METTL3 regulates hippocampal gene transcription via N6-methyladenosine methylation in sevoflurane-induced postoperative cognitive dysfunction mouse

The reduced total m6A RNA methylation in a POCD hippocampus

First, physiological parameters including blood glucose, SpO2, and IL-1β levels for the exclusion of probable diseases of respiratory depression, infection, or hypoglycemia were screened in the mice enrolled in this study (Table 1). Thirty mice were treated with 2% sevoflurane to establish a POCD model. Of these mice, seven were identified as having POCD based on the data from Morris water maze tests conducted before and after the sevoflurane treatments. Another twenty mice receiving normal air were considered as negative controls. Given the escaping latent period, the total distance of movement, and number of grids crossed, the POCD mice took a significantly longer time to seek their destination compared with the controls and non-POCD mice, indicating that POCD mice might have sustained a certain extent of substantial hippocampal memory damage (Table 1).

Total hippocampal RNA was used to further detect m6A RNA methylation levels by dot plot assays. We observed a compromised global m6A RNA methylation pattern in POCD compared with the control (F = 5.82, p = 0.017) and non-POCD mice (F = 3.46, p = 0.026) (Figure 1A). Next, RIP-Seq was conducted to investigate the dynamics of m6A RNA methylation to compare the hippocampuses of POCD mice with those of the controls. The clean reads and unique mapping ratio both indicated a high quality of sequencing data (Table 2). Consistently, the genome-scale m6A RNA methylation was significantly lower in the POCD mice than in the controls (p = 3.58E-13) (Figure 1B), which was mainly attributed to the compromised m6A RNA methylation especially at the 5′UTR (p = 7.84E-19) (Figure 1C). Taken together, our data indicated a reduced global m6A methylation in the POCD hippocampus.

The differential m6A RNA methylation especially on the 5′UTR in POCD

Next, we observed that m6A RNA methylation was significantly reduced in 1,244 genes and elevated only in 56 genes in POCD mice compared with the controls (Figure 2A). These genes were involved in central nervous system development and the DNA damage response, NADP (Figure 2B), containing BACE1, BDNF, IL17A, SOX2, and SYN1 (Figure 2C). Transcription of these genes was also validated by qPCR. It was noteworthy that the genes with robustly enriched m6A RNA methylation at the 5′UTR were all highly expressed, while genes with infrequent m6A RNA methylation at the 5′UTR were transcriptionally repressed (Figure 2D). Nevertheless, we failed to find any significant differences in the mRNA levels of CCND2, STAR, and TIGAR, which had differential m6A RNA methylation at the 3′UTR (Figure 2E and 2F). Taken together, we determined that abnormal m6A RNA methylation at the 5′UTR might contribute to transcription profiling in POCD.

The weakened METTL3 affinity for target genes in POCD

Next, we investigated the role of enzymes regulating m6A RNA methylation at the 5′UTR in POCD. Expressions of METTL3, METTL14, WTAP, FTO, ALKBH5, YTHDF1, IGF2BP1, and HNRNPA2B1 showed no significant differences between the POCD mice and the controls (Figure 3A). However, the affinity of METTL3 for target genes BDNF, SOX2, and SYN1 with reduced m6A RNA methylation at the 5′UTR was compromised in the POCD mice compared with the controls (Figure 3B). In turn, it was strange that enrichments of METTL3 on genes BACE1 and IL17A with elevated m6A RNA methylation at the 5′UTR in the POCD mice did not change compared with the controls (Figure 3C). Additionally, the interaction of METTL3 on genes CCND2, STAR, and TIGAR with the 3′UTR did not change in the POCD mice compared with the controls (Figure 3D). We also examined the binding ability of the three enzymes FTO, ALKBH5, and IGF2BP1 for regulating m6A RNA methylation on BDNF, SOX2, and SYN1 by RIP-qPCR, and we failed to find any differences in the 5′UTR (Figure 3E–3G). Collectively, the weakened METTL3 affinity was likely to contribute to the reduction of m6A RNA methylation at the 5′UTR of RNAs in POCD.

METTL3 function suppressed by sevoflurane-induced inactivated MAPK/ERK

Finally, we attempted to bridge the inactivated METTL3 and sevoflurane-induced POCD. Previous studies revealed that the serine phosphorylation of METTL3 played a crucial role in METTL3’s activity and was dependent on the MAPK/ERK signaling pathway [14]. To this end, IP experiments were conducted to capture phosphoserine, and reduced phosphorylated METTL3 was observed in the POCD models’ hippocampuses (Figure 4A). Due to the lack of a commercial phosphorylated METTL3 antibody, we indirectly found less bands of METTL3 over 65 kD in the POCD models’ hippocampuses using a regular METTL3 antibody by WB (Figure 4B), which indicated a potential reduced protein modification affected by sevoflurane. Moreover, the ERK activator TPA was used to treat the other three POCD mice by intracerebroventricular injection. The activity of the MAPK/ERK signaling pathway in hippocampal neurons was abrogated in the POCD mice compared with the controls and rescued by TPA (Figure 4C), suggesting that sevoflurane affected METTL3’s activity through MAPK/ERK inhibition. Expectedly, the target genes BDNF, SOX2, and SYN1 were observed to have highly enriched m6A RNA methylation at their 5′UTR after TPA treatment compared with the POCD mice (Figure 4D).

Our data supported the conclusion that sevoflurane inactivated the MAPK/ERK pathway, which caused dysfunction of METTL3. Fifty elderly mice were pretreated with the ERK activator TPA followed by treatment with sevoflurane to induce POCD. Compared with the fifty mice that were only treated with sevoflurane, the occurrence of POCD in the fifty mice pretreated with TPA was significantly decreased (χ² = 4.762, p = 0.027) (Figure 4E). Taken together, we determined that the activation of the MAPK/ERK pathway was beneficial to METTL3’s functional maintenance for m6A RNA methylation against POCD.

Animal study

A POCD mouse model was established as previously described [10, 11]. In brief, fifty 40-week-old outbred female C57BL/6 mice purchased from the experimental animal center of Xiangnan University were enrolled in this study. Peripheral blood isolated from the caudal vein was used to access the glucose and oxygen saturation (SpO2) levels via a biochemical analyzer (Beckman coulter, Brea, USA) as well as IL-1β levels by ELISA kits (R&D systems, Minneapolis, USA). All the mice were involved in the Morris water maze test by EthoVision XT working system (Noldus, Netherlands) as indicated by the manufacturer’s instructions. The escaping latent period of each mouse was recorded. Thirty mice were randomly selected and given 2% sevoflurane for 4 h in an automatic anesthetic chamber with a size of 24 cm × 12 cm × 18 cm (Rurui Technology Co., Ltd, Guangzhou, China). After natural resuscitation, POCD and non-POCD mice were picked based on their escaping latent periods. The other twenty mice received normal air and were treated as negative controls. Mice were pretreated with 5 mM/kg MAPK/ERK activator phorbol 12-myristate 13-acetate (TPA) (N2060, APExBIO, Houston, TX, USA) for 6 h before sevoflurane induction via intracerebroventricular injection. The mice were sacrificed via cervical dislocation, and their hippocampus were harvested immediately, washed with cold PBS, and RIPA buffer or Trizol Reagent were added accordingly. All the procedures were followed by the Institutional Animal Care and Use Committee of XiangNan University (animal protocol number 2017-32-166).

Dot plot

Dot plot was performed as previously described [20]. In brief, total RNA was isolated by the TRIzol method as described in the subsequent RT-qPCR section of the Materials and Methods. RNA samples dissolved in three times their volumes of RNA incubation buffer were denatured at 65°C within 5 min. Then the samples, divided into subgroups of 400 ng, 200 ng, and 100 ng, were loaded onto an Amersham Hybond-N+ membrane (GE Healthcare, Pittsburgh, PA, USA) installed in a Bio-Dot Apparatus (Bio-Rad, Richmond, CA, USA) with a mixture of ice-cold 20 × SSC buffer (Millipore, Billerica, MA, USA). The membrane was UV cross-linked for 5 min and washed with PBST. Then the membrane was stained with 0.02% methylene blue (Shanghai Sangon Biotechnology Company, Shanghai, China) and scanned to determine the total content of input RNA. After being blocked with 5% nonfat milk, the membrane was incubated with a specific m6A antibody (1:1000, Millipore) overnight at 4°C. Dot blots were hatched with HRP-conjugated anti-mouse immunoglobulin G (IgG) for 1 h before being visualized by an imaging system (Bio-Rad).

RNA immunoprecipitation (RIP) assay

The genome-wide m6A was detected by an RIP kit (Millipore). Hippocampal neuron cells were washed with pre-cooled PBS, and the supernatant was discarded. RIP Lysis Buffer was added, and the products were lysed on ice and collected. Then, the beads were washed with RIP Wash Buffer. After removing the supernatant, the beads were incubated with 1 μg antibodies of m6A (56593S, CST, Beverly, MA, USA), METTL3 (86132, CST), FTO (31687, CST), ALKBH5 (16837-1-AP, Proteintech, Rosemont, IL, USA), and IGF2BP1 (22803-1-AP, Proteintech) for 30 min at room temperature, and normal rabbit immunoglobulin G (Abcam, Cambridge, MA, USA) was used as the control. Subsequently, the cell lysate was incubated with RIP buffer containing magnetic beads. RNA was extracted after proteinase K digestion and used for subsequent sequencing or RT-qPCR assays.

For sequencing, the concentration and quality of purified RIP RNAs were measured by a Nanodrop 2000 (Thermo Fisher Scientific) and Agilent bioanalyzer 2100 (Agilent, Santa Clara, CA, USA). A total of 4 μg RNA in each group was used for library preparation by NEBNext Ultra Directional RNA Library Prep Kit for Illumina (NEB, Ipswich, MA, USA) following the manufacturer’s instructions and sequenced on an Illumina Hiseq platform.

The raw data, deposited into the ArrayExpress database (https://www.ebi.ac.uk/arrayexpress) with the accession number E-MTAB-10510, were adaptors-trimmed, the low quality reads were filtered out using Trimmomatic (non-default parameters: SLIDINGWINDOW:4:15 LEADING:10 TRAILING:10 MINLEN:35) [21], and the quality of the clean reads was checked using Fastqc [22]. Next, the clean reads were aligned to the latest mouse genome assembly 10 mm using Hisat2 v2.0.5 (non-default parameters: --rna-strandness RF --dta) [23]. The transcripts were assembled, and the expression levels were estimated with FPKM values using the StringTie algorithm (non-default parameters: --rf) [24]. Differential mRNA and lncRNA expression among the groups were evaluated using an R package Ballgown [25], and the significance of differences by the Benjamini and Hochberg (BH) p-value adjustment method was computed. Gene annotation was described by the Ensembl genome browser database (http://www.ensembl.org/index.html). The R package ClusterProfiler was used to annotate the differential genes with gene ontology (GO) terms and the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways [26]. The peaks of m6A of certain genes were browsed with an Integrative Genomics Viewer (IGV).

Reverse transcription quantitative PCR (RT-qPCR)

Total RNA was extracted from tissues and cells using TRIzol Reagent (Invitrogen). The total RNA and RNA pulled down from RIP were reversely transcribed into complementary DNA (cDNA) in accordance with the instructions of the Primescript™ RT Reagent Kit (TaKaRa Biotechnology, Dalian, China) and One Step PrimeScript miRNA cDNA Synthesis Kit (TaKaRa Biotechnology), respectively. RT-qPCR was performed on an ABI7500 Quantitative PCR instrument (Applied Biosystems, Foster City, CA, USA) using SYBR^® Premix Ex Taq™ Kit (TaKaRa Biotechnology) under the following conditions: 95°C for 2 min followed by 50 cycles at 95°C for 5 s, 60°C for 10 s, 72°C for 30 s, and 72°C for 10 min. The relative expression of products was calculated by the 2^−ΔΔCt method. The experiment was repeated three times. All primers synthesized by Shanghai Sangon Biotechnology Company are listed in Table 3.

Immunoprecipitation (IP)

The whole cell lysates or nuclear extracts were mixed with 1 μg phosphoserine antibody (ab9332, Abcam) or IgG Rabbit IgG antibody and 40 μl flurry IgA beads (Thermo Fisher Scientific) and rotated overnight at 4°C. Immunoprecipitates were washed by IP buffer (20 mM HEPES [pH 7.9], 350 mM NaCl, 0.1% NP-40, 1 mM DTT, 0.2 mM PMSF, 2 mg/ml leupeptin, and 2 mg/ml aprotinin) and western blotted for the METTL3 antibody.

Western blot

Brain tissue was homogenized in RIPA buffer solution and then centrifuged at 4°C at 13,000 rpm for 10 min. The protein quantity in the supernatant was determined using a BCA protein assay kit (Well-bio, China). Equal amounts of protein samples were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to polyvinylidene fluoride membranes. The membranes were then blocked in 5% nonfat milk and TBS for 90 min and incubated with the respective primary antibodies of METTL3 (1:2000), FTO (1:2000), ALKBH5 (1:2000), IGF2BP1 (1:2000), METTL14 (1:2000, 51104, CST), WTAP (1:2000, 56501, CST), YTHDF1 (1:2000, 17479-1-AP, Proteintech), HNRNPA2B1 (1:2000, 9304, CST), ERK1/2 (1:2000, 67170-1-Ig, Proteintech), p-ERK1/2 (1:500, 80031-1-RR, Proteintech), and GAPDH (1:5000, Beyotime Biotechnology, China) overnight at 4°C. Membranes were washed in TBST and incubated with rabbit anti-mouse and goat anti-rabbit IgG-HRP (1:10000 dilution; 20 ng/ml, Beyotime Biotechnology, China) at room temperature (RT) for 1 h. Membranes were then treated with an enhanced chemiluminescence detection kit (Millipore), and the intensity of each band was quantified by densitometry.

Statistical analysis

All experimental data were processed and analyzed using SPSS 22.0 statistical software (IBM Corp. Armonk, NY, USA). The measurement data were expressed as the mean ± standard deviation. One-way ANOVA was conducted for comparison among multiple groups. Chi-square analysis was conducted to compare POCD occurrence in mice with and without TPA pretreatment. A p-value less than 0.05 was considered statistically significant.