Early, sustained and broadly-tuned discharge of fast-spiking interneurons in the premotor cortex during action planning

Preparatory neural activity in premotor areas is critical for planning and execution of voluntary movements. Previous studies in monkeys and mice have revealed how the discharges of pyramidal, excitatory neurons (PNs) encode a motor plan for an upcoming movement (Afshar et al., 2011; Chen et al., 2017; Li et al., 2015). However, the contribution of GABAergic interneurons, specifically fast-spiking interneurons (FSNs), to voluntary movements remains poorly understood. Putative premotor areas involved in action planning have been demonstrated in rodents. In particular, in mice, a premotor area controlling voluntary licking has been identified in the anterior-lateral motor cortex (ALM) (Komiyama et al., 2010). Also, ALM partially overlaps with the rostral forelimb area (RFA), the previously defined premotor region involved in the control of paw movement in rats and mice (Rouiller et al., 1993; Tennant et al., 2011). To understand the excitatory-inhibitory microcircuit involved in action planning, here I compare directly the response properties of PNs and FSNs during licking behaviour and forelimb retraction in the mouse. Recordings are carried out with both acute electrodes and chronic microelectrode arrays from both the two premotor areas, i.e. the ALM – responsible for licking –, and RFA – involved in paw movement. Specifically, in a first set of experiments, I used head-restrained mice that spontaneously lick a reward delivered at random intervals from a drinking spout. Mice voluntary performed either single isolated or a burst of consecutive licks, which I categorized, a posteriori, in single (= 1 lick) and multiple licks (≥ 3 licks). During the task, I extracellularly recorded single units’ activity from ALM, using acute in vivo electrophysiology. I identified putative PNs and FSNs, based on well-established features of their waveforms, and investigated their functional properties during the movement. Unexpectedly, I report that optogenetically-verified FSNs showed an earlier and more sustained activation than PNs. In particular, most of the neurons’ activity anticipated the licking onset, consistently with an involvement of the ALM in movement planning. The majority of the neurons (~90%) increased their firing frequency in correspondence with the movement, but suppressive modulations were also observed in a subset of units. For both PNs and FSNs, I found significantly greater discharge during multiple than single licks and the peak discharge was significantly delayed for both subclasses during multiple licking events. However, FSNs modulated their activity about 100ms earlier than PNs. Furthermore, almost all FSNs showed a peak in their response before the beginning of the sequence of licks. Analysis of mean information content confirms that FSNs predict licking onset not only significantly better, but even earlier, than PNs. Chronic electrode arrays covering both the ALM and RFA were next used to simultaneously probe neural responses during (i) licking and (ii) forelimb pulling in a robotic device (Spalletti et al., 2017). I report that most of the FSNs respond with a stereotyped increase in their firing rates during both licking and pulling. In stark contrast, PNs show a variety of behaviours, dependent on movement type. At least for a minority of them, licking behaviour and forelimb retraction are represented as two different motor acts, reaching significant levels in the PNs. Accordingly, computational analysis shows that PNs carry more independent information than FSNs. Altogether, these data indicate that a global rise of GABAergic inhibition mediated by FSNs firing contributes to early action planning. Next, encouraged by the deeper understanding of the cortical microcircuits underlying movement planning in mice, I exploited this knowledge to explore more complex mechanisms, as action understanding. The neural circuits that integrate performed and observed actions have been found in the premotor cortex of monkeys and named as ‘mirror neurons system’ (di Pellegrino et al., 1992). Recently, the presence of mirror neurons have been demonstrated in rodents in the anterior cingulate cortex (Carrillo et al., 2019), but whether they could contribute to action understanding in the premotor cortex is still unclear. At behavioural level, the observation of actions can actually lead, in some cases, to the repetition of those same actions. This phenomenon has been named social facilitation, and the underlying motor program has been attributed to the mirror system (Ferrari et al., 2005). Here, I set up a behavioural task similar to the one exploited in monkeys to explore social facilitation in mice. I took advantage of licking behaviour to set up the social facilitation experiment. Therefore, head-restrained mice were allowed to lick water from a feeding needle. I found that mice can actually facilitated to lick more when another individual was engaged in the same action, supporting the hypothesis of a social facilitation in mouse. Altogether these results indicate that the observers’ behaviour was actually influenced by the demonstrators’ one, laying the groundwork for the study of mirror neurons in mice at cellular level.