Promoting endogenous associative plasticity in human primary motor cortex



As predicted by the Hebbian rule for associative plasticity, the strength of synaptic efficacy increases or decreases according to the specific time interval between pre- and postsynaptic activity due to spike timing-dependent plasticity (STDP). In healthy humans, one way to elicit cortical associative plasticity due to STDP is by applying paired associative stimulation (PAS), which consists of transcranial magnetic stimulation (TMS) delivered over the primary motor cortex (M1) combined with electric stimulation of a mixed peripheral nerve. When repetitive TMS pulses are delivered at 25 ms together with time-locked electric nerve stimulation at the contralateral wrist (PAS25), motor evoked potentials (MEPs) increase in size for approximately 60 min, whereas at 10 ms (PAS10) MEPs decrease in size (Stefan et al. 2000). As PAS implies repetitive activation of sensorimotor circuits at specific interstimulus intervals (ISIs), PAS-induced long-term changes in MEP amplitude reflect mechanisms of an associative Hebbian form of STDP (Stefan et al. 2000).

Several authors have recently developed new PAS protocols by coupling TMS over M1 with stimuli delivered over the contralateral M1 (Koganemaru et al. 2009; Rizzo et al. 2009), or targeting non-primary motor areas including the supplementary motor area (SMA) (Arai et al. 2011) and the ventral region of the premotor area (PMv) (Buch et al. 2011). More recently, we have developed a modified PAS protocol involving TMS paired with laser-induced nociceptive system activation (Laser-PAS). By delivering laser pulses that elicit laser evoked potentials (LEPs) over the skin, we found that when TMS follows laser stimulation at the specific ISI of the LEPs’ N1 component + 50 ms (Laser-PAS50), MEPs increase in size for approximately 60 min, reflecting cortical STDP arising from pain–motor integration processes (Suppa et al. 2012). Taken together, these studies demonstrate that Hebbian forms of associative plasticity operate in M1 as a general principle regardless of the nature of the specific stimuli coupled with TMS.

In addition to PAS protocols involving two exogenous stimuli (i.e. TMS and electric peripheral nerve activation), several authors have elicited associative plasticity in M1 by pairing exogenous stimuli with endogenous neural activity contributing to movement related cortical potentials (MRCPs) (Thabit et al. 2010; Mrachacz-Kersting et al. 2012). MRCPs are commonly recorded in the form of contingent negative variation (CNV) and Bereitschaftspotential (BP). The CNV consists of a slow negative potential that occurs between a ‘warning’ signal and a ‘go’ signal and can be divided in an early component (CNV1), 1.5–0.5 s before the ‘go’ signal, and a late component (CNV2), 0.5 s or less before the ‘go’ signal. By contrast, the BP precedes a self-initiated movement and consists of an early component (BP1), 1.5–0.5 s before movement onset, and a late component (BP2), 0.5 s or less before movement onset. Although the CNV and BP reflect different forms of neural activity, i.e. externally triggered versus self-triggered movement preparation, the late component of both CNV2 and BP2 is believed to reflect neural activity in M1 (Jankelowitz & Colebatch, 2002). Recently, Thabit et al. (2010) developed a new PAS protocol by delivering TMS over M1 paired with a visual cued simple-reaction time task consisting of repetitive self-triggered thumb abductions. They found that MEPs increased or decreased significantly in size according to specific ISIs; when TMS preceded EMG onset at 50 ms, MEPs increased in size, whereas when TMS followed EMG onset at 100 ms, MEPs decreased in size, reflecting associative plasticity in M1. Although Thabit et al. (2010) did not record MRCPs, they speculated that associative plasticity occurred because M1 neural elements activated by TMS overlapped, at least in part, those generating the late motor component of MRCPs. They did not, therefore, clarify which MRCP component related to the self-triggered thumb abductions was specifically associated with TMS in inducing the observed associative plasticity.

A recent study published in The Journal of Physiology (Mrachacz-Kersting et al. 2012) investigated the issue of associative plasticity elicited by pairing exogenous stimuli with endogenous neural activity contributing to MRCPs. Differently from Thabit et al. (2010), Mrachacz-Kersting et al. (2012) designed a new PAS protocol able to elicit changes in the amplitude of MEPs recorded from lower limb muscles by coupling CNV-related neural activity generated by imagined ankle dorsiflexion with electric stimulation of the common peroneal nerve. In their experimental protocol, after the ‘go’ signal, subjects were asked to imagine a ballistic right ankle dorsiflexion, to hold the imagined voluntary contraction for 2 s and then to release the imagined contraction. Electric stimulation of the right common peroneal nerve was delivered at specific ISIs before, during and after the CNV mean peak negativity. The authors (Mrachacz-Kersting et al. 2012) examined the effect of the associative protocol by measuring MEP amplitudes from the right tibialis anterior muscle before and after PAS. When peripheral electric stimuli were delivered during the CNV mean peak negativity, MEPs from the tibialis anterior muscle increased in size significantly, whereas the other ISIs left MEPs unchanged. Mrachacz-Kersting et al. (2012) concluded that MEP changes arose from associative plasticity in neural elements generating the late CNV (endogenous stimulus) and receiving afferent inputs from the common peroneal nerve activation (exogenous stimulus).

The study of Mrachacz-Kersting et al. (2012) is interesting since the proposed PAS protocol combining CNV and peripheral electric nerve stimulation at specific ISIs provides novel information on mechanisms underlying associative plasticity in human motor cortex. However, as Mrachacz-Kersting et al. (2012) did not apply TMS during PAS, the exact location where associative plasticity occurred remains unclear. Brain activation patterns induced by the execution of a real movement differ significantly from those induced by the imagination of a motor act since the activation of M1 is less prominent during an imagined motor act than during motor execution (Jankelowitz & Colebatch, 2002). Hence, it is difficult to determine whether the associative plasticity reported by the authors (Mrachacz-Kersting et al. 2012) was indeed related to M1 intrinsic activity or whether it reflected, instead, enhanced functional connectivity between M1 and non-primary motor areas, including the dorsal premotor cortex. In addition, it remains unclear whether the LTP-like plasticity in M1 reported by Mrachacz-Kersting et al. (2012) reflected STDP or rather arose from a ‘gating effect’ of the afferent sensory input driven by motor imagery-induced neural activity (Ziemann & Siebner, 2008). Differently from STDP, which requires precise timing and specific associativity between stimuli (Stefan et al. 2000), LTP-like plasticity might be also driven from afferent sensory inputs gated by a transient and unspecific enhancement in M1 excitability due to motor imagery (Ziemann & Siebner, 2008).

As associative plasticity due to STDP plays a crucial role in memory and learning processes, the various PAS protocols here reported might be helpful as a means of promoting functional recovery in patients with various types of motor disorders. In addition, it may in the future be possible to exploit PAS protocols that induce cortical associative plasticity in brain–computer interface (BCI) technologies to improve the quality of interaction with the surrounding environment in patients with severe motor disabilities.