & -6.0 anterior-posterior (AP) and 8 mm dorsal-ventral (DV), (total of 4 injections, 1 uL volume each, all coordinates relative to bregma.) Virus was infused at a rate of 0.1 ul/min, the needle was left in place for 5 mins post injection, raised 200 microns and left another 5 mins, then removed slowly.

 Electrode Implantations:

Custom made high-density silicon probes (NeuroNexus, Ann Arbor, USA) with four laminar shanks (6 mm long), having 16 recording sites per shank, were used to acquire multi-unit, and single unit spiking activity and the local field potential in vivo. Each probe was designed to target one of three areas in the right hemisphere: the prefrontal cortex, the striatum and the VTA. Fiber optic stubs (100 um diameter, 0.37 NA, Doric Lenses, Quebec, Canada) were used for optical stimulation of the VTA and SNC, implanted alongside the VTA electrode at a distance of 0.5 mm posterior and 0.5 mm dorsal to the tip of the electrodes.

Rats were implanted with the silicon probes by performing a craniotomy (+0.5 ML, +3.2 AP from bregma for the probe targeting the PFC, +2.0 ML, +1.0 AP from bregma for the probe targeting the striatum and +1.2 ML, -5.6 AP from bregma, for the probe targeting the VTA). Electrodes were referenced to a screw implanted over the cerebellum. 5 skull screws along the left hemisphere were also wired to provide ECoG recordings. The probes were fixed to the cranium with dental cement, and surrounded by a copper mesh Faraday cage (Vandecasteele et al., 2012). After recovery (minimum 7 days) electrodes were lowered in the dorsal-ventral axis to reach the target areas: 4 mm DV for PFC, 5 mm DV for striatum, and 8 mm DV for the VTA.

Electrophysiological recordings and optical stimulation:

For the optical stimulation of the VTA/SNC, 473 nm laser light was delivered to the targeted areas using fiber optic cables (100 um diameter, 0.37 NA, Doric Lenses, Quebec, Canada). Light pulses of 10 ms each were delivered during 2 seconds using frequencies between 1 and 60 Hz. The power of the laser, measured at the fiber optic tip was of 10-12 mW.

Stimulation was done in four blocks of 4 to 8 minutes. The first block consisted of 11 stimulation conditions between 1 and 6 Hz (spaced by 0.5 Hz), the second block, of 7 stimulation conditions between 8 and 20 Hz (spaced by 2 Hz), the third block, of 5 stimulation conditions between 22 and 30 Hz (spaced by 2 Hz) and the last block, of 5 stimulation conditions between 35 and 60 Hz

(spaced by 5 Hz). For each block, 10 trials of each condition were obtained. Trials within a block were randomized and separated by an intertrial period of 3 seconds.

Optical stimulation was paired with electrophysiological recordings, using OpenEphys software and Hardware, with a sampling rate of 30,000 Hz. A recording session lasted 40 minutes, adding a pre-stimulation period of 5 minutes, the stimulation blocks, and 4 minutes inter-block periods of no stimulation. Recordings were done in awake, behaving animals, during the light cycle, in a small recording chamber (30 x 40 cm). 15 session were done, one per day for two animals.

Data pre-processing:

Continuous data from each channel was imported into Matlab, downsampled to 1 KHz, and epoched into 4 second trials, with a 1 s pre stimulus and 3 s post-stimulus period (relative to laser onset).

The data from each session was visually inspected and cleaned from artifacts through trial and channel exclusion. Trial exclusion was blinded to the type of optical stimulation that was used. After cleaning, approximately 120 trials were kept per condition and 106 channels were used for analyses (58 prefrontal channels and 48 striatal channels). Trials over 15 sessions were combined and sorted into 18 conditions, between 1 and 20 Hz.

Additional sources of noise were identified and removed from the data through independent component analysis, using the Jade algorithm. The data was first downsampled to 250 Hz, and between 70 and 90 components were generated. Noise containing components, identified through visual inspection, were eliminated. Finally, common average referencing was applied, by subtracting the mean activity of all channels on the same electrode from the activity of each channel.

Time-Frequency decomposition for spectral analysis:

To assess the effects of optical stimulation of the VTA/SNc in terms of the oscillatory activity elicited in the prefrontal cortex and the striatum, time-frequency decomposition was performed, by convolving the signal with a set of complex Morlet wavelets (CMW), which are complex sine waves tapered by a Gaussian. Wavelets were generated using the following formula (Cohen, 2018):

𝑤 =

𝑒

_𝑒

−4ln (2)𝑡2 ℎ2

Where i is the imaginary operator, f is frequency in Hz, t is time in seconds and h is the full width at half maximum (FWHM) in seconds. The frequencies of the wavelets ranged between 1 and 100 Hz, with 80 wavelets increasing logarithmically in frequency within this range. Test frequencies is the term that will be used to refer to these frequencies. To set the width of the Gaussian, the full width at half maximum (FWHM) was fixed to 1 second for all wavelets, which corresponds to 2.6 cycles for the 1 Hz frequency, and 266.8 cycles for the 100 Hz frequency, according to the following formula:

𝑛 = ℎ𝜋𝑓 √2 𝑙𝑛 2

Where n is the number of cycles. Wavelets were normalized by setting the amplitude of the Fourier transform to have a peak of 1. After averaging power for trials, post-stimulus power values were

referenced to a pre-stimulus baseline -500 to -200 ms before the onset of stimulation. The change in power was converted to a decibel scale (dB).

Detection of peaks in power across the spectrum:

In order to determine the specific frequencies that had the largest response to optical stimulation the spectral power was averaged for all the channels of an electrode, in the timeframe 500 to 2000 ms (post laser onset) and for all trials of the same condition to obtain a single average power spectrum per condition and area. This averaged power spectrum was used to detect local peaks in power along the frequency spectrum from 1 – 100 Hz (as illustrated in figure 1) A peak in power was defined as a data sample which is larger than its two neighboring samples.

Figure 1. Three example power spectra obtained for the striatal channels using three stimulation conditions (1, 1.5 and 2 Hz).

Local peaks in power were detected along the frequency spectrum for each stimulation condition and each area (18*2). This data was used to describe the spectral effects in the PFC and the striatum of the stimulation using stimulation between 1 and 20 Hz.

Phase-amplitude coupling analysis:

Intra-area phase-amplitude coupling was tested on the channel that showed the third largest change in power on the peak frequencies per area, per condition, from the spectral decomposition analysis. This channel was selected instead of the first to reduce the chance of picking a channel with artefactual power changes. A time frequency decomposition using complex Morlet Wavelet convolution was performed to extract power from a frequency range between 15 and 80 Hz (in steps of 2 Hz) and to extract phase from a frequency range between 4 and 12 Hz (in steps of 1 Hz), on the selected channels of both areas. For the low frequency range, the width of the Gaussian was set to 5 cycles, which corresponds to a FWHM between 0.4 and 0.16 seconds, while the number of cycles for the high frequency range was defined within a range from 5 to 15 cycles, increasing linearly with an increase in frequency. This corresponds to a FWHM between 0.12 and 0.07 seconds.

For every low and high frequency combination a debiased phase-amplitude coupling measure was calculated using the following formula:

𝑑𝑃𝐴𝐶 = |1 𝑛∑ 𝑎𝑡(𝑒 𝑖𝜑𝑡_{− Φ)} 𝑛 𝑡=1 | Where Φ is defined by:

Φ = 1 𝑛 ∑

𝑒

𝑖𝜑𝑡 𝑛

𝑡=1

Here, n signifies the total number of time points, 𝑎𝑡 the amplitude at every time point of the time series for amplitude, and 𝑒𝑖𝜑𝑡 the vector at every time point of the time series for phase . dPAC is the magnitude of the average vector resulting from this operation. dPAC was calculated for the period between 500 and 2000 ms after the onset of stimulation.

The PAC measure is debiased by the subtraction of the average vector obtained from phase clustering of the time series for phase, from the euler transform of each phase-angle. The advantage of this method is that it corrects for the chance of finding spurious coupling when the signal is non-sinusoidal. This is done by subtracting phase clustering in time, which is different from 0 when the signal has this property (van Driel, Cox, & Cohen, 2015).

The selected striatal and frontal channels were combined in order to obtain: 1) intra-area PAC in the striatum, 2) intra-area PAC in PFC, 3) inter-area PAC with phase in the PFC and power in Striatum 4) and inter-area PAC with phase in the Striatum and power in the PFC. dPAC was averaged across trials of the same condition.

The observed dPAC values were contrasted to a null distribution of dPAC values that was generated by non-parametric permutation testing using 1000 surrogates. A standardized measure of dPAC was obtained, representing the distance from the distribution of phase-amplitude coupling expected by chance. Surrogates were generated for each channel pairing by re-sorting the power time series on a random time point per trial, while keeping the phase time-series unaltered.

Results

Spectral analysis

In order to assess the consequences of optical stimulation of the VTA/SNc complex on the oscillatory activity of the striatum and the prefrontal cortex, time-frequency decomposition was performed using high temporal smoothing to obtain high frequency precision. Then a power spectrum was generated for each condition and area (Supplementary figure 1 and supplementary figure 2), by averaging for channels, trials and a timeframe between 500 and 2000 ms post-stimulation. Power values are relative to the baseline period -500 to -200 ms before stimulation.

Power spectra were used to detect local maxima in power along the frequency range 1- 100 Hz. The frequencies that showed positive peaks in power were selected to describe the spectral effects

of the stimulation and to determine which frequencies had the largest power changes. Figure 2 summarizes the results of this analysis: along the y axis are the peaks that were detected from each power spectrum. Peaks are sized relative to the maximum power value that was found across the two areas (5.6 decibels in the striatum, to the stimulation condition of 5 Hz).

Figure 2 Scatter plots of the peak modulated frequencies between 1 and 100 Hz, for the prefrontal cortex and striatum. The

peak frequencies were taken from the power spectrum, obtained for each condition and area. In orange are the maximum frequency peaks for each stimulation frequency relative to the baseline (-500 to -200 ms), occurring in the range between 12 and 100 Hz. In blue are the maximum frequency peaks occurring in the range between 2 and 12 Hz. In black are other peaks detected in the power spectra for each condition with a power that is lower than the maximum. The size of the circle represents the size of the power change, relative to the size of the maximum peak across areas and conditions (striatum, 5Hz optical stimulation, 33 Hz peak, 5.6 dB).

The maximum peaks in each area and condition are given as an indication of where in the frequency spectrum the effect of the stimulation was the largest. Figure 2 shows that the maximum peaks were located between the beta and gamma band for both areas. To identify the maximum peaks restricted to the lower frequency bands, two ranges were defined, separating the test frequencies in a low range between delta, theta and alpha (1 to 12 Hz), and a high range between beta and gamma (12 to 100 Hz). Based on this division, the largest effects of the stimulation is set around the beta (12-30 Hz) frequency for the higher range, and alpha (8-12 Hz) for the lower range, in both the prefrontal cortex and the striatum, considering the median and the interquartile (IQR) ranges (table 1). Altogether, the effect across conditions showed little variance, considering the distribution of the peaks in the high and low range.

High range (12 to 100 Hz) Low range (12 to 100 Hz) Peak in

frequency /area

Median and IQR Mean and SD weighted by power Median and IQR Mean and SD weighted by power PFC 17.4 (23.2 – 16.4) 21 ± 6.6 Hz 8.6 (9.2 – 8.1) 8.8 ± 1.4 Hz Striatum 17.9 (19.5 – 16.4) 21.8 ± 8.7 Hz 8.1 (9.2 – 6.8) 8.1 ± 2.2 Hz

Table 1 Median and mean weighted by power of the major peaks across conditions, in the ranges between 12 and 100 Hz

and between 1 and 12 Hz. Medians are given with the interquartile range (IQR) and weighted means are given with the weighted standard deviation.

The IQR was restricted to beta for the range between 12 and 100 Hz. Yet, the prefrontal cortex showed a skewed distribution towards the higher values of beta. Interestingly, when considering the means weighted by power, this central value was 4 Hz over the median for both areas. This means that the distribution of weights was not even within the beta band, especially for the striatal channels which showed a more spread weighted standard deviation. This can be observed in figure 2, where the size of the peaks in the beta band tends to differ, with an apparent increase to the higher values. This means that although the distribution of maximum peaks is centered to 17 Hz, the maximum peaks above the median tend to weight more. As with the range between 1 and 12 Hz, the maximum peaks were somewhat central between 8 and 9 Hz, considering the median and the weighted average, although with a distribution skewed to the lower values in the case of the striatum.

Although the spread of values showed differences between the prefrontal cortex and the striatum, these were not significant. A Wilcoxon Signed-Ranks test indicated the maximum peaks in frequency across conditions, in the range between 12 and 100 Hz were not statistically different from the same peaks in the striatum (Z = 0.6, p = 0.5). Similarly, the same test showed that the peaks in the range between 1 and 12 Hz were not statistically different between areas (Z = 1.7, p = 0.08).

What does differentiate the effect of the stimulation over the prefrontal cortex and the striatum are the power values of the peak frequencies. In the prefrontal cortex, the maximum decibel change in the range between 12 and 100 Hz (where the biggest changes were observed) had a median of 2.4 (IQR= 2.9 – 1.9). Meanwhile, the maximum decibel change in the same range for the striatum had a median of 3.1 (IQR= 4.2 – 1.94). A Wilcoxon Signed-Ranks test showed that the maximum decibel change was significantly different between areas, when pairing by conditions (Z = 1.7, p = 0.003). This means that the effect of the modulation was stronger for the striatal channels.

Given that the modulation in terms of power was larger for the striatal channels, the results from this area were used to detect which were the conditions that showed to be more modulated. For the striatal electrodes, the biggest modulation was concentrated in the stimulation frequencies below 5 Hz, since the conditions that had maximum peaks above the third quartile were those using trains of 1.5, 2, 4, 5 and 12 Hz.

Figure 3 Spectral changes in the prefrontal cortex and the striatum after the optical stimulation at 5 Hz, with respect to a

baseline obtained for all trials of each condition (-500 to -200 ms). See figure 4 for plot details. Peak frequencies are 35 Hz for both electrodes.

Figure 4 Spectral changes in the prefrontal cortex and the striatum after the optical stimulation at 12 Hz, with respect to a

channels of an electrode, for the test frequencies between 1 and 100 Hz (PFC above, striatum below). Power changes are averaged on time (500 to 2000 ms). Channels are sorted in ascendant order, based on the power values on the peak frequency for area and condition (35 Hz PFC and 18 Hz striatum). Figures in the middle display time-frequency changes, averaged across all channels in an area. Optical stimulation begins at time 0. Figures on the right display the power spectrum on the time from 500 to 2000 ms, averaged for channels and trials.

The stimulation conditions using trains of pulses at 5 Hz and 12 Hz were then selected for further analyses, considering that stimulation at 5 Hz brought the highest modulation in the lower frequencies and that stimulation at 12 Hz is the condition that showed the highest modulation for the higher frequencies of stimulation.

Beginning with a more fine-grained inspection of the results from the spectral analysis, figure 3 and 4 gives the power changes by channels, by time, and in the power spectrum, along the frequency spectrum between 1 and a 100 Hz, for the optical stimulation at 5 Hz and 12 Hz.

An overall inspection of the power changes mapped by channels (figure 3 and 4, first column), considering both conditions, shows that activity in the prefrontal electrode is more localized to certain channels than the activity in the striatal channels. In the striatal channels, power changes appear to affect all locations, especially in the frequency where a power maximum was observed.

Yet, there is still homogeneity within the prefrontal channels because the local response occurred in the same frequencies between beta and low gamma, with a similar strength for all channels (figure 3 and 4, first column, above plots). For instance, for the stimulation condition at 5 Hz, channels 40/50 show at 17 Hz, power increases of 1.5/1.7 dB, at 26 Hz, power increases of 1.8/1.9 dB and at 35 Hz, power increases of 1.9/2,3 Hz. For these three frequencies, the maximum change was of 0.4 dB (figure 3, first column, above). On the other hand, although the striatal channels seem to give a global, similar response, there is still some degree of specificity on the frequencies that are higher or lower from the frequency where a power maximum was observed (figure 3 and 4, first column, below plots). For instance, for the stimulation condition at 5 Hz, channels 21/28 show at 17 Hz, power increases of 2.9/1.7 dB, at 35 Hz, power increases of 4.3/4.9 dB, and at 44 Hz, power increases of 1.3/3.5 dB. Then, while the change in the maximum peak frequency (35 Hz) is of 0.6 dB, the change in the lower frequency of 17 Hz is of 1.2 dB and in the higher frequency (44 Hz) of 2.2 dB (figure 3, first column, below). With regards to the power changes in time, one main observation that arises from an overall inspection considering both conditions and areas (figure 3 and 4, second column) is that despite the high temporal smoothing, there are clear changes in power occurring after the onset of stimulation, relative to the pre-stimulation period. Power changes confined to the time after stimulation were to be expected, as an expression of the effectiveness of the intervention. These power changes are distributed between alpha, beta and low gamma, which can also be observed with the peaks in the power spectrum (figure 3, third column).

Interestingly, these peaks seem to appear in a pattern where roughly every 8-9 Hz there is a power increase. This can be observed for all three types of plots, and for both conditions and areas. To illustrate, one can consider the effect of both types of stimulation over the prefrontal channels. In this electrode, both conditions elicit a very similar effect, with three major peaks in power occurring near 17, 26 and 35 Hz, i.e. at 9 Hz intervals. In the striatum peaks are not as equidistant as in the prefrontal cortex, but it is still possible to see that there are power increases rounding every 8 to 9 Hz (peaks at 8, 16, 26, and 35 Hz to the 5 Hz stimulation and at 9, 18, 26 and 35 Hz to the 12 Hz stimulation).

Contrary to the prefrontal channels, the effect of stimulation does seem to differ between the two conditions. The difference consists in that 5 Hz stimulation elicited the strongest changes in the beta, and more so the gamma band (up to 52 Hz), while the effect of the 12 Hz stimulation in this area is centered to beta. Furthermore, while there are positive changes in power to the lower frequencies for both conditions, this effect is stronger in for the 12 Hz stimulation. Additionally, there is strong suppression to the 5 Hz stimulation which affects all channels, occurring around 10 and 19 Hz, right after power increases around 8 and 16 Hz. Thus, stimulation with a 5 Hz train of pulses elicits power changes in the striatum in a higher band than the stimulation with 12 Hz, combined with strong suppression in the bands where the 12 Hz stimulation generates more changes.

Cross-frequency coupling analysis:

In order to assess if the optical stimulation resulted in coupling between the amplitude of a high frequency range between 15 and 80 Hz and the phase of a lower frequency between 4 and 12 Hz, an exploratory analysis of phase-amplitude coupling was made, considering all combinations between the low and high frequency range.

The selection of the channels per area and condition where coupling was examined, was made based on the spectral results. Since for both conditions, the frequency where a power maximum occurred in the PFC was at 35 Hz, the channel giving the third biggest power change in this frequency was selected for intra-area coupling on the PFC electrode. In the case of the striatum, the frequencies were a power maximum occurred were 35 Hz for the optical stimulation at 5 Hz and 18 Hz for the optical stimulation at 12 Hz. As with the PFC, the channel with the third biggest peak on these frequencies was selected for the analysis.

Figure 4 shows the z normalized debiased phase-amplitude coupling measure on each frequency combination. dPAC values 2 standard deviations above 0 indicate that there is more coupling than what would be expected by chance, considering a null distribution made with 1000 surrogates values for each frequency pair. On the other hand, dPAC values 2 standard deviations below 0 indicate that there is less coupling than would be expected by chance. Although there are frequency combinations that generated significant values going in both directions, these do not behave in a physiologically meaningful way. This is, they tend to be focalized to specific frequency combinations and they do not cluster, sparsely spanning only short ranges of frequencies for phase and amplitude. This is unlike physiological examples of cross-frequency coupling (López-Azcárate et al., 2013). Consequently, it could be predicted that significance would not survive to a correction for multiple comparisons. Thus, it can be concluded that for the selected conditions, channels and frequency ranges used for this exploratory analysis, cross frequency coupling does not occur.

Figure 5 Intra-area phase amplitude coupling (PAC) in the prefrontal cortex (upper row) and the striatum (bottom row) for

the post-stimulus period from 500 to 2000 ms using trains of 5 Hz (left column) and 12 Hz (right column) pulses. A z-normalized value of the debiased PAC is given for every frequency combination using phase on the low frequency range between 4 and 12 Hz and amplitude on the high frequency range between 15 and 80 Hz.

Discussion

The spectral analysis:

The spectral analysis revealed that the optical stimulation of the VTA/SNc using frequency trains from 1 to 20 Hz generates a multi-spectral effect in the prefrontal cortex and the striatum. This is, an effect that spans several frequency bands, and that appears to be sustained across several channels (figure 3). The multi-spectral effect shows slight differences between areas and conditions of stimulation regarding where in the spectrum the change in power is the largest, or how the power increases are distributed along the spectrum. Yet, there is an overall consistency in the effect spanning all frequencies of stimulation, where the largest power increases occurred in the theta and beta band.

 The largest power increases along the spectrum occurred in theta and beta:

The most consistent effect of the stimulation is an increase in power occurring around 8 Hz. This is not the largest increase in power along the spectrum for any of the conditions or electrodes, but is the effect within the lower range (from 1 to 12 Hz) that gave the maximum power values. Notably, this effect showed low variance between conditions, in contrast to the maximum peaks occurring in the higher range (table 1).

The occurrence of high theta power peaks in the cortico-striatal circuit has also been observed in freely behaving rats (DeCoteau et al., 2007; Nakhnikian et al., 2014; von Nicolai et al., 2014) and monkeys In the dorsolateral striatum and the primary sensorimotor cortex of rats, a peak of 8 Hz was observed both during rest and running (von Nicolai et al., 2014). Although this peak was more prominent during locomotion, it was not conditioned to active movement. In line with this observation, an increase of power in the dorsal striatum, appearing between 5 and 10 Hz, was also seen during free behaviors like exploration (moving and stopping to look around), alertness (being immobile but observing), and rearing (standing in the hind legs) (Nakhnikian et al., 2014), and between 7 and 14 Hz during spontaneous movement (DeCoteau et al., 2007). In the present study theta band activity also appears with free behavior in the cortical and striatal electrodes. Notably, the optical stimulation results in an increase of this activity with respect to the baseline period where there is no stimulation, suggesting that the excitation of dopaminergic neurons in the VTA/SNc complex is implicated in this power increase, as animals are able to freely move both before and during the stimulation.

The relation between dopamine and theta band activity was studied in Rhesus monkeys through the local injection of dopamine receptor DR1 and DR2 agonists and antagonists into the prefrontal cortex (Ott, Jacob, & Nieder, 2014). A power increase in theta/low beta band was induced both with an antagonist of DR1, in a range between 4 and 15 Hz and with an agonist of DR2, in a range between 7 and 19 Hz. Interestingly, when an agonist of the DR1 receptor was used, the power increase was observed in the high beta band (25 to 32 Hz). This study not only shows that dopaminergic activity can result on an increase in theta and beta band activity, but also that a multi-spectral effect of dopaminergic activation can be mediated by the dopamine receptor that is stimulated. Thus, it could be that the most prominent power increases observed in the present study, occurring both in the theta and beta band, were mediated by dopamine release affecting both types of receptors, and induced by the stimulation of the VTA/SNc complex. Indeed, the second most consistent effect of the stimulation occurred in the beta band, where the largest power increases were observed.

However, it is interesting that in relation to dopamine, an increase of beta power has been related to dopamine depletion, considering the case of Parkinson’s disease. In Parkinson’s patients, dopaminergic neurons in the SNc are degenerated and this is associated with an increase in beta power in the cortico-striatal pathway, to the point that beta band over-expression is seen as a marker or this condition (Singh, 2018). In support of this idea, a causal experiment in which rats were 6-OHDA lesioned to induce dopamine depletion showed that in the striatum, a clear peak in the power spectra appears between 18 and 22 Hz, which does not occur in dopamine control rats (West et al., 2018).

A contrasting view suggests that beta band activity is a predominant rhythm of the striatum, appearing under normal dopamine function. Recordings from the caudate nucleus and putamen of macaques during the off-task period of a visual task showed that beta power (15-20 Hz) fluctuates in time, increasing by transients lasting 600 ms on average (Courtemanche, Fujii, & Graybiel, 2003). Given that the transient power increases were observed across various locations of the striatum and during rest, it was suggested that beta band activity is an underlying rhythm of the area.

The waxing and waning of the beta band rhythm has also been observed in the neocortex of mice and humans (Sherman et al., 2016; Shin, Law, Tsutsui, Moore, & Jones, 2017), leading to the idea that power fluctuations in the beta band are generated by a “bursty” mechanism. More specifically, these transients were observed in human somatosensory cortex in a range between 18 and 27 Hz, for a duration of about three beta periods (Sherman et al., 2016). Notably, these transients appeared on a trial basis, yet, when averaged across trials, the increase in beta power appeared as continuous in the time domain.

Thus, it could be that the beta band activity that was seen to increase in the striatum and prefrontal cortex after the optical stimulation of the VTA/SNc complex is the expression of an underlying beta rhythm in this network. Given the present results there is no evidence to say that the observed power increase in the beta band occurs in transients, yet, it could be that the continuity in time of the beta power increase is also a result of averaging across trials, plus the use of high temporal smoothing. Thus, a way to observe if beta band activity appears in transients is to examine trials separately for a time-frequency decomposition with more temporal resolution (Jones, 2016). Regarding the relation of the increase in beta power and dopaminergic neuron activity, it appears that the optical stimulation resulted in the observed increase in beta power, contrasting the idea that an increase in power is related to dopamine depletion and suggesting that the underlying beta rhythm could be modulated by dopaminergic neuron activation.

All things considered, it could be said that the power increases in theta and beta that were observed in the prefrontal cortex and the striatum across conditions, have also been reported in the cortico-striatal circuit for awake, behaving animals which are not engaged with a task. In the present study these power changes were induced by the stimulation of the VTA/SNc, suggesting that the activation of the dopaminergic neurons that innervate the recorded areas have a role in the generation of the power changes in theta and beta. To elucidate how dopaminergic neuronal activation results in these effects, it would be useful to consider if the power increase in theta and or beta are mediated by a specific receptor. Also, it would be key to test if beta band changes are generated in transients, and the relation of these beta bursts to the timing of dopaminergic neuronal activation.

 The optical stimulation generated power changes that were distributed along the spectrum: Although the largest power changes were concentrated in the theta and beta band, in the higher frequency range some conditions generated a maximum peak around low gamma instead of beta, generating some variance between conditions. In the lower frequency range, variance occurred towards lower values of theta. More importantly, for all stimulation conditions, several peaks occurred along the spectrum, which had important power changes. These repeating peaks distributed the effect of the modulation along various test frequencies in theta, beta and gamma.

Previous studies have also shown that in the cortico-striatal circuit, power increases can occur at various frequencies distinct from high theta and beta. For instance, gamma was shown to occur in the striatum of rats (West et al., 2018), or theta combined with gamma in the striatum and somatosensory cortex of rats (von Nicolai et al., 2014), or theta combined with gamma and also delta (DeCoteau et al., 2007). Hence, multi-spectral increases of power are common to the system in study, supporting the presented observations.

Nonetheless, it should be taken into account that for the current results, increases in power appear to be somewhat equidistant from each other in the frequency spectrum, with peaks occurring roughly every 8 to 9 Hz. This could be taken as the expression of the harmonics of a fundamental frequency, where the fundamental frequency is that of the sinusoid that best explains the signal under

study, and the harmonics are the multiples of the fundamental frequency (Zhou, Melloni, Poeppel, & Ding, 2016).

When a time frequency decomposition is used to extract the waveform of an oscillation in the time series, the value of the fundamental frequency determines where in the frequency spectrum power increases can occur, which is at its harmonics. At the same time, the shape of the signal is what determines the size of the power changes occurring at the harmonics of the fundamental frequency (Zhou et al., 2016). Thus, the shape of the signal is what determines if power peaks are generated at the harmonics of the sinusoid that better fits the signal under study.

The outcome of using the common methods for time-frequency decomposition to extract the waveform of an oscillation that is not sinusoidal, is that important peaks in power can be observed at the harmonics of the fundamental frequency, giving a multi-spectral result (Jones, 2016; Kramer, Tort, & Kopell, 2008; Zhou et al., 2016). Notably, the frequency were the largest power change is generated is not necessarily the fundamental frequency, which has been demonstrated for non-sinusoidal signal that has sharp edges (Zhou et al., 2016). Having sharp edges means that the rise or decay of the signal occurs faster than the rise or decay of a sinusoid.

Taking the previous into account, the peaks in power observed in the striatal channels due to the optical stimulation of the VTA/SNc complex at 5 Hz (figure 3) could be due to the decomposition of a signal with sharp edges, because the largest power value occurs in gamma, while the fundamental frequency appears to be in theta. If this is true, it can be said that the optical stimulation resulted on the increase in power of an oscillation that repeats at 8 Hz, but that differs from theta in that it has sharp edges, and that the rest of the multi-spectral result is an artifact of the decomposition of this signal. One way to assess if the signal has these characteristics is to do an event related average of the unfiltered data (Kramer et al., 2008).

All in all, the presence of harmonics in the observed multi-spectral result, which were observed for both electrodes and across conditions is of utmost importance for the interpretation of the present results. Given the evidence in hand, it is likely that the peaks in power that were observed along the spectrum, including those that were maximum at beta or gamma, are an artifact of the decomposition of a signal which is not sinusoidal, but behaves the most similarly to theta. Consequently, the beta or gamma band activity that is observed after time-frequency decomposition would not have a physiological meaning, despite previous evidence showing that these oscillations can occur in the cortico-striatal system during the awake state. Meanwhile, the peak in power occurring in theta, which was the most consistent effect of the optical stimulation across conditions and areas, would be the most meaningful result from the spectral analysis.

As reviewed, oscillatory activity in the theta band is common to the cortico-striatal circuit and based on the observed, could be modulated by dopaminergic activity. Yet, it is worth stating that activity detected in the theta band which does not behave as a sinusoid can have a different function or significance in the circuit than the more sinusoidal-fitting theta. To illustrate, hippocampal theta can have a different shape depending of the cell layer where it is recorded from, or can change its waveform under the influence of one or other neurotransmitter (Cole & Voytek, 2017). Hence, considering the shape of the original signal in the time series before transforming it can bring valuable information regarding its functionality, and in the context of the present study, can be informative about the specific influence of dopaminergic neuronal activation over the activity at the local circuit in the innervated areas.

 The effect size of the stimulation shows differences between areas and conditions:

Although the effect of the stimulation was similar between areas regarding where in the spectrum the major effects were located, a difference between areas was observed regarding the size of the modulation. The difference was that the striatal electrode showed generally larger power changes, in comparison to the prefrontal electrode. The similarity between areas regarding where in the spectrum the effects of the stimulation are predominant has been observed before in the cortico-striatal circuit, both during behavior (von Nicolai et al., 2014) and after stimulating dopaminergic activity (López-Azcárate et al., 2013). Likewise, the difference between areas regarding the size of the modulation is to be expected, given that the striatum is more innervated by dopaminergic neurons than the prefrontal cortex, with a high density of connections coming from the SNc (Björklund & Dunnett, 2007). The difference in the density of innervations between the prefrontal cortex and the striatum could then also explain why power changes were not homogeneous across the prefrontal channels: if these neurons are less innervated by dopaminergic neurons, it could be that the effect of the modulation is localized to sets of neuronal groups.

Another difference observed, regarding the size of the modulation, was detected across conditions: here, the lower frequencies of stimulation tended to be the ones that generated the biggest power changes. Since the lower frequencies would most likely generate the tonic entrainment of the VTA/SNc, it could be thought that dopaminergic modulation in this experiment is more strongly achieved by means of tonic firing, than by means of phasic firing. In accordance to this, a previous study found that firing in the VTA occurred at rates between 3 and 6 Hz, in the case of a task with working memory demands (Duvarci et al., 2018). If it is the case that the modulation by optogenetic stimulation of the VTA/SNc is stronger when it induces tonic firing, and given that tonic signaling is said to work through D2 receptors (Schultz, 2007), this would further support the idea that the observed theta band activity is induced through the action of dopaminergic neurons over D2 receptors.

However, it cannot be discarded that the stimulation is also able to induce significant modulation by means of phasic firing, considering that phasic firing occurs naturally in bursts, and that cortical and striatal areas have shown to have transient responses, for instance in the beta band (Courtemanche et al., 2003; Sherman et al., 2016; Shin et al., 2017). Despite that high frequency stimulation was given in a sustained fashion in the present study instead of by bursts, it may be meaningful to test for the effects of the high frequency stimulation with more temporal resolution, and before averaging across trials (Jones, 2016), to see if the modulation, being restricted to shorter time periods, behaves differently than in the current results.

In sum, the differences in the size of the effect between areas could be explained by the disparity in the degree of innervation by dopaminergic neurons of the striatal and cortical area under study. Meanwhile, the difference in the size of the effect between conditions may be due to the stimulation being more effective by means of tonic dopaminergic activity.

Cross-frequency coupling analysis:

 Phase-amplitude cross frequency coupling did not occur in the selected channels and

conditions:

The cross-frequency coupling analysis, where coupling was explored between a set of the low frequencies from 4 to 12 Hz, and a set of the high frequencies from 15 and 80 Hz, did not show that the amplitude of the higher frequencies was modulated by the phase of the lower frequencies. Thus, phase-amplitude coupling was not a consequence of the stimulation for the selected conditions and the selected channels. Yet, it is premature to conclude that the stimulation is ineffective at generating intra-area cross-frequency coupling, given that not all frequencies of stimulation were considered, and coupling was tested with only one channel per area.

Still, an explanation for the lack of coupling can be given, based on the current evidence: since it is plausible that the high frequency power increase observed after the spectral analysis is an artifact of the decomposition of a non-sinusoidal signal with sharp edges, then the lack of coupling is because there is no high frequency activity to be modulated by the phase of a lower frequency. Notably, the measure used for phase-amplitude coupling in the present study is tailored to reduce the chances of finding spurious coupling, arising from the analysis of a non-sinusoidal signal (van Driel et al., 2015). Thus, it is possible that this measure helped to avoid reporting a false positive, considering that high frequency activity was present after the spectral analysis, and considering that channels were selected because they showed this high frequency activity.

Besides the advantage of using a phase-amplitude coupling measure that reduces the chances of spurious coupling, two more methodological considerations could be made, when testing for cross-frequency coupling in the system under study. The first, is that the selection of channels to test for coupling could be made considering the spatial distribution of the connections from the midbrain DA neurons reaching the cortical and striatal areas. The second, is that coupling could be tested by time-bins. These two considerations are made, first, taking into account that midbrain projections have spatial domains within their areas of influence (Haber, 2016), and second, that phase-amplitude coupling can also be expected to occur transiently in the brain (Canolty & Knight, 2010).

Conclusion

The current evidence indicates that the optical stimulation was successful at generating oscillatory changes in the innervated areas, where the biggest effects occurred in the theta and beta band. Theta and beta band activity have been shown to occur in the cortico-striatal system, and there is evidence to think that the stimulation of dopaminergic activity can result in the enhancement of the oscillatory activity at these frequencies. Yet, the multi-spectral result suggested that the frequency modulations that happened above the theta band were an artifact of the decomposition of a non-sinusoidal signal. If this were the case, it could be said that the effect of the stimulation is focused at the theta band, which is an effect that is sustained across conditions and areas. To corroborate this suggestion, it would be necessary to make an assessment of the shape of the signal in the untransformed data, not only to determine if the higher frequency power changes are artefactual, but also to determine the physiological meaning of having a signal in the theta band that differs from sinusoidal theta. Regarding

the cross-frequency coupling analysis, it can be partially concluded that intra-area phase amplitude coupling is not a consequence of the stimulation, which for this particular analysis could be connected to a lack of high frequency activity in the system.

Methodological considerations

After this initial assessment of the electrophysiological consequences of the optical stimulation of the VTA/SNc using laser trains from 1 to 20 Hz, some methodological considerations arose, which are presented as a recommendation for further analyses. The first two considerations are about the spatial and temporal dimensions of the data. Here, the spectral effects of the stimulation were assessed by averaging by channels and by time, aiming to get a starting picture of the global effects elicited in the areas of study. However, it could be that effects of the stimulation are spatially and or temporally restricted, meaning that variance across channels or across time should be accounted for. A third consideration is about the assessment of the shape of the signal in the untransformed data, which can influence the way that the data is processed from the time-frequency decomposition to the coupling analyses.

Supplementary material

Supplementary figure 1 Power spectra for every condition on the PFC channels, for the time between 500 and 2000 ms, with

respect to a baseline -500 to -200 ms. Power values are given at a decibel scale, for the 80 test frequencies between 1 and 100 Hz. Data is averaged by channels and trials. All peaks found in the power spectra are displayed.

Supplementary figure 2 Power spectra for every condition on the striatal channels, for the time between 500 and 2000 ms,

with respect to a baseline -500 to -200 ms. Power values are given at a decibel scale, for the 80 test frequencies between 1 and 100 Hz. Data is averaged by channels and trials. All peaks found in the power spectra are displayed.

References

Berke, J. D. (2018). What does dopamine mean? Nature Neuroscience, 21(6), 787–793. https://doi.org/10.1038/s41593-018-0152-y

Björklund, A., & Dunnett, S. B. (2007). Dopamine neuron systems in the brain: an update. Trends in Neurosciences, 30(5), 194–202. https://doi.org/10.1016/j.tins.2007.03.006

Canolty, R. T., & Knight, R. T. (2010). The functional role of cross-frequency coupling. Trends in Cognitive Sciences, 14(11), 506–515. https://doi.org/10.1016/j.tics.2010.09.001

Cohen, M. X. (2018). A better way to define and describe Morlet wavelets for time-frequency analysis. BioRxiv. https://doi.org/10.1101/397182

Cole, S. R., & Voytek, B. (2017). Brain Oscillations and the Importance of Waveform Shape. Trends in Cognitive Sciences, 21(2), 137–149. https://doi.org/10.1016/j.tics.2016.12.008

Courtemanche, R., Fujii, N., & Graybiel, A. M. (2003). Synchronous, Focally Modulated beta-Band Oscillations Characterize Local Field Potential Activity in the Striatum of Awake Behaving Monkeys. Journal of Neuroscience, 23(37), 11741–11752. Retrieved from

http://ovidsp.ovid.com/ovidweb.cgi?T=JS&PAGE=reference&D=emed6&NEWS=N&AN=200400 8410

DeCoteau, W. E., Thorn, C., Gibson, D. J., Courtemanche, R., Mitra, P., Kubota, Y., & Graybiel, A. M. (2007). Oscillations of local field potentials in the rat dorsal striatum during spontaneous and instructed behaviors. Journal of Neurophysiology, 97(5), 3800–3805.

https://doi.org/10.1152/jn.00108.2007

Draguhn, A., & Buzsaki, G. (2004). Neuronal Oscillations in Cortical Networks. Science, 304(5679), 1926–1929. https://doi.org/10.1126/science.1099745

Duvarci, S., Simpson, E. H., Schneider, G., Kandel, E. R., Roeper, J., & Sigurdsson, T. (2018). Impaired recruitment of dopamine neurons during working memory in mice with striatal D2 receptor overexpression. Nature Communications. https://doi.org/10.1038/s41467-018-05214-4 Fries, P. (2005). A mechanism for cognitive dynamics: Neuronal communication through neuronal

coherence. Trends in Cognitive Sciences, 9(10), 474–480. https://doi.org/10.1016/j.tics.2005.08.011

Fries, P. (2015). Rhythms for Cognition: Communication through Coherence. Neuron, 88(1), 220–235. https://doi.org/10.1016/j.neuron.2015.09.034

Fujisawa, S., & Buzsáki, G. (2011). A 4 Hz Oscillation Adaptively Synchronizes Prefrontal, VTA, and Hippocampal Activities. Neuron. https://doi.org/10.1016/j.neuron.2011.08.018

Graham V. Williams, & Patricia S. Goldman-Rakic. (1995). Modulation of memory fields by dopamine D1 receptors in prefrontal cortex. Nature, 376(6541), 572–575.

https://doi.org/10.1038/376572a0

Haber, S. N. (2016). Corticostriatal circuitry. Dialogues in Clinical Neuroscience, 18(1), 7–21. https://doi.org/10.1007/978-1-4939-3474-4_135

Jones, S. R. (2016). When brain rhythms aren’t ‘rhythmic’: implication for their mechanisms and meaning. Current Opinion in Neurobiology. https://doi.org/10.1016/j.conb.2016.06.010 Kramer, M. A., Tort, A. B. L., & Kopell, N. J. (2008). Sharp edge artifacts and spurious coupling in EEG

frequency comodulation measures. Journal of Neuroscience Methods. https://doi.org/10.1016/j.jneumeth.2008.01.020

Lohani, S., Martig, A. K., Deisseroth, K., Witten, I. B., & Moghaddam, B. (2018). Dopamine modulation of prefrontal cortex activity is manifold and operates at multiple temporal and spatial scales Abbreviated Title: Nature of dopamine modulation of cortical activity. BioRxiv. https://doi.org/10.1101/452862

López-Azcárate, J., Nicolás, M. J., Cordon, I., Alegre, M., Valencia, M., & Artieda, J. (2013). Delta-mediated cross-frequency coupling organizes oscillatory activity across the rat cortico-basal

ganglia network. Frontiers in Neural Circuits, 7. https://doi.org/10.3389/fncir.2013.00155 Nakhnikian, A., Rebec, G. V., Grasse, L. M., Dwiel, L. L., Shimono, M., & Beggs, J. M. (2014). Behavior

modulates effective connectivity between cortex and striatum. PLoS ONE, 9(3). https://doi.org/10.1371/journal.pone.0089443

Ott, T., Jacob, S. N., & Nieder, A. (2014). Dopamine receptors differentially enhance rule coding in primate prefrontal cortex neurons. Neuron, 84(6), 1317–1328.

https://doi.org/10.1016/j.neuron.2014.11.012

Roopun, A. K. (2008). Temporal interactions between cortical rhythms. Frontiers in Neuroscience, 2(2), 145–154. https://doi.org/10.3389/neuro.01.034.2008

Schultz, W. (2007). Multiple Dopamine Functions at Different Time Courses. Annual Review of Neuroscience, 30(1), 259–288. https://doi.org/10.1146/annurev.neuro.28.061604.135722 Sherman, M. A., Lee, S., Law, R., Haegens, S., Thorn, C. A., Hämäläinen, M. S., … Jones, S. R. (2016).

Neural mechanisms of transient neocortical beta rhythms: Converging evidence from humans, computational modeling, monkeys, and mice. Proceedings of the National Academy of Sciences of the United States of America, 113(33), E4885–E4894.

https://doi.org/10.1073/pnas.1604135113

Shin, H., Law, R., Tsutsui, S., Moore, C. I., & Jones, S. R. (2017). The rate of transient beta frequency events predicts behavior across tasks and species. ELife, 6, 1–31.

https://doi.org/10.7554/eLife.29086

Singh, A. (2018). Oscillatory activity in the cortico-basal ganglia-thalamic neural circuits in Parkinson’s disease. European Journal of Neuroscience, 48(8), 2869–2878.

https://doi.org/10.1111/ejn.13853

Tsai, H. C., Zhang, F., Adamantidis, A., Stuber, G. D., Bond, A., De Lecea, L., & Deisseroth, K. (2009). Phasic firing in dopaminergic neurons is sufficient for behavioral conditioning. Science, 324(5930), 1080–1084. https://doi.org/10.1126/science.1168878

van Driel, J., Cox, R., & Cohen, M. X. (2015). Phase-clustering bias in phase-amplitude cross-frequency coupling and its removal. Journal of Neuroscience Methods.

https://doi.org/10.1016/j.jneumeth.2015.07.014

Vandecasteele, M., Royer, S., Belluscio, M., Berényi, A., Diba, K., Fujisawa, S., … Buzsáki, G. (2012). Large-scale recording of neurons by movable silicon probes in behaving rodents. Journal of Visualized Experiments, (61), 1–6. https://doi.org/10.3791/3568

von Nicolai, C., Engler, G., Sharott, A., Engel, A. K., Moll, C. K., & Siegel, M. (2014). Corticostriatal Coordination through Coherent Phase-Amplitude Coupling. Journal of Neuroscience, 34(17), 5938–5948. https://doi.org/10.1523/JNEUROSCI.5007-13.2014

West, T. O., Berthouze, L., Halliday, D. M., Litvak, V., Sharott, A., Magill, P. J., & Farmer, S. F. (2018). Propagation of Beta/Gamma Rhythms in the Cortico-Basal Ganglia Circuits of the Parkinsonian Rat. Journal of Neurophysiology, 1608–1628. https://doi.org/10.1152/jn.00629.2017

Witten, I. B., Steinberg, E. E., Lee, S. Y., Davidson, T. J., Zalocusky, K. A., Brodsky, M., … Deisseroth, K. (2011). Recombinase-driver rat lines: Tools, techniques, and optogenetic application to

dopamine-mediated reinforcement. Neuron, 72(5), 721–733. https://doi.org/10.1016/j.neuron.2011.10.028

Zhou, H., Melloni, L., Poeppel, D., & Ding, N. (2016). Interpretations of frequency domain analyses of neural entrainment: Periodicity, fundamental frequency, and harmonics. Frontiers in Human Neuroscience, 10(June), 1–8. https://doi.org/10.3389/fnhum.2016.00274