Modulation of excitatory but not inhibitory synaptic inputs in the mouse primary motor cortex in the late phase of motor learning

Abstract

Motor skill training induces functional and structural changes in the primary motor cortex. New dendritic spines are formed with training and the horizontal connections in the layer II/III area of the primary motor cortex are strengthened. Here we investigated the functional synaptic properties of pyramidal neurons following motor skill training. We trained mice on a single forelimb-reaching task for five days and performed whole cell recordings from layer II/III pyramidal neurons in the forelimb representation area of the primary motor cortex in the ipsilateral (untrained) and contralateral (trained) hemispheres in acute brain slices. Success rate in the forelimb-reaching task rapidly improved over the first 3 days and stabilized on subsequent days. After five days of training, a time at which learning has peaked and synaptic strengthening with field potential recordings show enhancement, we observed an increase in mEPSC frequency while increases in mEPSC amplitudes was only observed in 20% of the cells. Increase in excitatory synaptic properties were correlated with improved motor skill. Measurement of miniature IPSC (mIPSC) after five days of training showed no difference in either frequency or amplitude between the trained and untrained hemispheres. Our present results indicate dynamic changes in excitatory but not inhibitory synapses in M1 layer II/III pyramidal neurons at the late stages of motor skill learning.

Introduction

Strengthening of synaptic connection is thought to underlie learning and memory formation [1]. The primary motor area (M1) is a cortical region important for skilled voluntary movements that also participates in learning motor skills. Motor skill training on a forelimb reaching task results in the strengthening of the synapses in the primary motor cortex as evidenced by increase in the field potentials evoked by stimulation of M1 horizontal connections [2–5]. Moreover, the amount of LTP that could be induced electrically or chemically in slices was reduced after motor skill training, suggesting that the observed strengthening of horizontal connections may involve an LTP-like mechanism [3–7]. The involvement of an LTP-like mechanism is further suggested by findings of an increase in size as well as the formation and stabilization of dendritic spines, the postsynaptic structures of excitatory synapses, and increase in synaptic AMPA receptor subunit GluA1 with motor skill training [4, 5, 8–11]. In addition, reduction of GABA inhibition has been shown to be necessary for LTP induction in M1 [12, 13] and recent studies have also demonstrated that motor skill training results in plasticity of the axonal boutons of inhibitory neurons in the motor cortex [14]. Although it is clear that motor skill learning leaves a structural and electrophysiological trace in M1, the synaptic properties of layer II/III pyramidal neurons following motor skill reaching task have not yet been examined. To explore the learning-induced changes in synaptic inputs onto layer II/III pyramidal neurons we have recorded miniature excitatory and inhibitory synaptic events from trained and untrained hemispheres of trained mice. We report an increase in the frequency of miniature excitatory postsynaptic currents (mEPSCs) after 5 days of training and an increase in amplitudes of mEPSCs in a subset of neurons. Interestingly, no changes were observed in inhibitory synaptic inputs after 5 days of training. Our data suggest that formation and stabilization of new spines is likely accompanied by strengthening of preexisting synapses and is likely to contribute to the synaptic strengthening observed at the late stages of motor skill learning.

Methods

Animals:

Mice were cared for in accordance with NIH guidelines for laboratory animal welfare. The mice were kept on a 12-hour light/dark cycle and provided ad libitum access to water and rodent chow (Teklad rodent diet 8656). Mice were housed in micro-isolator cages with corncob laboratory animal bedding. All experiments were approved by the University of Nebraska Medical Center Institutional Animal Care and Use Committee. Male C57BL/6 mice were used for behavioral and electrophysiology experiments.

Motor skill training:

Training was performed as previously described [5]. Four to five week old mice were food-restricted (85% of their free-feeding weight) in order to motivate them to perform the motor skill task. Food chow was removed the night before the first training session. Mice ate food pellets (Dustless Precision Pellets 20 mg, rodent grain-based diet, Bioserv Cat # F0163) during and after the training session and were food restricted in the evening. The weight of the mouse was monitored daily and if the weight dropped below 85% additional food pellets were given prior to training. Mice were placed in a Plexiglas box with an attached platform that could be reached through a thin slit on the front wall (Fig. 1A). Mice were trained to reach through the slit with their preferred forelimb and grasp and retrieve individual food pellets. An initial pre-training session determined forelimb preference, and pellets were then placed on the side that enabled the use of the preferred forelimb only. Mice had one training session per day that lasted 30 minutes or 100 reaches. Motor skill performance was quantified by the success rate (% of successful retrievals). The trained hemisphere (tr) is contralateral to the trained forelimb and the untrained hemisphere (utr), acts as an internal control.

Figure 1: Motor skill training paradigm using the forelimb reaching task.

A) A schematic of the training box. A mouse reaches for a food pellet through a slit with the preferred forelimb. The trained hemisphere is contralateral to the trained forelimb. B) Fraction of successful attempts on each of the 5 training days. n= 20 mice C) Experimental timeline for training and electrophysiology experiments. * P<0.05, ^**** P<0.0001 (Repeated-Measures ANOVA)

Electrophysiology:

Slices were prepared 12–15 hours after the last training session, and the experimenter was blind to the identity of the trained forelimb. Anesthetized mice (Avertin, 0.25mg/g body weight) were decapitated and their brains quickly removed and immersed in ice-cold oxygenated (95% O₂/5% CO₂) artificial cerebrospinal fluid (ACSF) containing (in mM) 126 NaCl, 3 KCl, 1.25 NaH₂PO₄, 4 MgSO₄, 2 CaCl₂, 26 NaHCO₃, and 10 dextrose (300 mOsm). Acute coronal slices containing the forelimb motor regions collected from 1.0 mm anterior to 0.5mm posterior to the bregma [15] were cut to 300μm using a vibratome. Slices were allowed to rest for at least 60 minutes before recordings started. For recording, slices were transferred to the submersion-type recording chamber and superfused at room temperature (22–24°C) with ACSF saturated with 95% O₂/5% CO₂ containing (in mM) 126 NaCl, 3 KCl, 1.25 NaH₂PO₄, 1 MgSO₄, 2 CaCl₂, 26 NaHCO₃, and 10 dextrose (300 mOsm). Somatic whole-cell recordings were obtained from upper layer II pyramidal cells of forelimb region of M1 (1 to 2 mm lateral to the midline) visualized with infrared differential interference contrast optics. Whole-cell pipette solution for mEPSC recordings contained (in mM): Potassium gluconate 130, KCl 5, HEPES 10, MgCl₂2, di- tris- Phosphocreatinine 10, NaATP 4, NaGTP 0.4 (pH adjusted to 7.2 with KOH, osmolarity 290–295 mOsM). For recording mIPSCs, the pipette solution contained (in mM): KCl 120, HEPES 10, MgCl₂2, di- tris- Phosphocreatinine 10, NaATP 4, NaGTP 0.4 (pH adjusted to 7.2 with KOH, osmolarity 290–295 mOsM). The morphological tracer Alexa Fluor 594 (20 μM) was also included in the pipette solution to help in the identification of pyramidal cells. Recordings were performed using an Axoclamp-2B amplifier (Axon Instruments), signals were filtered at 3kHz, digitized (Digidata 1322A; Molecular Devices), and sampled at 10 kHz using pClamp 9.0 (Molecular Devices). Data was analyzed offline using Mini-Analysis software (Synaptosoft). mEPSCs and mIPSCs were recorded at −70 mV. For mEPSC recordings, ACSF contained 1 μM TTX (Tocris) and 10 μM bicuculline methiodide (Sigma). For mIPSCs recordings, ACSF contained 1 μM TTX, 20 μM NBQX and 50 μM AP-5. The access resistance and leak current were monitored and recordings were rejected if these parameters changed significantly during the data acquisition. For analysis of mEPSCs, 5 minutes of recording was used. Analysis of mIPSCs was done on data recorded for 2 minutes.

Statistical analysis:

All values are means ± S.E.M. For behavioral analysis of success rate, repeated measures ANOVA was used. Normal distribution was tested using Kolmogorov-Smirnov test and variance was compared. Two-sided unpaired Student’s t tests, Mann Whitney test or one-way ANOVA with the Bonferroni method for post-hoc multiple comparison were used to determine whether synaptic current values were different between untrained and trained hemispheres. Numbers of cells or mice are indicated by n. Probability values (P) of less than 0.05 were considered to represent significant differences. Data were analyzed using GraphPad Prism software. Hierarchical clustering analysis was performed on data that included cells from both the trained and untrained hemispheres (Fig. 2D) using the Statistics and Machine Learning Toolbox function clusterdata in MATLAB.

Figure 2: Increased mEPSCs frequency and amplitude after 5 days of motor skill training.

A) Example traces of mEPSCs recorded from layer II/III pyramidal neurons in the forelimb M1 from untrained (utr) and trained (tr) hemispheres. B) Increased average mEPSCs frequency in the trained hemisphere. C) No change in average mEPSCs amplitude between the trained (n=18 cells) and untrained cells (n=14 cells) * P<0.05, unpaired t test. D) Each cell was plotted as a function of its averaged event amplitude and standard deviation. Few cells from trained-group had large averaged amplitudes and Standard-deviations. Using hierarchical clustering analysis the cells were divided into two groups (separated by the dotted line). E) The averaged amplitude of the small cluster from the trained hemisphere (n=4 cells) was significantly higher than that observed from the larger cluster of the trained hemisphere (n=14 cells) or from all the cells in the untrained hemisphere (n=14 cells). ^**** P<0.0001, One-way ANOVA.

Results

To determine how synaptic properties change with motor skill learning we trained mice on the forelimb-reaching task for five days (n=20 mice, Fig. 1). As previously described, success rate rapidly increased over the first 3 days after which most mice had little or no additional improvement. As compared to the first day of training, success rate was significantly increased on all other days (P=0.01 for day 2, P<0.0001 on days 3–5, Repeated-Measures ANOVA).

Coronal slices from the forelimb M1 were prepared 12–15 hours following 5 days of training on the forelimb reaching-task (Fig. 1C). Whole cell recordings from the superficial layer II/III pyramidal neurons were performed. In total, 14 mice were trained with recordings made from 11 slices from the untrained and trained hemispheres each. Following 5 days of motor skill training we found a 20% increase in frequency of the mEPSCs (untrained: 0.81±0.04 Hz, n=14 cells; trained: 0.97±0.06 Hz, n=18 cells. P=0.04, unpaired t-test. Fig. 2A,B). Analysis of the mEPSC amplitudes showed no change in the average amplitude (untrained: 16.53±0.7 pA, n=14 cells, trained: 18.31±1.28 pA, n=18 cells, P=0.2 unpaired t-test, Fig. 2A,C). These experiments demonstrate that number of synapses remains elevated in the trained hemisphere 5 days after training. Because only a subset of neurons is reliably activated with repeated training [16] we wanted to investigate if a subset of cells might show elevated mEPSC amplitudes. We plotted each cell as a function of its averaged event amplitude and standard deviation and found that a few (4/18) cells from the trained hemisphere had large averaged amplitudes and standard deviations. Hierarchical clustering analysis of all cells (from trained and untrained hemispheres) resulted in two clusters (small and large clusters) as shown in Fig. 2D. While all the cells in the untrained hemispheres clustered together (large cluster), the cells in the trained group segregated into two distinct sub-groups. 4 out of 18 cells in the trained hemisphere clustered separately (small cluster). When we compared the averaged mEPSC amplitudes of the small cluster to the cells from trained and untrained hemispheres of the large cluster we found that the small cluster cells had 70% larger mEPSC amplitude (tr small cluster: 26.88±1.08 pA, n=4 cells, tr large cluster: 15.85±0.77 pA, n=14 cells, P<0.0001; utr large cluster: 16.5±0.76 pA, n=14 cells, P<0.0001, F(2,29)= 25.64, one-way ANOVA, Fig. 2E). These results suggest that only a subset of neurons, possibly the ones repeatedly activated during the motor skill training, display a substantial increase in their synaptic strength.

The increase in number of excitatory synapses (increased frequency) as well as increase in the strength of excitatory synapses (increased amplitude) in a subset of neurons could contribute to the learning-induced increase in synaptic strength observed in the forelimb M1 in earlier studies [2, 4, 5]. Next, we asked if alteration in the inhibitory synapses could also be a contributing mechanism. To test this, we measured mIPSCs from the superficial layer II/III pyramidal neurons in the forelimb M1 following 5 days of training on the forelimb reaching-task (Fig. 3A). In total, 6 mice were trained and recordings were made from 9 and 8 slices from untrained and trained hemispheres respectively. Following 5 days of motor skill training we found no change in frequency of the mIPSCs (untrained: 4.61±0.31 Hz, n=12 cells; trained: 5.17±0.33 Hz, n=13 cells. P=0.23, unpaired t-test. Fig. 3A,B). Analysis of the mIPSC amplitudes also did not detect changes with training (untrained: 43.8±2.8 pA, n=12 cells, trained: 41.59±2.6 pA, n=13 cells, P=0.5 Mann-Whitney test, Fig 3A,C). These experiments demonstrate that motor skill training does not change the number and strength of inhibitory synapses on the layer II/III pyramidal neurons 5 days after training. Finally, we wanted to determine if there is relationship between increased mEPSC frequency or amplitude and learning. For each mouse we plotted the interhemisphere ratio (tr/utr) of average mEPSC frequency (Fig. 4A) or amplitude (Fig. 4B) and the amount of learning the mouse displayed (ratio of success rates on days 5 and 1). For the mice from which recordings were performed in both hemispheres we observed a positive correlation between amount of learning and interhemisphere ratio of mEPSC frequency (r²=0.6; P =0.12) and a weaker correlation with amplitude (r²=0.45; P=0.21) albeit these were not significant. Trained mice from which mIPSCs were recorded showed weak correlation between learning and mIPSCs frequency (r²=0.45, P=0.1) and no correlation between learning and interhemisphere mIPSCs amplitude (r²=0.1, P=0.45). These data suggest that there is a link between motor skill learning and enhanced synaptic properties similarly to what has been reported for dendritic spines [8].

Figure 3: No change in inhibitory synaptic inputs after 5 days of motor skill training.

A) Example traces of mIPSCs recorded from layer II/III pyramidal neurons in the forelimb M1 from untrained (utr) and trained (tr) hemispheres. B) No change in average mIPSCs frequency in the trained hemisphere. C) No change in average mIPSCs amplitude between the trained (n=13 cells) and untrained cells. (n=12 cells)

Figure 4: Correlation between mEPSCs and learning.

The degree of interhemisphere average mEPSC frequency (A) and amplitude (B) was positively correlated with the degree of learning the mice displayed (ratio of success rates on days 5 and 1). Although the degree of interhemisphere average mIPSC frequency (C) was weakly correlated with the degree of learning, this relationship was not observed for the mIPSC amplitude (D).

Discussion

The experiments described here were aimed to determine if structural synaptic changes that occur with motor skill learning are accompanied by alterations in functional synaptic inputs unto layer II/III pyramidal neurons in the primary motor cortex. While we found evidence for increased excitatory synaptic input, no change in inhibitory synaptic transmission has been observed.

Several studies have looked at modulations in excitatory and inhibitory synaptic plasticity by recording miniature synaptic events following different experience dependent plasticity paradigms [11, 17–20]. Following 5 days of training on the forelimb-reaching task we observed increased mEPSC frequency. Although untrained mice were not recorded from, these changes are likely to occur due to the behavioral training as previous studies have shown no interhemisphere differences in untrained mice [5]. The increases in mEPSC frequency that we observe following completion of motor skill learning could also be due to an increase in presynaptic glutamate release during the late phase of learning and could contribute to the enhanced evoked responses of the horizontal connection in M1.

After 5 days of training we also observed increased mEPSCs amplitudes but only in a subset of about 20% of the recorded neurons in the trained hemisphere. The increase in mEPSCs amplitude is consistent with previously reported increase in spine head widths after 5 days of forelimb reaching task [4, 21]. This too could contribute to the enhanced evoked responses following motor skill training [5]. The finding of only a subset of neurons affected is consistent with other studies where motor learning task induced substantial remodeling of spines in a small subpopulation of neurons (16%) in layer II/III [22]. In addition, large population of movement related neurons observed in the early phase of learning were shown to gradually reduce and result in a smaller and more stable population of movement related neurons towards the end of learning phase [16]. This emergence of a reproducible spatiotemporal activity of neurons in the late phase of learning is likely to result from strengthening of synapses on a subpopulation of neurons. Therefore, the changes in the mEPSC amplitudes observed in later stages of learning, most likely indicates that this subpopulation of neurons is reliably activated with repeated motor skill learning. Analysis of levels of glutamate receptors in dendritic spines will allow to reconcile the structural and functional synaptic changes that occur with motor skill learning [23]. Similar to our findings, rotarod training paradigm leads to increases in both mEPSC frequency and amplitude in layer II/III neurons following completion of the motor task [11]. Although the motor learning paradigm used in our study is different from the rotarod paradigm, it appears that similar effects in excitatory synaptic transmission are observed in the late phase of motor learning. These results indicate that motor training modulates excitatory synaptic inputs on layer II/III pyramidal neurons in the primary motor cortex. Furthermore, the enhanced synaptic properties we observed following 5 days of motor training was positively correlated with the degree of motor learning. Previous studies have shown that the degree of spine formation is positively correlated with the number of successful reaches during motor training [8][24]. These results together suggest that there is a link between learning and the synaptic changes observed.

We did not observe any changes in the mIPSC frequency and amplitude following 5 days of motor skill training. It is possible that a reduction in GABAergic inhibition occurs during the early phase (following 1 day training) and that these changes are restored to levels comparable to untrained conditions after completion of the motor skill task. Several other studies using other behavioral paradigms have reported changes in inhibitory synaptic transmission at different times following the experience [17–20]. Magnetic resonance spectroscopy studies have shown that modulation of GABA is associated with motor learning in humans [25] particularly a rapid short-term modulation during the acquisition of a motor learning task [26]. In a recent study, decreased mIPSC frequency and increased paired-pulse ratio for evoked IPSCs were observed in recordings performed following one day of rotarod training suggesting a decreased presynaptic GABA release probability [11]. There were no differences observed in the mIPSC frequency and the paired-pulse ratio recorded following the completion of the training sessions (following 2 day rotarod training) suggesting a transient reduction of GABAergic inhibition during early stages of learning [11]. In vivo imaging studies have also highlighted the contribution of distinct subtypes of inhibitory neurons involved in learning. The motor learning-induced spine dynamics that occurs in the distal branches of apical dendrites was shown to coincide with subtype-specific plasticity of inhibitory circuits [14]). These changes in inhibition were proposed to be necessary for spine formation and motor learning [14, 27]. In our study, the absence of changes in inhibitory synaptic transmission following 5 days of motor learning could be attributed to transient effects of inhibition during early stages of learning in the motor skill task.

In summary, our data are consistent with the structural analysis that has shown an increase in number of dendritic spines, and suggest that motor skill learning is associated with changes in excitatory synaptic inputs.

Highlights.

• Motor skill training results in increase in mEPSCs frequency in L2/3 neurons in the primary motor cortex.

• Increase in mEPSCs amplitude is only observed in a subset of L2/3 neurons.

• No changes in mIPSCs are observed following motor skill training.