Suppression of neuronal eEF2K alleviates cognitive deficits and apathy-like behavior in APP/PS1 AD model mice

Abstract

Alzheimer’s disease (AD) is a complex neurodegenerative disorder characterized by synaptic failure, cognitive impairment and neuropsychiatric symptoms (NPS). Apathy is the most common NPS seen in AD patients, and its underlying mechanisms remain unknown. Here, we investigated the roles of neuronal eukaryotic elongation factor 2 (eEF2) phosphorylation (by its kinase eEF2K) in AD-associated cognitive deficits and NPS. We performed a series of experiments using a multidisciplinary approach including genetics, behavioral assays, synaptic electrophysiology, and unbiased proteomics. The results demonstrated that neuron-specific inhibition of eEF2K and eEF2 phosphorylation can alleviate cognitive deficits, synaptic plasticity impairments, and apathy-like behavior in aged APP/PS1 AD model mice. Our findings indicate the therapeutic potential of targeting the eEF2K signaling in the treatment of dementia and NPS in AD and related dementias (ADRDs).

Introduction

Alzheimer’s disease (AD) is a complex neurodegenerative disorder that is characterized extracellular amyloid-β plaques and intracellular tau neurofibrillary tangles, among other pathological hallmarks, that contribute to synaptic dysfunction and cognitive decline. In addition to learning and memory impairments, AD is also characterized by neuropsychiatric symptoms (NPS) such as apathy, anxiety, and depression¹. Previous studies show that 97% of AD patients have at least one NPS^2,3. Compared to dementia, NPS in AD are significantly understudied, and their underlying mechanisms remain elusive. Apathy is the most common NPS in AD patients, which is clinically defined as decreased motivation and goal-directed behavior that impair aspects of daily functioning^4–6. Apathy has been associated with greater caregiver burden and faster cognitive and functional decline^4–6. Several studies have found that apathy is a predictor of future cognitive decline in many disease states, as well as the transition from mild cognitive impairment (MCI) to AD^7–15. Recent studies have found that apathy is also a common feature of mouse models of AD and Downs syndrome (DS), a genetic risk factor for AD^16–18. Currently, little is known about the underlying mechanisms of AD-associated apathy. Insight into the molecular underpinnings of this common neuropsychiatric disorder could have far-reaching implications in the management and treatment of AD and related dementias (ADRDs).

Dysregulation of protein synthesis (mRNA translation) is linked to cognitive impairments and neuropsychiatric disorders^19–23. Previous studies indicate that hyper-phosphorylation of eukaryotic elongation factor 2 (eEF2) by its kinase eEF2K plays a role in AD pathophysiology^22,24,25. eEF2 is a GTPase that mediates the translocation of the peptidyl-tRNA from the A site to the P site within the ribosome, thereby facilitating the elongation step of protein synthesis²⁶. When eEF2 is phosphorylated at T56 by eEF2K, eEF2 function is impaired and elongation is impaired thereby inhibiting general protein synthesis. Protein synthesis is essential for long-lasting forms of memory and synaptic plasticity^27,28. Consistently, impaired protein synthesis capacity is linked to AD pathogenesis^21,22,24,25. Notably, eEF2K is an integral target of several metabolic pathways that are also involved in AD pathophysiology including mammalian target of rapamycin complex 1 (mTORC1) and AMP-activated protein kinase (AMPK) signaling pathways²². The convergence of these pathways on eEF2K enables coupling of elongation to the nutrient and energy demands of the cell²².

Here, taking advantage of a novel conditional eEF2K knockout mouse model, we asked whether suppression of eEF2 phosphorylation in excitatory neurons can alleviate AD-associated cognitive impairment. Importantly, we investigated the roles of eEF2K/eEF2 signaling dysregulation in the development of NPS particularly apathy-like behavior in APP/PS1 mouse model of AD. We performed a battery of behavioral tests to look at hippocampal-dependent learning and memory, as well as multiple neuropsychiatric domains including apathy, anxiety, and anhedonia. We also evaluated the effect of inhibiting neuronal eEF2K in APP/PS1 mice on synaptic function. Lastly, we performed a proteomic analysis to determine proteins dysregulated in APP/PS1 mice that could be rescued by reduction of neuronal eEF2K.

Methods and Materials

Mice

All mice were housed at the Wake Forest School of Medicine barrier facility under the supervision of Institutional Animal Care and Use Committee. The facility operates in accordance with standards and policies of the US Department of Agriculture’s Animal Welfare Information Center (AWIC) and the NIH Guide for Care and Use of Laboratory Animals. Mice adhered to a 12-h light/12-h dark cycle, with regular feeding, cage cleaning, and 24-h food and water access. Both male and female mice 12–15 months old were used for experimentation. All genotyping was done by polymerase chain reaction (PCR).

The breeders of the floxed eEF2K mice were generously provided by Dr. Kobi Rosenbloom at University of Haifa in Israel. The floxed eEF2K mice were bred with CAMKII-cre mice to produce heterozygous Cre/eEF2K^+/− conditional knockdown mice. Cre/eEF2K^+/− mice were bred with APP/PS1 (purchased from Jackson Laboratory, Strain #: 034832) to generate littermate groups: WT, APP/PS1 (APP), eEF2K^+/−, and APP/PS1/eEF2K^+/− (APP/eEF2K^+/−). APP/PS1 mice express human transgenes for APP (KM670/671NL) and presenilin-1 (PSEN1 L166P)³⁰.

Western Blot

Mouse hippocampal tissue was dissected and flash frozen on dry ice. Lysis and gel electrophoresis were performed as previously described¹⁰⁵. Briefly, the tissue was sonicated in lysis buffer with protease and phosphatase inhibitors. Protein quantification was determined using the Bradford colorimetric Assay (Thermo Fisher Scientific, catalog #23227). Equal amounts of protein lysate from each sample were loaded on 4% − 12% Tris-glycine SDS-PAGE gels (Bio-Rad 18-well gel, catalog #5670184) followed by standard gel electrophoresis. After transfer, membranes were blocked for 10 min with SuperBlock TBS Blocking Buffer (Thermo Fischer Scientific, catalog #37535). All primary and secondary antibodies were diluted in 5% milk/TBST or 5% BSA/TBST. Blots were incubated with the following primary antibodies overnight at 4°C: phospho-eEF2 (Thr56) (1:1000, Cell Signaling Technology, catalog 2331); eEF2 (1:1000, Cell Signaling Technology, catalog 2332); β-Actin (1:5000, Sigma Aldrich, catalog A2228); 6E10(1:1000; BioLegend, catalog# 803001). Blots were then incubated in secondary antibodies for 2 h at room temperature. The ChemiDoc Imaging System (Bio-Rad) was used for protein visualization. Densitometric analysis was performed using ImageJ software (NIH). Phospho-proteins were normalized to the levels of total protein; total proteins were normalized to the housekeeping proteins β-Actin.

Aβ ELISA

Mouse hippocampal tissue was lysed as previously described and the supernatant was collected for ELISA^32,106. Aβ 1–42 (Thermo Fisher Scientific, catalog KMB3441) and Aβ 1–40 (Thermo Fisher Scientific, catalog KMB3481) sandwich ELISAs were performed according to the manufacturer’s instructions. 96-Well plates were read at 450 nm using an iMark microplate reader (Bio-Rad).

Immunohistochemistry

Immunohistochemistry was performed as previously described²⁴. Briefly, 5 μm thick paraffin-embedded sections were mounted on charged slides, cleared in xylene and rehydrated through a graded ethanol series. Sections were pretreated in boiling citrate buffer for 10 minutes and blocked in 3% H₂O₂ for 25 minutes. To reduce nonspecific signals, sections were blocked using the Vector M.O.M. Kit according to the manufacturer’s specifications (Vector Labs, catalog BMK-2202). Primary antibody 6E10 (mouse monoclonal; 1:200; BioLegend, catalog SIG-39320) or p-eEF2 (1:200; Invitrogen, PA5–38085) was incubated overnight at 4°C in a humid chamber. Following secondary antibody, sections were incubated in ABC reagent (Vectastain ABC Kit; Vector Labs, catalog PK-4000) followed by DAB solution (ImmPACT DAB; Vector Labs, catalog SK-4105) according to the manufacturer’s instructions. Sections were dehydrated through a graded ethanol series, cleared in xylene, and coverslipped with VectaMount Permanent Mounting Medium. Slides were imaged at ×2, x20 and ×40 on a Keyence BZ-X710 microscope. Hippocampal regions were blinded and quantified. Densitometric analysis was performed using either x20 (for p-eEF2) or ×40 (for 6E10) stitched images of hippocampus and ImageJ software.

Acute Hippocampal Slice Preparation

A Leica VT200S vibratome was used to prepare 400 μm transverse acute hippocampal slices as previously described¹⁰⁶. Slices were maintained in ACSF bubbled with 95% O₂ / 5% CO₂ at room temperature for 2 h before the experiment. ACSF contained the following: 118mM NaCl, 3.5mM KCl, 2.5mM CaCl2, 1.3mM MgSO4, 1.25mM NaH2PO4, and 15mM glucose.

Surface Sensing of Translation (SUnSET) Assay

Surface Sensing of translation was performed as previously described^32,46. Briefly, acute hippocampal slices were incubated in 1μg/mL puromycin for 1 h at 32°C in bubbling ACSF. Slices were flash frozen on dry ice, lysed, and standard gel electrophoresis run. The puromycin-labeled proteins were identified by antibody (1:10,000, Millipore, catalog #MABE343). Total lane density was used to determine protein synthesis levels and densitometry was done using ImageJ (NIH).

Electrophysiology

Acute hippocampal slices were prepared as described above. Following the 2 hr incubation period, slices were maintained at 32°C, and monophasic current stimuli of 100 μs were delivered with a bipolar silver electrode in the stratum radiatum of area CA3. Field excitatory postsynaptic potentials (fEPSPs) were recorded using a glass microelectrode from the stratum radiatum of CA1. LTP was induced using a high-frequency stimulation (HFS) comprised of two 1s, 100 Hz trains, with a 60s interval, delivered at 60–70% of evoked spike intensity. The input–output relationship was determined by increasing the magnitudes of stimuli from 0 to 10 mV at a step of 0.5 mV. Paired-pulse ratio was measured by delivering two identical stimuli separated by 25 to 200 ms at a step of 25 ms.

Behavioral Assays

For all behavioral assays, mice were handled prior to behavioral testing and were habituated to the testing room for 1hr before start of experiments. Experiments were performed during the 12-hr light cycle. Behavioral assays were ordered from least stressful to most stressful. Experimenter was blinded to all genotypes during experiments and analysis.

Morris Water Maze (MWM)

A 5-day MWM protocol was performed with 4 trials a day (60s maximum per trial, 15min interval between trials) as previously described³². Escape latency was measured for each trial. Two hours after training on the 5th day, a probe trial was performed. EthoVision XT Tracking Software (Noldus Information Technology) was used to track trajectories, time spent in maze quadrants, distance and velocity.

Visible Platform (VP)

VP was performed after MWM with a 2-day protocol consisting of 4 trials a day (60s maximum per trial, 15min interval between trials)³¹. The platform was marked by a visual cue and moved randomly among 4 locations. Escape Latency was measured for each trial.

Novel Object Recognition (NOR)

Mice were habituated to the experiment chamber 1 day before start of experiments. A 2-day familiarization protocol was used. On the first 2 days, mice were placed in a white plastic chamber (40cm × 40cm × 40cm) with 2 identical objects and allowed to explore for 5min. Twenty-four hours after familiarization, mice were tested in the chamber with one object replaced with a novel object and allowed to explore for 5min. Objects were randomly assigned to each mouse and location of the novel object was counterbalanced. Time spent exploring objects was measured manually and by EthoVision XT Tracking Software (Noldus Information Technology). Mice with < 10s total interaction time were excluded from analysis. Discrimination index was calculated as the novel object interaction time minus the familiar object interaction time divided by the total interaction time.

Nestlet Shredding (NS)

All mice were habituated in the testing room for 1 hour prior to start of testing. Group-housed mice were placed individually into a clean mouse cage with bedding and one piece of cotton fiber nestlet (5 cm × 5 cm, 5 mm thick) that was pre-weighed and placed on top of the bedding in each cage. Each mouse was left undisturbed in the cage with the nestlet for 30 minutes. After the test, the nestlet was left in open space overnight to dry and was weighed again. The last weight and the starting weight were used to calculate percentage of nestlet shredded, a smaller number will indicate more severe apathy-like behavior¹⁰⁷.

Nest Building (NB)

Nest building was performed after the nestlet shredding and marble burying tests. The group-housed mice were habituated with a cotton fiber nestlet (5 cm × 5 cm, 5 mm thick) in their home-cage overnight before the day of testing. On the test day, mice were individually housed in a new cage with fresh, unscented bedding with a piece of clean and pre-weighed nestlet. The mice were allowed to behave freely overnight and the nests were evaluated the following morning. The nest from each mouse was photographed and the nests were graded using scores ranging from 1 (very poor/no nest building) to 5 (optimal nest building) with half-point scores for nests falling between categories according to a previously published scoring scheme by Deacon¹⁰⁸. Similar to the nestlet shredding test, the amount of nestlet left unshredded was weighed and used to calculate the percentage of unshredded nestlet. A lower nest score and a higher percentage of unshredded nestlet indicates more severe apathy-like behavior.

Marble Burying Test (MBT)

Group-housed Mice were habituated in the testing room 1 hour before the. A standard rat cage (26 cm × 48 cm × 20 cm) was used with 5cm layer of unscented mouse bedding material. 15 standard glass toy marbles (assorted styles and colors/15mm diameter, 5.2g in weight) were gently put on the surface of the bedding in 3 rows of 5 marbles. Each mouse was gently placed in the cage in a corner away from marbles and allowed to behave freely, undisturbed for 30 minutes. The percent of total marble volume buried for each marble was recorded and the number of those with > 2/3 total volume buried were counted as buried. Fewer marbles buried indicates more severe apathy-like behavior¹⁰⁷.

Burrowing

Group-housed mice were habituated with a burrowing tube filled with food pellets in their home-cage overnight before the day of testing. On the testing day, the mice were individually housed in a new cage with fresh bedding. A clean burrowing tube was filled with 200 grams of food pellets and placed in each cage. For the 2-hour burrowing test, the filled burrowing tube was placed in the cage around 2:00–3:00pm and the mouse was allowed to behave freely for two hours. Following the 2-hour period, the amount of food left in the tube was weighed and used to calculate the amount burrowed. For the overnight burrowing test, the burrow tube was emptied and refilled with 200g of new food pellets and placed back in the cage overnight. On the following morning, the amount of food left in the tube was weighed and used to calculate the amount burrowed^109,110. Smaller amount burrowed will indicate greater apathy-like behavior.

Open Field (OF)

Mice were placed in a white plastic chamber (40cm × 40cm × 40cm) and allowed to explore for 15min. Time spent in the center and periphery of the chamber, as well as distance moved, and average velocity was measured using EthoVision XT Tracking Software (Noldus Information Technology). The percentage of total time spent in the periphery of the chamber was calculated.

Novelty-Induced Hypophagia (NIH)

NIH was performed as previously cited with minor adjustments¹¹¹. Mice were trained to drink vanilla Ensure^® in their home cage over the course of three days for 30 min each day. The Ensure^® was presented in a 50mL conical tube with a rubber stopper through the wire rack of the cage lid. The amount consumed was weighed in grams and the latency to drink was recorded for each training day. Following training, mice underwent home cage testing following the same procedure as during training. Training and home cage testing were performed in low light. The day after home cage testing, mice underwent novel cage testing where the mice were presented the ensure in a novel cage without a wire rack or bedding and in bright light conditions. The amount consumed was weighed in grams and the latency to drink was recorded for both testing days.

Novelty-Suppressed Feeding Test (NSFT)

NSFT was performed as previously described with minor adjustments¹¹². Mice were food deprived for 48 hours prior to behavioral testing with a 2-hour free feeding period after the first 24 hours. Following food deprivation, mice were placed in a white plastic chamber (40cm × 40cm × 40cm) that had 2cm of bedding in the bottom and a food pellet position in the middle of the arena and secured to a platform at the level of the bedding. The latency for the mouse to begin eating the food pellet was recorded. The mouse was then immediately returned to its home cage with a pre-weighed food pellet for 5 min and the weight of food consumed was recorded.

Sucrose Preference Test (SPT)

Mice were first habituated to two bottles of water in their home cage overnight. Water was presented in a 50mL conical tube fitted with a rubber stopper on the left and right side of the cage. The amount of water consumed was weighed after 24-hour access. One bottle was replaced with a 1% (w/v) sucrose solution. The position of the sucrose solution was randomly assigned for each mouse. The amount of sucrose consumed was weighed after 24-hour access. This was repeated for 2 more days and the position of the sucrose solution was alternated each consecutive day.

Composite z-score Analysis

Z-scores were calculated for each behavior test independently for each mouse relative to the mean and standard deviation (SD) of WT controls so that changes of ± 1 z-score unit represented ± 1 SD of the control sample’s behavior. To reduce the variance and enhance the reliability of our data, we further created composite scores (i.e., a combined z-score for each mouse across all tests) for 2 different behavioral domains: apathy-like behavior and anxiety-like behavior. The raw z-scores were corrected so that a higher value was indicative of greater anxiety or apathy, respectively. The composite score for apathy-like behaviors was calculated using the following equation: (z_i(NS) + z_i(MB) + z_i(NB % unshredded) + z_i(NB score) + z_i(2hr burrowing) + z_i(overnight burrowing))/6, which will indicate overall apathy-like behavior severity^16,55. The composite score for anxiety-like behaviors was calculated using the following equation: (z_i(OF) + z_i(NIH Latency) + z_i(NIH Consumption) + z_i(NSFT))/4, which will indicate overall anxiety-like behavior severity.

Proteomics

SP3 digestion

10ug of sample were diluted into 100 ul lysis buffer (2% SDS, 50 mM HEPEs, pH 8, 50 mM EDTA) was subjected to reduction with 5 mM DTT for 30 min at 60C, alkylation with 20mM iodoacetamide for an hour at room temperature in the dark and followed by SP3 beads digestion method described in Hughes et al¹¹³ with trypsin (sequencing grade, Thermo Scientific Cat#90058) in 100mM ammonium bicarbonate, 2 mM CaCl2 and incubated at 37°C for overnight. Peptides acidified with formic acid and 10% of the sample was analyzed by LC-MSMS.

Liquid chromatography-tandem mass spectrometry (LC-MSMS)

Samples were analyzed by LC-MS using Nano LC-MSMS (Dionex Ultimate 3000 RLSCnano System, Thermofisher) interfaced with Eclipse (Thermofisher).

Samples were loaded on to a fused silica trap column Acclaim PepMap 100, 75umx2cm (ThermoFisher). After washing for 5 min at 5 μl/min with 0.1% TFA, the trap column was brought in-line with an analytical column (Nanoease MZ peptide BEH C18, 130A, 1.7um, 75umx250mm, Waters) for LC-MS/MS. Peptides were fractionated at 300 nL/min using a segmented linear gradient 4–15% B in 30min (where A: 0.2% formic acid, and B: 0.16% formic acid, 80% acetonitrile), 15–25%B in 40min, 25–50%B in 44min, and 50–90%B in 11min. Solution B then returns at 4% for 5 minutes for the next run.

DIA (Data independent acquisition) workflow was used to analyze the eluted peptides. The MS scan range was set to 400–1200 m/z, with a resolution of 12,000, an AGC target of 3E6, and an automatic ion injection time. 8 m/z windows were used to sequentially isolate ions for 400–800 m/z and 20 m/z windows were used to isolate ions from 800–1200. (AGC target 4E5, automatic ion injection time), The ions were fragmented in the C-trap with a relative collision energy of 30. MS/MS spectra were recorded at a resolution of 30,000 in the orbitrap.

Raw data was analyzed using a predicted library from the UniProt human proteome FASTA file for a library-free search with DIA-NN 1.8.1 (https://github.com/vdemichev/DiaNN) under recommended settings. Protease was set as trypsin (C-terminus of K or R unless followed by P). Miss cut of 1 was allowed. Peptide length range from 7–30, precursor charge range of 1–4, precursor m/z range 300–1800 and fragment ion m/z range 200 − 18—were used to generate predicted library. Deep learning-based spectra and RTs and IM prediction were used for spectra matching evaluation. Decory spectra generation method was default as described in Demichev et al, 2020¹¹⁴. The results were filtered with a posterior error probability (PEP) filter of < 0.01 and a protein group Q value (PG.Q) filter of < 0.01. Protein group MaxLFQ values were used for quantification.

Normalized abundances were analyzed for all samples using R. Hierarchical clustering was used to determine sample outliers. One-way ANOVA with Tukey’s post hoc was performed for each protein reported. Proteins that had a corrected p-value < 0.05 and a fold change greater than 20% were reported. Gene ontology enrichment analysis was performed via the STRING database¹¹⁵.

Statistical Analysis

Data are presented as mean ± standard error of the mean (SEM). For comparisons between two groups, a two-tailed independent Student t-test was performed. For comparisons among more than two groups, one-way analysis of variance (ANOVA) was used with Tukey post hoc tests for multiple comparisons. If two levels of factors were involved in multiple group comparisons, two-way ANOVA was used with Tukey post hoc tests for multiple comparisons. P < 0.05 was considered statistically significant. Outliers were determined by Grubbs test. Statistics were performed using Prism 10 software (GraphPad).

Results

Neuronal Reduction of eEF2K prevents learning and memory impairments in APP/PS1 AD model mice

First, we generated eEF2K conditional knockout mice (eEF2K-cKO) by crossing the Camk2a-cre mice with a line of transgenic mice harboring loxP-flanked Eef2k gene²⁹. We then crossed the eEF2K-cKO mice with the APP/PS1 AD model mice (referred to as APP mice)³⁰ to generate the double mutant APP/eEF2K^+/− mice in which eEF2K expression and eEF2 phosphorylation are repressed in excitatory neurons of the forebrain (Supplemental Fig. 1). APP mice had a trending increase in p-eEF2 staining in the hippocampus that was significantly decreased in the APP/eEF2K^+/− mice (Supplemental Fig. 1). We assessed hippocampal-dependent spatial learning and memory using the Morris Water Maze (MWM) task where mice use spatial cues to locate a hidden platform^31,32. Compared to the WT mice, the APP mice showed spatial learning and memory deficits, as indicated by longer escape latencies over the 5 days of training (Fig. 1A) and significantly lower target occupancy (TQ) as well as fewer “platform” crossings in the probe trial phase (Fig. 1B–C). Notably, the APP/eEF2K^+/− mice had a similar learning curve to that of WT mice (Fig. 1A). During the probe trial, performance of the APP/eEF2K^+/− mice (TQ and “platform” crossing) was indistinguishable from the WT mice (Fig. 1B). Consistent with previous studies, the eEF2K^+/− mice exhibited normal spatial learning and memory²⁹ (Fig. 1A–C). In summary, suppression of neuronal eEF2K and eEF2 phosphorylation resulted in alleviation of spatial learning and memory impairments in APP mice. To control for swimming/locomotor ability or potential visual impairments, mice were subjected to the visible platform task (VP)³¹. There were no significant differences in escape latencies between groups for VP (Fig. 1D). The total distance traveled from the MWM probe trial was also analyzed and no significant differences between groups was observed suggesting that there were no visual or locomotor deficits (Fig. 1E).

Figure 1. Suppression of neuronal eEF2K prevents learning and memory impairments in the APP/PS1 AD model mice.

(A) Escape latency over 4 trials/day for 5 days of training in MWM test. APP mice took longer to locate the platform compared to all other groups. (B) Percent of time spent in target quadrant during MWM Probe trial was reduced in the APP mice but not in the APP/eEF2K^+/− mice. (C) Decrease in the number of platform crossings during MWM Probe trial in the APP mice but not in the APP/eEF2K^+/mice. (D) Escape latency over 4 trials/day for 2 days of VP was not different between groups. (E) Distance traveled during MWM probe trial was not significantly different among groups. (F) Ratio of time spent with familiar (empty) and novel (stripped) objects in the NOR task during the testing phase. Preference of less than 0.5 for novel object indicates cognitive impairment as seen in the APP but not in the APP/eEF2K^+/− mice. (G) Discrimination index, calculated as (time interacting with novel object – time interacting with familiar object) / total interaction time, was significantly decreased in the APP mice but not the APP/eEF2K^+/− mice. (H) Distance traveled in open filed (OF) test was not significantly different between groups. (I) Total interaction time in NOR was significantly reduced in the APP but not in the APP/eEF2K^+/− mice. WT, n=13; APP, n=12; eEF2K^+/−, n=12; APP/eEF2K^+/−, n=10. Error bars represent ± SEM. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001; One-way ANOVA with Tukey’s post hoc test.

We next used the novel object recognition task (NOR) to evaluate long-term recognition memory^32,33. In this task, mice must distinguish between a familiar and novel object where a cognitively normal mouse will spend more time exploring the novel object rather than the familiar ones. Consistent with previous studies, the APP mice failed to distinguish novel and familiar objects and showed negative discrimination index (time interacting with novel object – time interacting with familiar object) / total interaction time), indicating impairment of recognition memory (Fig. 1F and 1G). In contrast, the APP/eEF2K^+/− mice exhibited normal recognition memory as indicated by more interaction with the novel object and a positive discrimination index similar to WT group (Fig. 1F and 1G). Additionally, no differences in locomotor activity, as measured by distance traveled in the open field assay (OF), were observed (Fig. 1H). Taken together with the results from the MWM test, the data show that suppressing neuronal eEF2 phosphorylation can alleviate cognitive impairment in the APP/PS1 AD model mice. Interestingly, analysis of the total interaction time during NOR test³⁴ revealed that the APP mice had significantly lower total interaction time compared to the WT group, and such defects were also reversed by suppression of neuronal eEF2K (Fig. 1I).

Reduction of neuronal eEF2K mitigates apathy-like behavior (ALB) in APP/PS1 mice

That APP mice exhibited normal locomotor activities but significantly lower total interaction time with either object in the NOR task (Fig. 1H) suggests potential apathy-like behavior (ALB) associated with the AD model mice³⁴. We thus further assessed ALB using a battery of rodent typical behaviors including nestlet shredding (NS), nest building (NB), marble burying (MB), and burrowing (2 h and overnight)¹⁶.

There was a significant decrease in the amount of nestlet shredded in a 30-minute period in the APP mice that was not observed in the APP/eEF2K^+/− mice (Fig. 2A). In agreement, APP mice built lower scored-nests after an overnight period compared to WT and the APP/eEF2K^+/− mice (Fig. 2B–C). APP mice also left larger portions of the nestlet unshredded at the conclusion of the nest building task compared to all other groups (Fig. 2D). We next assessed marble burying behavior where the APP mice buried fewer marbles compared to wild type mice (Fig. 2E). Notably, a similar though non-significant decrease was seen in the eEF2K^+/− mice and a trending decrease was observed in the APP/eEF2K^+/− mice (Fig. 2E). We also evaluated burrowing behavior after a 2-h period and after overnight. After 2 h, APP mice burrowed significantly less than WT and APP/eEF2K^+/− mice (Fig. 2F). After the overnight period, APP mice still burrowed less compared to WT and eEF2K^+/− mice, while the performance of the APP/eEF2K^+/− mice was indistinguishable from WT mice (Fig. 2G).

Figure 2. Suppression of neuronal eEF2K alleviated apathy-like behavior in the APP/PS1 AD model mice.

(A) The percentage of nestlet shredded in a 30 min period was decreased in the APP mice but not in the APP/eEF2K^+/− mice. (B) Representative images of nests for nest building experiments (C) Nest building score was decreased in the APP mice but not in the APP/eEF2K^+/− mice. (D) APP mice left a higher percentage of the nestlet unshredded during the nest building task which was reversed in the APP/eEF2K^+/− mice. (E) The number of marbles buried in a 30 min period was decreased in the APP mice with a similar trending decrease in the APP/eEF2K^+/− mice. (F) APP mice burrowed less in the 2hr burrowing task compared to all other groups. (G) APP mice, but not the APP/eEF2K^+/− mice, burrowed less in the overnight burrowing task. (H) The APP mice had significantly higher combined apathy z-score compared to all other groups. WT, n=10; APP, n=11; eEF2K^+/−, n=11; APP/eEF2K^+/−, n=9. Error bars represent ± SEM. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001; One-way ANOVA with Tukey’s post hoc test.

Furthermore, z-score was calculated from the behavior of each mouse in each test and the values combined to reduce variance and enhance reliability¹⁷. A higher combined z-score is indicative of greater apathy-like behavior. In brief, the APP mice had a significantly higher combined z-score of ALB compared to the WT group, which was blunted in the APP/eEF2K^+/− group (Fig. 2H). Taken together, suppression of neuronal eEF2K expression and eEF2 phosphorylation can mitigate apathy-like behavior associated with the APP/PS1 AD model mice.

Besides apathy, anxiety and anhedonia are common neuropsychiatric symptoms in AD patients^1,2,35. We investigated the presence of anxiety-like and anhedonia behaviors in APP mice and then determined if this behavior was affected by neuronal suppression of eEF2K. The open field (OF) test was first used to assess anxiety-like behavior³⁶. There was no significant difference between WT and APP mice for the percentage of time spent in the periphery (Supplemental Fig. 2A). Additionally, there was no change in the OF test with eEF2K suppression. We also performed the novelty induced hypophagia task (NIH). For this task, mice are habituated to a novel food source in their home cage and then presented it again in a novel cage environment. In general, normal mice should take longer to drink, and drink less, in the novel environment. In accordance, all groups had a higher latency to drink in the novel cage (Supplemental Fig. 2B). However, APP mice had a significantly higher latency to drink in the novel cage compared to both WT and APP/eEF2K^+/− mice (Supplemental Fig. 2B). All groups consumed the same amount (as percentage of body weight) in the home cage (Supplemental Fig. 2C). While all groups consumed less in the novel environment compared to the home cage, there was no difference between groups in the novel cage (Supplemental Fig. 2C). The mice also underwent the novelty suppressed feeding task (NSFT) where they are food-deprived prior to testing and then the latency to eat in a novel arena is measured as well as the amount eaten in the 5 min immediately following the test. We saw no significant differences between groups in latency to eat in the NSFT (Supplemental Fig. 2D). However, APP mice did consume more food following the test compared to eEF2K^+/− mice but was not significantly different from WT (Supplemental Fig. 2E). We also used a combined z-score to determine the presence of anxiety-like behavior. There was no significant difference between groups suggesting that absence of an overt anxiety phenotype in the APP mice which was unchanged by eEF2K suppression (Supplemental Fig. 2F). Lastly, we examined potential anhedonia via the sucrose preference test (SPT)^37,38. We did not observe any significant differences between groups suggesting APP mice displayed no anhedonia behavior (Supplemental Fig. 2G). In summary, APP mice did not display an overt anxiety or anhedonia phenotype. Further, neuronal suppression of eEF2K had no effect on these behaviors.

Suppression of neuronal eEF2K alleviated impairment of hippocampal long-term potentiation (LTP) in the APP mice

Long-term potentiation (LTP) is a well-established form of synaptic plasticity and cellular model of learning and memory³⁹. A growing body of evidence demonstrates that LTP deficits are commonly present in neurodegenerative disorders (e.g. AD) and a number of neuropsychiatric disorders such as autism spectrum disorder, schizophrenia, bipolar disorder, and depression^40–42. We thus preformed synaptic electrophysiology experiments to evaluate the effect of genetic suppression of neuronal eEF2K on LTP and synaptic function in APP mice. As described, LTP was induced in acute hippocampal slices by high frequency stimulation (HFS)⁴³. Compared to WT slices in which robust LTP was induced and maintained, slices derived from the APP mice exhibited transient LTP (Fig. 3A–B). At 90 mins post HFS, APP slices had a significantly lower slope change compared to WT slices (Fig. 3C). Importantly, we found that neuronal suppression of eEF2K was able to reverse LTP deficits in APP mice, as indicated by normal LTP induction and maintenance in slices from the APP/eEF2K^+/− double mutant mice (Fig. 3A–C). Additionally, slices from the eEF2K^+/− mice showed normal LTP (Fig. 3A–C). We also analyzed input/output (I/O) curves and paired pulse facilitation (PPF) to assess basal synaptic transmission and presynaptic function^44,45 and did not find difference among all four groups (Fig. 3D–E). These findings are consistent with the behavioral data, demonstrating that synaptic plasticity impairments in the APP mice can be improved by neuronal suppression of eEF2 phosphorylation.

Figure 3. Suppression of neuronal eEF2K alleviated hippocampal long-term potentiation (LTP) deficits in the APP/PS1 AD model mice.

(A) Acute hippocampal slices were stimulated with high frequency stimulation (HFS, indicated by arrows) to induce LTP. Compared to sustained LTP in WT slices, APP slices showed transient, impaired LTP, which was improved in the slices from the APP/eEF2K^+/− mice. (B) Representative fEPSP traces. (C) Compared to the WT group, fEPSP slope 90 min after HFS was significantly decreased in the APP slices but not in the APP/eEF2K^+/− slices. WT, n=5 slices; APP, n=10 slices; eEF2K^+/−, n=7 slices; APP/eEF2K^+/−, n=6 slices. (D) Input/output curve assessing basal synaptic transmission was not significantly different between groups. WT, n=9 slices; APP, n=7 slices; eEF2K^+/−, n=9 slices; APP/eEF2K^+/−, n=7 slices. (E) Curve of Paired-pulse ratio with varying interval time as a measure of presynaptic transmission was not different between groups. WT, n=10 slices; APP, n=11 slices; eEF2K^+/−, n=10 slices; APP/eEF2K^+/−, n=8 slices. Error bars represent ± SEM. *p<0.05, ****p<0.0001. One-way ANOVA with Tukey’s post hoc test.

Effects of neuronal eEF2K reduction on proteome dysregulation in the APP/PS1 AD model mice

Given the behavioral and synaptic deficits observed in the APP mice that were blunted with neuronal suppression of eEF2K, we wanted to determine how eEF2K suppression affected the proteome of APP mice. Given that eEF2K negatively regulates the elongation phase of protein synthesis²⁶, we first measured de novo protein synthesis using the surface sensing of translation (SUnSET) assay⁴⁶. Here, acute hippocampal slices were incubated with a subthreshold dose of puromycin, a tRNA analogue, followed by western blot experiments using an anti-puromycin antibody⁴⁶. Neuronal suppression of eEF2K led to a trend for an increase in protein synthesis in the APP/eEF2K^+/− mice compared to APP mice (Supplemental Fig. 3).

Moreover, we performed mass spectrometry-based proteomics to understand how suppression of neuronal eEF2K in APP mice affected the proteome. In summary, over 7,889 proteins were detected, and after analysis with one-way ANOVA with Tukey’s post hoc test, 1,372 proteins were found to be significantly different across all group comparisons that are shown in a heatmap (Fig. 4A). Normalized abundances were analyzed for all samples using R. Hierarchical clustering was used to determine sample outliers (Supplemental Fig. 4). Proteins were considered to be dysregulated if they met the criteria for a fold change greater than 20% with a p-value less than 0.05. When compared to WT, APP mice had 63 upregulated proteins and 32 downregulated proteins (Supplemental Fig. 5A, Supplemental Table 1). The increased proteins included APP, PSEN1, Nicastrin, APOE, and GFAP. Gene ontology enrichment analysis confirmed increases in proteins related to amyloid pathology and immune processes (Supplemental Fig. 5B).

Figure 4. Effects of eEF2K reduction on protein profiling in APP/PS1 AD mocel mice.

(A) Heatmap of proteins significantly dysregulated between groups. WT, n=4; APP, n=4; eEF2K^+/−, n=4; APP/eEF2K^+/−, n=4. One-way ANOVA with Tukey’s post hoc test. (B) Volcano plot showing the distribution of differently regulated proteins in the APP/eEF2K^+/− group compared to the APP group. Gray circles are the proteins that were not significantly altered. Green circles represent proteins that had a fold change greater than 20% but were not significant. Blue circles represent proteins that were significantly changed but had a fold change less than 20%. Red circles represent the proteins that were significant and had a fold change greater than 20%. (C) Gene ontology enrichment analysis of all proteins that met the p-value and fold change requirements. (D) Venn diagram of upregulated proteins in APP vs WT (n = 59, green), downregulated proteins in APP vs WT (n = 27, pink), upregulated proteins in APP/eEF2K^+/− vs APP (n = 97, blue), and downregulated proteins in APP/eEF2K^+/− vs APP (n = 26, yellow). Two proteins were upregulated in APP mice but were downregulated by reduction of eEF2K, and 5 proteins were downregulated in APP but were upregulated by reduction of eEF2K. (E) Table showing the p-value and fold change of the 7 proteins from D that were dysregulated in APP vs. WT group but were no different from WT levels in APP/eEF2K^+/− group. Green cells indicate a p-value that is less than 0.05. Yellow cells indicate a trending p-value between 0.1 and 0.05. Blue cells indicate a positive fold change greater than 20%. Red cells indicate a negative fold change greater than 20%.

In comparing the APP/eEF2K^+/− and APP mice, 97 proteins met the criteria and were found to be upregulated while there were 26 downregulated proteins (Fig. 4B, Supplemental Table 2). Gene ontology enrichment analysis revealed that a large number of the differentially regulated proteins were associated with transport and localization (Fig. 4C). These included subunits for ion transporters (kcnmb4, Kcnip4), and many proteins involved in transmembrane transport (Slc14a1, Slc5a3, Slc7a14, Slc7a5, Slc7a8, Slc38a7, Slc4a7, Slc31a1, Slc37a4). Another group of proteins was associated with oxidative phosphorylation (Cox7a2, Cox7c, Uqcrh, Ndufb8) and proton transport (Atp5j, Atp6v1f, Atp6v1b2, Atp6v0c, mt-Atp8, mt-Atp6). A further cluster of proteins included several G protein gamma subunits (Gng2, Gng3, Gng4, Gng5, Gng7, and Gng12). Lastly, several members of the SNARE complex, which mediates vesicle-membrane fusion and is involved in neurotransmitter release, were also differentially regulated (Snap25, Vamp2, Cplx1, Cplx2)⁴⁷.

We further compared the differentially regulated proteins in the APP/eEF2K^+/− vs. APP group to those dysregulated in the APP vs. WT group (Fig. 4D). There were 2 proteins that were up in APP vs. WT but down in APP/eEF2K^+/− vs. APP (Cbx7 and Drp2). There were another 5 proteins that were down in APP vs. WT but up in APP/eEF2K^+/− vs. APP (Atp6v0c, CD47, Gpr26, Shisa4, and Slc37a4) (Fig. 4E). We verified that these proteins returned to WT levels with suppression of eEF2K in the APP mice (Fig. 4E).

In summary, APP mice exhibited changes in the proteome compared to WT that largely consisted of amyloid-related pathways and immune processes. Neuronal suppression of eEF2K in the APP mice resulted in differential regulation of levels of proteins that were involved in localization and transport but did not affect any amyloid-related proteins. Lastly, 7 proteins that were dysregulated in the APP mice compared to WT, returned to WT levels in the APP/eEF2K^+/− mice (Fig. 4E).

We investigated the effects of suppression of neuronal eEF2K/eEF2 phosphorylation on brain Aβ pathology in APP/PS1 AD model mice. By performing immunohistochemical experiments, we examined brain Aβ plaque and found no difference between APP and APP/eEF2K^+/− mice (Supplemental Fig. 6A-B). Western blot analysis using the 6E10 antibody showed increased APP protein as well as Aβ monomer and dimer in the APP mice (Supplemental Fig. 6C-F). However, there was no significant difference between the APP and the double mutant APP/eEF2K^+/− mice (Supplemental Fig. 6C-F). We then performed ELISA experiments to determine whether neuronal eEF2K inhibition regulated brain levels of Aβ₁₋₄₀ and Aβ₁₋₄₂. The results of ELISA did not reveal any difference between the APP and APP/eEF2K^+/− mice (Supplemental Fig. 6G-I). In brief, these data indicate that neuronal suppression of eEF2K could alleviate cognitive deficits and synaptic plasticity impairments in AD model mice independent of an effect on brain Aβ pathology.

Discussion

In this study, we utilized a conditional suppression of neuronal eEF2K in APP mice to better assess the potential of eEF2K as a therapeutic target for AD-associated cognitive dysfunction and apathy-like behavior. This differs from our previous studies that used global knockout of eEF2K in AD model mice^24,48. Importantly, use of the neuronal-specific suppression of eEF2K eliminates the possibility of broad systemic or peripheral effects. We demonstrated that neuronal suppression of eEF2K in APP mice alleviated cognitive deficits in the MWM and NOR tasks. This is in line with previous work from our lab that demonstrated how genetic or pharmacological reduction of eEF2K activity in AD mice rescued learning and memory impairments^25,32,48. We also demonstrated a robust apathy phenotype in the APP mice that was corrected with neuronal suppression of eEF2K. Additionally, APP mice demonstrated LTP failure that was reversed in APP/eEF2K+/− mice. Lastly, neuronal eEF2K suppression was able to correct the levels of a number of dysregulated proteins in the APP mice back to WT levels; this may be the basis for the positive behavioral and synaptic effects associated with suppressing eEF2K. Additionally, we found that genetic reduction of eEF2K had no significant effect on Aβ pathology and may be eEF2K-independent as previously reported⁴⁸. These findings further confirm the therapeutic potential of eEF2K in the treatment of AD.

Apathy is the most common NPS experienced by AD patients although the underlying mechanism is still unclear^4–6. Only recently has apathy been assessed in preclinical models associated with AD and dementia^16–18. Additionally, there is no consensus or “gold standard” behavior test to measure apathy-like behavior in animal models. Most commonly, a decrease in exploratory behavior, namely decreased movement in open field or decreased interaction time with novel objects, has been used as a measure of apathy^34,49–51. Some studies have also used nest building as a measure of apathy^18,34,52. Other studies utilized operant conditioning with a fixed or progressive ratio to determine motivation, but this is rarely done in AD models^50,53,54 since these tests rely on cognitive ability and can be confounded by the presence of cognitive impairments. In contrast to these, previous studies, we aimed to use multiple different behavior paradigms to assess apathy-like behavior in AD-model mice.

Here, we used a battery of rodent typical behaviors that are generally thought to be independent of cognitive abilities, given AD mice have known learning and memory impairments^16,17. We first wanted to confirm whether the APP mice have an apathetic phenotype. We observed that the APP mice had decreased total interaction time in NOR which may be indicative of apathy-like behavior³⁴. Following with the other rodent typical behaviors, we found that APP mice had marked deficits in nestlet shredding, nest building, marble burying, and burrowing which indicate a robust apathy phenotype. We used a z-score analysis to be able to compare groups across multiple different behaviors. This approach provides greater sensitivity and reliability when assessing complex behaviors such as apathy^55,56. To our knowledge, this is the first-time apathy-like behavior has been robustly characterized in the APP/PS1 mice. Since eEF2K has been implicated in both AD and other neuropsychiatric disorders, we evaluated whether genetic reduction of eEF2K in excitatory neurons could improve apathy-like behavior seen in the APP mice. APP/eEF2K^+/− mice showed improved nest building and burrowing when compared to APP mice. This is in concurrence with other studies that found deficits in nest building behavior in APP/PS1 mice^52,57. There was an increase in nestlet shredding in the APP/eEF2K^+/− mice though not significantly different from the APP group. The APP/eEF2K^+/− mice were also not significantly different from WT mice which may indicate a modest improvement. The exception was marble burying where there was no difference between the APP and the APP/eEF2K^+/− mice. This may most likely be explained by a deficit in the eEF2K^+/− mice. Genetic reduction of eEF2K in excitatory neurons seems to affect marble burying behavior differently. This effect may also be due to the non-specificity of the marble burying task which is sensitive to a host of different conditions, drugs, and neurotransmitter systems⁵⁸. Interestingly, one study found an increase in the number of marbles buried in APP/PS1 mice compared to WT⁵⁷. However, in that study, the APP/PS1 mice were younger (6 months of age compared to our 12 months of age). This could suggest age-dependent changes in marble burying behavior in the APP/PS1 mice⁵⁷. Although more work is needed to understand this relationship. Lastly, we used a combined z-score that was averaged from the individual z-scores from each test. This allowed us to have greater sensitivity and reliability in determining apathy-like behavior and allows for the summary and comparison of multiple behavior tasks^55,56. The combined z-score from each individual test and averaged by subject showed a robust apathy phenotype in the APP mice that was significantly reduced with genetic reduction of eEF2K.

We next wanted to look at anxiety-like behavior as it may affect both cognitive and other NPS-related behaviors. Anxiety is also a common feature of AD. We used OF, NIH, and NSFT to assess anxiety behavior in APP mice. We did not observe any significant differences between groups in OF. In NIH, however, we did see an increase in latency to drink in the novel cage in APP mice that was significantly lower in the APP/eEF2K^+/− mice. We did not see a similar trend in the NSFT where there was no difference between groups in latency to eat. The combined anxiety z-score confirmed that there was no overt anxiety phenotype in any of the groups compared to WT mice. From the literature, it is inconclusive if APP mice have an anxiety phenotype⁵⁹. It may depend heavily on the specific behavioral paradigm used. We tried to mitigate this effect by using a number of different behavioral paradigms. We lastly looked at anhedonia via SPT and did not see any significant differences between groups. Few studies have assessed anhedonia in APP/PS1 mice. One study found a decrease in sucrose preference in young (6 month old) APP/PS1 mice⁵⁷. This too points to the potential of age-dependent changes in sucrose preference in APP mice. One other study found that 6-month old APP/PS1 mice exhibited decreased sucrose preference under a chronic mild stress paradigm, although there was no difference between WT and APP mice before the stress procedure⁶⁰.

We performed unbiased proteomic analysis to identify those proteins whose levels are dysregulated in APP/PS1 mice but are normalized, wholly or partially, by reduction of eEF2K, since such proteins may contribute to the cellular and molecular mechanisms underlying the rescue effect of the behavioral phenotypes. When compared to WT, APP mice had a significant increase in proteins associated with amyloid processing and regulation. These included increased expression of Amyloid precursor protein (APP), Presenilin1 (Psen1), Aph-1 Homolog B (Aph1b) and Nicastrin (Ncstn), the last three of which are γ-secretase components and likely increase the amyloidogenic processing of APP^61,62. Several others that were also increased in APP mice, including Apolipoprotein E (ApoE), Low-density lipoprotein receptor (Ldlr), and Clusterin alpha chain (Clu) are involved in the negative regulation of amyloid fibril formation and Aβ clearance^63–65. Additionally, ATP-binding cassette sub-family G member 1 (Abcg1) may negatively regulate γ-secretase activity⁶⁶. Interestingly, the APP mice had a higher number of upregulated proteins than downregulated ones. This could be due to mRNA subtypes and changes to the rate of translation or degradation. This may explain the lack of significant differences in the SUnSET assay between APP and WT mice.

When comparing APP/eEF2K^+/− vs. APP samples, a total of 123 proteins were found to be significantly different. One of the largest and most significantly enriched category of proteins was associated with transport. Within the large transport family were subgroups associated with transport of ions, organic substances, and nitrogen compounds. Another group of proteins was associated with mitochondrial processes linked to ATP production (Upregulated: Cox7a2, Cox7c, Uqcrh, Ndufb8, Atp5j, mt-Atp8, mt-Atp6, Ndufb8, Atp5j, mt-Atp8, mt-Atp6, Ampd3; Downregulated: Gapdh, Atp6v1b2). Mitochondrial dysfunction is a key feature of AD^67,68. Some of the most notable perturbations are impaired activity of Complex I and Complex IV of the electron transport chain as well as decreased ATP production and increased reactive oxygen species (ROS) production⁶⁸. Ndufb8 is a critical structural subunit of Complex I and an increase in its expression could promote complex I assembly or stability⁶⁹. Cox7a2 is a nuclear encoded structural subunit of complex IV and Cox7c is a mitochondrial encoded catalytic subunit of complex IV. Increased expression of these complex IV subunits may lead to increased complex IV activity and potentially address the deficit seen in APP mice⁷⁰. Additionally, ATP production and the energy status of the cell are tied to eEF2K signaling through AMPK. AMPK is a cellular energy sensor that directly phosphorylates and activates eEF2K to tune the rate of elongation and protein synthesis to the cellular energy state²². eEF2K inhibition may indirectly affect AMPK activity since a greater rate of translation utilizes more ATP and alters the ATP:AMP ratio in the cell, thereby activating AMPK signaling as a feedback loop⁷¹.

Another cluster of proteins included several G protein gamma subunit isoforms (Upregulated: Gng2, Gng3, Gng4, Gng5, Gng7, and Gng12). G proteins, which are regulated by G protein coupled receptors, are composed of three subunits (α, β, and γ), and are important intracellular signaling molecules. There are 13 different γ subunit isoforms and here, we find that 6 of them are upregulated in APP mice with decreased eEF2K expression. This is significant and could affect a wide range of signaling cascades⁷². Lastly, several members of the SNARE complex were also differentially regulated (Upregulated: Snap25, Vamp2, Cplx1, Cplx2). SNARE complex dysfunction has been associated with AD and other neurodegenerative disorders as well as neuropsychiatric disorders like schizophrenia and depression⁷³. An increase in these proteins may promote vesicular fusion with the synaptic membrane and neurotransmitter release⁷³.

We further compared the differentially regulated proteins in the APP/eEF2K^+/− vs. APP group to those dysregulated in the APP vs. WT group. Two proteins were up in APP vs. WT but down in APP/eEF2K^+/− vs. APP (Cbx7 and Drp2) mice. Another 5 proteins that were down in APP vs. WT but up in APP/eEF2K^+/− vs. APP (Atp6v0c, CD47, Gpr26, Shisa4, and Slc37a4). When these proteins were compared in APP/eEF2K^+/− vs. WT, they were not significantly different. This suggests that neuronal suppression of eEF2K in APP mice returned these dysregulated proteins to WT levels.

Chromobox homolog 7 (Cbx7) is an epigenetic regulator of gene expression and is specifically involved in gene silencing. Inhibition of Cbx7 has been shown to activate nuclear factor-erythroid 2-related factor 2 (Nrf2) signaling pathway that responds to oxidative stress^74,75. In this context, decreased Cbx7 may affect the system’s capacity to respond to excessive ROS as seen in APP mice. Our group and others have previously shown that inhibition of eEF2K promotes Nrf2 signaling and decreases markers oxidative stress in AD model mice^48,76. Another group showed that inhibition of Cbx7 and subsequent activation of Nrf2 signaling improved cognitive deficits in a model of ischemia⁷⁷. Additionally, several studies in humans have linked an increase in oxidative stress to apathy, though these were not AD-related^78–81. CD47 is a ubiquitously expressed protein with a variety of known functions including cell adhesion, phagocytosis, cell migration and proliferation, and apoptosis among other immune functions⁸². Interaction of CD47 with signal regulatory protein (SIRP) is crucial for neuronal communication⁸³. In AD, CD47 may have a protective role by binding Aβ and mitigating some of its toxic effects through nitric oxide signaling⁸⁴. In terms of phagocytic activity, CD47 acts as a “don’t eat me” signal for microglia and mast cells^85,86. Ding et al. also found that CD47 was decreased in AD brains and that a decrease in CD47 can result in excessive synaptic pruning in AD⁸⁵. Another study found that CD47 expression decreases with age and is associated with cognitive impairment in aged rhesus macaques⁸⁷. Interestingly, CD47 has been associated with autism spectrum disorder and one study found that neuronal overexpression of CD47 improved spatial memory in MWM and increased Marble burying behavior⁸⁸.

Gpr26 is an orphaned G protein coupled receptor whose natural ligand is still unknown⁸⁹. Importantly, decreased Gpr26 has been associated with an increase in anxiety-like behavior in OF and elevated plus maze, as well as increased despair-like behavior in the forced swim and tail suspension tests⁹⁰. Interestingly, Gpr26 deficiency did not affect spatial memory in MWM⁹⁰.

Several of these rescued proteins seem to converge on the wingless-related integration site (Wnt)/βcatenin signaling pathway. This is notable, as Wnt/β-catenin signaling has been implicated in AD pathogenesis as well as neuropsychiatric disorders^91,92. Briefly, canonical Wnt signaling involves binding of Wnt ligands, like Wnt3a, to Frizzled and Low-density lipoprotein receptor-related protein 5/6 (LRP5/6) receptors which inhibit the destruction complex that phosphorylates and promotes the degradation of β-catenin. This allows for the translocation of β-catenin to the nucleus where it can activate Wnt target genes like cellular myelocytomatosis oncogene (c-Myc) and Cyclin D1. The destruction complex is composed of Axin, adenomatous polyposis coli (APC), glycogen synthase kinase 3 β (GSK3β), and Casein Kinase 1 (CK1)⁹¹. Wnt signaling promotes synaptogenesis, dendritic spine formation, synaptic plasticity, neuronal survival, and neurogenesis^93–96.

Cbx7 increases transcription of dickkopf-1 (DKK1), an inhibitor of Wnt signaling, decreases nuclear localization of β-catenin, and decreases expression of the downstream target ZEB1. DKK1 is increased in AD brains and plays a role in neuronal apaoptosis and synapse loss, promotes activation of the tau kinase GSK3β^97,98. CD47 expression is upregulated by c-Myc⁹⁹. Vacuolar H+-adenosine triphosphatase subunit V0C (ATP6V0C) drives proton transport and organelle acidification (i.e. lysosomes) which is necessary for Wnt signaling¹⁰⁰. Shisa proteins negatively regulate Wnt signaling by interacting with immature form of the Frizzled receptor and prevents its transport to the cell surface^101,102.

In summary, reduction of eEF2K in APP mice may affect multiple processes related to AD pathogenesis such as response to oxidative stress, the endolysosomal pathway, and phagocytic synaptic pruning. Future studies are warranted to understand the potential roles of these proteins in AD pathogenesis.

The current standard of care for the treatment of apathy in AD patients is methylphenidate (Ritalin). However, in a clinical trial assessing the effect of methylphenidate in AD patients, this agent had no effect on cognition or patient quality of life¹⁰³. In our study, we were able to show that suppression of eEF2K improved apathy-like behavior as well as learning and memory. Importantly, eEF2K and its downstream target eEF2 have a 1:1 relationship¹⁰⁴. eEF2K also belongs to a small group of atypical α kinases whose catalytic domains do not share homology with other conventional kinases¹⁰⁴. This allows for specific inhibition and greatly limits the possibility of off-target effects. Indeed, there are several eEF2K inhibitors currently available²⁵. This makes eEF2K a promising therapeutic target for the treatment of AD with potential improvements in cognition and apathy that could greatly improve patient outcomes and caregiver burden.

Supplementary Material

Supplementary Files

This is a list of supplementary files associated with this preprint. Click to download.

• SupplementalFiguresandtables.docx