NMDAR mediated dynamic changes in m6A inversely correlates with neuronal translation

Epitranscriptome modifications are crucial in translation regulation and essential for maintaining cellular homeostasis. N6 methyladenosine (m6A) is one of the most abundant and well-conserved epitranscriptome modifications, which is known to play a pivotal role in diverse aspects of neuronal functions. However, the role of m6A modifications with respect to activity-mediated translation regulation and synaptic plasticity has not been studied. Here, we investigated the role of m6A modification in response to NMDAR stimulation. We have consistently observed that 5 min NMDAR stimulation causes an increase in eEF2 phosphorylation. Correspondingly, NMDAR stimulation caused a significant increase in the m6A signal at 5 min time point, correlating with the global translation inhibition. The NMDAR induced increase in the m6A signal is accompanied by the redistribution of the m6A marked RNAs from translating to the non-translating pool of ribosomes. The increased m6A levels are well correlated with the reduced FTO levels observed on NMDAR stimulation. Additionally, we show that inhibition of FTO prevents NMDAR mediated changes in m6A levels. Overall, our results establish RNA-based molecular readout which corelates with the NMDAR-dependent translation regulation which helps in understanding changes in protein synthesis.

m 6 A is a widespread, abundant and well conserved internal RNA modification, which plays a pivotal role in regulating multitude of physiological and pathological processes. It is present on diverse classes of RNA molecules such as rRNAs, mRNAs, tRNAs, snRNAs, miRNAs, and long non-coding RNAs. In mammalian mRNA, m 6 A is primarily located near the stop codon, 3' UTR, and long internal exons 1 . It is known to critically regulate several aspects of RNA metabolism, such as RNA stability, splicing, translation, translocation, localization, degradation, and transport [2][3][4] . This modification is highly enriched in the brain and global levels are developmentally regulated 5 .
In mammals, the m 6 A mark on RNA is dynamically and reversibly regulated by the action of RNA methyltransferases (writers) and demethylases (erasers) 6 . Methyl groups are added co-transcriptionally onto the adenosine nucleotide through a multicomponent methyltransferase complex. This complex consists of a catalytic core protein Methyltransferase-like 3 (METTL3), along with the adaptor proteins Methyltransferase-like 14 (METTL14) and Wilms tumour 1 associated protein (WTAP) 6 . This m 6 A mark can be reversed by the action of m 6 A demethylases, Fat mass and obesity-associated (FTO) 7 and alkB homolog 5 (ALKBH5) 8 . These demethylases primarily differ in their substrate preference and localisation. Alkbh5 is localised in the nucleus and demethylates m 6 A on DRACH motif. Whereas, FTO is localised both in the cytosol and the nucleus and acts on broad spectrum substrates 8 . Some studies have defined the role of FTO in neuron present in cytosol and dendrites, where it regulates local translation of mRNA 9,10 . Thus, these studies provide basis for the FTO to study the role of activity mediated translation response. The m 6 A mark is recognised by a set of the proteins known as readers, which preferentially binds to the methylated RNA and mediate the downstream effector functions of m 6 A modification 11 . The YTH domain-containing proteins YTHDF1-3, YTHDC1-2 are the most common m 6 A readers in mammals, and are proposed to play a role in determining the stability and translational efficiency of the bound transcripts [12][13][14] .
Several recent studies have uncovered the role of m 6 A modification in the nervous system. The m 6 A modified transcripts are highly enriched in the brain 5,15 , and their levels are developmentally regulated 16,17 . Studies involving knockdown and overexpression of m 6 A writers and erasers have demonstrated the role of m 6 A in the process of neurogenesis, axon morphogenesis, synapse morphogenesis and synaptic plasticity 18,19 . Furthermore, studies have also shown the preferential synaptic localization of m 6  www.nature.com/scientificreports/ modulation of m 6 A levels 3 . The dynamic and reversible nature of m 6 A modifications makes it an excellent candidate for regulation protein synthesis upon synaptic activity. Surprisingly, despite the recent surge in studies focusing on synaptic m 6 A enrichment and experience dependent m 6 A modulation, the dynamic regulation of m 6 A in response to synaptic activity, and its relation to translation remains unexplored 20 . In our previous studies, we have shown the importance of translation regulation 21,22 and its kinetics 23 on synaptic stimulation. In this study, we probed the temporal dynamics of m 6 A in response to NMDAR stimulation. Briefly, we observe a robust increase in the total m 6 A levels, temporally coinciding with the global translation inhibition phase of NMDAR stimulation in cultured cortical neurons. This is facilitated by the reduction in the somato-dendritic expression of the m 6 A demethylase FTO along with the redistribution of m 6 A marked transcripts in the non-translating fractions of the polysome profile.

Results and discussion
NMDAR stimulation leads to increase in m 6 A levels on RNA which correlates with the global translation inhibition. In this study, we have used a well-established NMDAR stimulation paradigm to understand the change in m 6 A levels on RNA in response to synaptic activity in neurons 24 . NMDAR is known to elicit a dynamic translational response in neurons, involving rapid and robust inhibition of global translation for a short term period (1-5 min), followed by the activation of global translation at delayed period (20 min) 21,23,24 . At first, we validated our NMDAR stimulation paradigm by measuring the translation response in cultured rat cortical neurons upon NMDA treatment for 1, 5, and 20 min. We used the phosphorylation status of eEF2 to measure global translation response downstream of NMDAR stimulation. We measured the levels of p-eEF2 in cortical neurons using immunostaining analysis after 1, 5, and 20 min of treatment with 20 µM NMDA. We observed a robust, threefold increase in p-eEF2 levels upon 1-min NMDAR stimulation (Fig. 1A,B). By 5 min, the p-eEF2 levels had reduced compared to 1 min time point; however, it remained significantly higher compared to unstimulated condition (Fig. 1A,B). By 20 min, there was a maximum reduction in p-eEF2 levels, bringing it lower than the untreated condition ( Fig. 1A,B). In addition, we have quantified the phosphorylation of eEF2 through western blotting and we observed a significant increase in p-eEF2 levels at 1 min NMDAR stimulation ( Supplementary Figs. 1A and B). These results are in accordance with the previous reports 23, 24 which show a rapid translation inhibition followed by a delayed translation activation upon NMDAR stimulation. Next, to understand the NMDAR mediated changes in m 6 A kinetics, we used the RNA dot blot method to measure the total m 6 A levels upon 1, 5 and 20 min of NMDA treatment of cultured cortical neurons. Brief methodology and steps involved in RNA dot blotting are indicated in Fig. 1C. Total RNA samples extracted from the DIV15 cultured neurons treated with NMDA (1,5 and 20 min) are indicated in the dot blot (Fig. 1D). We have also shown that the observed m 6 A signal is majorly contributed by RNA and not by DNA ( Supplementary  Fig. 1C). Blots were probed for m 6 A and methylene blue (colorimetric read out) which was used as a loading control (Fig. 1D). We observed a significant increase in total m 6 A levels on 5-min and 20 min of NMDAR stimulation, while the 1-min time point did not show any significant change as compared to the untreated condition (Fig. 1D,E). These results suggest that, at 5 min NMDAR stimulation, the increased m 6 A levels correlate with the translation inhibition (increased phosphorylation of eEF2). Hence, we propose that measured m 6 A levels are inversely correlative to NMDAR translation response at the 5 min time point. Interestingly, there is a time delay in the peaking of m 6 A levels (peaks at 5 min) as compared to eEF2 phosphorylation (peaks at 1 min). This indicates that the change in m 6 A levels is likely to be downstream of the kinase activation upon NMDAR stimulation. Further, measured m 6 A levels at 20 min shows a decreasing trend compared to 5 min NMDAR stimulation, but remains significantly high in comparison to basal condition. These results suggest that the m 6 A mark on RNA can be used as a potential marker to understand the temporal profile of NMDAR-dependent translation response.
We further validated our dot-blot results by investigating the NMDAR mediated changes in neuronal m 6 A levels using immunostaining analysis. Similar to our previous experiment, we stimulated DIV15 cultured cortical www.nature.com/scientificreports/ neurons with 20 µM NMDA for 1, 5 and 20 min time points followed by immunostaining with m 6 A and MAP2 antibodies. In accordance with our previous results from dot-blot analysis, our image quantification revealed that m 6 A levels were not altered at 1 min, but significantly increased at 5 min and 20 min time points of NMDA treatment (Fig. 1F,G).To validate the total neuronal m 6 A measurements in synaptic compartments, we used synaptoneurosomal preparations. Cortical synaptoneurosomes were prepared from P30 rats and stimulated them with 40 µM NMDA for 1, 5, and 20 min and investigated the changes in m 6 A levels using dot-blot analysis. The synaptoneurosomal preparation was validated by the enrichment of post-synaptic protein PSD95 using immunoblotting analysis (Fig. 1H). We used dot-blot to understand the m 6 A methylation pattern in synaptoneurosomes. Quantification of m 6 A dot blots from synaptoneurosome indicated that m 6 A levels increased upon NMDA treatment, peaking at 5 min time point and showed a significant reduction at 20 min compared to 5 min (Fig. 1I,J). This is consistent with our previous observations from cultured cortical neurons (Fig. 1B,E). Overall, it suggests that NMDAR mediated changes in the m 6 A signal is dynamic and inversely-correlated to translation response (initial increase and later decrease of m 6 A coincides with the initial translation inhibition followed by translation activation), both in the whole neuron and in the synaptic compartments.

NMDAR stimulation induces changes in nuclear and cytosolic levels of m 6 A demethylase FTO.
In the previous section, we showed the distinct temporal profile of m 6 A upon NMDAR stimulation.
Next, we wanted to explore the possible mechanism behind NMDAR mediated changes in m 6 A levels. The dynamic changes in m 6 A levels is primarily determined by the action of designated methyltransferases and demethylases 6 . The m 6 A methyltransferases primarily act in the nucleus in a co-transcriptional manner 25 , while the m 6 A demethylases are known to function in both nuclear and cytoplasmic compartments 9,26,27 . Since we observed the NMDAR induced changes in the m 6 A levels in the cell body as well as the synapto-dendritic compartments, we hypothesized that m 6 A demethylases are the primary determinant of NMDA-induced changes in m 6 A levels. Among the two widely known m 6 A demethylases, ALKBH5 and FTO, ALKBH5 is known to primarily localize in the neuronal nucleus and its levels are low in the adult brain 26 . On the other hand, FTO is widely studied in neurons and is shown to be expressed in the nucleus, dendrites and dendritic spines of CA1 pyramidal neurons 9 . Hence, we speculated that FTO is the primary driver of mediating NMDAR induced changes in m 6 A levels. To test this, we performed immunostaining to determine the nuclear and cyto-dendritic changes in the FTO levels on 1, 5, and 20 min treatment with 20 µM NMDA. In accordance to the previous studies 9, 28 , we observed that FTO was primarily localized in neuronal nucleus ( Fig. 2A). Notably, we were also able to detect the FTO staining in the cytosolic and dendritic compartments ( Fig. 2A). As we find high levels of FTO in the nucleus in comparison to the cyto-dendritic compartment, we imaged nuclear and cyto-dendritic FTO under different imaging parameters (Supplementary Fig. 2A and B). The cyto-dendritic quantification revealed that the levels of FTO significantly decreased upon 1 min NMDAR stimulation and recovered to the basal levels by 5 min time point (Fig. 2B). We did not observe any significant difference in the FTO levels between untreated and 20 min NMDA treated neurons (Fig. 2B). When a similar analysis was done for the nuclear FTO levels, we again found a significant reduction in the FTO levels in 1 min NMDA treated neurons in comparison to the untreated neurons (Fig. 2C,D). However, in contrast to the cyto-dendritic FTO levels, the nuclear FTO levels remained significantly low even after 5 min of NMDA treatment. On 20 min of NMDA treatment, the FTO levels had recovered and was significantly higher compared to the untreated condition (Fig. 2D). To determine the total FTO levels we performed immunoblot on total protein lysate and we observed that significantly low amount of FTO at the 5 min (Fig. 2E,F). Thus, we observed a temporal delay between NMDAR induced reduction in FTO levels versus NMDAR mediated increase in m 6 A levels. A significant and consistent increase in m 6 A levels was observed by 5 min of NMDAR stimulation, whereas the reduction in the total FTO protein levels were consistently low at NMDA 5 min time point.
To understand the importance of FTO in NMDA mediated changes of m 6 A levels, we used an FTO specific inhibitor Meclofenamic acid 29 (MA) and compared the m 6 A levels at basal and NMDA stimulation conditions (Fig. 2G). Treatment with MA (120 μmolar) for 2 h caused a significant increase in m 6 A levels compared to www.nature.com/scientificreports/ basal condition validating the inhibition of FTO (Fig. 2H). Further, 5-min NMDAR stimulation in the presence of FTO inhibitor MA did cause significant changes in the m 6 A levels as compared to the mock (MA) treatment (Fig. 2I). Finally, as a control, we recaptured the increase in m6A signal upon 5-min NMDAR treatment in the absence of MA (Fig. 2J). Thus, we show that FTO is a critical player mediating the dynamic changes of m 6 A levels upon NMDAR stimulation.
In the FTO imaging analysis, we observed differential dynamics of nuclear versus cyto-dendritic changes in FTO levels upon NMDAR stimulation. We speculate that this is primarily caused by redistribution of FTO between these compartments, along with the differential decay kinetics of nuclear and cyto-dendritic FTO. NMDAR induced changes in FTO levels could be attributed to the transcription, translation as well as degradation pathways. We summarise our findings in the representative graph shown in Fig. 2K. The m 6 A readout follows a similar trend as eEF2 phosphorylation, but with delay in reaching the peak (Fig. 2K). The reduction of the total FTO levels show a good correlation with increase in total m 6 A levels (Fig. 2K). It is likely that the reduction in the FTO levels on 1 min and 5 min of NMDAR stimulation is caused by ubiquitin-mediated degradation, as NMDAR activation is reported to cause widespread degradation at acute time points 23 . Further, NMDA treatment is also known to cause a delayed translation activation response, providing a possible explanation for the increase in FTO levels on 20 min treatment 23,24 .

NMDAR mediated increase in m 6 A levels is accompanied with the shift of m 6 A marked RNA from polysome to non-polysome fractions.
Our previous experiments clearly demonstrated the increase in m 6 A levels upon 5 min of NMDA treatment. From our previous observation, it is known that NMDAR stimulation elicits an overall translation inhibition response at 5 min time point 24 . We wanted to understand if the NMDAR mediated increase in m 6 A levels drives the translation repression of m 6 A marked RNAs. To investigate this, we used polysome profiling technique to monitor the distribution of m 6 A marked RNAs in ribosomal/ polysomal pool versus non-ribosomal pool on NMDA treatment. Briefly, the DIV15 cultured cortical neurons were treated with 20 µM NMDA for 1, 5 and 20 min and the lysates were separated on 15-45% linear sucrose gradient. The steps involved in the sample treatment, polysome profiling and pooling strategy are depicted in Fig. 3A. A representative profile (A254) is shown in Fig. 3B. Additionally, we show the 18S rRNA distribution in the translating and non-translating fractions in the basal condition ( Supplementary Fig. 3A).
As shown in schematic Fig. 3B, we have labelled the pooled samples from F1-6 as non-translating and F7-12 as translating pool, indicated in purple and red colour respectively. An equal volume of each fraction was taken for RNA isolation and subjected to dot blot with m 6 A antibody and methylene blue staining to generate the calorimetric signal (loading control) (Fig. 3C). We observed a uniform staining in methylene blue treated samples, indicating approximately equal loading of RNA on the membrane. We observed a significant increase in m 6 A levels in the non-translating pool upon 5 min NMDA treatment (Fig. 3D). In contrast, quantification of the translating pool of ribosomes showed a decrease in m 6 A signal at 5 min of NMDA treatment (Fig. 3E). Further, quantification of the ratio of m 6 A levels in the translating to non-translating pool showed a significant decrease of the m 6 A levels at 5 min NMDAR stimulation compared to basal condition (Fig. 3F). This decrease in m 6 A levels in the translating fractions suggests that the m 6 A marked pool of ribosomes and mRNAs associated with them have shifted towards the non-translating pool, indicative of translation inhibition. Altogether this data suggests that the NMDAR induced increase in overall m 6 A signal is accompanied by the redistribution of m 6 A marked RNA from translating to non-translating fractions. This shift in signal from translating to non-translating pool could potentially be contributed by both rRNA and mRNA and we are yet to identify the factors driving this shift. In contrast, we observe that the m 6 A levels does not alter significantly at 1 and 20 min of NMDA treatment. Further, since the m 6 A mark on these RNAs could be potentially removed by demethylases, these RNAs could shift back to the translating pool. Another interesting possibility is that the pool of RNAs present in m 6 A marked inhibitory complex could be targeted for enzymatic degradation 30,31 . Thus, from our observations, we conclude that m 6 A signal dynamically distributed across the polysome fractions upon NMDAR stimulation.
Apart from mRNA m 6 A modification, there are reports suggesting m 6 A role in non-coding RNA modification and known to affect gene expression 32,33 . Non-coding RNA like microRNA, tRNA, rRNA and lncRNA are m 6 A methylated and their changes are implied in diseases such as cancer 33,34 . In case of microRNA, presence of m 6 A is known to reduce the duplex stability between the 3'UTR and miRNA seed region 33,35 . Other prime example is from the rRNA, where 18S and 28S rRNA carry one m 6 A mark each which is shown to regulate protein synthesis 36 . While our interpretation is mainly focused on m 6 A modifications on mRNA, we cannot rule out the changes in m 6 A mark on other RNAs contributing to our results.
Overall, we have shown that m 6 A levels change dynamically upon NMDAR stimulation. At 5 min NMDAR stimulation, we observe an increase in m 6 A levels which is correlated with a phase of translation inhibition. Correspondingly, m 6 A signal is also shifted from the translating fractions towards the non-translating pool at 5 min NMDAR stimulation; further supporting the correlation with translation inhibition. Interestingly, the levels of m 6 A demethylase FTO is decreased at 5 min NMDAR stimulation which is responsible for the increase in m 6 A levels. Inhibition of FTO prevents the dynamic changes of m 6 A levels upon NMDAR stimulation indicating that FTO is a key player in this regulation.

Methods
Ethics statement. The study was carried out in compliance with the ARRIVE guidelines. We performed all the animal work in accordance to the guidelines approved by the Institutional Animal Ethics Committee (IAEC) and the Institutional Biosafety Committee (IBSC), InStem, Bangalore, India. All experiments were performed with a minimum of three independent neuronal cultures. All our experiments were performed with cultured neurons and synaptoneurosomes preparation derived from Sprague Dawley (SD) rats. Rat colonies were main- www.nature.com/scientificreports/  For the all the replicative immunoblots of FTO and eEF2 blots were cut at their corresponding molecular weight makers and probed later with respective antibodies. This method of blot imaging was done for simultaneous imaging of blots and to reduce the variance in the assay.
Polysome profiling. The DIV15 rat cortical neurons were stimulated with NMDA (20 μM) for 1, 5 and 20 min. After treatment cells were lysed in lysis buffer (20 mM Tris-HCl, 100 mM KCl, 5 mM MgCl 2 , 1% NP40, 1 mM dTT, 1X Protease inhibitor cocktail, RNAse inhibitor, 0.1 mg/mL cylohexamide and 1X Phosphotase inhibitor) and centrifuged at 20 K RCF at 4 °C for 20 min. The supernatant was loaded onto 15-45% linear sucrose gradient prepared in buffer (20 mM Tris-HCl, 100 mM KCl, 5 mM MgCl 2 , 0.1 mg/mL Cycloheximide). The gradient was loaded with cell lysate and centrifuged at 39 K RPM at 4 °C for 90 min. Gradient fractions were collect using Brandel fractionation collector instrument and equal volume of each pooled fractions were processed for RNA isolation and dot blot.

RNA isolation and dot blot.
The DIV 15 rat cortical neurons and synaptoneurosomal, RNA was isolated using standard TRIzol RNA extraction method (Thermo Fisher Scientific Cat No. 15596018). Isolated RNA was finally resuspended in milliQ water and its concentration was measured using Qubit (Invitrogen) and equal an concentration of RNA was used for dot blot analysis. For the dot blot, the Nitrocellulose membrane (Cat No. 10600002, Sigma-Aldrich) was cut according to the requirement and rinsed first with milliQ water followed with 20X SSC buffer (Cat.no-AM9763, Thermo Fisher Scientific) and air dried. Extracted RNA was diluted to 250 ng in the RNA dilution buffer (6X SSC buffer and 7.5% para-formaldehyde) and heated to 65 °C for 5 min and kept on ice for 5 min. RNA was spotted on to nitrocellulose membrane and UV crosslinked. The membrane was blocked in TBST with 5% BSA for 1 h at room temperature. Further, membrane was incubated overnight at 4 O C with 1:1000 dilution of m 6 A antibody (Cat No. 202111, Synaptic systems) Subsequently, the membrane was washed three times in TBST for 10 min of interval. Anti-rabbit-HRP secondary antibody was incubated at 1:10,000 dilution for 1 h at room temperature.
Methylene-blue staining. RNA  www.nature.com/scientificreports/ Afterwards the membrane was airdried until the loading of samples. The samples were prepared by diluting the extracted RNA to a final concentration of 250 ng in the RNA dilution buffer followed by heating of diluted samples at 65 °C for 5 min. Afterwards the samples were incubated on ice for 5 min. The diluted RNA sample were spotted on the activated membrane and crosslinked using UV cross linker. After crosslinking, the membrane was incubated with the methylene blue staining solution (0.3 M sodium acetate and 0.03% methylene blue) for 5 min, followed by washes with distilled water to remove the background signal. The processed membrane was then imaged using Image Quant (LAS 4000/Amersham imager 600).
Quantitative PCR. Isolated RNA from the NMDA treated sample was processed for cDNA synthesis with reverse transcriptase and without reverse transcriptase. 200 ng of RNA was taken and cDNA synthesis was done with random hexamer and while doing the reverse transcription enzyme (M-MLV cat.no 28025013, Invitrogen) was excluded (minus reverse transcription) and included (plus reverse transcription) to the master mix and cDNA synthesis was done according to manufacturer protocol. Quantification of PSD95, PTEN and actin was measured by qPCR using TAKARA SYBR green (Cat.no -RR82WR).
Synapto-neurosome preparation. Rat cortical synaptoneurosomes were prepared using the filtration method from Sprague Dawley (SD) rat 21,22 . Briefly, the dissected cortices were resuspended in the synaptoneurosome buffer (118 mM NaCl, 5 mM KCl, 1.2 mM MgSO 4 , 2.5 mM CaCl 2 , 1.53 mM KH 2 PO 4 , 212.7 mM Glucose, 1X Protease Inhibitor Cocktail, pH 7.5) followed by homogenization on ice. The obtained homogenate was filtered by passing through three 100 µm nylon filters (NY1H02500, Merck Millipore) and one 11 µm nylon filter (NY1102500, Merck Millipore). The filtrate was then centrifuged at 1500 RCF for 15 min at 4 O C. The pellet obtained was resuspended in 2 mL synaptoneurosome buffer and used for NMDA treatment (40 µM) for different time points. After NMDA treatment, the resuspended synaptoneurosomes were subjected to a brief spin and the obtained pellets were resuspended in lysis buffer (50 mM Tris pH 7.4, 150 mM NaCl, 5 mM MgCl 2 , 1% Triton X-100, supplemented with EDTA-free protease inhibitor complex and phosphatase inhibitor cocktail) and centrifuged at 16,000 RCF for 30 min at 4 °C. The obtained lysates were used for western blotting and RNA isolation as per previously described protocol.