Abbreviations AD Alzheimer disease ADRD Alzheimer disease and related dementias AKT protein kinase B AMPK AMP-activated protein kinase BDNF brain-derived neurotrophic factor CaMKIII calmodulin-dependent protein kinase III CPKs conventional protein kinases DG dentate gyrus eEF2 eukaryotic elongation factor 2 eEF2K eukaryotic elongation factor 2 kinase eIF2α eukaryotic initiation factor 2 alpha ERK extracellular signal-regulated kinase LTD long-term depression LTP long-term potentiation mTORC1 mammalian target of rapamycin complex 1 MWM Morris water maze NOR novel object recognition PERK protein kinase RNA-like endoplasmic reticulum kinase ROS reactive oxygen species SUnSET surface sensing of translation α-kinases alpha-kinases or atypical kinases 1 INTRODUCTION

Elucidation of the molecular basis for cognitive function and synaptic plasticity (persistent change in neuronal circuits) would help reveal mechanisms underlying the pathogenesis of cognitive syndromes such as Alzheimer disease (AD). Findings from such studies could in turn help identify novel therapeutic strategies as well as diagnostic/prognostic biomarkers for neuronal disorders characterized with cognitive impairments. Extensive studies over the past few decades have established that protein synthesis (mRNA translation) is indispensable for long-lasting forms of synaptic plasticity and memory (Costa-Mattioli et al., 2009; Kelleher et al., 2004). Further, dysregulation of mRNA translation is linked to a variety of human diseases including many neuronal diseases (Scheper et al., 2007; Tahmasebi et al., 2018).

As a fundamental cellular process critical for almost all biologic functions, protein synthesis involves complex machinery and consumes a significant amount of energy and nutrients. Accordingly, the process of protein synthesis is subject to tight control to help maintain cellular homeostasis (Hershey et al., 2019; Tahmasebi et al., 2019). Overall protein synthesis takes place in three phases: initiation, elongation, and termination. Each phase requires the action of specific translational protein factors to regulate the process. Most (>95%) of the energy and amino acids utilized in the mRNA translation are consumed during the elongation phase (Browne & Proud, 2002; Kenney et al., 2014). Consistently, accumulating evidence indicates that translation elongation control is critical in the modulation of protein synthesis, particularly during cellular responses to deficiency of nutrients and energy (Biever et al., 2019; Sutton & Schuman, 2006). A key mechanism involved in elongation control is through phosphorylation of eukaryotic elongation factor 2 (eEF2). Phosphorylation of eEF2 on the Thr56 site by its specific kinase eEF2K hinders the engagement of eEF2 with ribosomes and consequently inhibition of mRNA translation (Kenney et al., 2014; Liu & Proud, 2016; Ryazanov, 2002).

There are excellent review articles on basic biochemistry and biology of eEF2K/eEF2 signaling (Drennan & Ryazanov, 2004; Kenney et al., 2014). In this review, we mainly discuss studies on the functional roles of eEF2K/eEF2 signaling in the context of regulation/dysregulation of synaptic plasticity and cognitive function. We specifically focus on the discussion of recent evidence indicating suppression of eEF2K signaling as a potential novel therapeutic strategy for AD and related dementias (ADRDs).

2 THE eEF2K SIGNALING: UPSTREAM AND DOWNSTREAM

The vast majority of the eukaryotic protein kinases possess homologous catalytic domains (e.g., serine/threonine kinases and tyrosine kinases) and are usually referred to as the conventional protein kinases (CPKs). Unlike the CPKs, eEF2K belongs to an unusually small group of kinases (six members in the human genome) termed “alpha-kinases” (α-kinases), whose catalytic domains are distinct from those in the CPKs (Drennan & Ryazanov, 2004; Kenney et al., 2014; Ryazanov et al., 1999). Among all six members of the α-kinases family, eEF2K is the only one whose activity is associated with Ca2+ signaling. Indeed, eEF2K is originally known as Calmodulin-dependent protein kinase III (CaMKIII) (Nairn et al., 2001). One potential mechanism underlying eEF2K activation by Ca2+/CaM involves autophosphorylation at the Thr348 and Ser500 sites (Tavares et al., 2012). In addition to autophosphorylation induced by Ca2+/CaM, the activity of eEF2K is regulated through phosphorylation by multiple upstream signaling pathways under various situations. Particularly, eEF2K activity is inhibited by the mammalian target of rapamycin complex 1 (mTORC1) signaling pathway and stimulated by the AMP-activated protein kinase (AMPK) signaling (Kenney et al., 2014). (Figure 1).

Schematic model that depicts the upstream regulators and downstream effectors of eEF2K. Small molecule eEF2K antagonists (NH125 and A-484594) currently available are indicated. Arrows denote activation, and blunted lines indicate inhibition. AKT, protein kinase B; ERK, extracellular signal-regulated kinase; AMPK, AMP-activated protein kinase; mTORC1, mammalian target of rapamycin complex 1; eEF2K, eukaryotic elongation factor 2 kinase; eEF2, eukaryotic elongation factor 2

The mTORC1 signaling is involved in a wide range of cellular processes such as autophagy and protein synthesis (mainly through translation initiation) (Hall, 2016; Yang & Guan, 2007). Mounting evidence indicates an important role of the mTORC1 signaling in synaptic plasticity and memory formation (Graber et al., 2013; Hoeffer & Klann, 2010). Studies from the Proud laboratory revealed that eEF2K is phosphorylated at the Ser366 site by S6 kinase1 (S6K1 or p70S6K), an established substrate of mTORC1 (Wang et al., 2001). A later study from the same group discovered that eEF2K is phosphorylated at the Ser78 site in a mTORC1-dependent manner (Browne & Proud, 2004). Phosphorylation of eEF2K at either Ser366 or Ser78 site by mTORC1 results in inhibition of eEF2K activity, leading to dephosphorylation of eEF2 and subsequently increase of general protein synthesis (Kenney et al., 2014). A most recent study suggested that a scaffolding protein Homer1b/c regulates eEF2K activity, and the process is associated with the mTORC1 pathway through phosphorylation of eEF2K at the Ser396 site (Gladulich et al., 2021). Previous studies in rodents indicate that the involvement of mTORC1 in the regulation of long-term hippocampal synaptic plasticity requires inputs from multiple kinases including AKT (protein kinase B) and ERK (Ma et al., 2011; Tsokas et al., 2007). In addition, studies from non-neuronal systems reveal that ERK can inactivate eEF2K through direct or indirect phosphorylation (Liu & Proud, 2016) (Figure 1).

AMPK is a central molecular sensor critical for the maintenance of cellular energy homeostasis (Hardie, 2011). In a low energy state (e.g., decreased levels of ATP and increased levels of AMP), AMPK is activated to phosphorylate a range of substrates to reduce energy consumption and increase energy supply (Hardie et al., 2012). It is perhaps unsurprising that AMPK regulation is linked to eEF2K signaling given that the protein synthesis process consumes a substantial amount of energy as mentioned above (Hershey et al., 2019). Activation of AMPK stimulates eEF2K through at least two mechanisms: 1) inhibition of the mTORC1, which is a negative regulator of eEF2K as discussed above (Horman et al., 2002) and 2) direct phosphorylation of eEF2K (mainly at the Ser398 site), which is independent of the mTORC1 signaling (Browne et al., 2004). Stimulation of eEF2K by AMPK leads to phosphorylation of eEF2 and inhibition of general protein synthesis (Figure 1).

3 ROLES OF eEF2K IN SYNAPTIC PLASTICITY AND COGNITION: KNOCKDOWN AND OVER-EXPRESSION

A plethora of studies have demonstrated that maintenance of synaptic plasticity and memory is dependent on de novo protein synthesis (Klann & Dever, 2004; Richter & Klann, 2009; Rosenberg et al., 2014; Sossin & Costa-Mattioli, 2019). As also mentioned above, energy consumption during elongation is dominant for the protein synthesis process. Moreover, regulation of protein synthesis via elongation is particularly important in cellular environments (e.g., neuronal dendrites), where both initiation and elongation processes need to be up-regulated to fulfill the substantial requirements of new protein synthesis associated with synaptic plasticity (Sutton & Schuman, 2006; Tsokas et al., 2005). Consistently, it was demonstrated that treatment of neurons with brain-derived neurotrophic factor (BDNF), a key player in neuronal plasticity, reduced eEF2 phosphorylation, leading to the promotion of translation elongation and general protein synthesis (Takei et al., 2009). A more recent study from the Proud group using the proteomics approach reported that eEF2K/eEF2 signaling controls the synthesis of microtubule-related proteins, which play an important role in regulating dendritic spine morphology that is modified during synaptic plasticity, learning, and memory (Kenney et al., 2016). On the other hand, a previous study performed in cultured neurons revealed that inhibition of eEF2K resulted in decreased dendritic BDNF protein expression (Verpelli et al., 2010). Another interesting finding, in light of the role of eEF2K signaling and general protein synthesis in synaptic plasticity discussed above, is that activation of the NMDA receptor led to increased protein expression of alphaCamK II via phosphorylation of eEF2 (Scheetz et al., 2000). At least two outstanding review articles summarized the biochemical and molecular alterations associated with the roles of eEF2 phosphorylation during synaptic plasticity or memory formation (Sossin & Costa-Mattioli, 2019; Taha et al., 2013). For this review, we focus on a discussion of functional neuronal phenotypes associated with the genetic manipulation of eEF2K in rodents.

The majority of studies in the neuroscience field on genetic eEF2K manipulation are carried out in transgenic mice with global eEF2K knockout (eEF2K KO) originally developed by the laboratory of Alexey Ryazanov or Christopher Proud (Heise et al., 2017; Moore et al., 2015; Ryazanov, 2002). As expected, phosphorylation of eEF2 is dramatically decreased in the eEF2K KO mice (Heise et al., 2017; Zimmermann et al., 2018). Thus, it is predicted that neuronal protein synthesis shall be increased in the eEF2K KO mice given the roles of eEF2 in mRNA translation described above. Using the non-radioactive assay SUnSET (surface sensing of translation), we found that overall de novo protein synthesis is increased in acute living hippocampal slices from the eEF2K KO mice (Zimmermann et al., 2018). Another study also used the SUnSET method, however, did not reveal significant alterations of general protein synthesis in cultured neurons from the eEF2K KO mice (Heise et al., 2017). Such findings indicate that protein synthesis might be differently regulated by the eEF2K/eEF2 signaling, possibly depending on different brain regions or subcellular locations (e.g., cell bodies or dendrites/synapses). It is worth mentioning that measurement of de novo protein synthesis for both studies was carried out under basal resting state. Regulation of protein synthesis via eEF2K and eEF2 phosphorylation might be further exemplified during intense neuronal activities (e.g., induction of synaptic plasticity). Alternatively, the inconsistent findings on protein synthesis could be related to the limitation of the SUnSET method in precisely detecting specific pools of proteins. More comprehensive studies using advanced techniques such as those in cellular imaging combined with large-scale proteomics analysis shall shed light on this topic.

eEF2K KO mice are well characterized and they are “normal” as assessed by numerous basic measurements such as life span, body weight, food intake, gross brain morphology (Heise et al., 2017). Further, comprehensive behavioral studies have been performed by Heise and colleagues (Heise et al., 2017) to assess a broad range of cognitive functions in the eEF2K KO mice. Overall, adult eEF2K KO mice exhibit normal cognitive behaviors in most assays including open field (OF) and elevated plus maze (EPM) for testing locomotor activities and anxiety-like behavior, hidden-platform Morris water maze (MWM) for testing spatial learning/memory, novel object recognition (NOR) for testing long-term recognition memory, marble burying for testing repetitive and perseverative-like behavior, and reversal MWM task for testing cognitive flexibility (Heise et al., 2017). Notably, we recently reported that age-related long-term recognition memory impairments in mice, assessed by the NOR test, were alleviated in eEF2K KO mice (Gosrani et al., 2020). In comparison, age-related spatial learning and memory deficits evaluated by the MWM test were not affected by eEF2K KO (Gosrani et al., 2020). Both the NOR (for assessing long-term cognitive function) and MWM tests are established as hippocampal-dependent behavioral assays. Meanwhile, other brain regions such as the prefrontal cortex are also involved (Akirav & Maroun, 2006; Buzsáki & Moser, 2013). Recent studies indicate that the protein synthesis process in specific cell types (e.g., inhibitory or excitatory neurons) is associated with different neuronal phenotypes (Sharma et al., 2020; Shrestha, Ayata, et al., 2020; Shrestha, Shan, et al., 2020). Together with the above discussion on the SUnSET findings, it would be important for future studies to elucidate potential distinct roles of eEF2 phosphorylation in the regulation of cognitive functions that are associated with different brain regions and/or neurons.

With elegant modification in some specific behavioral paradigms, Heise and colleagues revealed that the eEF2K KO mice exhibited impaired long-term trace fear conditioning, which is considered to be heavily dependent on the function of hippocampal Dentate Gyrus (DG) (Heise et al., 2017). Performance of the eEF2K KO mice assessed by regular fear conditioning protocols is largely normal (Heise et al., 2017). The researchers also performed immunohistochemical experiments to probe for phosphorylated eEF2 in the hippocampus and found higher levels of eEF2 phosphorylation in the area CA3 and DG compared with the staining in the area CA1 (Heise et al., 2017). Another study from the group of Kobi Rosenblum reported that cortical-dependent associate taste learning was impaired in a line of transgenic mice with eEF2K repression (Gildish et al., 2012).

In further pursuing DG-related roles of the eEF2K/eEF2 signaling in the regulation of cognition function, the Rosenblum group generated a “conditional” eEF2K knockout mice (eEF2K cKO) by injecting adeno-associated viruses (AAVs) containing CaMKII-Cre to the DG subregion of the eEF2K floxed mice (Taha et al., 2020). They reported that CaMKII-Cre-injected mice showed enhanced hippocampal neurogenesis. They further performed a series of behavioral tests to characterize cognitive phenotypes associated with the selective repression of eEF2K in DG. They found that knockdown of eEF2K in DG was able to enhance performance in multiple hippocampal-dependent behavioral assays including a weak protocol MWM, pattern separation of contextual fear memory, and spatial pattern completion (Taha et al., 2020). Furthermore, they showed in aged mice that the DG-specific knockdown of eEF2K improved contextual fear memory 4 weeks after training (Taha et al., 2020). To the best of my knowledge, this is the only report on studying cognitive effects in rodents with brain region-specific suppression of eEF2K. It would be important for future studies to determine the behavioral and synaptic phenotypes in transgenic mice with conditional eEF2K suppression in other brain regions. It is also important to explore the detailed mechanisms underlying the different behavioral phenotypes in these mice compared with those in the global eEF2K KO mice.

Long-term potentiation (LTP) and long-term depression (LTD) in the hippocampus are probably the most intensively studied forms of synaptic plasticity, coordination of both is considered as a key molecular mechanism underlying learning and memory (Malenka & Bear, 2004). A previous study from the laboratory of Paul Worley reported that late LTP induced by a strong electrical protocol [four trains high-frequency stimulation (HFS)] was facilitated in hippocampal slices from the eEF2K KO mice (Park et al., 2008). Meanwhile, we found that hippocampal late LTP induced by a weaker protocol (two trains HFS) was unaltered in the eEF2K KO mice (Zimmermann et al., 2018). The mechanisms underlying the different LTP performances remain unclear.

LTD is broadly divided into two categories: N-methyl-D-aspartate receptor (NMDAR)-dependent and metabotropic glutamate receptor (mGluR)-dependent LTD. NMDAR-dependent. LTD is usually induced by electrical low-frequency stimulation (LFS) and requires activation of NMDARs, whereas mGluR-dependent LTD requires activation of group 1 mGluRs (includes mGluR1 and 5) and is typically induced by DHPG [(RS)-3,5-dihydroxyphenylglycine], a selective agonist for group 1 mGluR (Lüscher & Huber, 2010; Malenka & Bear, 2004). Park and colleagues reported that NMDAR-dependent LTD induced by LFS protocol was unaltered in hippocampal slices from eEF2K KO mice. In comparison, mGluR-dependent LTD elicited by DHPG treatment was significantly impaired in hippocampal slices derived from the eEF2K KO mice (Park et al., 2008).

Compared to the multiple studies with suppression of eEF2K and eEF2 phosphorylation, there is only one report that I am aware of investigating neuronal functions in transgenic mice with over-expression of eEF2K (Im et al., 2009). Taking advantage of a Cre/loxP recombination system, Im and colleagues developed the hip-eEF-2K-tg mice in which the eEF2 gene was over-expressed specifically in the hippocampus, resulting in hyper-phosphorylation of eEF2 and consequently inhibition of general protein synthesis (Im et al., 2009). They further examined the memory phenotypes in the transgenic mice, mainly by using the regular fear conditioning test. The hip-eEF-2K-tg mice behaved normally in short-memory performance evaluated by both contextual and cued fear conditioning tasks. Examination of long-term memory revealed that the freezing response of the hip-eEF-2K-tg mice was impaired during contextual fear conditioning but unaltered during cued conditioning (Im et al., 2009). Further, late LTP was impaired in the hippocampus of the hip-eEF-2K-tg mice (Im et al., 2009). More comprehensive and systematic studies (both in vivo and in vitro) are needed to investigate neuronal phenotypes associated with elevated eEF2K activities and eEF2 hyper-phosphorylation.

4 HYPER-PHOSPHORYLATION OF eEF2 IN ALZHEIMER DISEASE: A DOUBLE-EDGE SWORD?

Alzheimer disease (AD) is an age-related neurodegenerative disease and the most common form of dementia syndromes (Holtzman et al., 2011; Querfurth & LaFerla, 2010; Alzheimer's Association, 2019; Roberson & Mucke, 2006). AD has become a global threat for public health mainly because of the significant increase of the aging population. The two classical hallmarks of AD brain pathology are Aβ plaques and hyper-phosphorylated tau tangles. Meanwhile, multiple studies indicate that synaptic deficits are better correlated with clinical symptoms in AD patients, and synaptic failure might be a key and early event during AD pathogenesis (Ma & Klann, 2012; Selkoe, 2002; Teich et al., 2015). Despite decades of research, the detailed molecular signaling mechanisms underlying AD-associated cognitive impairments and synaptic deficits remain elusive, hindering the identification of effective therapeutic strategies and novel diagnostic/prognostic biomarkers for the disease.

Studies from our and other groups described abnormal hyper-phosphorylation (Thr56) of eEF2 in postmortem brain tissues from human AD patients and hippocampus of multiple transgenic mouse models of AD including APP/PS1, 3xTg, and Tg19959 mice (Beckelman et al., 2019; Jan et al., 2017; Li et al., 2005; Ma et al., 2014). By doing immunohistochemical experiments in human AD brain samples, Jan and colleagues also revealed that the increase of eEF2 phosphorylation is more pronounced in the hippocampus compared with the staining in the cortex (Jan et al., 2017). Importantly, examination of human AD brain tissues showed significantly increased immunostaining of phospho (Thr56)-eEF2 in neurons (Beckelman et al., 2019; Jan et al., 2017). We also reported increased levels of eEF2 phosphorylation in the hippocampus of old mice (19–21 months) compared with mice of 3–5 months old (Yang et al., 2019). Interestingly, our study with human samples suggested that eEF2 hyper-phosphorylation could be specific to AD-related pathologic processes since we did not observe the similarly elevated levels of eEF2 phosphorylation in postmortem brain samples from human patients with other neurodegenerative diseases such as frontotemporal dementia (FTD) and Lewy body dementia (LBD) (Beckelman et al., 2019).

Substantial evidence has indicated that energy metabolism dysregulation and chronic oxidative stress induced by excessive reactive oxygen species (ROS) play a central role in the pathogenesis of many neurodegenerative diseases including AD (Dumont et al., 2009; Lin & Beal, 2006; Ma, Hoeffer, et al., 2011; Ma & Klann, 2012, 2014; Massaad et al., 2009). Another related common feature of many neurodegenerative diseases is the accumulation of misfolded proteins, which induces the unfolded protein response (UPR) and significant cellular stress (Ma & Klann, 2014; Wek & Cavener, 2007). Numerous biologic processes are stimulated to coordinate the stress response, one important component of which is inhibition of global protein synthesis through regulation of various translational factors including phosphorylation of initiation factor eIF2 α subunit (eIF2α) and elongation factor eEF2 (Wek, 2018; Wek et al., 2006). The process of UPR involves activation of PERK (protein kinase RNA-like endoplasmic reticulum kinase), which phosphorylates eIF2α and results in inhibition of translation initiation (Trinh & Klann, 2013; Wek, 2018; Wek & Cavener, 2007). On the other hand, AD-related oxidative stress and low energy state induce activation of AMPK, which in turn actives eEF2K and eEF2 phosphorylation, leading to inhibition of translation elongation (Ma et al., 2014; Vingtdeux et al., 2011; Wang et al., 2019, but see Jan et al., 2017). Transient repression of global protein synthesis in response to cellular stress could be neuroprotective because it allows neurons to conserve energy resources (protein synthesis is a high energy-consuming process) meanwhile enhancing the translation of specific stress-related mRNAs, thus reconfiguring gene expression to cope with the cellular stress (Ma & Klann, 2014; Paschen et al., 2007) (Figure 1). Nevertheless, the severe and chronic cellular stress associated with pathologic conditions in many neurodegenerative diseases (e.g., AD) could result in continuous hyper-phosphorylation of eEF2 (via activation of AMPK and eEF2K) and consequently long-lasting repression of general protein synthesis. An enhanced and prolonged eEF2 hyper-phosphorylation would eventually impair neuronal plasticity and cognitive function because integral mechanisms of de novo protein synthesis are essential for memory consolidation and maintenance of long-term synaptic plasticity (Richter & Klann, 2009; Rosenberg et al., 2014; Sossin & Costa-Mattioli, 2019) (Figure 1). Interestingly, studies in cultured neurons and C. elegans demonstrated that inhibition of eEF2K activity was able to blunt the increased ROS induced by Aβ oligomers (Jan et al., 2017). Taken together, there may exist a retrospective relationship between the eEF2K signaling and the cellular stress.

5 REPRESSION OF eEF2K AS A THERAPEUTIC STRATEGY FOR AD-ASSOCIATED SYNAPTIC FAILURE AND COGNITIVE IMPAIRMENTS: PROMISES AND CAVEATS

From a general view of translational medicine, there are several appealing features of targeting eEF2K as a potential therapeutic strategy for AD and other aging-related dementia syndromes. First, as described above, eEF2K is an unusual kinase that belongs to a very small group of kinases termed “α-kinases,” whose catalytic domains are distinct from those in the CPKs (e.g., serine/threonine kinases and tyrosine kinases), which consists of the vast majority of the eukaryotic protein kinase superfamily (Drennan & Ryazanov, 2004). Thus, compounds targeting eEF2K, that is, eEF2K inhibitors, if designed appropriately targeting the α-kinase-specific domains, are unlikely to affect activities of other CPKs that play important roles for a broad spectrum of biological processes. Second, multiple lines of studies suggested that eEF2K activity is not required for development or cell survival under physiologic conditions. As discussed, transgenic mice with global homozygous knockout (or heterozygous knockdown) of the eEF2K gene appear normal in all measurements on basic biologic functions during different stages of development (Beckelman et al., 2019; Gosrani et al., 2020; Heise et al., 2017; Ryazanov, 2002). Wild-type mice with the chronic treatment of various eEF2K inhibitors are also “normal” based on numerous measurements including life span, weight change, gross morphology of hippocampus, locomotor activities, and cognitive performance (Ma lab, unpublished data). Therefore, it is promising for future studies in a clinical setting that eEF2K inhibitors can be given safely without causing severe side effects (but see below discussion on the caveats of eEF2K inhibitors). The safety issue is critically important for patients with AD and related dementia syndromes who usually need to keep taking medicine over a long period. Third, there is a one-on-one relationship between eEF2K and eEF2 in that eEF2K is the only known upstream kinase for eEF2 and eEF2 is the only known substrate for eEF2K (phosphorylation at the Thr56 site) (Proud, 2015; Ryazanov & Davydova, 1989). Such specificity can further strengthen the selectivity and efficiency of the eEF2K inhibitors for therapeutic purposes.

To investigate the roles of eEF2K/eEF2 signaling dysregulation in AD pathophysiology, we genetically reduced AD-related eEF2 hyper-phosphorylation by crossing the eEF2K KO mice with two different lines of AD model mice. In AD mice with heterozygous knockdown of eEF2K, we found that AD-associated impairments of hippocampal de novo protein synthesis (assessed by SUnSET and analysis of the dendritic polyribosome assembly) were restored (Beckelman et al., 2019). Importantly, aging-related long-term memory deficits (assessed by multiple behavioral tasks) and synaptic plasticity (including LTP and mGluR-LTD) impairments in both AD model mice were significantly improved with genetic suppression of eEF2K (Beckelman et al., 2019; Yang et al., 2021) (Figure 2). Notably, neither brain Aβ deposit nor tau phosphorylation in AD model mice was affected with genetic reduction of eEF2K, adding another line of evidence supporting potential therapeutic strategies through targeting mechanisms independent of Aβ or tau (Beckelman et al., 2019; Kametani & Hasegawa, 2018; Pimplikar et al., 2010).

Repression of eEF2K activity as a potential therapeutic strategy for Alzheimer disease. (a) A working model describing the link between dysregulation of eEF2K signaling and AD-associated synaptic failure and dementia syndromes. (b) Cartoon diagram describing a genetic approach to suppress AD-associated eEF2 hyper-phosphorylation. Representative western blot images showed restoration of eEF2 phosphorylation in the hippocampus of Tg19959 AD model mice. Adapted from Beckelman et al. (2019). (c) Genetic suppression of eEF2K and eEF2 phosphorylation alleviated cognitive impairments displayed in AD model mice. Results from novel object recognition (NOR) behavioral assay were shown. Adapted from Beckelman et al. (2019). Please also refer to the main text for details. AD, Alzheimer disease; AMPK, AMP-activated protein kinase; eEF2K, eukaryotic elongation factor 2 kinase; eEF2, eukaryotic elongation factor 2; KO, knockout

What are exactly the molecular mechanisms underlying the improvement of synaptic and cognitive phenotypes in AD mice with eEF2K suppression? Given the critical role of eEF2K/eEF2 signaling in controlling translation elongation, and in light of the evidence showing deficits of protein synthesis capacity in AD (Ma et al., 2013), the parsimonious answer for the above question would be enhancement of protein synthesis. However, this will instantly bring another outstanding question: what are the particular proteins (whose synthesis is facilitated by suppression of eEF2K and eEF2 phosphorylation) that are responsible for the rescuing phenotypes? We tried addressing this question (to a certain extent) by performing mass spectrometry (MS)-based proteomics analysis in brain samples of AD mice with genetic or pharmacologic reduction of eEF2K (Beckelman et al., 2019)(Ma lab, unpublished data). The proteomics data showed multiple proteins whose expression levels were dysregulated (both increased and decreased) in AD mice and restored with eEF2K inhibition. Further functional analysis using bioinformatics tools revealed proteins that were involved in synaptic function, cytoskeletal dynamics, mitochondrial function, calcium buffering, and cell adhesion (Beckelman et al., 2019)(Ma lab, unpublished data). Future in-depth investigation at the functional level would help provide insights into the exact roles of these proteins in AD pathogenesis. Moreover, it should be cautioned that most if not all proteomic studies in the AD field are designed to examine overall protein profiling alterations and did not really differentiate “newly synthesized proteins,” which are critical for the maintenance of long-term forms of memory and plasticity (Alberini, 2008; Klann & Dever, 2004; Sutton & Schuman, 2006).

Decades of research in identifying particular “memory proteins” or “plasticity-related proteins (PRPs)” have not yielded consensus among neuroscientists (Okuda et al., 2020). Here we propose an alternative concept that the “general” protein synthesis up-regulation per se could be the main mechanism for improvement of AD pathophysiology with eEF2K inhibition. This concept is also supported by previous studies demonstrating that the boost of general protein synthesis through regulation on the initiation phase of protein synthesis, either genetically or pharmacologically, resulted in the improvement of multiple aspects of AD pathophysiology (Ma et al., 2013; Oliveira et al., 2021; Yang et al., 2016).

Through electrophysiology studies, Heise and colleagues found that the synaptic activities mediated by the GABAergic inhibitory neurons, which contribute to the tonic inhibition in the hippocampus, were significantly up-regulated in hippocampal slices derived from the eEF2K KO mice (Heise et al., 2017). Dysfunctional GABAergic neurons and reduced tonic inhibition are some of the most frequent causes of brain epileptic/seizure activities (Pavlov & Walker, 2013). In line with the ex vivo electrophysiologic data, the authors further uncovered with in vivo studies that the eEF2K KO mice were more resistant to seizure activities induced by pentylenetetrazol (PTZ) (Heise et al., 2017). Interestingly, multiple lines of evidence suggest a link between epileptic seizure-like activities and the development of AD pathophysiology (Palop & Mucke, 2016).

Besides AD, aberrant eEF2K activity has been linked to several other human diseases including cardiovascular disease and cancer (Liu & Proud, 2016). The development of eEF2K inhibitors was initially motivated by the concept that targeting aberrant protein synthesis associated with eEF2K signaling dysregulation might be a viable anti-cancer strategy (Liu & Proud, 2016). NH125 is among the first identified small-molecule antagonists of eEF2K through studies in cancer cell lines (Arora et al., 2003). We have shown in ex vivo studies that Aβ-induced LTP failure in mouse hippocampal slices was reserved by treatment of NH125 (Ma et al., 2014). Another study showed that in vivo NH125 treatment was able to reduce eEF2 hyper-phosphorylation and seizure phenotype in an epilepsy mouse model (Heise et al., 2017). It needs to be cautioned that two studies conducted in non-neuronal cell lines argued the specificity and inhibition potency of NH125 on eEF2K as well as the actual effects on eEF2 phosphorylation (Chen et al., 2011; Devkota et al., 2012). Investigation on potential eEF2K-independent mechanisms underlying the beneficial effects of NH125 on electrophysiologic and cognitive phenotypes in neuronal disorders warrants further studies.

In addition to NH125, several other small molecule compounds are shown to inhibit eEF2K activities such as A-484954 and Jan-384 (Kenney et al., 2016; Liu & Proud, 2016). Very few studies have been performed to test the effects of these inhibitors on AD-related abnormalities, especially at the functional level. The compound A-484954 is considered to be more potent and selective than NH125 based on studies from cell culture (Chen et al., 2011). Research from our laboratory revealed that treatment with A-484954 (in two different AD model mice) was able to rescue AD-associated cognitive dysfunction assessed by multiple behavioral assays including MWM and NOR. Similar to the results from the genetic manipulation, the therapeutic effects of the eEF2K inhibitor appear to be independent of effects on brain Aβ and tau pathology. Furthermore, A-484954 treatment alleviated AD-associated defects in dendritic spine morphology, postsynaptic density formation, protein synthesis, and dendritic polyribosome assembly (Ma lab, unpublished data). Our proof-of-principle study suggests translational implication of inhibiting eEF2K for AD and related dementia syndromes.

There are some caveats in proposing eEF2K inhibitors as a potential therapy for AD and related dementia syndromes. First, as indicated in the discussion above, short-term activation of eEF2K and eEF2 phosphorylation might be an important mechanism that is actually beneficial for neurons to cope with cellular stress (Figure 1). Further, too much or too little protein synthesis can both lead to neuronal diseases with impairments of synaptic and cognitive functions (Rosenberg et al., 2014; Scheper et al., 2007; Tahmasebi et al., 2018). It is thus critical to optimize the treatment paradigm (dose, duration, etc.) to “restore” (but not over-suppress) levels of eEF2 phosphorylation and the machinery for mRNA translation. Second, overall, there is a lack of understanding of the three-dimensional structure and binding mechanisms of the current eEF2K inhibitors (e.g., A-484594), which hampers the development of eEF2K inhibitors with optimal selectivity and potency (Liu & Proud, 2016; Wang et al., 2015). It is urgent to identify more selective and potent antagonists of eEF2K, which would be a useful tool to help investigate the roles of eEF2 phosphorylation in cognitive function and neuronal plasticity.

6 CONCLUSIONS AND FUTURE PERSPECTIVES

Regulation of eEF2K activity and eEF2 phosphorylation is involved in the process of memory formation and synaptic plasticity, as evidenced in ample research studies from various model systems. A number of recent studies have also indicated that targeting dysregulation of eEF2K/eEF2 signaling could be a viable therapeutic strategy for AD and related dementia syndromes. Meanwhile, the use of the global eEF2K knockout model makes it challenging to tease out peripheral effects of eEF2 dephosphorylation. For a more detailed understanding of the roles of eEF2K/eEF2 signaling in the central nervous system under physiologic or pathologic conditions (e.g., AD), it is critical to develop advanced model systems with region-specific manipulation (up- or down-regulation) of eEF2K and eEF2 phosphorylation. Given all the arguments around the compounds currently available for inhibition of eEF2K, there is also an urgent need for the development of better eEF2K inhibitors to facilitate both basic and translational research related to the eEF2K signaling. Finally, there are plenty of studies indicating a pivotal role of eEF2K signaling and related protein synthesis regulation in tumorigenesis (Liu & Proud, 2016). Interestingly, previous studies in the human population using epidemiologic approaches suggest that there might be an inverse relationship between the incidence of AD and cancer (Driver et al., 2012; Musicco et al., 2013; Realmuto et al., 2012). It would an intriguing topic for future studies to elucidate whether dysregulation of the eEF2K/eEF2 signaling represents one of the common pathologic mechanisms at the molecular level contributing to the development of both cancer and AD.