Synapses are communication areas between neurons (or between a neuron and a muscle cell) where chemicals are transmitted through the cleft between the pre- and the postsynaptic membranes. Over the past 40 years, the concept of a synapse as a simple site of transfer of information that once established does not change along life has been revolutionized by the discovery of its extremely plastic properties (Bliss and Lomo 1973; Bliss and Collingridge 1993). This synaptic plasticity is central to understanding the mechanisms of learning and memory. The synaptic “strength” (Colgin et al. 2009; Ermentrout et al. 2008; Li et al. 2014) results from the sum of pre-and postsynaptic responses of stimulated neurons, in other words, action potentials firing. The sequence and the temporal precision of the spikes in the central nervous system (CNS) have been shown to be linked with the strength of long-term plastic changes (Rutishauser et al. 2010) and to be especially related to cognitive function of the brain and its regulation. Long-term potentiation is the predominant form of synaptic plasticity in the brain, has been shown in the amygdala, the hippocampus CA1 neurons, and has been described especially in learning and memory processes. It is also considered as serving as the cellular model for chronic pain (Zhuo 2004, 2007, 2008, 2013; Sandkühler 2007; Ikeda et al. 2006; Costigan et al. 2009). Peripheral noxious inputs in injured afferent neurons or associated cell bodies trigger LTP in dorsal horn neurons of the spinal cord and contribute to spinal sensitization (allodynia and hyperalgesia). Nociceptive information is then transmitted to the thalamus and to central structures (the prefrontal cortex (PFC) and the anterior cingulate cortex (ACC)) where potentiation contributes to pain central sensitization and impacts on other brain functions including fear and emotion (Sandkühler and Lee 2012).

Ji et al. (2003) described striking similarities in the synaptic plasticity involved in pain central sensitization and memory. However, hippocampal LTP reflects only synaptic strengthening, whereas central sensitization might also reflect other cellular mechanisms. LTP has been mainly studied in the hippocampus and other cortical areas but may be induced in the spinal cord and has been reported in sensory pain-related central synapses, spinal cord, and cortical areas involved in pain perception (Zhuo 2007, 2008; Sandkühler 2007). It is complex and characterized by successive phases of induction, consolidation, and maintenance in the CNS. It requires the synaptic activation (by glutamate, the main neurotransmitter in nociceptive pathways) of N-methyl d-aspartate receptors (NMDARs) (Collingridge et al. 1983), tetra-heteromeric assemblies made up of two GluN1 (NR1) and two GluN2 (NR2) subunits, but it is not a single process as thought for a long time (Volianskis et al. 2013). Different subtypes of NMDARs are involved during induction of different temporal phases of synaptic plasticity. High-frequency stimulation of NMDAR relieves the physiological magnesium block from the NMDAR, leading to an increase in calcium ions and to induction of potentiation (Bliss and Collingridge 1993). A number of other receptors (AMPA, kainate, G-protein-coupled metabotropic, neurokinin-1…) are involved and nociceptor inputs induce the phosphorylation of NR1, NR2A, and NR2B by serine/threonine and tyrosine kinases. Activation of the mitogen-activated protein kinase (MAPK) cascade leads to triggering in the nucleus of the cell of the transcription of genes that encode a number of related products (cAMP response element-binding protein CREB) that are critical for synaptic potentiation in central areas and in the ACC. The ACC is a key cortical region for pain perception and has been shown to be activated in brain imaging studies with healthy volunteers and with chronic pain patients. It is also involved in emotion, cognition, executive function, social pain, and in emotional pain situations and is a nodal point in pain and cognitive-affective domains (Chen et al. 2014).

Memory traces of pain result from neuronal mechanisms and plasticity. This memory aspect must be distinguished from the recollection of a painful experience (with its location, nature, intensity, duration, and environmental components) that is encoded in the explicit memory.

Erasure of chronic pain has been the center of interest of many papers and is the hope of patients suffering from long-standing pain and associated deleterious impact on everyday life. Sandkühler and Lee (2012) stress the importance of the balance between LTP and depotentiation of LTP (a less studied area of neuronal plasticity), a balance between the formation of new and the erasure of old memory traces, that is disrupted in chronic pain states. After the induction phase of plastic changes (including LTP) in response to noxious stimuli at the first synapse, the consolidation and the maintenance phases of LTP in nociceptive pathways will leave a long-term pain trace via de novo protein synthesis.

However, the development of lasting pain in the consolidation phase may be aborted if the insult is antagonized in an adequate time window that may range from a few hours (Dableh et al. 2011) to a few weeks (Eaton et al. 1999) depending of the type of trauma. Drdla-Scutting et al. (2012) succeeded to erase pain memory traces in animals with a very high dose of remifentanil (450 μg/kg for 1 h), a short-acting mu opioid receptor agonist, by modifying the phosphorylation state of AMPA. Millecamps et al. (2007) showed in rats that neuropathic pain was reduced for weeks after injection of d-cycloserine (a partial NMDA receptor agonist) into limbic areas, with encouraging results in refractory orofacial pain (Antal and Paulus 2011).

Protein kinase M zeta (PKM ζ) has been shown to play an important role in LTP, in declarative, procedural memory, and also at some but not all synapses and not for all memory traces of pain (Migues et al. 2010; Laferrière et al. 2011). In the ACC, the presence of PKM ζ is required for the expression of neuropathic pain, and ζ pseudo-substrate inhibitory peptide (ZIP) may block its activity and erase some traces of pain (Sandkühler and Lee 2012; Price and Ghosh 2013). However, its deleterious disruptive effect on normal functions like fear responses may limit its interest in chronic pain.

Erasure of chronic pain is still a challenge and the use of compensatory mechanisms has been discussed (Sandkühler and Lee 2012), but from the model of fear memories, the competition between reconsolidation and extinction phases appears not to be helpful to complement pain erasure processes.

Another approach to prevent memory traces of pain is to prevent the whole process of LTP especially consolidation and maintenance. Preemptive/preventive analgesia before and at the time of surgery aim at aborting chronic pain development in the postoperative period (Kissin 2000). The large literature in this field gives different results for the prevention of chronic neuropathic pain, with positive and negative outcomes that depend on multiple factors (Katz et al. 2011; Dualé et al. 2009).

Focusing on the NMDA receptor, a novel hypothesis suggests that memantine administration long before the surgical trauma (rather than on the day of the trauma) might prevent the development of central sensitization in an animal neuropathic pain model (Pickering et al. 2014). Memantine, prescribed in Alzheimer’s disease to maintain cognitive function, has minimal side effects at doses within the therapeutic range, probably because of its specific mechanism of action as it is an uncompetitive antagonist with moderate affinity, strong voltage dependency, and rapid unblocking kinetics (Morel et al. 2013). Prevention of neuropathic pain with memantine administered for 4 days before surgery was successful, with no mechanical hypersensitivity and tactile allodynia with no increase of de novo synaptic proteins (especially tyrosine kinases pTyr¹⁴⁷²NR2B), and was also efficient in maintaining spatial memory. In a translational approach, a recent clinical trial (Pickering et al. 2014) in postmastectomy patients confirmed these findings and suggests that the plasmatic presence of this NMDA receptor antagonist 2 weeks before and after surgery (and not only on the day of surgery) might be a promising strategy to abort central sensitization and diminish the burden of disease in oncology. This approach must now be studied in relation to the different phases of LTP.

Finally, neurons have been the focus of attention to understand how central sensitization and LTP may help to find therapeutic options to treat chronic pain; accumulating evidence demonstrates that glial cells (microglia and astrocytes of the CNS and satellite glial cells of the dorsal root and trigeminal ganglia) are not static entities and are activated in chronic pain (Ji et al. 2013) with neuroglial interactions. Glial cell activation is complex, may be different in different neuropathic pain etiologies and timings. Considering that a human astrocyte is estimated to contact up to two million synapses (Oberheim et al. 2009) and that glial mediators modulate synaptic transmission, the role of the glia may have been so far underestimated: Ji et al. (2013) recently suggested that chronic pain could be a result of a “gliopathy at peripheral and central levels.”