m⁶A RNA methylation regulates mitochondrial function

Abstract

RNA methylation of N⁶-methyladenosine (m⁶A) is emerging as a fundamental regulator of every aspect of RNA biology. RNA methylation directly impacts protein production to achieve quick modulation of dynamic biological processes. However, whether RNA methylation regulates mitochondrial function is not known, especially in neuronal cells which require a high energy supply and quick reactive responses. Here we show that m⁶A RNA methylation regulates mitochondrial function through promoting nuclear-encoded mitochondrial complex subunit RNA translation. Conditional genetic knockout of m⁶A RNA methyltransferase Mettl14 (Methyltransferase like 14) by Nestin-Cre together with metabolomic analysis reveals that Mettl14 knockout-induced m⁶A depletion significantly downregulates metabolites related to energy metabolism. Furthermore, transcriptome-wide RNA methylation profiling of wild type and Mettl14 knockout mouse brains by m⁶A-Seq shows enrichment of methylation on mitochondria-related RNA. Importantly, loss of m⁶A leads to a significant reduction in mitochondrial respiratory capacity and membrane potential. These functional defects are paralleled by the reduced expression of mitochondrial electron transport chain complexes, as well as decreased mitochondrial super-complex assembly and activity. Mechanistically, m⁶A depletion decreases the translational efficiency of methylated RNA encoding mitochondrial complex subunits through reducing their association with polysomes, while not affecting RNA stability. Together, these findings reveal a novel role for RNA methylation in regulating mitochondrial function. Given that mitochondrial dysfunction and RNA methylation have been increasingly implicate in neurodegenerative disorders, our findings not only provide insights into fundamental mechanisms regulating mitochondrial function, but also open up new avenues for understanding the pathogenesis of neurological diseases.

m6A RNA methylation, mitochondria, neurological disease, metabolomics, transcriptomics

Introduction

m⁶A RNA methylation has recently been shown to be important in regulating essentially every aspect of RNA biology [1, 2], ranging from transcription [3, 4], nuclear-cytoplasmic transport [5, 6], splicing [7], translation [8, 9] to degradation [10]. As the most prevalent internal modification on mRNA and long non-coding RNA in eukaryotes [11], m⁶A tags on conserved RACH sequence motifs (R = G or A; H = A, C, or U) are preferentially distributed around stop codons and enriched at 3′ untranslated regions (3′ UTRs) in the transcriptomes of human and mice [12, 13]. m⁶A RNA methylation can be dynamically added by “writer” RNA methyltransferase protein complexes consisting of METTL3 (Methyltransferase-like 3) and METTL14 as the primary methyltransferase components [14], and removed by “eraser” proteins including FTO (Fat mass and obesity-associated protein) and ALKBH5 (AlkB Homolog 5) [15, 16]. Different from traditional epigenetic modifications of DNA methylation and histone methylation that regulate gene expression, epitranscriptomic (or RNA epigenetic) modification by m⁶A primarily controls protein production, directly impacting biological phenotypes. Therefore, RNA methylation provides a form of regulation that bypasses the time-consuming RNA transcription process to modify dynamic cellular processes such as neuronal activities and energy metabolism.

Mitochondria regulate bioenergetics by synthesizing ATP via oxidative phosphorylation (OXPHOS) and are deeply involved in cellular metabolism, signaling and cell death [17–20]. Mitochondria exert their functions through a set of over 1100 mitochondrial proteins (in humans) localized on mitochondrial outer membrane, inner membrane, and two compartments created by them—the intermembrane space and matrix [21, 22]. Among these mitochondrial proteins, 13 are encoded by their own independent genome—the mitochondrial DNA (mtDNA), with all of the rest encoded by nuclear genomic DNA [23]. The expression of mitochondrial proteins by nuclear- encoded genes is regulated by traditional epigenetic modifications including the methylation of DNA and histones [24]. On the other hand, mitochondrial metabolism can change epigenetic modification, chromatin accessibility and gene transcription by affecting the availability of substrates for chromatin-modifying enzymes, including acetyl group donor acetyl coenzyme A [19, 24]. In addition, some intermediate mitochondrial metabolites, such as α-ketoglutarate, are needed both for histone demethylation and DNA demethylation catalyzed by jumonji-domain histone demethylases and the ten-eleven translocation (TET) family of dioxygenases, respectively [25, 26]. However, the relationship between RNA methylation and mitochondrial function hasn’t been explored.

Here we show that m⁶A RNA methylation regulates mitochondrial function through promoting nuclear-encoded mitochondrial electron transport chain (ETC) subunit RNA translation. Using Nestin-Cre mediated genetic knockout of RNA methyltransferase Mettl14 to deplete m⁶A in neurons and glia, we performed transcriptome-wide RNA methylation profiling in mouse brain and metabolomic analysis. We found that m⁶A methylation is enriched in mitochondria-related RNA and m⁶A depletion significantly downregulates metabolites related to energy metabolism. In addition, loss of m⁶A leads to a significant reduction in mitochondrial respiratory capacity and membrane potential. These functional defects are paralleled by the reduced expression of mitochondrial ETC complexes, as well as decreased assembly and activity of mitochondrial super-complexes. Mechanistically, we found that m⁶A depletion reduces the translational efficiency of mitochondrial ETC subunit RNA transcripts while not affecting RNA stability. Altogether, our findings reveal a novel role for RNA methylation in regulating mitochondrial function in neuronal cells. Defects in mitochondrial function contribute to neurodegeneration [27–33], and increasing evidence is implicating dysregulation of RNA methylation in the pathogenesis of neurological disorders [5, 34–39]. Therefore, our findings not only provide insights into fundamental mechanisms regulating mitochondrial function mediated by RNA methylation, but also open up new avenues for understanding the pathogenesis of neurological diseases.

Results

Genetic knockout of Mettl14 depletes m⁶A RNA methylation and changes the neural metabolome

m⁶A RNA methylation is particularly abundant in mammalian brain and the methylome is conserved in mammals [12, 13]. To study the function of RNA methylation in the nervous system, we generated tissue-specific Nestin-Cre;Mettl14^f/f conditional knockout mice to genetically remove m⁶A methyltransferase Mettl14 in Nestin-expressing neural stem cells that give rise to neurons and glia. From these Mettl14 knockout and wild type control mice, we derived neural progenitor cells (NPCs) that can be differentiated into neurons and glia (Fig. 1A). Using mass spectrometry and dot blot analyses we confirmed that m⁶A levels are reduced by ~50%–60% in Mettl14 knockout mouse brains and NPCs compared to those of wild type mice (Fig. 1B and C).

Figure 1

Genetic knockout of m⁶A methyltransferase Mettl14 depletes m⁶A RNA methylation and leads to changes in the neuronal metabolome. (A) Immunostaining of mouse neural progenitor cell (NPC)-differentiated neurons and glia with antibodies recognizing neuronal marker β-III tubulin (TuJ1) and glial marker glial fibrillary acidic protein (GFAP), respectively. (B) LC-MS/MS quantification of m⁶A levels in purified mRNA from WT and Mettl14 KO brain, showing m⁶A depletion in Mettl14 KO mice (P = 0.0063). (C) RNA dot blot quantification of m⁶A levels using m⁶A antibody on 60 ng and 30 ng of purified mRNA from WT and Mettl14 KO NPCs (P < 0.0001). Data shown are mean ± SEM, and are from three independent experiments, two-tailed t-test. (D) Volcano plot showing the relationship between the p-value and fold change (FC) of each individual metabolite in WT and Mettl14 KO NPCs. Fold change is the ratio of each metabolite’s abundance in KO/WT. Metabolites that have a p-value less than 0.05 or have a FC less than or more than 2 are highlighted and labeled with their compound IDs. Metabolites highlighted with blue are downregulated in the KO, while those highlighted with red are upregulated in the KO when compared to the WT. (E) Pathway enrichment analysis showing the top ten significant KEGG pathways enriched for metabolites with a fold change of ±35% (54 metabolites) between WT and Mettl14 KO NPCs. (F) Heatmap of three WT and three Mettl14 KO samples showing the relative abundance of the top 50 changed metabolites. Metabolites are listed by their KEGG IDs on the right. Relative abundance is shown using a red/blue color gradient, with red indicative of higher abundance and blue indicative of lower abundance.

The nervous system is characterized by high metabolic activities. Therefore, we first examined how m⁶A depletion by genetic knockout of Mettl14 may affect the metabolome. Comprehensive metabolomic analysis showed that metabolites are significantly changed following m⁶A depletion in Mettl14 knockout NPCs (Fig. 1D–F). Of these metabolites, the vast majority (41 out of top 50) are downregulated in Mettl14 knockout cells when compared to those in wild type (Fig. 1D–F), indicating that RNA methylation promotes neural metabolism. Specifically, metabolites involved in energy metabolism pathways and biosynthesis pathways are significantly changed (Fig. 1E), suggesting that mitochondrial function may be altered by depletion of m⁶A RNA methylation.

Transcriptome-wide RNA methylation profiling shows that m6A methylation is enriched on mitochondria-related RNA

To explore the regulation of molecular function by RNA methylation in the nervous system, we performed transcriptome-wide RNA methylation profiling of brains (cortices and hippocampi) from wild type and Nestin-Cre;Mettl14^f/f knockout mice. We immunoprecipitated methylated mRNA using an anti-m⁶A antibody and followed by high throughput sequencing (m⁶A-Seq) (Fig. 2A). Analysis of m⁶A-Seq data showed that most m⁶A modifications on mRNA are localized around the junction between the CDS (coding sequence) and 3′UTR (untranslated region) (Fig. 2B and C), consistent with previous reports [1, 12, 13]. In addition, the consensus sequences that we found tagged with m⁶A methylation in mouse brain mRNA (Fig. 2D) is also very similar to the previously identified methylation motif R(m⁶A)CH (R = G or A; H = A, C, or U) [1, 12, 13]. The number of m⁶A peaks in RNA from Mettl14 knockout mouse brains is reduced by more than 50% compared to that from wild type mouse brains (Fig. 2E), which is consistent with our mass spectrometry quantification (Fig. 1B). More importantly, gene ontology analysis of the 5637 genes, which are m⁶A tagged in the wild type brains but not in the Mettl14 knock out brains, showed that multiple mitochondria-related cellular component terms are significantly enriched (Fig. 2F, Supplemental Table 1). These findings suggest that mitochondrial function may be regulated by m⁶A RNA methylation.

Figure 2

Mitochondria-related RNAs are enriched in m⁶A tags as revealed by transcriptome-wide RNA methylation profiling. (A) Schematic illustration of the m⁶A-Seq procedure. (B) Pie chart depicting the distribution of m⁶A tags across the 5′ UTR (untranslated region), CDS (coding sequence), ncRNA (non-coding RNA), and 3′ UTR throughout the transcriptome of wild type mouse brain (cortices and hippocampi). (C) Meta-gene analysis showing the relative frequency of m⁶A tag distribution. (D) Motif analysis of m⁶A sites identified by m⁶A-Seq. (E) Venn diagram illustrating the number of m⁶A-tagged genes identified in WT and Mettl14 KO mouse brains. (F) Gene ontology analysis of m⁶A-tagged genes methylated in WT but not Mettl14 KO mouse brains, showing the significantly enriched mitochondria-related cellular component pathways.

Genetic knockout of Mettl14 and depletion of m⁶A reduce mitochondrial bioenergetics and membrane potential

To test whether mitochondrial bioenergetics are altered by genetic depletion of m⁶A, we investigated mitochondrial respiration and oxygen consumption in Mettl14 knockout cells using an Agilent Seahorse Bioanalyzer. We found significant decreases in basal oxygen consumption rate (basal OCR), as well as coupled OCR which is used by mitochondria for generating ATP, in Mettl14 KO NPCs compared to WT NPCs (Fig. 3B–D). We also observed significant decreases in the maximum respiratory capacity of mitochondria (maximal OCR) and OCR specifically used to counteract mitochondrial proton leakage (Fig. 3B, E, and F), indicating compromised overall mitochondrial bioenergetics in Mettl14 KO cells. Because defective mitochondria commonly show decreased membrane potential, which are linked to impaired energy production and mitochondrial function, we also measured the membrane potential of mitochondria in Mettl14 KO and WT NPCs. Using mitochondrial membrane potential-dependent fluorescent dye tetramethylrhodamine ethyl ester (TMRE), we found that Mettl14 KO cells have significantly reduced membrane potential (Fig. 3G). Together with reduced mitochondrial respiration, these data demonstrate that genetic depletion of m⁶A leads to robust mitochondrial functional defects, consistent with the down regulation of metabolites found in our metabolomics study (Fig. 1).

Figure 3

Genetic knockout of Mettl14 and depletion of m⁶A reduces mitochondrial bioenergetics and membrane potential. (A) Schematic illustration of an Agilent seahorse assay. (B) Oxygen consumption rates (OCRs) of WT and Mettl14 KO NPCs measured by an Agilent seahorse XF96 analyzer. (C–F) Quantification of basal OCR (P < 0.0001), coupled OCR/ATP synthesis (n = 5, P < 0.0001), maximal OCR (n = 5, P < 0.0001), and proton leak OCR (n = 5, P < 0.0001) shows that all OCRs are significantly decreased in Mettl14 KO. (G) Mitochondrial membrane potential measured by TMRE fluorescence shows significantly decreased mitochondrial membrane potential in Mettl14 KO (n = 3, P = 0.0014). Data shown are expressed as mean ± SEM. ^**P < 0.01 and ^****P < 0.0001, two-tailed t-tests.

Mettl14 KO and m⁶A depletion decrease mitochondrial electron transport chain complex expression and activity

m⁶A RNA methylation regulates many aspects of RNA metabolism, including RNA transcription, nuclear-cytoplasmic transport, translation and degradation [3–6, 8–10]. To understand how Mettl14 KO and m⁶A depletion lead to defective mitochondrial function, we first checked the impact of Mettl14 KO on the expression of mitochondrial electron transport chain (ETC) complexes. Western blot analysis using antibodies recognizing specific subunits of mitochondrial ETC complexes showed that mitochondrial complex expression is downregulated in Mettl14 KO NPCs (Fig. 4A and B). All five mitochondrial ETC complexes show significant reduction in protein levels, while the number of mitochondria remain unchanged as measured by Vdac1 expression (Supplemental Fig. 1). Similar to Vdac1, mtDNA-encoded ETC components also remain unchanged in Mettl14 KO cells when compared to WT cells at both the mRNA and protein levels (Supplemental Fig. 1). Mitochondrial ETC complexes can assemble into super-complexes to facilitate electron transfer efficiency and reduce reactive oxygen species production [40]. Therefore, we next examined whether ETC super-complex levels are also altered following m⁶A depletion, in addition to individual ETC complex changes. Utilizing Blue Native gel electrophoresis (BN-PAGE) and an anti-OXPHOS antibody cocktail that recognizes ETC complexes, we found significant decrease in the levels of all ETC super-complexes (Fig. 4C and D; Supplemental Fig. 2). Furthermore, in order to test whether m⁶A depletion has an impact on the activity of these complexes, we checked how native ETC complex functional activity is affected in Mettl14 KO cells. In a native in-gel activity assay, complex I on its own showed significantly decreased activity, and the super-complex containing complexes I + III also displayed significantly reduced activities (Fig. 4E). Collectively, these data show that m⁶A depletion through Mettl14 KO in NPCs leads to reduced expression of mitochondrial ETC complexes and decreased mitochondrial ETC complex activities.

Figure 4

Mitochondrial electron transport chain (ETC) complex expression and activity are reduced following Mettl14 KO and m⁶A depletion. (A) Representative Western blots of mitochondrial ETC complex subunits in WT and Mettl14 KO NPCs. β-actin is used as a loading control in (A). (B) Quantification of Western blots in (A). Complex I (n = 3, P < 0.001), complex II (P = 0.031485), complex III (P < 0.001), complex IV (P = 0.00174), complex V (P = 0.000114). (C and D) Blue-native PAGE blot and quantification of ETC supercomplexes in WT and Mettl14 KO mitochondria. Complex I + III (P = 0.002759), complex I + III + II (P = 0.002190), complex I (P = 0.000793), complex V + III + IV (P = 0.004923), complex III + IV (P = 0.000416), and complex II (P = 0.003884). Total mitochondrial protein amount was used for normalization (Supplemental Fig. 2). (E) Mitochondrial ETC supercomplex activity level measured by clear-native PAGE in-gel activity assays. Complex I + III (P = 0.0005), complex I + III + II (P = 0.1314), and complex I (P = 0.0452). Activity levels were normalized to total mitochondria protein level, and to levels of corresponding super-complexes measured in BN-PAGE. Data shown are expressed as mean ± SEM, and are from at least three independent experiments. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001, two-tailed t-tests.

RNA methylation regulates mitochondrial function in NPC-derived neurons and glia

Previous studies have shown that m⁶A RNA methylation is critical for neural development and function [5, 13, 35, 41–43], and m⁶A is particularly abundant in mammalian brain [12, 13]. So, we next tried to determine if the regulation of mitochondrial function by RNA methylation we identified above is a universal property in neurons and glia as well. We generated NPC-derived neurons and glia (Fig. 1A), and then investigated how m⁶A depletion by Mettl14 KO impacts mitochondrial respiratory capacity, membrane potential, and ETC complex expression. Similar to neural progenitor cells, Mettl14 KO NPC-derived neurons and glia showed significantly reduced respiratory capacity when compared to their wild type counterparts (Fig. 5A–F). We also found significantly decreased basal OCR, proton leak OCR, ATP synthesis/coupled OCR, as well as maximal OCR (Fig. 5A–F) in Mettl14 KO NPC-derived neurons and glia. Furthermore, Mettl14 KO neurons and glia had significantly lower mitochondrial membrane potential than that of WT cells, as measured by relative TMRE fluorescence (Fig. 5F). Lastly, Mettl14 KO neurons and glia showed dramatically reduced expression of mitochondrial ETC complex subunits, similar to Mettl14 knockout NPCs (Fig. 5G and H). These mitochondrial defect profiles exhibited by NPC-derived neurons and glia following m⁶A depletion indicate that the regulation of mitochondrial function by m⁶A RNA methylation is likely a common mechanism conserved in diverse cell types.

Figure 5

RNA methylation regulates mitochondrial function in NPC-derived neurons and glia. (A) Oxygen consumption rates (OCRs) of WT and Mettl14 KO NPC-differentiated neurons and glia, measured by an Agilent seahorse XF96 analyzer. (B–E) Quantification of basal OCR (P < 0.0001), coupled OCR/ATP synthesis (P < 0.0001), maximal OCR (P < 0.0001), and proton leak OCR (P < 0.0001) shows all OCRs are significantly decreased in Mettl14 KO. (F) Mitochondrial membrane potential measured by TMRE fluorescence shows significantly reduced mitochondrial membrane potential in Mettl14 KO (P = 0.0308). (G and H) Western blot analysis and quantification of different mitochondrial ETC complex subunits shows reduced expression in Mettl14 KO NPC-derived neurons and glia. Complex I (P < 0.003107), complex II (P = 0.308947), complex III (P = 000046), complex IV (P = 0.000088), complex V (P = 0.000863). Data shown are mean ± SEM, and are from at least three independent experiments. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001, ^****P < 0.0001, two-tailed t-tests.

Mettl14 KO and m⁶A depletion results in decreased translational efficiency of nuclear-encoded mitochondrial complex subunit RNAs without affecting RNA stability

Next, we asked what is the mechanism by which m⁶A RNA methylation regulates mitochondrial function. It has been shown that m⁶A methylation modulates RNA degradation or translation [8–10]. To distinguish between these possibilities, we first checked if depleting m⁶A leads to RNA translational changes using polysome profiling that measures the amount of actively translating RNA associated with polysomes. We fractionated nuclear-encoded mRNA from WT and Mettl14 KO NPCs using 10%–50% sucrose density gradient ultracentrifugation to separate polysomes and monosomes, and then measured their quantities (Fig. 6A). We found no significant differences in overall RNA translation between WT and Mettl14 KO cells, as shown by similar monosome and polysome profiles (Fig. 6A). We next isolated RNA from actively translating polysomal fractions and measured the amount of specific m⁶A-tagged transcripts by RT-qPCR. Interestingly, we saw significantly decreased levels of Atp5a1, Cox4i1, and Ndufa13, which are methylated and encode mitochondrial ETC complex subunits, in Mettl14 KO polysomal fractions when compared to WT fractions (Fig. 6B). But for m⁶A-tagged transcripts Hes6 and Foxk2, which do not encode mitochondrial complexes, we found no difference between their levels in Mettl14 KO and WT polysomal fractions (Fig. 6B). These data indicate that RNA methylation specifically regulates the polysomal association and translational efficiency of RNA transcripts encoding mitochondrial complexes.

Figure 6

Mettl14 KO and m⁶A depletion leads to reduced translational efficiency of nuclear-encoded mitochondrial complex subunit RNAs without changing RNA stability. (A) Polysome profiles of WT and Mettl14 KO NPCs showing OD254nm as a function of gradient depth. (B) RT-qPCR quantification of m⁶A-tagged polysome-associated RNA in WT and Mettl14 KO NPCs, normalized to the amount of unfractionated total RNA. Levels of nuclear-encoded mitochondrial complex subunits Atp5a1, Cox4i1, and Ndufa13 are significantly reduced in Mettl14 KO polysomal fractions, but not non-mitochondria-related Hes6 and Foxk2. (C–H) Relative mRNA levels following actinomycin D treatment in WT and Mettl14 KO NPCs. Cells were treated with 5 μM actinomycin D for 5 h, and RNA at each time point was collected and measured by RT-qPCR. Data shown are mean ± SEM, and are from four independent experiments. ^***P < 0.001, two-tailed t-test.

RNA methylation has been shown to regulate the degradation of m⁶A-tagged mRNA transcripts [10]. We therefore tested whether m⁶A depletion in Mettl14 KO condition also affects the stability of mitochondrial complex encoding RNA, thus leading to mitochondrial defects. We treated both WT and Mettl14 KO NPCs with actinomycin D to inhibit transcription of new RNA, and collected RNA at both 0 h and 5 h post treatment. Quantification of RNA using RT-qPCR showed that the stability of β-Tubulin, which has been shown to be m⁶A-tagged in our RNA methylation profiling study (Fig. 2) and by others [44], was significantly affected by Mettl14 KO and m⁶A depletion (Fig. 6C). However, the stability of other methylated RNA transcripts including Atp5a1, Ndufa13, Cox4i1, Hes6, and Foxk2, regardless of whether they encode mitochondrial complexes, was not significantly affected by Mettl14 KO and m⁶A depletion (Fig. 6D–H). In particular, mitochondrial complex encoding transcripts Atp5a1, Ndufa13 and Cox4i1 were very stable (Fig. 6D–F), with little degradation during the 5 h in both WT and Mettl14 KO conditions. Taken together, these data show that m⁶A RNA methylation does not affect the stability of mitochondrial complex-encoding RNA, but rather regulates their translational efficiency (Figure 7).

Figure 7

Schematic illustration of the proposed mechanism underlying m⁶A regulation of mitochondrial function. Under wild type condition with normal levels of m⁶A, translation-promoting m⁶A readers recognize mitochondria-related RNAs and localize them to polysomes, leading to normal expression of mitochondrial electron transport chain proteins and healthy mitochondrial function. Genetic knockout of Mettl14 depletes m⁶A, which causes decreased m⁶A reader binding and reduced polysomal association of nuclear-encoded mitochondrial complex subunit RNAs. This decreases mitochondrial electron transport chain complex expression and leads to defective mitochondrial function.

Discussion

m⁶A RNA methylation is emerging as a critical regulator of every aspect of RNA biology, directly impacting protein production to achieve quick and efficient regulation of dynamic biological processes. It is not known whether m⁶A regulates mitochondrial function and bioenergetics, in particular in neuronal cells which require a high energy supply and quick reactive responses. Using genetic knockout of the m⁶A RNA methyltransferase Mettl14 together with genome wide RNA methylation profiling and metabolomic analysis, we identified a critical role for RNA methylation in regulating mitochondrial function. Genetic depletion of m⁶A leads to reduced mitochondrial respiration at the basal, maximal, and ATP production levels, as well as compromised mitochondrial membrane potential. These functional defects are paralleled by the reduced expression of mitochondrial ETC complexes and their activities. Mechanistically, RNA methylation promotes nuclear-encoded mitochondrial ETC complex subunit RNA translation while not affecting RNA stability.

Traditional epigenetic modification of DNA and histone methylation regulates gene expression by modifying chromatin accessibility. The epitranscriptomic (or RNA epigenetic) modification of RNA by m⁶A affects essentially every aspect of RNA biology, including splicing, nuclear cytoplasmic transport, translation, and stability. As cellular functions are controlled by proteins, RNA methylation directly impacts protein expression and biological activities without going through the time-consuming transcription process. Therefore, RNA methylation is believed to be a layer of regulation particularly important for dynamic cellular functions. Energy production by mitochondria is one of these highly dynamic processes, yet very little is known about the role that RNA methylation plays in regulating mitochondrial function.

Neuronal cells have high energetic demand due to frequent synaptic activities, and the need to respond quickly to environmental stimuli. Here we identify for the first time the regulation of mitochondrial function by m⁶A RNA methylation and the underlying mechanism. Our transcriptome-wide RNA methylation profiling of mouse brain by m⁶A-Seq showed that m⁶A-tagged nuclear-encoded RNA transcripts are significantly enriched in multiple mitochondria-related cellular component gene ontology terms (Fig. 2F). Depletion of m⁶A by genetic knockout of Mettl14 leads to reduced basal, maximal, and coupled (ATP-generating) mitochondrial respiration, suggesting that RNA methylation promotes mitochondrial respiration. This is consistent with our finding that RNA methylation upregulates neural metabolism, as metabolites are significant down regulated upon m⁶A depletion in Mettl14 knockout cells (Fig. 1). Interestingly, in plant Arabidopsis, mitochondria and chloroplast transcriptomes have been found to be tagged with m⁶A modifications [45]. In yeast, upon the depletion of m⁶A, mitochondrial cytochrome c oxidase activity as well as citrate synthase activity were decreased, while mitochondrial fragmentation was significantly increased [46], suggesting a link between m⁶A and mitochondrial function in yeast. In addition, our analyses of published m⁶A-Seq data from other cell types show that RNA methylation is also enriched in mitochondria-related RNA in other cell types, including adult mouse hippocampal tissue, HEK293 human embryonic kidney cells, and human hematopoietic stem cell-derive monocytes [13, 44, 47], indicating that the regulation of mitochondrial function by m⁶A RNA methylation is a general mechanism conserved in different species and cell types.

To test the role of RNA methylation in mitochondrial function in neuronal cells, we generated tissue-specific Nestin-Cre;Mettl14^f/f conditional knockout mice to genetically remove Mettl14 gene expression in Nestin-expressing neural stem cells that will give rise to neurons and glia. Mettl14 is part of a RNA methyltransferase core complex consisting of the catalytic subunit Mettl3, and the essential component Mettl14 that facilitates RNA substrate binding [14, 48, 49]. Our mass spectrometry quantification and dot blot analysis show 50%–60% reduction of m⁶A in Mettl14 knockout cells, which is consistent with m⁶A decrease observed in Hela cells with Mettl14 knockdown [14]. This partial depletion of m⁶A is likely because Mettl3 still retains some level of methyltransferase activity in the absence of Mettl14, or due to functional compensation by other putative RNA methyltransferases. Among those METTL (methyltransferase like) proteins, METTL4 has been suggested to mediate the methylation of mammalian mitochondrial DNA (mtDNA) on methyldeoxyadenosine (⁶mA). Mitochondria have their own independent genome—the mitochondrial DNA, which encodes 13 proteins that are components of the mitochondrial electron transport chain for OXPHOS [23]. The rest of the over 1100 mitochondrial proteins (in humans) are encoded by nuclear genomic DNA [21]. METTL4-mediated ⁶mA modification on mtDNA leads to attenuated mtDNA transcription and reduced mtDNA copy number [50]. In contrast, we found here that the mitochondria-related RNAs that are regulated by m⁶A are all nuclear encoded. In addition, we revealed that m⁶A modification enhances mitochondrial function through promoting the translation of mitochondrial complex encoding RNA transcripts without affecting their stability (Figs 3–6).

The effects of m⁶A modification on target RNA are mediated by reader proteins that recognize and bind methylated RNA motifs. YTH domain-containing family (YTHDF) protein YTHDF1 and YTHDF2 have been shown to function as m⁶A readers to modulate methylated RNA translation and degradation, respectively [8, 10]. In addition, m⁶A reader YTHDF3, YTHDC2 can also regulate RNA translation, while m⁶A readers insulin growth factor 2 binding proteins 1/2/3 (IGF2BP1/2/3) can modify the stabilization and translation of methylated RNA [2, 51, 52]. YTHDF1 preferentially binds to methylated target RNA transcripts, and recruits them to active ribosomes and polysomes to promote their translation [8]. Interestingly, among YTHDF1 binding RNA targets identified in Hela cells [8], our mining of the published data reveal that mitochondria-related transcripts are significantly enriched, when we cross-referenced YTHDF1 targets with the MitoMiner list of mitochondria related genes [53]. These findings suggest that the modulation of mitochondrial complex RNA translation by m⁶A may be mediated by the YTHDF family readers of methylated RNA.

Mitochondria profoundly impact neuronal functions by regulating bioenergetics, cellular metabolism, signaling and cell death [17–20]. Defects in mitochondrial function have been established as a major contributor to neurodegeneration [27–30]. Furthermore, increasing evidence implicates dysregulation of RNA methylation in the pathogenesis of neurological disorders [5, 34–38]. Therefore, our findings not only reveal a fundamental mechanism regulating mitochondrial function by RNA methylation, but also provide a new angle for understanding the pathogenic mechanism and therapeutic development for neurological disorders.

Materials and methods

Mettl14 conditional knockout mice and neuronal cell culture

Neuronal tissue-specific Nestin-Cre;Mettl14^f/f conditional knockout mice were generated by crossing the Mettl14^f/f mouse line with the Nestin-Cre line to genetically remove m⁶A methyltransferase Mettl14 in Nestin-expressing neural stem cells. Cortices of E14.5 wild type and Mettl14 KO embryos were isolated, digested with Accutase (Fisher) and dissociated by trituration, yielding cortical neural progenitor cells (NPCs). NPCs were cultured in DMEM/F12 culture media, supplemented with B27 (Gibco), N2 (Gibco), and Glutamax (Gibco) on cell culture dishes treated with matrigel (Corning #35234). EGF (20 ng/ml) and FGF (20 ng/ml) growth factors together with Heparin (5 μg/ml) were also added fresh to the culture media. Cells were housed in 37°C and 5% CO₂ incubators. For differentiation, 1.5 million cells per well were seeded onto poly-L-lysine (PLL) (20 μg/ml) and Laminin (8 μg/ml) coated 6-well plates in culture media without EGF/FGF/Heparin. On day in vitro 1 (DIV1), the media was exchanged for fresh media with the addition of BDNF (50 μg/ml), CNTF (50 μg/ml), GDNF (25 μg/ml), and Forskolin (10 μM) to promote differentiation. Half media was exchanged every two days until DIV7. NPC-differentiated neurons and glia were harvested or used on DIV7 for different assays.

Metabolomic analysis

Approximately 5 million cells were used for each replicate. The cells were first washed twice with 5 ml ice-cold normal saline, and then put on dry ice. To extract metabolites, 1 ml of 80% methanol pre-chilled to −80°C was added to the plates and then incubated at −80°C for 20 min. Then cells were then scraped on dry ice, transferred to a 1.5 ml tube, and frozen in liquid nitrogen. The samples were then subjected to three freeze-thaw cycles between liquid nitrogen and 37°C, with vertexing between each cycle. Then lysates were centrifuged at 20 000 g for 15 min at 4°C. The supernatant containing the metabolites was transferred to new 1.5 ml tube and dried to a pellet using a SpeedVac vacuum concentrator. Samples were then resuspended into MS-grade H₂O, and submitted to the Metabolomics Core Facility at Robert H. Lurie Comprehensive Cancer Center of Northwestern University. Data visualization and post-processed analysis was performed with R and metaboanalyst.

m⁶A-Seq

Total RNA was isolated from WT and Mettl14 KO postnatal day 5 (P5) mouse cortices and hippocampi using Trizol. mRNA was then isolated using the Ambion Dynabeads mRNA Direct Kit (Fisher), with two rounds of oligo-dT Dynabeads purification to eliminate rRNA contaminants. 5 μg of mRNA was then fragmented by heating at 94°C for 4 min in fragmentation buffer (100 mM Tris-HCl, 100 mM ZnCl₂), followed by the addition of EDTA to a concentration of 45 mM, and was purified using the Zymo RNA Clean and Concentrator kit (R1013). The resulting mRNA was then immunoprecipitated using m⁶A antibody (Synaptic Systems #202-003) pre-conjugated Protein A Dynabeads, for 2 h at 4°C. The beads were then washed 3X with 500 ul IP buffer (10 mM Tris-HCl, 150 mM NaCl, 0.1% NP-40) supplemented with RNase inhibitors, and eluted by incubating with 100 ul of elution buffer (10 mM Tris-HCl, 150 mM NaCl, 0.1% NP-40, 6.67 mM m⁶A) supplemented with RNase inhibitors for 1 h at 4°C. The supernatant was removed and the elution was performed again on the beads in order to ensure complete elution. The eluates were combined and purified using the Zymo RNA Clean and Concentrator Kit, and used for library building with the Illumina TruSeq Stranded mRNA Library Prep Kit (Illumina #20020594). The libraries were sequenced on an Illumina HiSeq 4000 with single-end 50 bp read length. Read alignment was performed using STAR mapping to the GRCm38 mouse genome. m⁶A peaks consistent across three biological replicates were identified using the exomePeak package. Motif identification was performed using Meme-CHIP, ontology analysis was performed using DAVID, and meta-analysis was performed using MetaPlot. Figures were generated using R and Prism.

RT-qPCR and RNA stability assay

Total RNA was isolated using Trizol following the manufacturers protocol. 1 μg of total RNA was reverse transcribed into cDNA using the QuantBio qScript cDNA Supermix (VWR), and then subjected to qPCR analysis using either the QuantBio PerfeCTa SYBR Green Supermix (VWR) or the Applied Biosystems PowerUp SYBR Green Master Mix (Thermo Fisher). An Applied Biosystems QuantStudio 3 platform was used for all reactions. The delta-delta CT method was used for fold-change quantification. For RNA stability assays, WT and Mettl14 KO NPCs were treated with 5 g/ml of actinomycin D for 0 h and 5 h. Then cells were harvested, and total RNA was isolated using Trizol and relative mRNA levels were analyzed using RT-qPCR as described above.

mRNA isolation and m⁶A dot blot analysis

Total RNA was isolated as described above. mRNA was isolated from total RNA using the Ambion Dynabeads mRNA Direct Kit (Fisher). To ensure complete ribosomal RNA removal, the mRNAs were subjected to two rounds of oligo-dT Dynabeads selection as described in the manufacturers protocol. Serial dilutions of double selected mRNAs were heated at 95°C for 3 min and placed immediately on ice. 2 μl per dilution were dotted onto a Hybond H⁺ positively charged nylon membrane (GE Healthcare). The membrane was then air dried for ~15–20 min, crosslinked twice at 125 mJ, and washed in 0.02% PBST. It was then blocked with milk for one hour at room temperature and incubated overnight at 4°C in primary anti-m⁶A antibody (Synaptic Systems #202-003, 1:1000). After removing the primary antibody, the membrane was washed three times in PBS with 0.05% Triton X-100 (PBST) followed by a one hour incubation with an anti-rabbit HRP-linked secondary antibody (1:10000). The membrane was then treated with ECL, imaged and quantified.

m⁶A quantification by liquid chromatography tandem mass spectrometry (LC-MS/MS)

Double selected mRNA was isolated using the Ambion Dynabeads mRNA Direct Kit (Fisher) as described in the manufacturers protocol, and eluted in MS-grade H₂O. 125 ng of double-selected mRNA was digested using Nuclease P1 (Wako Chemicals) at 42°C for 2 h. FastAP alkaline phosphatase (ThermoFisher) was then added to the reaction and incubated at 37°C overnight. The reaction was then diluted to 5 μl in mass spectrometry-grade H₂O and filtered through a 0.22 μm syringe driven filter (Millipore-Sigma). The samples were subjected to LC-MS/MS using an Agilent 6460 Triple Quad LC-MS coupled to a 1290 UHPLC in MRM mode. The nucleosides were separated using a C18 column (Agilent, 927700-902). The mobile phase was H₂O + 0.1% Formic Acid and Methanol +0.1% Formic Acid. Standard solutions of adenosine, guanosine, cytidine, uridine, and N⁶-methyladenosine (Cayman Chemicals) were used to generate standard curves and quantify nucleoside concentrations.

Western blot analysis

Cell lysates were prepared in RIPA buffer and supplemented with Roche cOmplete Protease Inhibitor Cocktail (Roche). Samples were run by SDS-PAGE and transferred to PVDF membranes. The membranes were then blocked in 0.05% PBST with 5% milk for 1 h at room temperature and incubated overnight at 4°C with Invitrogen OxPhos Blue Native WB Antibody Cocktail (#45-7999). The next day, the membranes were washed three times in 0.05% PBST and incubated in respective secondary antibodies for 1 h at room temperature, followed by another three washes in 0.05% PBST. They were then treated with ECL, imaged and quantified.

BN-PAGE and mitochondrial Supercomplex activity assay

Mitochondrial BN-PAGE and supercomplex activity assays were performed as described [54]. Briefly, three confluent 10 cm plates of WT or Mettl14 KO NPCs were scraped in native isolation buffer (IB) (200 mM Sucrose, 10 mM Tris, 1 mM EGTA, pH 7.4) supplemented with protease inhibitors, homogenized, and centrifuged at 600 g for 10 min at 4°C. The supernatant was then placed in a new tube and centrifuged at 7000 g for 10 min at 4°C. The pellet was then washed by resuspending it in IB buffer and centrifuged at 10 000 g for 10 min at 4°C. The pellet containing the native mitochondria was resuspended again in IB buffer. Protein concentration was calculated by BCA method. Either 35 μg or 50 μg of mitochondria was pelleted, resuspended in 20 μl native sample buffer containing 8 g digitonin/g mitochondria, and incubated on ice for 30 min. These samples were then centrifuged at 20 000 g for 10 min at 4°C. The supernatant was then removed and Coomassie G-250 was added to ¼ the digitonin concentration. This was then loaded onto a pre-cast 3%–12% Bis-Tris native gel (Thermo Fisher). NativePAGE Running Buffer (Novex) was used as the outer buffer, whereas the inner buffer was the same but supplemented with 0.04 g/200 ml Coomassie G-250. The gel was then run at 150 V for 45 min at 4°C. Following electrophoresis, the inner buffer was replaced with Native PAGE Running Buffer supplemented with 0.02 g/200 ml Coomassie G-250, and the gel was run again at 250 V for 90 min at 4°C. Prior to transfer, the Coomassie-stained gel was imaged to quantify total mitochondrial protein level for normalization. The gel was then transferred to a PVDF membrane at 300 mA for 110 min at 4°C. The membrane was fixed in 10% acetic acid, air dried, washed with methanol and H₂O, and blocked with 5% milk for 1 h at room temperature. Following blocking, the membrane was incubated overnight in primary OXPHOS antibody cocktail (ThermoFisher, #45-7999) at 4°C. The next day, the membrane was washed in 0.05% PBST, incubated in secondary antibody, washed in 0.05% PBST again, and developed. Mitochondrial Supercomplex activity assays were run similarly to the BN-PAGE detailed above; however, instead of supplementing the inner buffer with 0.04 g/ml Coomassie G-250 in the first run, 0.02 g/ml is used. After the first run, the inner buffer is replaced with NativePAGE Running Buffer with no Coomassie supplementation. The second run is also run for 150 min instead of 90 min. Following electrophoresis, the gel is placed in ice cold water. Then the gel is incubated at room temperature in Complex I substrate (2 mM Tris-HCl pH 7.4, 0.1 mg/ml NADH, 2.5 mg/ml Nitrotetrazoleum Blue chloride) for 30 min. The reaction was stopped with 10% acetic acid, and the gel was imaged with a visible protein stain gel imager.

Mitochondrial oxygen consumption rate analysis

Mitochondrial OCR was measured using an XF96 Seahorse Biosciences Extracellular Flux Analyzer. Wild type and Mettl14 KO NPCs were seeded onto a Seahorse 96 well plate at 30 000 cells per well 24 h prior to assay in NPC culture medium as described in the Cell Culture section. Culture media was changed to 80 μl of Agilent DMEM/F12 without phenol red 30 min before assay. Analyzer injection ports contained 1 μM oligomycin A, 7.5 μM FCCP, or 2 μM Antimycin and Rotenone. Basal OCR was calculated by subtracting the OCR following antimycin A and rotenone treatments from OCR following FCCP treatment. Maximal OCR was calculated by subtracting the OCR following antimycin A and rotenone treatment from the OCR following FCCP treatment. Coupled OCR was calculated by subtracting the OCR following oligomycin A treatment from Basal OCR. Following the assay, cells were immediately fixed with 4% PFA, stained with DAPI, and normalized based on cell number per well.

Mitochondrial membrane potential measurement

Mitochondrial membrane potential was measured using the potential-dependent fluorescent dye tetramethylrhodamine ethyl ester (TMRE). 30 000 wild type or Mettl14 KO NPCs per well were seeded onto a black with clear bottom 96 well plate 24 h prior to experiment. Cells were stained with 200 nM TMRE for 30 min, washed with PBS + 0.2% BSA, and the fluorescent intensity of TMRE was then measured. The cells were then washed and treated with 40 μM FCCP to measure background fluorescent. Following the assay, cells were immediately fixed with 4% PFA, stained with DAPI, and normalized based on cell number per well.

Polysomal profiling and fractionation

Cells were treated for 30 min with 100 μg/ml cycloheximide, washed twice with Dulbecco’s phosphate buffered saline (DPBS) and then lysed in lysis buffer (0.5% Triton X-100, 0.5% sodium deoxycholate, 5 mM Tris pH 7.5, 2.5 mM MgCl2, 1.5 mM KCl, 100 μg/ml cycloheximide, 2 mM DTT, protease inhibitor and 1unit/μl RNase inhibitor). Lysates were then centrifuged at 20 000 g for 20 min at 4°C and supernatants were collected and snap frozen in liquid nitrogen. To isolate ribosomal fractions, lysates were layered on a sucrose gradient of 5 to 50%. Samples were centrifuged at 4°C for 120 min at 35 000 rpm in a Beckman SW41-Ti rotor. Absorbance was measured at 254 nm continuously in an ISCO density gradient fractionator with the following settings: pump speed, 1.5 ml/min; fraction size, 10 drops per fraction; chart speed, 150 cm per hour; sensitivity, 0.5; peak separator, off; noise filter, 0.5 s. Fluorinert ™ FC-40 (Sigma-Aldrich #F9755) solution was used to set the baseline in an UA-6 detector for all experiments. Polysomal fractions for each sample were then pooled together, and RNA isolation was performed using AllPrep RNA/Protein Kit (Qiagen #80404) according to the manufacturer’s instructions.

Acknowledgements

The research reported in this manuscript was made possible in part by the generous support of the Agape Foundation and the Children’s Research Fund. Metabolomic analyses were performed by the Metabolomics Core Facility at Robert H. Lurie Comprehensive Cancer Center of Northwestern University. Sequencing was performed at the NUSeq core of Northwestern University’s Center for Genetic Medicine. We thank Zhongyu Zou and The University of Chicago Mass Spectrometry Core Facility for their assistance in LC-MS/MS quantification of m⁶A.

Author contributions

MK and YCM conceptualized the study. MK, ZX and SA performed experiments and investigation. ZX, BME, SA, MK and YCM maintained the Mettl14 knockout mice and generated the NPC lines. MF performed polysomal fractionation. ACZ, LCP and CH provided critical resources and protocols. XZ generated the Mettl14 knockout mouse line. MK and YCM wrote, reviewed, and edited the manuscript.

Conflict of interest statement: The authors declare no competing interests.

Funding

YCM was supported by grants from the NIH (R01NS094564, R21NS106307, RF1AG077451). MK was supported in part by the Mary J. C. Hendrix Outstanding Graduate Student Award from the Stanley Manne Children's Research Institute and the Children's Research Fund.

Data availability

The sequencing datasets presented in this manuscript have been deposited and can be retrieved using GEO accession number GSE252477.