MEK/mTOR-dependent D1 dopamine receptor activation induces local protein synthesis via eEF2 dephosphorylation in neurons

Neuromodulators in general, and dopamine in particular, define brain and neuronal states in different ways including regulation of global and local mRNA translation. Yet, the signaling pathways underlying the effects of dopamine on mRNA translation are not clear. Here, using genetic, pharmacologic, biochemical, and imaging methods, we tested the hypothesis that dopamine regulates phosphorylation of the eukaryotic elongation factor 2 (eEF2). We found that activation of dopamine receptor D1 but not D2 leads to rapid dephosphorylation of eEF2 at Thr56 in cortical primary neuronal culture and in vivo in a time-dependent manner. Additionally, NMDA receptor, mTOR, and ERK pathways are upstream to the D1 receptor-dependent eEF2 dephosphorylation and essential for it. Furthermore, D1 receptor activation resulted in a major reduction in dendritic eEF2 phosphorylation levels together with a correlative increase in local mRNA translation. These results reveal the role of eEF2 in dopamine regulation of local mRNA translation in neurons. One-sentence summary D1 receptor activation increases protein synthesis in dendrites by inactivating eEF2K in an ERK2/mTOR-dependent manner.


Introduction
Dopamine is an important neuromodulator involved in motivation, addiction, and learning and memory, which modulates neuronal activity and synaptic plasticity in different brain areas (Chafee andGoldman-Rakic, 1998, Waelti et al., 2001). It exerts its complex effect via regulation of cellular and molecular networks. Modulation of long-term memory by dopamine requires de novo local protein synthesis (Schicknick et al., 2008), a process controlled by translation factors.
The translation factor elongation factor 2 (eEF2) and its kinase (eEF2K)play a central role in the elongation phase of mRNA translation. eEF2K phosphorylates eEF2 on Thr 56 and thereby inactivates it, leading to reduction in the rate of mRNA translation. Since eEF2K activity is regulated by Ca 2+ /calmodulin, elevation in intracellular calcium by synaptic activation receptors such as NMDA receptor or G-coupled receptors (e.g. metabotropic glutamate receptors) results in induced neuronal activity-dependent phosphorylation of eEF2 (Proud, 2015, Barrera et al., 2008, Heise et al., 2017, Im et al., 2009, Sutton et al., 2007. Furthermore, general translation is reduced in dendrites due to eEF2K activity, while certain synaptic proteins are selectively translated at the synapse (Heise et al., 2014). eEF2K can be deactivated by other kinases such as p70S6 kinase (S6K) or p90 ribosomal kinase (p90 RSK), which are activated in response to changes in mammalian target of rapamycin-signaling (mTOR) or extracellular-regulated kinase (ERK) signaling respectively. These kinases inactivate eEF2K by phosphorylation on its Ser366 residue (Browne et al., 2004, Redpath et al., 1993, Wang et al., 2001. Regulation of protein synthesis by neurotransmitters, especially at the phase of elongation, has been shown as a mechanism to explain the effects of antidepressants. For example, ketamine, a NMDA receptor non-competitive antagonist, has a fast, potent, and long-lasting antidepressant effect mediated by the reduction in eEF2 phosphorylation via eEF2K (Adaikkan et al., 2018, Duman andVoleti, 2012). Interestingly, drugs targeting the function of dopamine receptors and their interaction with NMDA receptors have become an important field of research to develop better antidepressant medications (Miller et al., 2014, Raab-Graham andNiere, 2017). However, the mechanism by which dopamine receptors regulate mRNA translation during the elongation phase remains poorly understood.
In this work, we found that D1 but not D2 receptor activation increases local protein synthesis in dendrites by eEF2 dephosphorylation. This increase in local protein synthesis in cortical neurons is mediated by the D1 receptor and requires NMDA receptor-dependent activation of the MEK/mTOR signaling pathways that lead to inactivation of eEF2K by phosphorylation on Ser 366 residue resulting in eEF2 dephosphorylation. Furthermore, using eEF2K knockout mice, we show that eEF2K/eEF2 is the main pathway for D1dependent increase in local protein synthesis.
Following the results we obtained in primary cultures, we further examined whether D1 activation induces dephosphorylation of eEF2 in-vivo as well.
Western blot analyses of both cortex and hippocampus extracts from SKF38393-injected C57BL/6J mice (i.p. injection) showed reduction in eEF2 phosphorylation after 15 minutes ( Figure 1C and figure 1D), similar to the data obtained in primary cultures ( Figure 1A). In agreement with previous results, ERK2 activation was increased in both hippocampus and cortex of these SKF38393-treated mice compared to controls (Figure 1 supplement 1 C and D). These results suggest that dopamine regulates eEF2 phosphorylation both in vivo and in vitro via D1 but not D2 receptor.

NMDA receptor
Following the correlation between ERK2 activation as indicated by its phosphorylation state and eEF2 dephosphorylation (Figure 1 supplement 1 A and B), and since ERK is a coincidence detector of dopamine and the NMDA receptor (Kaphzan 2006, David 2014, we further asked whether the NMDA receptor has a role in D1 receptor-dependent dephosphorylation of eEF2. To test this, primary cortical neuronal cultures were pretreated for 30 min with APV, a NMDA antagonist, followed by 15 or 60 min treatment of D1 receptor agonist SKF38393. Treatment of the cells with APV prevented the dephosphorylation of eEF2 following SKF38393 treatment, both after 15 and 60 min (Figure 2A).
Since eEF2K inhibits eEF2 activity, and is negatively regulated by phosphorylation on Ser 366 , we further asked whether treatment with SKF38393 increases eEF2K phosphorylation on Ser 366 at the same time points when we observed eEF2 dephosphorylation. Indeed, treatment of primary cultures with SKF38393 clearly induced phosphorylation of eEF2K on Ser 366 , which was blocked by pre-incubation with NMDA receptor antagonist, APV ( Figure 2B). As expected, APV also abolished ERK1/2 activation as shown in previous studies (Kaphzan et al., 2006;David et al., 2014) (Figure 2 supplement 1A). Furthermore, eEF2 dephosphorylation showed negative correlation with eEF2K phosphorylation (r=0.76, p<0.05, Pearson's correlation) (Figure 2 C). These results suggest that NMDA receptor activation plays a role in D1 receptor-dependent regulation of eEF2 and eEF2K activities.

Dopamine D1 receptor activation induces eEF2 dephosphorylation in an ERK-and mTOR-dependent manner
Since both MEK-ERK and mTOR can lead to cross-activation and pathway convergence on substrates, as in the case of eEF2K phosphorylation (Redpath and Proud, 1993a;Wang et al., 2001;Browne et al., 2004;Browne and Proud, 2004;Lenz and Avruch, 2005), we sought to differentiate between these pathways, and test whether one of them is more dominant in the inhibition of eEF2K and inducing D1 receptor-dependent eEF2 dephosphorylation in neurons. Pre-incubation of cortical neurons with U0126 compound, a MEK inhibitor, blocked the phosphorylation of eEF2K at Ser 366 , inhibited eEF2 dephosphorylation ( Figure 3A), and reduced dopamine D1 receptor-dependent activation of ERK1/2 and the phosphorylation of S6K at Thr 389 ( Figure 3C). Interestingly, incubation of the cells with rapamycin alone resulted in enhancement of eEF2 phosphorylation and reduction in S6K phosphorylation at the Thr 389 residue ( Figure 4A and 4B) without affecting ERK1/2 phosphorylation ( Figure 4B). Pre-incubation with U0126 markedly reduced the phosphorylation of ERK1/2 and S6K ( Figure 3B). These findings imply that the MEK-ERK pathway is upstream to the mTOR pathway following D1 receptor stimulation.

Dopamine D1 receptor dephosphorylates eEF2 Thr56 predominantly in dendrites
Regulation of eEF2K activity and reduction of eEF2 phosphorylation by synaptic receptors such as the NMDA receptor serves as one possible way of mRNA translation in neuronal dendrites (Autry et al., 2011). To examine whether dopaminergic transmission can also regulate dendritic eEF2 phosphorylation, we performed immunocytochemical analysis of cortical neurons from wild-type mice treated with SKF38393 for 15 minutes. The results revealed a reduction in phospho-eEF2 immunoreactivity in most of the neurons analyzed. The reduction was measured in both neuronal soma show that D1 receptor activation reduces eEF2 phosphorylation in dendrites, suggesting its potential role in local protein synthesis.

Dopamine D1 receptor activation enhances total protein synthesis in cultured cortical neurons
In light of our immunocytochemistry results, we further asked whether the dopamine D1 receptor-dependent eEF2 dephosphorylation coincides with enhanced protein synthesis. For this purpose, we utilized SUnSET, a nonradioactive method of monitoring global protein synthesis in cultured cells that uses puromycin to tag nascent proteins (Schmidt et al., 2009). Timecourse experiments showed that puromycin incorporation was significantly increased after 1.5 and 4 hours of SKF38393 treatment ( Figure 6A). In addition to measuring total levels of mRNA translation, we measured expression levels of BDNF and synapsin 2B proteins, which are known to be induced by D1 receptor and affected by eEF2 activity (Chong et al., 2006, Williams andUndieh, 2009). Prolonged stimulation of the cells with SKF38393 resulted in increased levels of BDNF and synapsin 2B after 4 hours of incubation ( Figure 6B and 6C). Treatment of cortical neurons with SKF38393 did not change 4E-BP phosphorylation. However, rapamycin treatment did, as expected ( Figure 6 supplement 1 A and B), suggesting that the D1 receptor-dependent increase in protein synthesis is mediated by eEF2 dephosphorylation.
In addition to its negative regulators (MEK-ERK and mTOR), eEF2K has also a positive regulator which is downstream to the D1 receptor, AMP kinase (AMPK). Examining the phosphorylation of AMPK at its activation site, Thr 172 , revealed no difference in the phosphorylation state at the same time points when D1 receptor inhibits eEF2K at Ser 366 and leads to eEF2 dephosphorylation (15 and 60 min). This suggests that MEK-ERK and mTOR are indeed the dominant pathways for D1 receptor-dependent eEF2 dephosphorylation and increased protein synthesis.

D1 receptor activation increases protein synthesis in neurons by eEF2K inactivation
To test directly if D1 receptor activation increases protein synthesis in an eEF2K/eEF2 pathway-dependent manner, we used primary cultures from eEF2K-KO mice. These mice show no eEF2 phosphorylation, but display normal phosphorylation of other translation factors (Adaikkan et al., 2018).
Cortical cultures from eEF2K-KO mice and their wild-type littermates were treated for 4 hours with SKF38393 and global protein synthesis was analyzed by SUnSET. Immunocytochemistry and western blot analyses of puromycin incorporation showed a significant increase in wild type but not in eEF2K-KO cultures following SKF38393 incubation ( Figure. 7A and B, Figure 7 supplement 1 A-C). To probe whether the D1 receptor activation-induced increase in translation is dependent on ERK/mTOR/eEF2K/eEF2 signaling, we pre-treated cortical neurons derived from wild type and eEF2-KO mice with or without MEK inhibitor U0126, followed by SKF38393 treatment. We found that the SKF38393-mediated increase in translation was reduced by treatment with U0126 in wild type neurons, while no change in puromycin incorporation was detected in eEF2K-KO mouse-derived cultures treated with U0126. These results suggest that the MEK/ERK and eEF2 pathways mediate the D1 receptor-dependent increase in protein synthesis in cortical neurons.

Discussion
In summary, our data show that dopamine D1 receptor activation modulates eEF2 phosphorylation and increases protein synthesis in primary cortical neurons. We found that NMDA receptor is required for the D1 receptordependent induction of MEK/mTOR activity that leads to inactivation of eEF2K. The reduction in eEF2K activity is sufficient to induce a marked eEF2 dephosphorylation in neurons, especially in dendrites, and increased global mRNA translation.
These data support the view that dopamine D1 receptor activation influences protein synthesis, an important component of memory and synaptic plasticity formation (Andre and Manahan-Vaughan, 2015, Fenu et al., 2001, Hikind and Maroun, 2008, Nader and LeDoux, 1999, Roffman et al., 2016, Schicknick et al., 2012. In addition, these data provide insight into the kinetics and mechanism of the signaling pathways involved in regulation of eEF2K activity in neurons. Our data provide further decoding of the well-known relationship between D1 and NMDA receptors and its effect on signal transduction (David et al., 2014, Dunah and Standaert, 2001, Lee et al., 2002, Martina and Bergeron, 2008, Pei et al., 2004, Stramiello and Wagner, 2008. Finally, our results indicate that, in addition to the canonical calcium-calmodulin-dependent activation of eEF2K following NMDA receptor stimulation, NMDAR-D1 partnership induces translational changes via ERK and S6K signaling cascades. This includes eEF2 dephosphorylation in neurons, in the soma and especially in dendrites, which is likely to contribute to the observed increase in polypeptide elongation and expression of BDNF and synapsin 2B (Adaikkan et al., 2018, Autry et al., 2011, Belelovsky et al., 2009, Gildish et al., 2012, Im et al., 2009, Ma et al., 2014, Park et al., 2008, Taha et al., 2013. The data obtained using cortical cultures derived from eEF2K-KO mice led to the notion that the eEF2K pathway accounts for the increase in protein synthesis following dopamine D1 receptor activation. eEF2K has been identified as a biochemical sensor that is tuned to the pattern of neuronal stimulation (Heise et al., 2014, McCamphill et al., 2015, Sutton et al., 2007. Perturbation of intracellular calcium concentration by NMDA and neuronal activity produces eEF2K/eEF2-dependent changes of dendritic proteome (Ehlers, 2003, Lazarevic et al., 2011, Perez-Otano and Ehlers, 2005, Turrigiano, 2008, Turrigiano and Nelson, 2004, Virmani et al., 2006. Although eEF2K activity is controlled principally by Ca +2 -Calmodulin complex, it is also regulated by phosphorylation. mTOR-and MEK-dependent phosphorylation of eEF2K reduces its activity, while PKA and AMPK mediated phosphorylation does the opposite (Heise et al., 2014, Taha et al., 2013, Wang et al., 2001. Given this complex regulation of its function, it is tempting to speculate that eEF2K works as a signaling hub, linking synaptic activity to protein synthesis. Conversely, eEF2 dephosphorylation occurs by spaced 5-HT activity and the PKA stimulation (McCamphill et al., 2015). eEF2K activity can be regulated by the type of neurotransmission in which eEF2K acts as a biochemical sensor to discriminate between evoked action potential and spontaneous miniature synaptic transmission (Sutton 2007).
In a similar manner, optimal levels of dopamine D1 receptor stimulation in dendrites work to gate out "noise", while high levels, e.g., during stress, suppress delay firing (Arnsten et al., 2015). For instance, maintenance of synaptic strength in hippocampal slices treated with low concentrations of dopamine D1 agonist requires MEK and CaMKII activation, while in slices treated with high concentrations maintenance of synaptic strength is dependent only on MEK activation (Barcomb et al., 2016). Therefore, it is not surprising that dopamine D1 receptor agonist serves potentially as an antidepressant (D'Aquila et al., 1994). This effect may be due to the ability of dopamine to regulate neuronal function via the eEF2K pathway.
On the other hand, mRNA translation following synaptic activation can also be regulated by other pathways such as the mTOR/4E-BP pathway (Scheetz et al., 2000). It was shown that, in primary striatal neurons, the antipsychotic drug haloperidol activates translation-related pathways mediated by mTOR, which results in increased phosphorylation of 4E-BP and S6K1 (Bowling et al., 2014). Moreover, our results show that the phosphorylation status of 4E-BP does not change following D1 receptor agonist treatment, but it does affect the eEF2K pathway, which leads to D1 receptor-dependent increased protein synthesis. Recent studies involving manipulation of eEF2K activity have been a field of intense research, ranging from addiction Caron, 2015, Werner et al., 2018) and depression. Indeed, eEF2K serves as a target for the antidepressant ketamine, and has a role also in the progression of neurodegenerative diseases (Adaikkan et al., 2018, Heise et al., 2017, Jan et al., 2018, Kokkinou et al., 2018, Zang et al., 2018.
Our findings establish a connection between dopamine D1 receptor activation and eEF2K, and open a door to better understanding the molecular mechanism underlying the role of neurotransmitters in synaptic plasticity, addiction, and antidepressants. Future studies using neurodegenerative disease animal models, combined with pharmacological and behavioral paradigms will better link the eEF2 pathway and a potential target for therapy.

Material and Methods
Mice eEF2K-KO mice, in which coding exons 7, 8, 9, and 10 of eEF2K were deleted, were generated by the laboratory of Christopher G. Proud. We derived eEF2K wild-type (WT) and KO littermates by crossing heterozygous mice as previously described (Heise et al., 2017), eEF2K mice were bred from colonies maintained at the University of Haifa. C57BL/6 mice were obtained from local vendors (Envigo RMS, Jerusalem, Israel) and after acclimation to the facility were used for experiments. Animals were provided ad libitum with standard food and water and were maintained on a 12/12 h light/dark cycle.

Cortical cell culture
Primary cortical neuronal cultures were isolated from P0 or P1 C57BL/6J or eEF2K-WT or KO mice of either sex as previously described (Ounallah-Saad et al., 2014). Briefly, both hippocampi were removed, and cortical regions were taken. The tissue was chemically dissociated by trypsin and DNase, and mechanically, using a siliconized Pasteur pipette. Cells were plated onto round coverslips coated with 20µg/ml Poly-L-ornithine and 3 µg/ml laminin (Sigma), placed in 6-well plates (300,000 cells per well) or 12-well plates (150,000 cells per well). Culture medium consisted of MEM (Gibco), 25 µg/ml insulin (Sigma), 27.8 mM glucose (Sigma), 2 mM l-glutamine (Sigma), and 10% horse serum (Biological Industries, Israel). Cultures were maintained at 37°C in a 95% air/5% CO 2 humidified incubator. Half the volume of the culture medium was replaced at days 8 and 11 with feeding medium containing glutamine 2mM, insulin 25 µg/ml, and 2% B-27 supplement (Gibco).

In vivo experiments: Injection of mice with SKF38393
Adult male C57BL/6 mice were injected i.p. with either 0.5mg/Kg SKF38393 or vehicle (0.9% NaCl saline) and returned to their cage. The animals were sacrificed 15 minutes later by cervical dislocation, the prefrontal cortex and the hippocampus were collected on ice, and the tissue was immediately frozen in liquid nitrogen. The tissue was stored at -80°C until further use.

Pharmacological treatments
After 14

Immunocytochemistry quantification
Signal intensity quantification of phosphorylated eEF2 (peEF2) in the cell soma or dendrites was done by NIS Element Advanced Research (Ar) 4.5 (Nikon Japan) software in MAP2 labeled neurons. Confocal images were acquired using a Nikon 63X immersion oil objective at a resolution of 1024x1024 pixels. Each image was a Z series projection of 7 to 10 images, taken at depth intervals of 0.5 µm. To define the region of interest for quantification, cell bodies and dendrites close to the cell body were manually traced using NIS Element AR software on the MAP2 channel. peEF2 signal intensity (mean pixel intensity) was estimated as the peEF2 integrated fluorescence intensity divided by the area marked by the MAP2 signal.

Puromycin immunocytochemistry and quantification
Cells for immunofluorescence were plated on coverslips coated with Poly-Lornithine and laminin coating. After 14 DIV cells were pre-treated with U0126 (20 µM) or vehicle (DMSO) for 30 minutes and then treated with SKF38393 (25 µM) for 4 hours. After SKF38393 treatment, cells were further incubated with 10µg/ml puromycin for 10 min in the same medium and fixed in 4% paraformaldehyde. Cells were stained with anti-puromycin (1:1000) and MAP2 (1:1000) antibodies, following the same procedure for peEF2Thr 56 immunocytochemistry. Images were taken as Z series projection of 5 to 9 images at depth intervals of 0.25 µm at x60 magnification with an Olympus IX81 microscope using Olympus cellSens1.16 software. Quantification of puromycin incorporation was done by selecting neurons randomly in MAP2 labeled neurons and estimating the puromycin signal as mean intensity divided by the area marked by MAP2 signal using ImageJ 1.51J software. Quantification was done in a blind manner based on three independent experiments for each condition.

SUnSET
Protein synthesis was measured by the SUnSET method. Cortical neurons were isolated and maintained for 14 days in culture, as described above.

Western blotting
Samples in SDS sample buffer were subjected to SDS-PAGE (7.5-10%) and Western blot analysis. Lanes were loaded with an equal amount of protein. (1:10,000) (Jackson ImmunoResearch). Immunodetection was accomplished with the Enhanced Chemiluminescence EZ ECL kit (Biological Industries).

Quantification of immunoblots was performed with a CCD camera and
Quantity One 4.6 software (Bio-Rad). Each sample was measured relative to the background. Phosphorylation levels were calculated as the ratio of phosphorylated protein and a total amount of protein.

Statistical analysis
Graphs were prepared using GraphPad Prism 6.01, InStat Software (GraphPad Software, CA, USA). Data are expressed as mean ± SEM.
Statistical analysis was performed using SPSS version 24. Statistical significance was determined with one-way ANOVA followed by Tukey's post hoc test. Student's t-test or Mann-Whitney were used to examine the differences between groups.

Figure 1 supplement 1. D1 receptor activation induces ERK1/2 phosphorylation both in vitro and in vivo and correlates with eEF2 dephosphorylation. (A)
Representative blots and quantification of the ratios of phosphorylated to total ERK1/2 in cortical primary cultures incubated with D1 receptor agonist SKF38393 (25µM) for the indicated time points (from five independent cultures).           Figure 3B.