Lipids and phosphates at odds in synaptic depression

Abstract

Long-term depression (LTD) is a reduction in the efficacy of neuronal synapses, but the molecular basis of LTD signaling and how these signals lead to phenotypic outcomes, such as the shrinkage of synaptic regions, is not clear. In a new report, Woolfrey et al. use chemically-induced LTD and a multitude of in vitro biochemical assays to provide evidence that synaptic removal of the scaffolding protein AKAP79/150 promotes LTD-induced spine shrinkage. The further identification of CaMKII, a kinase primarily associated with long-term potentiation (LTP), as a requirement for AKAP79/150 removal, uncovers unexpected interplay between different post-translational modifications and points to a new model of LTD.

Introduction

Long-term potentiation (LTP)² and long-term depression (LTD) are two principal categories of long-term synaptic plasticity and represent the molecular basis for the strengthening or weakening of synapses, respectively. Neuroplasticity plays an extensive role in learning and memory, psychiatric disorders, and drug addiction (1). Investigations into this process have the potential to inform our most basic understanding of how connections in our brains are remodeled and to provide new directions for the development of novel therapeutics. LTD may last for hours or longer and occurs in response to patterned electrical activity. The scaffolding and signaling proteins of the postsynaptic region are highly organized, and multiple molecular events have been implicated in the mechanisms underlying LTD. However, an encompassing model of LTD has not yet been established (2). For instance, although scaffold protein–trafficking and protein kinase activity are known to play substantial roles in LTD (3), how protein-trafficking is coordinated and what the causal relationships are between these molecular steps is unclear. In addition, Ca²⁺/calmodulin (CaM)-dependent protein kinase II (CaMKII), which has long been associated with LTP, has recently been shown to be involved in LTD as well (4). However, how CaMKII acts upon other mediators of LTD has not yet been fully characterized. A new report by Woolfrey et al. (5) sheds light on these questions, elucidating the functional role of CaMKII in LTD and facilitating the development of a cohesive model of LTD events.

LTP and LTD are induced via distinct neuronal activity patterns mediated through the excitatory glutamate receptors, α-amino-3-hydroxy-5-methylisoxazole-4-propionic acid (AMPA) and N-methyl-d-aspartate (NMDA) receptors. These receptors are present on dendritic spines, which are protrusions located on neuronal dendrites that mediate signaling between neurons, and their trafficking and insertion into the membrane regulate synaptic strength. Under basal conditions, the AMPA receptor subunit GluA1 is phosphorylated by PKA, which stabilizes the receptor's position in the postsynaptic membrane. Initiation of LTD-associated stimuli leads to moderate calcium influx through NMDA receptors, which then activates calcineurin and promotes dephosphorylation of GluA1 (Fig. 1a). LTD-mediated removal of the PKA scaffold, protein kinase A anchor protein AKAP79/150 (human/rat), prevents GluA1 rephosphorylation by PKA. Without PKA-mediated rephosphorylation of GluA1, AMPA receptors undergo endocytosis, which significantly weakens the strength of the synapse (6). AKAP79/150 also serves indirectly as a scaffold for the protein phosphatase calcineurin, which is localized near NMDA receptors through interaction of AKAP79/150 with postsynaptic density protein 95 (PSD-95) (7). Thus, AKAP79/150 serves a pivotal role in LTD. However, the mechanism underlying synaptic removal of AKAP79/150 and whether AKAP79/150 removal is a requirement for or consequence of LTD is not known.

Figure 1.

CaMKII-regulated delipidation and phosphorylation of AKAP79/150 are required for LTD-induced synaptic removal of AKAP79/150 and subsequent spine shrinkage. a, Ca²⁺ enters through the NMDA receptor and activates CaM, which initiates a protein phosphatase cascade mediated through CaN positioned near the NMDA receptor by a PSD-95 and AKAP79/150 complex. b, CaN indirectly activates protein phosphatase 1 (PP1), which dephosphorylates the GluA1 subunit of the AMPA receptor and PSD-95, promoting AMPA receptor endocytosis and subsequent LTD. c, CaMKII mediates autonomous phosphorylation and depalmitoylation of the AKAP79/150 TD. The AKAP79/150 TD is protected by Ca²⁺/CaM when calcium concentrations are high, and this protection is relinquished when calcium levels are low, allowing for CaMKII phosphorylation of AKAP79/150. d, CaMKII-mediated depalmitoylation and phosphorylation allow for synaptic removal of AKAP79/150. Synaptic removal of AKAP79/150 prevents PKA-mediated re-phosphorylation of AMPAR, promoting endocytosis of AMPAR, which weakens the strength of the synapse as a result.

To study this system, Woolfrey et al. (5) tested the outcome of NMDA-initiated chemical LTD (cLTD) when it is paired with administration of the pharmacological inhibitors of CaMKII, tatCN21 or KN93. The authors observed that CaMKII inhibition impeded cLTD-induced removal of AKAP150 from synaptic spines in hippocampal neuron cultures. As actin reorganization is required for AKAP79/150 removal, the authors tested whether inhibition of CaMKII affects actin organization. However, CaMKII inhibition did not prevent F-actin redistribution induced by cLTD, implicating a different mechanism at play. Next, Woolfrey et al. (5) tested whether synaptic removal of AKAP79/150 may be mediated through direct phosphorylation of the AKAP79/150 membrane–cytoskeletal targeting domain (TD) by CaMKII. Surprisingly, AKAP79/150 was not only phosphorylated by CaMKII, but phosphorylation was much more effective in the absence of Ca²⁺ and CaM rather than in their presence. This so-called “autonomous” activity, which occurs without CaM, is distinct from the Ca²⁺- and CaM-stimulated CaMKII phosphorylation of conventional CaMKII substrates. Autonomous phosphorylation of the TD by CaMKII decreased the interaction between AKAP150 and F-actin, indicating a mechanism through which CaMKII may mediate LTD. These findings led the authors to determine the phosphorylation sites that mediate synaptic AKAP79 removal following LTD. Phosphomimetic mutation of the Thr-87 and Ser-92 sites in AKAP79 reduced its binding to F-actin; however, only a nominal difference was observed in localization of AKAP79 to spines when phosphomimetic (Glu) and phosphodeficient (Ala) mutants were expressed. Mutation of the Ser/Thr sites in other AKAP79/150 regions decreased CaMKII-mediated phosphorylation. Some of these sites, found in the TD, were found to contain Ca²⁺/CaM-binding motifs that block CaMKII phosphorylation during basal conditions but become exposed during Ca²⁺-depleted LTD, explaining this end result. Having tested the role of individual sites, the authors constructed a full phosphomimetic mutation of AKAP79/150. Unexpectedly, this construct promoted, but was not sufficient for, LTD-mediated removal of AKAP79/150 from spines, indicating additional mechanisms at play.

Depalmitoylation was previously reported to promote synaptic removal of AKAP79, without being sufficient (8). Based on this, Woolfrey et al. (5) used an acyl–PEGyl exchange gel shift assay to investigate whether phosphorylation of AKAP79 affects its depalmitoylation. The full phosphomimetic AKAP79 mutant showed significantly decreased palmitoylation compared with the constructs with partial phosphomimetic mutations. The combination of the full phosphomimetic mutations with palmitoylation-incompetent C36S/C129S mutations generated a construct localized in the cytosol under basal conditions. Taken together, these data suggest that LTD induction leads to AKAP150 depalmitoylation, which can be blocked through inhibition of CaMKII. Finally, the authors added an N-terminal myristoylation consensus sequence on AKAP79, creating a mutant that imitates irreversible palmitoylation. This persistent myristoylation prevented LTD-induced AKAP79 removal, suggesting that depalmitoylation is required for LTD-mediated synaptic removal of AKAP79. Surprisingly, myristoylation also prevented spine shrinkage following induction of LTD, providing tantalizing evidence that AKAP79 trafficking plays a causal role in this process.

The findings presented in Woolfrey et al. (5) elucidate new functions for CaMKII and AKAP in LTD. Their data show that CaMKII phosphorylation promotes, but is not required for, AKAP79/150 removal following LTD. Surprisingly, CaMKII plays a significant role in the regulation of AKAP79/150 depalmitoylation, which is required for LTD-mediated synaptic removal of AKAP79/150. Their findings strongly suggest that CaMKII-mediated phosphorylation and depalmitoylation are required for the synaptic removal of AKAP79/150 and for the accompanying spine shrinkage, which takes place after the induction of LTD. These conclusions greatly expand the model behind the mechanistic basis of LTD (Fig. 1). This new report is significant, as, until now, the role CaMKII has been largely associated with the mechanisms underlying LTP (2). It is likely that the different roles of CaMKII in LTP and LTD are mediated through the ratio of autonomous to Ca²⁺/CaM-dependent CaMKII activity. The identification of other high-autonomy CaMKII substrates, in addition to AKAP79/150, will further illustrate the important role of CaMKII in both LTD and LTP and facilitate the formation of an encompassing mechanistic model behind the two opposing processes.