Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2018 Nov;563(7730):249-253.
doi: 10.1038/s41586-018-0666-1. Epub 2018 Oct 31.

M 6 A Facilitates Hippocampus-Dependent Learning and Memory Through YTHDF1

Affiliations
Free PMC article

M 6 A Facilitates Hippocampus-Dependent Learning and Memory Through YTHDF1

Hailing Shi et al. Nature. .
Free PMC article

Abstract

N6-methyladenosine (m6A), the most prevalent internal RNA modification on mammalian messenger RNAs, regulates the fates and functions of modified transcripts through m6A-specific binding proteins1-5. In the nervous system, m6A is abundant and modulates various neural functions6-11. Whereas m6A marks groups of mRNAs for coordinated degradation in various physiological processes12-15, the relevance of m6A for mRNA translation in vivo remains largely unknown. Here we show that, through its binding protein YTHDF1, m6A promotes protein translation of target transcripts in response to neuronal stimuli in the adult mouse hippocampus, thereby facilitating learning and memory. Mice with genetic deletion of Ythdf1 show learning and memory defects as well as impaired hippocampal synaptic transmission and long-term potentiation. Re-expression of YTHDF1 in the hippocampus of adult Ythdf1-knockout mice rescues the behavioural and synaptic defects, whereas hippocampus-specific acute knockdown of Ythdf1 or Mettl3, which encodes the catalytic component of the m6A methyltransferase complex, recapitulates the hippocampal deficiency. Transcriptome-wide mapping of YTHDF1-binding sites and m6A sites on hippocampal mRNAs identified key neuronal genes. Nascent protein labelling and tether reporter assays in hippocampal neurons showed that YTHDF1 enhances protein synthesis in a neuronal-stimulus-dependent manner. In summary, YTHDF1 facilitates translation of m6A-methylated neuronal mRNAs in response to neuronal stimulation, and this process contributes to learning and memory.

Figures

Extended Data Figure 1 |
Extended Data Figure 1 |. Generation and evaluation of Ythdf1-KO mice.
a, Schematic diagram of the targeting strategy for generating Ythdf1-KO mice using CRISPR/Cas9. Two sgRNAs (red) were designed to target the 4th exon (E4) of Ythdf1. PAM sequence, underlined, green; F1 and R1, genotyping primers. b, Genotyping PCR products of the seven founders co-injected with 20 ng Cas9 mRNA and the two sgRNAs (5 ng each). c, Genotypes of sequenced mice. PCR products were cloned and sequenced. Founder #4 with a 179-bp deletion was crossed with C57BL/6 wild-type mice for further analysis. d, Representative genotyping PCR products of off-spring mice with different genotypes. e, Validation of Ythdf1 knockout by western blot using mouse hippocampal tissues. For gel source data, see Supplementary Figure 1. f, Representative images of Ythdf1 immunostaining in the mouse basal lateral amygdala (BLA) and the cortex.
Extended Data Figure 2 |
Extended Data Figure 2 |. Ythdf1-KO mice are normal in hippocampal neurogenesis, cortical morphology, motor activities, anxiety-like behavior, and depressive-like behavior
a, b, Representative images of Doublecortin (DCX, a marker of neurogenesis) immunostaining (a) and quantification of the number of DCX+ cells (b) in the dentate gyrus (DG) region of Ythdf1-KO and wild-type control mice at different postnatal development stages. Scale bar, 100 μm. c, Representative images of cortical morphology staining using Hoechst in adult control and Ythdf1-KO mice. Scale bar, 200 μm. d, Representative confocal immunostaining of Ctip2 (a marker for deep layer cortical neurons) and Satb2 (a marker for upper layer cortical neurons) in the cortex of adult control and Ythdf1-KO mice. Scale bar, 200 μm. e-h, Motor activities measured by various parameters as listed in the open-field test. i-l, Anxiety-like behavior measured by the light-dark box transition test (i, j) and the elevated-plus maze test (k, l). m, Depressive-like behavior measured by tail suspension test. P values, two tailed t-test. Numbers in bars, numbers of mice. Error bars, mean ± s.e.m.
Extended Data Figure 3 |
Extended Data Figure 3 |. Morris water maze (MWM) tests and fear conditioning tests in Ythdf1-KO mice.
a, Schematics of procedure of MWM trainings and MWM probe tests. b, c, Number of crossings over previous platform location (b) and swimming velocity (c) of control and Ythdf1-KO mice in MWM probe tests. d, Schematics of the fear conditioning procedures (left) and freezing responses measured at different stages (right). e, Titration curves of the freezing level of wild-type mice 24 hours after trainings with different foot shock intensities. The conditioning protocols used in later experiments (moderate protocols) are indicated by arrows. f, Learning curves for auditory fear conditioning under moderate (left) or strong (right) training protocols. The training sessions were separated into two parts: baseline (base) and tone periods (tone). g, h, Auditory fear memory of control and Ythdf1-KO mice assessed 24 hours (g) and 2 hours (h) after the indicated trainings. P values, two-way repeated measures ANOVA (d) and two tailed t-test (b, c, f-h). Numbers in bars, numbers of mice. Error bars, mean ± s.e.m.
Extended Data Figure 4 |
Extended Data Figure 4 |. Paired-pulse ratios (PPR), spine morphology, and total protein levels of various LTP-related genes in Ythdf1-KO mouse hippocampus, related to Figure 2.
a, b, PPR with different inter-stimulus intervals in CA1 neurons from wild-type control and Ythdf1-KO mice. c, d, Representative images of Lucifer Yellow staining (c) and statistical analyses of spine density (d, left) and spine size (d, right) in CA1 neurons from adult control and Ythdf1-KO brain. e, Uncropped western blot images for Fig. 2g. f, Total protein levels of a set of LTP-related genes in control and Ythdf1-KO mouse hippocampus. For gel source data, see Supplementary Figure 1. P values, two-way repeated measures ANOVA with post hoc two-tailed t-test (a) and two tailed t-test (b, d, f). Numbers in bars, numbers of slices (b), neurons/mice (d, left), spines (d, right), or mice (f). Error bars, mean ± s.e.m.
Extended Data Figure 5 |
Extended Data Figure 5 |. Viral targeting in Ythdf1-KO mouse hippocampus and behavioral analyses of Ythdf1-KO mice injected with AAV virus, related to Figure 3.
a, Representative fluorescence images of brain slices from rostral to caudal positions dissected from a mouse injected with AAV-Ythdf1 virus. Hoechst, blue; Ythdf1 co-expressed with mCherry, red. b, Representative images of virus expression (mCherry, red) and Ythdf1 immunostaining (green) in the mouse hippocampus after AAV-control or AAV-Ythdf1 infection. Hoechst, blue. c, Ythdf1 protein overexpression level indicated by immunofluorescent signal intensity in the CA1 and DG regions. d, Number of crossings over the previous platform location for Ythdf1-KO mice injected with AAV-Ythdf1 or AAV-control in MWM probe tests. e, Anxiety-like behavior of the injected mice measured as open-arm durations in elevated-plus maze. f-h, Motor activities of the injected mice measured as total distance (f), number of moves (g), and average velocity (h) in the open-field test. P values, two tailed t-test (c-h). Numbers in bars, numbers of mice. Error bars, mean ± s.e.m.
Extended Data Figure 6 |
Extended Data Figure 6 |. Impaired spatial learning and memory after selective knockdown of Ythdf1 in the hippocampus of WT mice.
a, Schematics of the AAV construct expressing Ythdf1 shRNA. b, Western blot and quantification of protein expression level of YTH proteins in N2A cells after Ythdf1-shRNA (RNAi) or control vector (Ctrl) transfection. For gel source data, see Supplementary Figure 1. c, Spatial learning curves in the hidden-platform MWM training sessions for RNAi (red) and control (gray) mice. d-f, Spatial memory performances measured by quadrant time (%) (d) and number of platform crossings (e), and motor activities (f) of RNAi (red) and control (gray) mice in MWM probe tests. g, i, Contextual (g) and auditory (i) fear memories assessed 24 hours after fear conditioning in RNAi and control mice. h, Anxiety level of mice assessed by open-arm durations in elevated-plus maze. P values, two-way repeated measures ANOVA with post hoc two-tailed t-test (c), two-way ANOVA with two-tailed t-test (comparison between group or to “Target”) (d), and two tailed t-test (b, e-i). Numbers in bars, numbers of biologically independent samples (b) and mice (d-i). Error bars, mean ± s.e.m.
Extended Data Figure 7 |
Extended Data Figure 7 |. Impaired spatial learning and memory after acute knockdown of Mettl3 in the hippocampus of WT mice.
a, Representative western blot (left) and quantification (right) of Mettl3 protein level in N2A cells transfected with Mettl3-shRNA (RNAi) or control vector (Ctrl). For gel source data, see Supplementary Figure 1. b-d, Spatial learning curves in the hidden-platform MWM training sessions (b), and spatial memory performances measured by quadrant time (%) (c) and the number of platform crossings (d) in MWM probe tests, for Mettl3-RNAi and control mice. e, Contextual (left) and auditory (right) fear memories measured by freezing levels 24 hours after fear conditioning in Mettl3-RNAi and control mice. f, Motor activities of mice accessed in the open-field test. P values, two-way repeated measures ANOVA with post hoc two-tailed t-test (b), two-way ANOVA with two-tailed t-test (comparison between groups or to “Target”) (c), and two tailed t-test (a, d-f), Numbers in bars, numbers of biologically independent samples (a) and mice (c-f). Error bars, mean ± s.e.m.
Extended Data Figure 8 |
Extended Data Figure 8 |. Ythdf1 binding sites and m6A sites in the hippocampus of adult mice, and Ythdf1-mediated effects of m6A on hippocampal transcriptome and proteome.
a, Peak overlap among three biological replicates of Ythdf1-CLIP-seq. b, Validation of immunoprecipitation efficiency for Ythdf1-CLIP-seq. The position of the gel slice cut during the step of protein-RNA complex size selection was indicated in red (see Methods). c, Consensus motif and its P value generated by HOMER40 of the three sets of hippocampal m6A sites from biological replicates of m6A-CLIP-seq. d, e, Distribution of m6A-CLIP peaks along the different regions of transcripts (d) and genome (e). f, Functional annotation of m6A-modified transcripts in the adult mouse hippocampus (number of mutations in m6A-CLIP-seq >= 5, n = 2,922). g, Peak overlap between high-confidence Ythdf1-CLIP peaks and high-confidence m6A-CLIP peaks. The percentage of Ythdf1-CLIP peaks overlapped is indicated. h, IGV screenshots of the piled mutated reads for the each of the biological triplicates of Ythdf1-CLIP-seq (red) and m6A-CLIP-seq (blue). Three examples of synaptic plasticity transcripts were presented; the overlapped peak regions are highlighted in orange. i, j, Box-plots of mRNA abundance (i) and protein abundance (j) log2 fold changes (Δ) between Ythdf1-KO hippocampus and wild-type control for all expressed genes (black), non-Ythdf1-CLIP transcripts (gray), Ythdf1-CLIP targets (red), transcripts with overlapped Ythdf1-CLIP peaks and m6A-CLIP peaks (pink), and m6A-modified transcripts (blue). Box-plot elements: center line, median; box limits, upper and lower quartiles, whiskers, 1–99%; P values, two-sided unpaired Kolmogorov-Smirnov test; number of genes and 95% CI of mean are indicated for each box (i, j).
Extended Data Figure 9 |
Extended Data Figure 9 |. Effects of Ythdf1 on nascent protein synthesis in cultured hippocampal neurons in response to KCl stimulus.
a, Additional representative images of nascent protein (Nascent-P) synthesis in cultured wild-type control and Ythdf1-KO hippocampal neurons before (sham) and 2 hours after KCl depolarization, related to Fig. 4e-f. b, c, Representative images (b) and quantification (c) of Nascent-P in wild-type control and Ythdf1-KO hippocampal neurons before (sham) and 4 hours after KCl depolarization. d, e, Representative images (d) and quantification (e) of Nascent-P in AAV-control and AAV-Ythdf1-RNAi treated hippocampal neurons before (sham), 2 hours, and 4 hours after KCl depolarization. Intensities of Nascent-P were normalized to that of wild-type control (c) or AAV-control (e) neurons under the sham condition. P values, two-tailed t-test (c, e). Numbers in bars, numbers of images/biologically independent samples. Error bars, mean ± s.e.m.
Extended Data Figure 10 |
Extended Data Figure 10 |. Neuronal-stimulus-dependent functions of Ythdf1 in the mouse hippocampus and potential underlying mechanisms.
a, Representative western blot (left) and quantification (right) of the protein levels of Bsn (top) and Camk2a (bottom), in the whole hippocampus and the PSD fraction, respectively, before (Mock) and 2 hours after fear conditioning (FC). The protein quantification was normalized to the Mock condition for each genotype separately. For gel source data, see Supplementary Figure 1. b, c, Representative western blot (b) and quantification of Ythdf1 protein level in the hippocampal postsynaptic density (PSD) fraction (c, left) and the whole hippocampus (c, right) before (Mock) and 2 hours after FC. For gel source data, see Supplementary Figure 1. d, Schemes of the experimental design to quantify the change in the extent of m6A methylation for each transcript in the DG region before (Mock) and 1 hour after electroconvulsive treatment (ECT). e, A proposed mechanism for how Ythdf1 contributes to memory formation: Ythdf1 promotes translation of m6A-modified target transcripts, including synaptic transmission and LTP-related ones, in response to learning stimulus, thus facilitating synapse strength adequately for a memory to occur. P values, two-tailed t-test (a, c). Numbers in bars, numbers of biologically independent samples. Error bars, mean ± s.e.m.
Figure 1 |
Figure 1 |. Impaired spatial learning and memory in Ythdf1-KO mice.
a, b, Representative images of Ythdf1 immunostaining (a) and Hoechst (b) in the control and Ythdf1-KO hippocampus. DG, dentate gyrus; P30/P120, postnatal day 30/120. c, d, Learning curves of control (blue) and Ythdf1-KO (red) mice in Morris water maze (MWM) tests in visible (c) and hidden (d) platform training. e, Quadrant time (%) (left) and representative swimming paths (right) of control and Ythdf1-KO mice in the MWM probe test. The red dash line represents the chance level (25%). f, Learning curves of control (blue) and Ythdf1-KO mice (red) for contextual fear conditioning (FC) in moderate (left) or strong (right) training sessions. Base, baseline; ITI, inter-trial interval. g, h, Contextual fear memory assessed 24 hours (g) or 2 hours (h) after the indicated FC. P values, two-way ANOVA with two tailed t-test (relative to “Target” or between genotypes) (e), two-way repeated measures ANOVA with post hoc test (c, d, f), and two-tailed t-test (g, h). Numbers in bars, numbers of mice. Error bars, mean ± s.e.m.
Figure 2 |
Figure 2 |. Deficient basal transmission and plasticity in Ythdf1-KO hippocampal synapses.
a, b, Representative traces (a) and quantification of amplitude (b, left) and frequency (b, right) of spontaneous miniature excitatory postsynaptic currents (mEPSCs) in control and Ythdf1-KO hippocampal CA1 neurons. c, d, Summary plots (c) and average amplitude (d) of long-term potentiation (LTP) induced by 2 × high frequency stimulation (HFS) in the CA1 region of control and Ythdf1-KO acute slices. fEPSP, field excitatory postsynaptic potential. e, f, Summary plots (e) and average amplitude (f) of late phase LTP induced by 4 × HFS. Top panels, sample traces taken at time points 1 and 2 indicated above the summary plots; scale bars, 10 ms (horizontal) and 0.2 mV (vertical) (c, e). g, h, Representative western blots (g) and quantification (h) of a number of LTP-related proteins in the control and Ythdf1-KO hippocampal postsynaptic density (PSD) fraction. P values, Kolmogorov-Smirnov test for cumulative distributions followed by comparisons with Mann–Whitney U test (b) and two-tailed t-test (d, f, h). Numbers in bars, numbers of neurons/mice (b), slices/mice (d, f), and mice (h). Error bars, mean ± s.e.m.
Figure 3 |
Figure 3 |. Selective Ythdf1 re-expression in the hippocampus rescues defects in memory and synaptic plasticity.
a, Schematics of AAV constructs overexpressing Ythdf1 (AAV-Ythdf1) or control (AAV-control). ITR, inverted terminal repeats; CMV, cytomegalovirus promoter; WPRE, Woodchuck Hepatitis Virus Posttranscriptional Regulatory Element. b, Illustration of bilateral viral injections into the mouse hippocampus. The mouse brain in this figure has been reproduced with permission from the atlas of Paxinos and Franklin 2004.e. c, Representative fluorescence images of the mouse hippocampus after AAV infection, Hoechst, blue; Ythdf1 co-expressed with mCherry, red. d-g, Learning curves in MWM hidden-platform trainings (d), quadrant time (%) in MWM probe tests (e), and contextual (f) and auditory (g) fear memories assessed 24 hours after fear conditioning, of Ythdf1-KO mice injected with AAV-control (red) or AAV-Ythdf1 (blue), compared to uninjected wild-type (WT, green). h-j, Representative traces (h), summary plots (i), and average amplitude (j) of LTP induced by 2 × HFS in acute slices from Ythdf1-KO mice injected with AAV-Ythdf1 or AAV-control. Sample traces (h) were taken at the time points 1 and 2 indicated the summary plots (i). P values, two-way repeated measures ANOVA with post hoc two-tailed t-test (horizontal P values, AAV-Ythdf1 relative to AAV-control; vertical P values, comparisons between curves) (d), two-way ANOVA with post hoc two-tailed t-test (comparison within group or with “Target”) (e), one-way ANOVA with post hoc Fisher test (f, g), and two-tailed t-test (j). Numbers in bars, numbers of mice (e-g) and slices/mice (j). Error bars, mean ± s.e.m.
Figure 4 |
Figure 4 |. Ythdf1 facilitates translation of m6A-modified targets in response to neuronal stimuli.
a, b, Distributions of high-confidence Ythdf1-CLIP peaks in different regions of genome (a) and transcripts (b). c, Functional annotation of Ythdf1-CLIP targets (n = 1,032) in the adult mouse hippocampus. d, Box-plots of the number of m6A-CLIP peaks (left) and the log2 number of m6A-CLIP-seq mutations (right) on m6A-modified transcripts, non-Ythdf1-CLIP transcripts, and Ythdf1-CLIP targets. e, f, Representative images (e) and quantification (f) of nascent protein (Nascent-P) synthesis in cultured control and Ythdf1-KO hippocampal neurons before (sham) and 2 hours after KCl depolarization. Nascent-P signals were normalized to that of control neurons under sham condition. g, Schematics of a tether reporter system that mimics the binding between Ythdf1 and 3’UTR m6A sites of target transcripts. Ythdf1-N, truncated N-terminal mouse Ythdf1 (1–389 aa); F-Luc/R-Luc, firefly/Renilla luciferase. h, Normalized F-Luc reporter expression in cultured hippocampal neurons tethered with Ythdf1-N or control, before (sham) and after KCl depolarization. i, Box-plots of transcript abundance log2 fold change between electroconvulsive treated (ECT) and untreated (Mock) dentate gyrus, for m6A-modified transcripts, m6A-modified non-Ythdf1-CLIP transcripts, and transcripts with overlapped Ythdf1-CLIP and m6A-CLIP peaks, in “Input” (left) and m6A-enriched “RIP” (right) RNA-seq libraries. Dash lines, median log2 fold change of all reliably detected transcripts (rpkm > 1). Box-plot elements: center line, median; box limits, upper and lower quartiles; whiskers, 1–99%; error bars, 95% CI of mean; number in parentheses, number of genes (d, i). P values, two-sided unpaired Kolmogorov-Smirnov test (d, i) and two-tailed t-test (f, h). Numbers in bars, numbers of images/mice (f) and biologically independent samples (h). Error bars, mean ± s.e.m (f, h).

Similar articles

See all similar articles

Cited by 26 articles

See all "Cited by" articles

References

    1. Roundtree IA, Evans ME, Pan T & He C Dynamic RNA modifications in gene expression regulation. Cell 169, 1187–1200 (2017). - PMC - PubMed
    1. Wang X et al. N6-methyladenosine-dependent regulation of messenger RNA stability. Nature 505, 117–20 (2014). - PMC - PubMed
    1. Wang X et al. N6-methyladenosine modulates messenger RNA translation efficiency. Cell 161, 1388–1399 (2015). - PMC - PubMed
    1. Xiao W et al. Nuclear m6A reader YTHDC1 regulates mRNA splicing. Mol. Cell 61, 507–519 (2016). - PubMed
    1. Roundtree IA et al. YTHDC1 mediates nuclear export of N6-methyladenosine methylated mRNAs. Elife 6, e31311 (2017). - PMC - PubMed

Publication types

MeSH terms

Feedback