Reward-timing-dependent bidirectional modulation of cortical microcircuits during optical single-neuron operant conditioning

Hira, Riichiro; Ohkubo, Fuki; Masamizu, Yoshito; Ohkura, Masamichi; Nakai, Junichi; Okada, Takashi; Matsuzaki, Masanori

doi:10.1038/ncomms6551

Article
Published: 24 November 2014

Reward-timing-dependent bidirectional modulation of cortical microcircuits during optical single-neuron operant conditioning

Riichiro Hira^1,2,
Fuki Ohkubo^1,2,
Yoshito Masamizu^1,2,
Masamichi Ohkura³,
Junichi Nakai³,
Takashi Okada⁴ &
…
Masanori Matsuzaki^1,2

Nature Communications volume 5, Article number: 5551 (2014) Cite this article

4358 Accesses
18 Citations
4 Altmetric
Metrics details

Subjects

Abstract

Animals rapidly adapt to environmental change. To reveal how cortical microcircuits are rapidly reorganized when an animal recognizes novel reward contingency, we conduct two-photon calcium imaging of layer 2/3 motor cortex neurons in mice and simultaneously reinforce the activity of a single cortical neuron with water delivery. Here we show that when the target neuron is not relevant to a pre-trained forelimb movement, the mouse increases the target neuron activity and the number of rewards delivered during 15-min operant conditioning without changing forelimb movement behaviour. The reinforcement bidirectionally modulates the activity of subsets of non-target neurons, independent of distance from the target neuron. The bidirectional modulation depends on the relative timing between the reward delivery and the neuronal activity, and is recreated by pairing reward delivery and photoactivation of a subset of neurons. Reward-timing-dependent bidirectional modulation may be one of the fundamental processes in microcircuit reorganization for rapid adaptation.

You have full access to this article via your institution.

Download PDF

A behavioural correlate of the synaptic eligibility trace in the nucleus accumbens

Article Open access 04 February 2022

Kenji Yamaguchi, Yoshitomo Maeda, … Sho Yagishita

Single-cell activity tracking reveals that orbitofrontal neurons acquire and maintain a long-term memory to guide behavioral adaptation

Article 03 June 2019

Vijay Mohan K. Namboodiri, James M. Otis, … Garret D. Stuber

Dopamine facilitates associative memory encoding in the entorhinal cortex

Article 22 September 2021

Jason Y. Lee, Heechul Jun, … Kei M. Igarashi

Introduction

Animals can rapidly adapt to environmental changes. This is accomplished by the reorganization of neuronal activity. When the adaptation requires body movement, activity changes occur in many motor cortex neurons associated with motor commands and proprioceptive feedback¹. By contrast, when the adaptation does not require body movement and the cortical activity is reinforced with repeated reward deliveries, primates can volitionally control the cortical activity at the level of single neurons^1,2,3,4,5. Single-neuron operant conditioning (SNOC) provides a framework for studying the reorganization of neuronal activity during rapid adaptation because it is accompanied by modulation of the activity of non-target neurons, and the associations between non-target and target neurons can be unambiguously determined. Theoretically, cortical reorganization during both SNOC and brain-machine and brain-computer interface learning can be explained by reward-timing-dependent plasticity, in which the activity change in an individual neuron depends on the activity timing relative to a global reward signal^6,7,8. In addition to the inputs associated with the global reward signal, cortical neurons receive strong synaptic inputs from surrounding neurons. Consistent with this, non-target neurons whose activity is highly correlated with that of the target neuron tend to change their firing rate with the target neuron^9,10. However, individual neurons recorded from the same electrode can be separately controlled in SNOC^2,3,9,11. It is difficult to determine the spatial and temporal modulation in cortical activity in fine-scale (<500 μm) microcircuits using electrical recordings.

Two-photon calcium imaging of cortical neurons has been used to identify multineuronal activity within a motor cortical microcircuit in rodents^{12,13,14,15,16}. In contrast to SNOC in the primate, SNOC in the rodent^11,17 has not been performed with identification of the relations of individual neurons to physical movement. Here, we pre trained mice to perform a lever-pull movement using the right forelimb to obtain water. Then, we conducted two-photon calcium imaging of multiple neurons in layer 2/3 (L2/3) motor forelimb areas and determined whether individual neurons were related to lever-pull movement or not. Immediately after that, we conditioned the mice by reinforcing the calcium transients of a single target neuron, but not the lever-pull movement, with water delivery (SNOC by two-photon calcium imaging; 2pSNOC). We found that when the target neuron was not related to the lever-pull movement, the activity of the target neuron rapidly increased during 15 min of 2pSNOC. This increase was not accompanied by an increase in the frequency of lever-pull movements. During 2pSNOC, a subset of non-target neurons that showed high activity synchronous with reward delivery increased their activity, whereas a subset that showed high activity 2–4 s after reward delivery decreased their activity. This upward and downward modulation was recreated by repetitive channelrhodopsin-2 (ChR2) photostimulation¹⁸ of neurons 250 ms before and 2.5 s after the reward delivery, respectively. The results indicate that reward-timing-dependent bidirectional modulation is a fundamental process in L2/3 microcircuit reorganization during fast adaptation to novel environments.

Results

Physical and neuronal operant conditioning

To introduce physical operant conditioning before 2pSNOC, head-restrained mice were trained to perform a self-initiated lever-pull task using the right forelimb^13,16,19. A 4-μl water drop was provided as a reward when the lever was pulled for 0.4 s after being left in the wait position for >3 s (Fig. 1a)¹³. An adeno-associated virus (AAV) encoding a calcium indicator, GCaMP7²⁰, was injected into the left motor forelimb areas^13,21,22 1–2 weeks before the start of the lever-pull task-training sessions. After 5–14 lever-pull task-training sessions (one session per day), two-photon calcium imaging of L2/3 motor cortical neurons was performed while mice performed 2pSNOC. A total of 24 2pSNOC sessions were recorded in seven mice (one to eight sessions per mouse). Each 2pSNOC session consisted of three periods (Fig. 1b): a pre-conditioning period (10 min), a conditioning period (15 min) and a post-conditioning period (10 min). 2pSNOC was performed during the conditioning period.

**Figure 1: Single-neuron operant conditioning.**

The pre-conditioning period was used to classify the imaged neurons according to their association with the pre-trained lever-pull movement and to select a conditioned neuron (target neuron). Mice could freely pull the lever, but the reward corresponding to a successful lever-pull for the lever-pull task (>0.4 s pull with >3 s wait) was omitted to avoid satiation. A successful lever-pull that did not directly induce the reward was defined as a virtual hit. Virtual hits occurred at a frequency of 1.9±0.1 min⁻¹ during the pre-conditioning period (n=24 sessions from seven mice), possibly because of partial habituation and/or expectation of reward. Many neurons exhibited large calcium transients related to the lever-pull movement, similar to our previous report¹³. According to the value of normalized non-negative deconvolution²³ of each relative fluorescent change (ΔF/F) trace (‘activity’; see Methods), active neurons during the pre-conditioning period were reconstructed (35.0±16.4 neurons, n=24 sessions from seven mice) and classified into two groups: neurons with high activity during lever-related periods (from 1 s before the onset of lever-pull to 3 s after the end of lever-pull; ‘lever-related neurons’) (Fig. 1c), and other neurons (‘lever-unrelated neurons’) (Fig. 1d). Most reconstructed neurons were assumed to be excitatory neurons because GCaMP expression was under the control of synapsin I promoter, which considerably limits the expression to excitatory neurons^16,24. One neuron was targeted in the conditioning period. This target neuron was a lever-related neuron in 13 sessions and a lever-unrelated neuron in 11 sessions.

During the conditioning period, the mouse performed a 2pSNOC task. During the 2pSNOC task, two 4-μl water drops were delivered to the mouse immediately after the ΔF/F of a single target neuron increased above a threshold (Fig. 1a,e–g). A water reward was delivered after a calcium transient in the target neuron, irrespective of whether the target neuron was lever-related or lever-unrelated (Fig. 1c,d). Calcium transients in nearby neurons did not trigger reward delivery (Fig. 1f,g). Ninety-eight percent of the water rewards were given immediately after the calcium transient in the target neuron exceeded the threshold (1,000 successful rewards from 1,020 total rewards in 24 sessions from seven mice). The remaining 2% of rewards were delivered when ΔF/F exceeded the threshold without an obvious calcium transient. The latency from the calcium transient in the target neuron to the corresponding reward delivery was 212±122 ms (n=1,000 successful rewards), which, when compared with values reported in a previous study, was sufficiently fast for operant conditioning²⁵.

Rapid activity changes in target neurons during SNOC

First, we examined whether reward delivery corresponding to calcium transients in lever-related and lever-unrelated target neurons resulted in rapid changes in the activity of the target neuron during the 15 min conditioning period. For lever-unrelated target neurons, activity increased by ~50%, whereas, for lever-related target neurons, activity did not change (Fig. 2a,b). Activity gain was defined as the ratio of the mean activity in the last 5 min of the conditioning period (late conditioning period) to the mean activity in the first 5 min of the conditioning period (early conditioning period) minus 1. When activity gain was averaged across sessions and/or neurons, the logarithmic mean of the ratio was used (Fig. 2c). For lever-unrelated target neurons, the activity gain was substantially above zero (Fig. 2d and Table 1). For lever-related target neurons, activity gain was not different from zero (Fig. 2d and Table 1). This pattern remained even after exclusion of the mouse in which eight sessions were performed, with a mean activity gain of 55.3% and −0.73% in lever-unrelated and lever-related target neurons, respectively.

**Figure 2: Neuronal activity during optical single-neuron operant conditioning.**

Table 1 The mean activity gain of each type of target neuron.

Full size table

Next, we determined whether operant conditioning of the target neuron increased the frequency of reward delivery induced by calcium transients above the threshold. As expected, the reward frequency increased by ~50% when lever-unrelated neurons were targeted, and did not change when lever-related neurons were targeted (Fig. 3a). Similar to activity gain, reward gain was defined as the ratio of the reward frequency in the late conditioning period to the reward frequency in the early conditioning period minus 1, and the logarithmic mean of the ratio was used for averaging (Fig. 3b). The reward gain was substantially above zero when lever-unrelated neurons were targeted, and was not different from zero when lever-related neurons were targeted (Fig. 3c). The activity gain and the reward gain were correlated in both types of target neurons (Fig. 3d). When lever-unrelated neurons were targeted, activity gain was positive in all 11 sessions (from five mice) and reward gain was positive in 8 of the 11 sessions; therefore, the mouse steadily increased the target neuron activity and reward frequency.

**Figure 3: Reward frequency and behaviours during optical single-neuron operant conditioning.**

Next, we examined the specificity of the target neuron. The mean activity gain of the non-target neurons was slightly larger than zero when a lever-unrelated neuron was targeted and slightly smaller than zero when a lever-related neuron was targeted (Table 2). To examine whether the relative change in activity of the target neuron from the early to the late conditioning period exceeded that in the non-target neurons, we ranked the activity of the target neuron among all reconstructed neurons, including target and non-target neurons. The rank was normalized between 1 (top) and 100 (bottom) in the early and late conditioning periods during each session. When a lever-unrelated neuron was targeted, the normalized rank of the target neuron significantly rose from the early to the late conditioning period (change in normalized rank, 18.0±6.4; P=0.0078, Wilcoxon signed-rank test; n=11). When a lever-related neuron was targeted, the normalized rank of the target neuron was similar in the early and late conditioning periods (change in normalized rank, 4.5±4.1; P=0.56, Wilcoxon signed-rank test; n=13). Thus, both of the activity and rank of the target neuron significantly increased from the early to the late conditioning period when a lever-unrelated neuron was targeted. These results indicate that the rapid operant conditioning of single lever-unrelated neurons was successful even though the activity of non-target neurons slightly changed. By contrast, the rapid operant conditioning of single lever-related neurons was not successful.

Table 2 The mean activity gain of each type of non-target neuron.

Full size table

In lever-related neurons, calcium transients would coincide with a lever-pull. In the 2pSNOC, mice were rewarded for calcium transients in the target neuron. Therefore, when lever-related neurons were targeted, pulling the lever would result in the delivery of a reward, meaning that the mice did not have to change the strategy that they adopted in the lever-pull task to get rewards during the conditioning period. In support of this, the amount of time spent pulling the lever and the frequency of virtual hits in the conditioning period were approximately twofold larger when lever-related neurons were targeted than when lever-unrelated neurons were targeted (Fig. 3e,f), although licking frequency was not different (Fig. 3g). Irrespective of the target neuron type, lever-pull behaviours did not change from the early to the late conditioning period. These results suggest that, when lever-related neurons were targeted, the mouse performed goal-directed lever-pull movements from the onset of the conditioning period without recognition of SNOC. This may explain why the activity gain for lever-related target neurons was not significantly positive. In the following analyses of the microcircuit reorganization during 2pSNOC, only data from sessions with a lever-unrelated target neuron were used (11 sessions from five mice; n=381 non-target neurons).

Reward-timing-dependent bidirectional modulation

Next, we examined whether non-target neurons associated with the target neuron changed their activity more markedly than other non-target neurons. The activity gain was similar for lever-related non-target neurons and lever-unrelated non-target neurons (Fig. 4a and Table 2). The activity gain of non-target neurons was not associated with the distance to the target neuron (Fig. 4b). By contrast, the activity gain of non-target neurons was weakly related to the pairwise correlation with the target neuron during the conditioning period (excluding the reward-related periods from 0 to 5 s after each reward delivery; spontaneous pairwise correlation) (Fig. 4c). This indicates that non-target neurons whose activity was temporally, but not spatially, associated with the target neuron increased their activity, although the strength of the spontaneous pairwise correlation was negatively associated with the cellular distance (Fig. 4d), as previously reported in the mouse motor cortex^12,13,15.

**Figure 4: Modulation of non-target neurons.**

Although the change in the activity of non-target neurons was weakly associated with the correlation with the target neuron, non-target neurons should have been affected by the water delivery because it was the reinforcer. Therefore, we next investigated whether the activity gain was affected by the activity timing of the non-target neurons relative to the reward delivery (Fig. 5a). For each non-target neuron, the ratio of the sum of the activity in a 0.33-s time bin t s after reward delivery during the first 10 min of the conditioning period to the sum of the remaining activity in the same bin during the first 10 min of the conditioning period was defined as the reward synchronization index (t) (RSI(t); see Methods for details). For each time bin, activity gain was averaged across the non-target neurons with RSI values in the top 5% (Fig. 5b). The activity gain of non-target neurons with RSI in the top 5% was significantly larger than zero when t ranged from −0.33 s to 0.33 s, and significantly smaller than zero when t was 2.3, 3 and 4 s (Fig. 5c). We refer to this phenomenon as reward-timing-dependent bidirectional modulation (RTBM), and refer to neurons that had at least one RSI (−0.33 s to 0.33 s) in the top 5% as reward-synchronous active (SR) neurons (n=34) and neurons that had at least one RSI (2–4 s) in the top 5% as after-reward active (AR) neurons (n=39). Neurons that met both criteria were categorized according to the time bin with the highest RSI. The activity gain of SR and AR neurons was substantially positive and negative, respectively (Fig. 5d). RTBM was also detected when RSI values that determined SR and AR neurons ranged from the top 2% to the top 10% (Fig. 5e). In addition, RTBM was observed when neurons with RSI in the top 20% for each time bin in each session were selected; the activity gain of neurons with RSI (0 s) in the top 20% (SR_20% neurons) averaged across-sessions was substantially higher than zero and that of neurons with RSI (2.3 s) in the top 20% (AR_20% neurons) averaged across-sessions (Fig. 5f). In 9 of 10 sessions with >2 SR_20% neurons and >2 AR_20% neurons (five, two, one, and one out of five, two, two and one sessions from four mice, respectively), the mean activity gain of SR_20% neurons was higher than that of AR_20% neurons; thus, the timing of neuronal activity relative to the reward was a critical factor in the modulation of the two groups of non-target neurons.

**Figure 5: Reward-timing-dependent bidirectional modulation.**

We next examined the association between SR and AR neurons and the target neuron. The percentage of lever-related and lever-unrelated neurons was similar between SR and AR neurons (Fig. 6a) and the distance of SR and AR neurons from the target neuron was similar to the distance of other neurons from the target neuron (Fig. 6b). The pairwise correlation with the target neuron was higher for SR neurons than for AR neurons and other neurons during both the conditioning and pre-conditioning periods, that is, without water delivery (Fig. 6c). This is consistent with the result shown in Fig. 4c and indicates that SR neurons were frequently active together with the target neuron because the reward delivery was induced by a calcium transient in the target neuron beyond the threshold with only ~200 ms delay. Thus, SR neurons, which were highly correlated with the target neuron, exhibited an increase in activity irrespective of whether the activity was lever-related, or the distance from the target neuron. By contrast, AR neurons did not appear different from other non-target neurons in terms of their associations with the target neuron.

**Figure 6: Characteristics of neurons with reward-timing-dependent bidirectional modulation.**

Neither SR nor AR neurons were specifically related to licking because licking frequency around the reward (–1 to 1 s) or after the reward (2 to 4 s) did not change from the early to the late conditioning period (Fig. 7a). Likewise, neither SR nor AR neurons were specifically related to lever-pull in response to the reward delivery because the percentage of lever-pull time and the probability of lever pulling around the reward (–1 to 1 s) and after the reward (2 to 4 s) did not change from the early to the late conditioning period (Fig. 7b,c). Therefore, the activity changes in AR or SR neurons cannot be explained by a gradual loss of the pre-trained memory for the mouse to pull the lever 3 s after the reward delivery.

**Figure 7: Behaviours around and after reward delivery and the distribution of reward-related activity index.**

We also estimated whether reward-related activity accounted for the activity gain of SR and AR neurons. The reward-related activity index (RRAI) was defined as the mean activity during the reward-related period divided by the mean activity during the non-reward-related periods of the conditioning period. Reward-related neurons were defined as the neurons with the top 5% of RRAI values because these neurons contributed to the skewed distribution of RRAI (Fig. 7d). As expected, the reward-related neurons tended to have positive activity gain (27.0%, P=0.14, n=19, one-sample t-test). The proportion of SR and AR neurons that were reward related was similar (14.8% for SR neurons, 23.1% for AR neurons; P=0.36, χ²-test). The proportion of lever-related and lever-unrelated neurons that were reward related was also similar (5.8% for lever-related neurons, 3.9% for lever-unrelated neurons; P=0.41, χ²-test). Thus, it was the timing of the activity relative to the reward delivery rather than the magnitude of activity induced by reward delivery that was critically involved in the microcircuit reorganization. The inter-reward interval time was 26.8±3.7 s (n=11 sessions), which was much longer than the interval between the reward and the AR neuron activity (~2.5 s), indicating that it was unlikely that the decrease in AR neuron activity from the early to the late conditioning period occurred because these neurons had the weakest relation to the reward of all non-target neurons. Rather, these results suggest that the activity of AR neurons decreased owing to the timing (~2.5 s) of the activity relative to reward delivery.

Pairing of ChR2 photostimulation and reward recreated RTBM

To validate whether the activity timing relative to the reward delivery is sufficient to induce the neuronal activity changes, we performed repetitive pairing of the activity of a random set of neurons and reward delivery with different time intervals. We expressed GCaMP6f²⁶ in a majority of the neurons and ChR2-mCherry¹⁸ in a subset of neurons to be photostimulated (Fig. 8a; see Methods). The photostimulation was performed in mice that had not been trained to perform the lever-pull task. In each photostimulation session, 60 photostimuli were delivered during a 15 min period to induce firing in ChR2-mCherry-expressing neurons in the field of view and a reward (8 μl of water) was given 0.25 s after or 2.5 s before each photostimulation (potentiation protocol and depression protocol, respectively) (Fig. 8b). Photostimulation did not induce any observable movements. To prevent visual associative learning between the photostimulation and the reward, a blue-light-emitting diode was illuminated in front of both eyes every 2–3 s. Two-photon calcium imaging was conducted during the photostimulation period. For each reconstructed neuron, photostimuli-induced activity gain was quantified as the ratio of the mean ΔF/F in the last 5 min of the photostimulation period to that in the first 5 min minus 1. We identified neurons that showed significantly higher activity in response to the ChR2 photostimuli than to the light-emitting diode illuminations (P<0.05; one-tailed two-sample t-test) and termed them photostimuli-responsive neurons (Fig. 8c). In the potentiation protocol, 56% of imaged neurons were photostimuli-responsive neurons (78 out of 139 neurons in four sessions from four mice), whereas, in the depression protocol, 30% of imaged neurons were photostimuli-responsive neurons (40 out of 132 neurons in four sessions from four mice). Although the photostimuli-responsive neurons in these two protocols had almost identical ΔF/F at the beginning of the sessions, the photostimuli-induced activity gains of these neurons were significantly positive and negative in the potentiation and depression protocols, respectively (Fig. 8d,e). The activity gain of the other neurons was not significantly different from zero in either protocol (potentiation protocol, P=0.059, n=61; depression protocol, P=0.20, n=92; one-sample t-test). The different modulation of photostimuli-responsive neurons in the potentiation and depression protocols was not due to a difference in licking behaviours because the licking frequency around the reward (–1 to –1 s) and after the reward (2 to 4 s) did not change from the first 5 min to the last 5 min in either protocol (Fig. 8f). In addition, 60 photostimuli delivered without any water delivery did not induce substantial changes in the activity gain in the photostimuli-responsive neurons (Fig. 8e); thus, repetitive pairing of neuronal activity with reward delivery at different time intervals was sufficient to recreate RTBM.

**Figure 8: Repetitive pairing of the reward and photostimulation altered the activity of photostimuli-responsive neurons.**

Discussion

Here, we demonstrated rapid operant conditioning of single neurons in the rodent motor cortex by two-photon calcium imaging. The physical reward contingency task was introduced in advance, and the mice rapidly learned to increase the activity of target neurons that were not related to the pre-trained movement. The increase in the activity of these neurons was robust; the mean activity increase was ~50% of the first 5 min activity. The successful operant conditioning of lever-unrelated neurons is consistent with the fact that cortical neurons with no discernible relation to muscles can be volitionally modulated in the primate motor cortex⁹. In addition, we showed, for the first time, that the timing of neuronal activity relative to the reward is one of the critical factors for the activity change in non-target neurons in microcircuits. The modulation of non-target neurons was bidirectional and was reproduced by repetitive pairing of ChR2 photostimulation with reward.

Recently, operant conditioning of two ensembles in the mouse motor cortex by two-photon calcium imaging was reported, whereby the mouse was conditioned to change the activity of the two ensembles in opposite directions²⁷. The temporal dynamics of the activities in the two target ensembles were different to each other, and the conditioning depended on the correlation between them. Therefore, for each non-target neuron, the spatial and temporal associations with the target ensembles were inevitably affected not only by each target ensemble, but also by the relation between the two target ensembles. By contrast, in the present study, the target neuron was a single neuron; therefore the spatial and temporal associations with the target neuron could be unambiguously determined for each non-target neuron. Our results clearly showed that non-target neurons that exhibited high activity synchronous with reward delivery (SR neurons) were reinforced and non-target neurons whose activity followed the reward after 2–4 s (AR neurons) were repressed. The correlation between SR neurons and the target neuron was higher than that between other non-target neurons and the target neuron; however, the distance to the target neuron was similar across SR, AR and other non-target neurons. This suggests that the microcircuit has the capability to strengthen the activity of neurons that were previously included in the same ensemble without strong spatial constrains, which can explain the variety of behaviours of non-target neurons observed during SNOC in primates^2,3,9,10. During fast adaptation, which requires changes in behaviours or internal models²⁸, novel ensembles may be broadly generated based on the structure of pre-existing ensembles. However, our finding that lever-related and lever-unrelated non-target neurons were similarly modulated during the conditioning period suggests that pre-existing ensembles that are not relevant to a novel reward contingency do not necessarily restrict the microcircuit reorganization. We found that a subset of non-target neurons that were not correlated with the target neuron decreased their activity over the conditioning period. The changes in the activity of SR and AR neurons were not due to changes in licking or lever-pull behaviours that might be related to reward prediction. The counterbalance of microcircuit activity may play a role in maintaining a constant overall level of network activity for homeostasis^14,16,29. Our results suggest that RTBM is possibly one of many processes that underlie rapid reorganization of the L2/3 cortical microcircuit during fast adaptation to environmental changes, which occurs during brain-machine and brain-computer interface learning, motor adaptation and skill learning.

RTBM was recreated by repetitive pairing of the reward with direct stimulation of a random set of neurons in the motor cortex. Although licking behaviours did not change in either protocol, the bidirectional activity changes might be related to reward prediction and reward-absence prediction signals without appearance of any behaviours. However, the proportion of imaged neurons that were photostimuli responsive (56% and 30% in the potentiation and depression protocols, respectively) was larger than the proportion of neurons with high RRAI (~5%; Fig. 7d). In addition, reward-modulated neurons were very rarely observed (~3%) in the rat forelimb M1 in a self-initiated lever-pull–push task¹⁹. It is therefore unlikely that the reward-prediction activity had a large effect on the bidirectional changes in the microcircuit. Thus, our results indicate that the motor cortex can be a causal origin of the RTBM without signals from the prefrontal and parietal cortices, which are presumably required for cognitive strategies^5,10. In the auditory cortex, forward conditioning of pure tone stimulation and electrical stimulation of dopaminergic ventral tegmental neurons increased the size of the cortical area responding to the tone frequency, and backward conditioning reduced the size of the area^30,31. This bidirectional remodeling may share the same underlying mechanisms as RTBM in the motor cortex, although RTBM in the motor cortex was faster. The dopaminergic projections to the motor cortex are necessary for motor learning³². Dopamine uptake in the prefrontal cortex takes up to 5 s³³; therefore, increased dopamine concentration in the motor cortex may last for the observed time interval between reward delivery and activity in AR neurons. The corticostriatal synapses also show bidirectional plasticity with dopamine release³⁴, which is necessary for neuroprosthetic skill learning³⁵. Dopaminergic projections to the motor cortex and striatum may cooperatively induce bidirectional modulations in L2/3 neurons.

Methods

Animals

All animal experiments were approved by the Institutional Animal Care and Use Committee of National Institutes of Natural Sciences, Japan. Male and female C57BL/6 mice (aged 2–4 months) were utilized for the present study. All mice were provided with food and water ad libitum and housed in a 12:12 h light–dark cycle. Mice were not used for other experiments before the study. Mice were anesthetized by intraperitoneal injection of ketamine (74 mg kg⁻¹) and xylazine (10 mg kg⁻¹) before an incision was made in the skin covering the neocortex. After the exposed skull was cleaned, a head plate (Tsukasa Giken, Sizuoka, Japan) was attached to the skull using dental cement (Fujiryu-to BC; GC, Tokyo, Japan, Bistite II; Tokuyama Dental, Tokyo, Japan). The surface of the intact skull was coated with dental adhesive resin cement (Super bond; Sun Medical, Shiga, Japan) to prevent drying. Mice were allowed to recover for 1 day before virus injection. The number of mice per cage was 2–5 before the head plate was attached. After that, mice were single-caged to avoid damage to the head plate and the glass window.

Virus production

GCaMP7 cDNA was cloned from the GCaMP7 vector²⁰ and inserted into pAAV (Addgene plasmid 26973; a kind gift from Dr K. Deisseroth, Stanford University) to obtain recombinant AAV type 2/1 (rAAV2/1) expressing GCaMP7. The vector plasmid (pAAV-syn-GCaMP7) contained GCaMP7 cDNA and the woodchuck hepatitis virus post-transcriptional regulatory element, which was expressed under control of the pan-neuronal human synapsin I promoter (syn-GCaMP7). rAAV2/1-syn-GCaMP7 (AAV-syn-GCaMP7) was produced following triple-transfection of HEK293 cells with pAAV-syn-GCaMP7, an adenoviral helper plasmid pAdeno and a chimeric helper plasmid encoding AAV2 rep/AAV1 cap genes (pAAV2-1, gifted to us from Dr J. M. Wilson, University of Pennsylvania)³⁶, which was mediated by calcium phosphate co-precipitation with active gassing³⁷. rAAV2/1-syn-GCaMP7 was purified according to previous studies^38,39. All rAAV2/9 were obtained from the University of Pennsylvania Gene Therapy Program Vector Core.

Virus injection

Thirty minutes before surgery, dexamethasone sodium phosphate (1.32 mg kg⁻¹; to prevent cerebral edema), the antibiotics sulfadiazine (24 mg kg⁻¹) and trimethoprim (4.8 mg kg⁻¹) and the anti-inflammatory carprofen (6 mg kg⁻¹) were administered intraperitoneally. Isoflurane (1%) inhalation was used for anesthesia throughout surgery. A 2-mm diameter circular craniotomy was made over the left rostral motor forelimb area (circle centered ~2.4 mm anterior and ~0.9 mm lateral to bregma) or the left caudal motor forelimb area (circle centered ~0.2 mm anterior and ~1.2 mm lateral to bregma). The dura mater was removed. Before virus injection, a pulled glass pipette (broken and beveled to an outer diameter of ~25–30 μm; Sutter Instruments, California, USA) and a 5 μl Hamilton syringe were back-filled with mineral oil (Nacalai Tesque, Kyoto, Japan) and front-loaded with virus solution. Either 50 nl of AAV-syn-GCaMP7 (7.1 × 10¹² vector genomes ml⁻¹) or 50 nl of mixed solution containing AAV-syn-GCaMP6f (9.0 × 10¹² vector genomes ml⁻¹), AAV-flex-ChR2-mCherry (3.8 × 10¹² vector genomes ml⁻¹) and AAV-CMV-Cre (2.2 × 10¹⁰ vector genomes ml⁻¹) was injected via a syringe pump at a rate of 0.1 μl min⁻¹ (KDS310; KD Scientific, Massachusetts, USA) to a depth of 300 μm from the cortical surface. The pipette was inserted vertically and maintained in place for 10 min after the injection. The injection was repeated at 1–3 sites within the rostral motor forelimb area or the caudal motor forelimb area. After the pipette was slowly withdrawn, 4% (w v⁻¹) agarose L (Nippon Gene, Tokyo, Japan) was placed over the craniotomy, a 4.5-mm-diameter glass coverslip (number 0 thickness; Matsunami Glass, Osaka, Japan) was pressed onto the agarose surface¹³, and the edges were sealed with dental adhesive resin cement. Mice were then returned to their cages. Mice were allowed to recover and to express sufficient doses of GCaMP7 or GCaMP6f and ChR2 proteins for 2–4 weeks before imaging experiments were performed.

Lever-pull task

Seven AAV-syn-GCaMP7-transduced mice were used for the self-initiated lever-pull task, which was modified from a previous study¹³. In brief, water-deprived mice were trained to pull a lever for a distance of 5 mm over a time period of 400 ms to obtain a 4 μl drop of water. The water delivery was accompanied by a small click from the water valve. The animals were required to leave the lever at its original position and wait >3 s before pulling the lever again to receive the next water drop.

Optical SNOC

The seven mice injected with AAV-syn-GCaMP7 and trained at the lever-pull task were used for SNOC by two-photon calcium imaging (2pSNOC). Two-photon calcium imaging was conducted using a FV1000-MPE system (Olympus, Tokyo, Japan) and a mode-locked Ti:sapphire laser (MaiTai HP; Spectra Physics, California, USA) at a wavelength of 910 nm. A 690-nm shortpass dichroic mirror (Olympus) was used to separate the excitation laser and the emitted fluorescence. Fluorescence emissions were collected using a GaAsP photomultiplier tube (PMT; Hamamatsu Photonics, Shizuoka, Japan). The laser intensity was adjusted to 5–36 mW to maintain a constant baseline level of fluorescence in the neurons. The imaged field was 128–512 × 128–268 μm (mean, 380 × 199 μm; n=24 fields from seven mice) and the depth of the imaging plane was 162.5±45.2 μm below the cortical surface (n=24). The frame duration was 216–311 ms (mean, 292 ms, n=24). The fluorescence intensity of multiple neurons on the monitor of the FV1000-MPE was continuously detected with a machine vision camera (Basler, Ahrensburg, Germany) connected to a personal computer, and was analyzed in real time using a custom-developed and -deployed program written with the assistance of the LabVIEW vision module (National Instruments, Texas, USA). A region of interest (ROI) containing a single neuron was specified, and the mean fluorescence intensity of this ROI was continually measured. When the fluorescence intensity of the ROI remained under the manually determined threshold (3–6 s.d. of the baseline fluorescence intensity) for >1 s and then exceeded the threshold, the mouse was rewarded with two 4-μl drops of water from a spout near the mouth.

Simultaneous two-photon imaging and ChR2 photostimulation

Six mice injected with AAV-syn-GCaMP6f, AAV-flex-ChR2-mCherry and AAV-CMV-Cre were used for the simultaneous two-photon calcium imaging and ChR2 photostimulation experiments. Relatively low-titered AAV-CMV-Cre (2.2 × 10¹⁰ vector genomes ml⁻¹) was stochastically transduced to a limited number of neurons, and relatively high-titered AAV-flex-ChR2-mCherry (3.8 × 10¹² vector genomes ml⁻¹) was used to enable strong expression of ChR2 in a sparse subset of neurons. The water-deprived mice were acclimatized to the head restraint and delivery of water from the spout under a microscope for 2 days before the start of the photostimulation experiments. Two-photon calcium imaging was performed in the same way as in the 2pSNOC experiments, except that multialkali PMTs were used instead of GaAsP PMTs to prevent photodamage. The imaged field was fixed to 320 × 160 μm (n=11 from six mice) and the depth of the imaged plane was 128.5±17.5 μm below the cortical surface (n=11). The frame duration was 294 ms. A blue-light source (cat. no. MBL-H-473, OptoEngine LLC, Utah, USA) was used for photostimulation^21,40 and bifurcated 473-nm light-emitting diodes (LEX2-B; BrainVision, Tokyo, Japan) were set in front of both eyes for masking. The frames that included either photostimulation or masking and one subsequent frame were discarded from the analysis. A 480-nm shortpass dichroic mirror (Olympus) was used to separate the excitation blue light and the emitted fluorescence.

Offline processing of imaging data

Analyses were performed using ImageJ software (version 1.45s; National Institutes of Health, Maryland, USA) and MATLAB software (versions R2010a, R2011b, MathWorks, Massachusetts, USA). Image sequences were corrected for focal plane displacements by applying the ImageJ plug-in, TurboReg⁴¹. ROIs were manually specified. Averaging the fluorescence of all pixels within each ROI resulted in a raw time series, from which time-averaged background fluorescence measured in unstained blood vessels was subtracted. The detrended relative change in fluorescence (ΔF/F) was calculated from the subtracted time series using the eighth percentile value from 15 s before to 15 s after each sample time point^12,13. Non-negative deconvolution²³ of each ΔF/F trace was performed using a decay time constant of 1 s and divided by the median absolute deviation of the deconvoluted trace × 1.3826 × 2 (2 σ). A value of 1 was subtracted from these values. Then, if the values were negative, they were set to zero. The final values were defined as ‘activity’. The ROIs whose activity demonstrated skewness >1 were defined as reconstructed neurons¹³. Further analysis was limited to the reconstructed neurons.

In the experiments involving simultaneous two-photon calcium imaging and ChR2 photostimulation, the ΔF/F trace without deconvolution was used for analysis because the light stimulation interrupted continuous monitoring of fluorescence and, therefore, precluded deconvolution. The ΔF/F values in interrupted frames were removed from the analysis. For each neuron, the ΔF/F value 588 ms (two frames) after photostimulation was compared with the ΔF/F value 588 ms after masking stimulation using a one-tailed two-sample t-test. If ΔF/F after photostimulation was significantly larger (P<0.05) than ΔF/F after masking stimulation, the neuron was defined as a photostimuli-responsive neuron. The number of photostimuli-responsive neurons was 14.8±9.7 (n=11 fields) per field of view (320 × 160 μm) and this accounted for 43.9% of reconstructed neurons. mCherry fluorescence was detected in 4.9±2.2 cells (n=8 fields in which mCherry fluorescence was recorded) per field of view (320 × 160 μm) by in vivo two-photon imaging and this accounted for 15.0% of reconstructed neurons; thus, neurons without sufficiently strong mCherry fluorescence should have been included in photostimuli-responsive neurons. ChR2-mCherry-expressing neurons out of the focal plane might also play a role in activating photostimuli-responsive neurons. In 90.2% (156/173) of the photostimuli-responsive neurons, the peak fluorescence value on the first trial was more than half of the peak of the mean fluorescence value on all trials. Thus, the photostimulation immediately evoked activity in the photostimuli-responsive neurons, which eliminates the possibility that the light itself gradually affected the neurons and induced conditioned light responses.

Definition of lever-related and lever-unrelated neurons

Eight hundred and forty reconstructed neurons were categorized according to their association with the lever movement during the pre-conditioning period. The period from 1 s before the onset of the lever-pull movement to 3 s after the lever had returned to the original position was defined as the lever-related period. During the lever-related period, neurons showing activity that was significantly larger (P<0.05 by one-tailed two-sample t-test) than that outside the lever-related period were defined as lever-related neurons. The remaining neurons were defined as lever-unrelated neurons.

Definition of RSI

The 5 s after water delivery was defined as the reward-related period. Spontaneous pairwise correlation was the correlation between the activity of two neurons in reward-unrelated periods of the conditioning period. Activity in reward-related periods was excluded from this analysis to remove any effect of reward-related activity and/or behaviours. The neuronal activity in the first 10 min of the conditioning period was used to calculate RSI(t). Activity in the final 5 min of the conditioning period was excluded to remove any effect of activity gain and reward gain. RSI(t) was defined as the sum of activity in a 0.33 s bin t s after reward delivery divided by the sum of activity in reward-unrelated periods. If the time bin t was included in the reward-related period, RSI(t) was defined as the sum of activity in a 0.33 s bin t s after reward delivery divided by the sum of activity t s after reward delivery and activity in reward-unrelated periods. The duration of the time bin was 0.33 s, which was approximately the duration of a single frame. Because the activity calculated by non-negative deconvolution was ≧0, RSI(t) ranged from 0 to 1 by definition.

Analysis of licking behaviour

Licking behaviour was monitored at 30 Hz with an infrared video camera during the 2pSNOC experiments and by electrophysiological methods during the photostimulation experiments. To quantify the number of licks per second from the video data, an ROI that included the tongue was selected within the movie and the number of times that the mean intensity of the ROI exceeded a threshold (the mean of the maximum and minimum values in a data set) was counted. To quantify the number of licks per second from the electrophysiological data, the electropotential difference between the spout and the head of the mouse was measured and sampled by the FV1000-MPE. The number of times that the electropotential exceeded a threshold (the mean of the maximum and minimum values in a data set) was counted.

Statistics

Data are presented as mean±s.d. Student’s t-test, the Wilcoxon signed-rank test, the Wilcoxon rank sum test, Spearman’s correlation test, Pearson’s correlation test and χ²-test were used for statistical comparisons. Pairwise comparisons were two-tailed unless otherwise noted. Error bars in graphs represent the s.e.m. ANOVA and post-hoc Tukey’s HSD (honestly significant difference) test were used for multiple comparisons. No statistical tests were run to predetermine sample size. Blinding and randomization were not performed.

Additional information

How to cite this article: Hira R. et al. Reward-timing-dependent bidirectional modulation of cortical microcircuits during optical single-neuron operant conditioning. Nat. Commun. 5:5551 doi: 10.1038/ncomms6551 (2014).

References

Green, A. M. & Kalaska, J. F. Learning to move machines with the mind. Trends Neurosci. 34, 61–75 (2011).
Article CAS Google Scholar
Moritz, C. T., Perlmutter, S. I. & Fetz, E. E. Direct control of paralysed muscles by cortical neurons. Nature 456, 639–642 (2008).
Article ADS CAS Google Scholar
Fetz, E. E. Operant conditioning of cortical unit activity. Science 163, 955–958 (1969).
Article ADS CAS Google Scholar
Fetz, E. E. Volitional control of neural activity: implications for brain–computer interfaces. J. Physiol. 579, 571–579 (2007).
Article CAS Google Scholar
Schafer, R. J. & Moore, T. Selective attention from voluntary control of neurons in prefrontal cortex. Science 332, 1568–1571 (2011).
Article ADS CAS Google Scholar
Izhikevich, E. M. Solving the distal reward problem through linkage of STDP and dopamine signaling. Cereb. Cortex 17, 2443–2452 (2007).
Article Google Scholar
Legenstein, R., Pecevski, D. & Maass, W. A learning theory for reward-modulated spike-timing-dependent plasticity with application to biofeedback. PLoS Comput. Biol. 4, e1000180 (2008).
Article ADS MathSciNet Google Scholar
Legenstein, R., Chase, S. M., Schwartz, A. B. & Maass, W. A reward-modulated hebbian learning rule can explain experimentally observed network reorganization in a brain control task. J. Neurosci. 30, 8400–8410 (2010).
Article CAS Google Scholar
Fetz, E. E. & Baker, M. A. Operantly conditioned patterns on precentral unit activity and correlated responses in adjacent cells and contralateral muscles. J. Nuerophysiol 36, 179–204 (1973).
Article CAS Google Scholar
Hwang, E. J., Bailey, P. M. & Andersen, R. A. Volitional control of neural activity relies on the natural motor repertoire. Curr. Biol. 23, 353–361 (2013).
Article CAS Google Scholar
Arduin, P.-J., Frégnac, Y., Shulz, D. E. & Ego-Stengel, V. ‘Master’ neurons induced by operant conditioning in rat motor cortex during a brain-machine interface task. J. Neurosci. 33, 8308–8320 (2013).
Article CAS Google Scholar
Dombeck, D. A., Graziano, M. S. & Tank, D. W. Functional clustering of neurons in motor cortex determined by cellular resolution imaging in awake behaving mice. J. Neurosci. 29, 13751–13760 (2009).
Article CAS Google Scholar
Hira, R. et al. Spatiotemporal dynamics of functional clusters of neurons in the mouse motor cortex during a voluntary movement. J. Neurosci. 33, 1377–1390 (2013).
Article CAS Google Scholar
Lütcke, H., Margolis, D. J. & Helmchen, F. Steady or changing? Long-term monitoring of neuronal population activity. Trends Neurosci. 36, 375–384 (2013).
Article Google Scholar
Komiyama, T. et al. Learning-related fine-scale specificity imaged in motor cortex circuits of behaving mice. Nature 464, 1182–1186 (2010).
Article ADS CAS Google Scholar
Masamizu, Y. et al. Two distinct layer-specific dynamics of cortical ensembles during learning of a motor task. Nat. Neurosci. 17, 987–994 (2014).
Article CAS Google Scholar
Gage, G. J., Ludwig, K. A., Otto, K. J., Ionides, E. L. & Kipke, D. R. Naive coadaptive cortical control. J. Neural Eng. 2, 52–63 (2005).
Article ADS Google Scholar
Gradinaru, V. et al. Targeting and readout strategies for fast optical neural control in vitro and in vivo. J. Neurosci. 27, 14231–14238 (2007).
Article CAS Google Scholar
Isomura, Y. et al. Reward-modulated motor information in identified striatum neurons. J. Neurosci. 33, 10209–10220 (2013).
Article CAS Google Scholar
Ohkura, M. et al. Genetically encoded green fluorescent Ca²⁺ indicators with improved detectability for neuronal Ca²⁺ signals. PLoS One 7, e51286 (2012).
Article ADS CAS Google Scholar
Hira, R. et al. In vivo optogenetic tracing of functional corticocortical connections between motor forelimb areas. Front. Neural Circuits 7, 55 (2013).
Article Google Scholar
Tennant, K. A. et al. The organization of the forelimb representation of the C57BL/6 mouse motor cortex as defined by intracortical microstimulation and cytoarchitecture. Cereb. Cortex 21, 865–876 (2011).
Article Google Scholar
Vogelstein, J. T. et al. Fast nonnegative deconvolution for spike train inference from population calcium imaging. J. Neurophysiol. 104, 3691–3704 (2010).
Article Google Scholar
O'Connor, D. H., Peron, S. P., Huber, D. & Svoboda, K. Neural activity in barrel cortex underlying vibrissa-based object localization in mice. Neuron 67, 1048–1061 (2010).
Article CAS Google Scholar
Schultz, W. Multiple dopamine functions at different time courses. Annu. Rev. Neurosci. 30, 259–288 (2007).
Article CAS Google Scholar
Chen, T.-W. et al. Ultrasensitive fluorescent proteins for imaging neuronal activity. Nature 499, 295–300 (2013).
Article ADS CAS Google Scholar
Clancy, K. B., Koralek, A. C., Costa, R. M., Feldman, D. E. & Carmena, J. M. Volitional modulation of optically recorded calcium signals during neuroprosthetic learning. Nat. Neurosci. 17, 807–809 (2014).
Article CAS Google Scholar
Gandolfo, F., Li, C.-S., Benda, B., Schioppa, C. P. & Bizzi, E. Cortical correlates of learning in monkeys adapting to a new dynamical environment. Proc. Natl Acad. Sci. USA 97, 2259–2263 (2000).
Article ADS CAS Google Scholar
Abbott, L. F. & Nelson, S. B. Synaptic plasticity: taming the beast. Nat. Neurosci. 3, 1178–1183 (2000).
Article CAS Google Scholar
Bao, S., Chan, V. T. & Merzenich, M. M. Cortical remodelling induced by activity of ventral tegmental dopamine neurons. Nature 412, 79–83 (2001).
Article ADS CAS Google Scholar
Bao, S., Chan, V. T., Zhang, L. I. & Merzenich, M. M. Suppression of cortical representation through backward conditioning. Proc. Natl. Acad. Sci. USA 100, 1405–1408 (2003).
Article ADS CAS Google Scholar
Molina-Luna, K. et al. Dopamine in motor cortex is necessary for skill learning and synaptic plasticity. PLoS One 4, e7082 (2009).
Article ADS Google Scholar
Garris, P. & Wightman, R. Different kinetics govern dopaminergic transmission in the amygdala, prefrontal cortex, and striatum: an in vivo voltammetric study. J. Neurosci. 14, 442–450 (1994).
Article CAS Google Scholar
Shen, W., Flajolet, M., Greengard, P. & Surmeier, D. J. Dichotomous dopaminergic control of striatal synaptic plasticity. Science 321, 848–851 (2008).
Article ADS CAS Google Scholar
Koralek, A. C., Jin, X., Long, J. D. II, Costa, R. M. & Carmena, J. M. Corticostriatal plasticity is necessary for learning intentional neuroprosthetic skills. Nature 483, 331–335 (2012).
Article ADS CAS Google Scholar
Xiao, W. et al. Gene therapy vectors based on adeno-associated virus type 1. J. Virol. 73, 3994–4003 (1999).
CAS PubMed PubMed Central Google Scholar
Okada, T. et al. Large-scale production of recombinant viruses by use of a large culture vessel with active gassing. Hum. Gene Ther. 16, 1212–1218 (2005).
Article CAS Google Scholar
Okada, T. et al. Scalable purification of adeno-associated virus serotype 1 (AAV1) and AAV8 vectors, using dual ion-exchange adsorptive membranes. Hum. Gene Ther. 20, 1013–1021 (2009).
Article CAS Google Scholar
Masamizu, Y. et al. Local and retrograde gene transfer into primate neuronal pathways via adeno-associated virus serotype 8 and 9. Neurosci 193, 249–258 (2011).
Article CAS Google Scholar
Hira, R. et al. Transcranial optogenetic stimulation for functional mapping of the motor cortex. J. Neurosci. Methods 179, 258–263 (2009).
Article Google Scholar
Thevenaz, P., Ruttimann, U. E. & Unser, M. A pyramid approach to subpixel registration based on intensity. IEEE Trans. Image Process 7, 27–41 (1998).
Article ADS CAS Google Scholar

Download references

Acknowledgements

We thank Ms J. Saito and M. Himeno for technical assistance, Dr T. Chiyo for assistance with AAV purification and Dr Y.R. Tanaka for helpful discussion. We are grateful to the Functional Genomics and the Spectrography and Bioimaging Facility at NIBB for allowing us to use the equipment. We thank Dr L.L. Looger at HHMI for providing the GCaMP6f vector, Dr K. Deisseroth at Stanford University for providing pAAV and Dr J.M. Wilson at the University of Pennsylvania for providing helper plasmids pAAV2-1 and pAAV2-9. This work was supported by Scientific Research on Innovative Areas ‘Mesoscopic Neurocircuitry’ (No. 22115005 to M.M.), a Scientific Research grant (No. 23300148 to M.M.), Grants-in-Aid for Young Scientists (No. 26830020 to Y.M. and No. 26830021 to R.H.), the Strategic Research Program for Brain Sciences ‘Construction of System for Spread of Primate Model Animals’ (to M.M.), a Research Activity Start-up fund (No. 25893287 to R.H.) from the Ministry of Education, Culture, Sports, Science and Technology, and Mitsubishi Foundation, Takeda Foundation and Toyoaki Foundation grants to M.M.

Author information

Authors and Affiliations

Division of Brain Circuits, National Institute for Basic Biology and the Graduate University of Advanced Studies (Sokendai), Myodaiji, Okazaki, Japan
Riichiro Hira, Fuki Ohkubo, Yoshito Masamizu & Masanori Matsuzaki
Japan Science and Technology Agency, CREST, Saitama, 332-0012, Japan
Riichiro Hira, Fuki Ohkubo, Yoshito Masamizu & Masanori Matsuzaki
Brain Science Institute, Saitama University, Saitama, 338-8570, Japan
Masamichi Ohkura & Junichi Nakai
Department of Biochemistry and Molecular Biology, Nippon Medical School, Tokyo, 113-8602, Japan
Takashi Okada

Authors

Riichiro Hira
View author publications
You can also search for this author in PubMed Google Scholar
Fuki Ohkubo
View author publications
You can also search for this author in PubMed Google Scholar
Yoshito Masamizu
View author publications
You can also search for this author in PubMed Google Scholar
Masamichi Ohkura
View author publications
You can also search for this author in PubMed Google Scholar
Junichi Nakai
View author publications
You can also search for this author in PubMed Google Scholar
Takashi Okada
View author publications
You can also search for this author in PubMed Google Scholar
Masanori Matsuzaki
View author publications
You can also search for this author in PubMed Google Scholar

Contributions

R.H. and M.M. designed the experiments. R.H. and F.O. conducted the experiments. M.O. and J.N. generated GCaMP7. Y.M. and T.O. generated AAV1-GCaMP7. R.H. analyzed the data. R.H. and M.M. wrote the paper, along with comments from all authors.

Corresponding author

Correspondence to Masanori Matsuzaki.

Ethics declarations

Competing interests

The authors declare no competing financial interests.

Rights and permissions

Reprints and permissions

About this article

Cite this article

Hira, R., Ohkubo, F., Masamizu, Y. et al. Reward-timing-dependent bidirectional modulation of cortical microcircuits during optical single-neuron operant conditioning. Nat Commun 5, 5551 (2014). https://doi.org/10.1038/ncomms6551

Download citation

Received: 01 July 2014
Accepted: 13 October 2014
Published: 24 November 2014
DOI: https://doi.org/10.1038/ncomms6551

This article is cited by

Existing function in primary visual cortex is not perturbed by new skill acquisition of a non-matched sensory task
- Brian B. Jeon
- Thomas Fuchs
- Sandra J. Kuhlman
Nature Communications (2022)
Optical deep-cortex exploration in behaving rhesus macaques
- Masanori Matsuzaki
- Teppei Ebina
Nature Communications (2021)
Operant conditioning of motor cortex neurons reveals neuron-subtype-specific responses in a brain-machine interface task
- Martha Gabriela Garcia-Garcia
- Cesar Marquez-Chin
- Milos R. Popovic
Scientific Reports (2020)
Calcium imaging reveals glial involvement in transcranial direct current stimulation-induced plasticity in mouse brain
- Hiromu Monai
- Masamichi Ohkura
- Hajime Hirase
Nature Communications (2016)

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.