|How Can Ca2+ Influx Through NMDARs Induces
LTP, LTD, Extinction and Retrieval
It has been well established that the two opposing synaptic modulations, long-term potentiation (LTP) and long-term depression (LTD), are controlled primarily by the Ca2+ influx through NMDA receptors (NMDARs). The mechanisms of memory extinction and retrieval are less clear. Accumulating evidence suggests that the Ca2+ ions entering NMDARs also play a pivotal role. How can Ca2+ ions induce opposing modulations, not only LTP vs. LTD but also extinction vs. retrieval?
An NMDAR consists of two GluN1 (also known as NR1) subunits and two additional subunits which are predominately either GluN2A (NR2A) or GluN2B (NR2B). Other subunits, GluN3, GluN2C and GluN2D, are relatively rare. Decades of intensive research has revealed that either GluN2A- or GluN2B-containing NMDARs are capable of supporting LTP induction, provided they can mediate sufficient Ca2+ influx (Shipton and Paulsen, 2013). However, GluN2B is critical for the NMDAR-dependent LTD as the GluN2B-selective antagonist, ifenprodil or Ro 25-6981, abolishes LTD, but not LTP (Liu et al., 2004; France et al., 2017). GluN2B, not GluN2A, has also emerged as a central player in addictive behavior, which is closely related to memory extinction (Hopf, 2017; Otis et al., 2014). Intriguingly, both tubulin and CRMP2 also interact with GluN2B, not GluN2A (van Rossum et al., 1999; Brustovetsky et al., 2014). Therefore, only the GluN2B-containing NMDARs can be inhibited by tubulin. These results are consistent with the fundamental postulate proposed in Chapter 7: memory extinction occurs when NMDARs are inhibited by tubulin.
This chapter will explain how the NMDAR-mediated Ca2+ ions can induce opposing synaptic modulations. An important factor is the kinetics of NMDAR currents: upon activation, the GluN2B-containing NMDAR currents rise and fall much slower than GluN2A-containing NMDAR currents (Erreger et al., 2005). Another crucial factor that leads to bi-directional plasticity is the location of A-kinase anchoring protein.
The Roles of AKAP79/150
AKAP79/150 refers to either A-kinase anchoring protein 79 (AKAP79) in humans or AKAP150 (also called AKAP5) in rodents. They have essentially the same functions. It is a scaffold protein that organizes protein kinase A (PKA), protein kinase C (PKC) and calcineurin (CaN) at a specific subcellular location to restrict their substrate targeting. For instance, the PKA anchored to a synaptic AMPA receptor (AMPAR) should be more effective in phosphorylating the AMPAR than the PKA diffusing randomly in the cytoplasm. AKAP79/150, together with anchored enzymes, can be recruited to the postsynaptic membrane by binding phosphatidylinositol-4,5-bisphosphate (PIP2) which is regulated by PKC phosphorylation (Dell'Acqua et al., 1998). It also interacts with membrane-associated guanylate kinase (MAGUK) such as PSD-95 and SAP97 (Sanderson and Dell'Acqua, 2011). Importantly, Ca2+ ions can enhance the activities of all three anchored enzymes: PKA, PKC and CaN.
AKAP79/150 has been demonstrated to associate with the GluA1 (old name: GluR1) subunit of AMPARs and the GluN2B subunit of NMDARs (Colledge et al., 2000). LTP or LTD is fundamentally determined by the increase or decrease in synaptic AMPARs, respectively. This in turn depends on the phosphorylation state of S845 and S831 in the GluA1 subunit of AMPARs. S845 is the target of both PKA and CaN while S831 can be phosphorylated by PKC and calcium/calmodulin-dependent protein kinase II (CaMKII). Phosphorylation on S845 or S831 stimulates synaptic incorporation of AMPARs, thus promoting LTP. CaN can dephosphorylate S845, resulting in LTD (Henley and Wilkinson , 2013; Woolfrey and Dell'Acqua, 2015, Figure 2).
How the Brief Tetanic Stimulation Induces LTP
Experimentally, LTP can be induced by several different protocols (Shipton and Paulsen, 2013, Table 1). One of them, referred to as "tetanus", applies strong high frequency (~100 Hz) stimulation on the presynaptic neuron for about 1 second. This leads to postsynaptic potentiation as monitored by field excitatory postsynaptic potentials (f-EPSPs) (Chapter 14).
Recalling that the GluN2B-containing NMDAR currents have slow kinetics. Therefore, the Ca2+ influx triggered by the brief tetanic stimulation should pass through mainly the GluN2A-containing NMDARs which are NOT associated with AKAP79/150. Hence, the major enzyme that induces LTD, CaN, is not significantly affected. The other two enzymes anchored by AKAP79/150, PKA and PKC, will also remain mostly inactive. In this case, LTP should arise mainly from phosphorylation on AMPARs and stargazin by CaMKII as described in Chapter 5.
Without contributions from PKA and PKC, a single tetanic stimulation typically generates lower levels of potentiation (Huganir and Nicoll, 2013). Multiple tetanic stimulations separated by a few minutes may induce PKA-dependent LTP via phosphorylation of GluA1 at S845, leading to the synaptic incorporation of calcium-permeable AMPARs (Park et al., 2016).
How the Prolonged Low Frequency Stimulation Induces LTD
The most commonly used protocol to induce LTD is a weaker low frequency (~ 1 Hz) stimulation on the presynaptic neuron for about 15 minutes. The prolonged weaker low frequency stimulation (LFS) will be able to trigger substantial Ca2+ influx through GluN2B-containing NMDARs (Erreger et al., 2005; Shipton and Paulsen, 2013). These Ca2+ ions should have significant impact on the activities of anchored enzymes. However, there are two competing processes on synaptic plasticity. CaN catalyzes the dephosphorylation, while PKA stimulates the phosphorylation, of S845. Which will win? The fact that prolonged LFS induces LTD indicates that CaN dominates. How?
AKAP79/150 binds both PIP2 and dendritic F-actin in a calmodulin- and PKC-regulated manner. The Ca2+ influx through NMDARs may activate PKC to release AKAP79/150 from the membrane (Dell'Acqua et al., 1998). The translocation of AKAP79/150 away from the membrane also involves CaN (Gomez et al., 2002). It has been shown that PKA moves with AKAP79/150 to the cytoplasm of dendrite shafts and the soma (Sanderson and Dell'Acqua, 2011), where PKA may trigger the expression of plasticity-related genes (e.g., BDNF) by phosphorylating cAMP response element-binding protein (CREB). Critically, CaN remains at the synapse so that it can dephosphorylate synaptic AMPARs. This key finding explains why prolonged NMDAR activation favors LTD (Smith et al., 2006).
The above mechanism is consistent with LTD at different developmental stages. It has been known for many years that LFS produces robust LTD in hippocampal slices from very young rodents (mice or rats), bu not from adult animals (Kemp et al., 2000; Milner et al., 2004). The hippocampal neurons express predominately GluN2B at birth, which then decreases into adulthood while GluN2A increases with age (Dong et al., 2006). Therefore, the young rodents, but not aged, may contain sufficient GluN2B-associated CaN to produce LTD.
Regulation of Extinction and Retrieval by Ca2+
Chapter 15 proposes that memory extinction and retrieval could be fundamentally determined by CaN and PKA, competing for the phosphorylation state of S1166 in GluN2B. Phosphorylation of S1166 by PKA facilitates retrieval whereas dephosphorylation of S1166 by CaN promotes extinction, because the dephosphorylated state allows tubulin to bind to the GluN2B-containing NMDARs, consequently inhibiting their opening and suppressing neuronal firing. Hence, the regulation of extinction and retrieval by Ca2+ resembles the regulation of LTP and LTD. They differ primarily in the phosphorylation target: GluN2B S1166 for extinction/retrieval and GluA1 S845 for LTP/LTD. Further details on memory extinction and retrieval will be presented in later chapters.
Author: Frank Lee