Decreased METTL3 in atrial myocytes promotes atrial fibrillation

Atrial fibrillation, m⁶A modification, Atrial myocytes, Calcium handling

Translational perspective

Our findings suggested that the CREB1/USP9X/METTL3/protocadherin gamma subfamily A, 10 axis in atrial myocytes plays a crucial role in atrial fibrillation (AF) development by influencing calcium handling. These findings provide insights into potential mechanism-based therapeutic targets for AF treatment.

What’s new?

Atrial fibrillation (AF) is associated with decreased methyltransferase like 3 (METTL3) protein in atrial myocytes.
Knockdown of METTL3 in atrial myocytes results in abnormal calcium handling and increased susceptibility to AF.
Protocadherin gamma subfamily A, 10 (PCDHGA10) is regulated by METTL3 and interacts with SR Ca²⁺-ATPase Type 2a (SERCA2a) in atrial myocytes.

Introduction

Atrial fibrillation (AF) is the most common sustained cardiac arrhythmia,¹ significantly increasing the risk of ischaemia stroke and all-cause mortality.^2,3 The development of AF is typically associated with multiple pathophysiological abnormalities, including atrial structural remodelling, electrical remodelling, and calcium-handling disorder in atrial myocytes.⁴ However, the molecular and cellular mechanisms underlying AF remain incompletely understood.⁵

In recent years, research into epigenetic regulatory mechanisms has revealed potential therapeutic targets for AF.⁶ N6-methyladenosine (m⁶A), the most abundant modification of mRNAs in eukaryotes, plays crucial roles in various cardiovascular diseases.⁷ The m⁶A modification is catalysed by the methyltransferase complex composed of methyltransferase like 3 (METTL3), methyltransferase like 14 (METTL14), and Wilms’ tumour 1–associating protein (WTAP), while it is erased by fat-mass and obesity-associated protein (FTO) or ALKB homologue 5 (ALKBH5), and read by proteins such as YTH N6-methyladenosine RNA-binding protein F2 (YTHDF2), YTHDF3, and Hu-antigen R (HuR).⁸ In the fibrotic heart, METTL3 expression has been reported to increase, further promoting cardiac fibroblast proliferation.⁹ However, the expression patterns of METTL3 in fibrillating atrial myocytes and its effects on calcium handling, electrophysiology and AF development are still unclear.

Here, we found that METTL3 and m⁶A modification levels were down-regulated in the left atrial appendages (LAAs) of patients with AF and experimental AF models. Genetic inhibition of METTL3 promoted AF development in mice by affecting calcium handling of atrial myocytes. Our findings may identify METTL3 as a potential therapeutic target for AF, with important clinical implications.

Methods

Methods are described in full detail in the Supplementary material.

Results

Patient characteristics

The clinical characteristics of patients with sinus rhythm (SR, n = 19) and persistent AF (n = 19), who were used for dot blot, reverse transcriptase–quantitative polymerase chain reaction (RT–qPCR), western blotting, and immunofluorescence analyses, are summarized in Table 1. There were no significant differences between the two groups in terms of age, gender, or comorbidities, such as hypertension and coronary artery disease. However, patients with AF had a larger left atrium. Since these patients underwent a radiofrequency maze procedure, their cardiopulmonary bypass time was significantly longer.

Table 1

Open in new tab

The basic clinical information of included patients

		SR (n = 19)	AF (n = 19)	P-value
Age, years (mean ± SD)		54.58 ± 11.29	60.32 ± 8.83	0.09^a
Male (n, %)		13, 68.4%	11, 57.9%	0.501^b
BMI (mean ± SD)		24.24 ± 3.96	23.96 ± 2.90	0.812^a
Hypertension (n, %)		6, 31.0%	5, 26.3%	0.721^b
Diabetes (n, %)		1, 5.3%	1, 5.3%	1^b
CAD (n, %)		4, 21.1%	2, 10.5%	0.374^b
Smoke (n, %)		3, 15.8%	2, 10.5%	0.631^b
Drink (n, %)		1, 5.3%	2, 10.5%	0.547^b
LAD, cm (mean ± SD)		4.18 ± 0.60	4.86 ± 0.69	0.006^a
EF, % (mean ± SD)		56.09 ± 8.66	55.56 ± 4.19	0.817^a
Blood transfusion, mL [median (IQRs)]		25 (0–612.5)	0 (0–300)	0.205^c
Operation time, min [median (IQRs)]		220 (168.75–242.50)	250 (210.00–295.00)	0.086^b
CPB time, min [median (IQRs)]		108.5 (77.25–144.25)	132 (127.00–158.00)	0.045^c
NYHA	Ⅰ (n, %)	2, 10.5%	2, 10.5%	0.455^b
	Ⅱ (n, %)	9, 47.4%	5, 26.3%
	Ⅲ (n, %)	8, 42.1%	11, 57.9%
	Ⅳ (n, %)	0, 0%	1, 5.3%
CRP, mg/L [median (IQRs)]		3.1 (2.175– 4.125)	2.9 (2.450–4.600)	0.13^c
Hb, g/L [median (IQRs)]		129.5 (114.25–145.00)	147.0 (133.00–154.00)	0.307^c
ALT, U/L [median (IQRs)]		18.8 (14.050–30.575)	19.1 (14.375–51.625)	0.353^c
TBA, μmol/L [median (IQRs)]		2.95 (2.550–5.275)	4.45 (2.025–7.500)	0.702^c
Scr, μmol/L [median (IQRs)]		66 (55.50–77.75)	57 (55.25–69.50)	0.87^c
BUN, mmol/L [median (IQRs)]		6.4 (5.425–6.925)	5.75 (5.425–6.400)	0.382^c
BNP, pg/mL [median (IQRs)]		115.5 (30.35–247.00)	175.0 (66.00–270.25)	0.237^c

		SR (n = 19)	AF (n = 19)	P-value
Age, years (mean ± SD)		54.58 ± 11.29	60.32 ± 8.83	0.09^a
Male (n, %)		13, 68.4%	11, 57.9%	0.501^b
BMI (mean ± SD)		24.24 ± 3.96	23.96 ± 2.90	0.812^a
Hypertension (n, %)		6, 31.0%	5, 26.3%	0.721^b
Diabetes (n, %)		1, 5.3%	1, 5.3%	1^b
CAD (n, %)		4, 21.1%	2, 10.5%	0.374^b
Smoke (n, %)		3, 15.8%	2, 10.5%	0.631^b
Drink (n, %)		1, 5.3%	2, 10.5%	0.547^b
LAD, cm (mean ± SD)		4.18 ± 0.60	4.86 ± 0.69	0.006^a
EF, % (mean ± SD)		56.09 ± 8.66	55.56 ± 4.19	0.817^a
Blood transfusion, mL [median (IQRs)]		25 (0–612.5)	0 (0–300)	0.205^c
Operation time, min [median (IQRs)]		220 (168.75–242.50)	250 (210.00–295.00)	0.086^b
CPB time, min [median (IQRs)]		108.5 (77.25–144.25)	132 (127.00–158.00)	0.045^c
NYHA	Ⅰ (n, %)	2, 10.5%	2, 10.5%	0.455^b
	Ⅱ (n, %)	9, 47.4%	5, 26.3%
	Ⅲ (n, %)	8, 42.1%	11, 57.9%
	Ⅳ (n, %)	0, 0%	1, 5.3%
CRP, mg/L [median (IQRs)]		3.1 (2.175– 4.125)	2.9 (2.450–4.600)	0.13^c
Hb, g/L [median (IQRs)]		129.5 (114.25–145.00)	147.0 (133.00–154.00)	0.307^c
ALT, U/L [median (IQRs)]		18.8 (14.050–30.575)	19.1 (14.375–51.625)	0.353^c
TBA, μmol/L [median (IQRs)]		2.95 (2.550–5.275)	4.45 (2.025–7.500)	0.702^c
Scr, μmol/L [median (IQRs)]		66 (55.50–77.75)	57 (55.25–69.50)	0.87^c
BUN, mmol/L [median (IQRs)]		6.4 (5.425–6.925)	5.75 (5.425–6.400)	0.382^c
BNP, pg/mL [median (IQRs)]		115.5 (30.35–247.00)	175.0 (66.00–270.25)	0.237^c

Italics indicate that the P < 0.05, indicating a significant difference.

ALT, alanine transaminase; BMI, body mass index; BUN, blood urea nitrogen; CAD, coronary heart disease; CPB, cardiopulmonary bypass; CRP, C-reactive protein; EF, ejection fraction; Hb, haemoglobin; IQRs, 25th–75th percentile; LAD, left atrial diameter; Scr, serum creatinine; TBA, total bile acid.

^aComparisons were made using unpaired Student’s t-test

^bComparisons were made using the χ² test or the Fisher’s test.

^cComparisons were made using non-parametric Mann–Whitney test.

Table 1

Open in new tab

The basic clinical information of included patients

		SR (n = 19)	AF (n = 19)	P-value
Age, years (mean ± SD)		54.58 ± 11.29	60.32 ± 8.83	0.09^a
Male (n, %)		13, 68.4%	11, 57.9%	0.501^b
BMI (mean ± SD)		24.24 ± 3.96	23.96 ± 2.90	0.812^a
Hypertension (n, %)		6, 31.0%	5, 26.3%	0.721^b
Diabetes (n, %)		1, 5.3%	1, 5.3%	1^b
CAD (n, %)		4, 21.1%	2, 10.5%	0.374^b
Smoke (n, %)		3, 15.8%	2, 10.5%	0.631^b
Drink (n, %)		1, 5.3%	2, 10.5%	0.547^b
LAD, cm (mean ± SD)		4.18 ± 0.60	4.86 ± 0.69	0.006^a
EF, % (mean ± SD)		56.09 ± 8.66	55.56 ± 4.19	0.817^a
Blood transfusion, mL [median (IQRs)]		25 (0–612.5)	0 (0–300)	0.205^c
Operation time, min [median (IQRs)]		220 (168.75–242.50)	250 (210.00–295.00)	0.086^b
CPB time, min [median (IQRs)]		108.5 (77.25–144.25)	132 (127.00–158.00)	0.045^c
NYHA	Ⅰ (n, %)	2, 10.5%	2, 10.5%	0.455^b
	Ⅱ (n, %)	9, 47.4%	5, 26.3%
	Ⅲ (n, %)	8, 42.1%	11, 57.9%
	Ⅳ (n, %)	0, 0%	1, 5.3%
CRP, mg/L [median (IQRs)]		3.1 (2.175– 4.125)	2.9 (2.450–4.600)	0.13^c
Hb, g/L [median (IQRs)]		129.5 (114.25–145.00)	147.0 (133.00–154.00)	0.307^c
ALT, U/L [median (IQRs)]		18.8 (14.050–30.575)	19.1 (14.375–51.625)	0.353^c
TBA, μmol/L [median (IQRs)]		2.95 (2.550–5.275)	4.45 (2.025–7.500)	0.702^c
Scr, μmol/L [median (IQRs)]		66 (55.50–77.75)	57 (55.25–69.50)	0.87^c
BUN, mmol/L [median (IQRs)]		6.4 (5.425–6.925)	5.75 (5.425–6.400)	0.382^c
BNP, pg/mL [median (IQRs)]		115.5 (30.35–247.00)	175.0 (66.00–270.25)	0.237^c

		SR (n = 19)	AF (n = 19)	P-value
Age, years (mean ± SD)		54.58 ± 11.29	60.32 ± 8.83	0.09^a
Male (n, %)		13, 68.4%	11, 57.9%	0.501^b
BMI (mean ± SD)		24.24 ± 3.96	23.96 ± 2.90	0.812^a
Hypertension (n, %)		6, 31.0%	5, 26.3%	0.721^b
Diabetes (n, %)		1, 5.3%	1, 5.3%	1^b
CAD (n, %)		4, 21.1%	2, 10.5%	0.374^b
Smoke (n, %)		3, 15.8%	2, 10.5%	0.631^b
Drink (n, %)		1, 5.3%	2, 10.5%	0.547^b
LAD, cm (mean ± SD)		4.18 ± 0.60	4.86 ± 0.69	0.006^a
EF, % (mean ± SD)		56.09 ± 8.66	55.56 ± 4.19	0.817^a
Blood transfusion, mL [median (IQRs)]		25 (0–612.5)	0 (0–300)	0.205^c
Operation time, min [median (IQRs)]		220 (168.75–242.50)	250 (210.00–295.00)	0.086^b
CPB time, min [median (IQRs)]		108.5 (77.25–144.25)	132 (127.00–158.00)	0.045^c
NYHA	Ⅰ (n, %)	2, 10.5%	2, 10.5%	0.455^b
	Ⅱ (n, %)	9, 47.4%	5, 26.3%
	Ⅲ (n, %)	8, 42.1%	11, 57.9%
	Ⅳ (n, %)	0, 0%	1, 5.3%
CRP, mg/L [median (IQRs)]		3.1 (2.175– 4.125)	2.9 (2.450–4.600)	0.13^c
Hb, g/L [median (IQRs)]		129.5 (114.25–145.00)	147.0 (133.00–154.00)	0.307^c
ALT, U/L [median (IQRs)]		18.8 (14.050–30.575)	19.1 (14.375–51.625)	0.353^c
TBA, μmol/L [median (IQRs)]		2.95 (2.550–5.275)	4.45 (2.025–7.500)	0.702^c
Scr, μmol/L [median (IQRs)]		66 (55.50–77.75)	57 (55.25–69.50)	0.87^c
BUN, mmol/L [median (IQRs)]		6.4 (5.425–6.925)	5.75 (5.425–6.400)	0.382^c
BNP, pg/mL [median (IQRs)]		115.5 (30.35–247.00)	175.0 (66.00–270.25)	0.237^c

Italics indicate that the P < 0.05, indicating a significant difference.

^aComparisons were made using unpaired Student’s t-test

^bComparisons were made using the χ² test or the Fisher’s test.

^cComparisons were made using non-parametric Mann–Whitney test.

RNA m⁶A modification and METTL3 protein levels were down-regulated in the left atrial appendages of atrial fibrillation patients and the experimental atrial fibrillation models

To investigate the relationship between m⁶A modification and AF, we first performed dot blot analysis on patient samples. The result showed that the m⁶A modification level of total RNA was significantly decreased in the LAAs of patients with AF (Figure 1A). HL-1 cells were subjected to tachypacing at indicate frequencies, which caused DNA damage (see Supplementary material online, Figure S1A), and were commonly used as the atrial myocytes model of AF.^10,11 Dot blot and immunofluorescence results showed that the m⁶A modification of HL-1 cells progressively decreased with increasing stimulation frequency (Figure 1B and see Supplementary material online, Figure S1B).

M6a modification and METTL3 were down-regulated in AF. (A) Dot blot analysis was used to assess the m6A modification levels in total RNA (500 ng) from LAA tissues of patients with SR (n = 15) and AF (n = 18). Data are presented as mean ± SD. ***P < 0.001, comparison made using an unpaired Student’s t-test. m6A, anti-m6A antibody; MB, methylene blue staining. (B) Dot blot analysis of m6A modification levels in total RNA from HL-1 cells at varying stimulation frequencies, with loading amounts of 250 ng (up) and 125 ng (down). The graph below shows the statistical analysis results at RNA loading amount of 250 ng. Data are presented as mean ± SD. 0 Hz (n = 3), 5 Hz (n = 3), and 10 Hz (n = 3). *P < 0.05, **P < 0.01, comparisons made using ordinary one-way ANOVA. (C) Protein expression of methyltransferases METTL3, METTL14, WTAP, and demethylase FTO, ALKBH5 in LAA tissue of patients with AF. Data are presented as mean ± SD. Sinus rhythm (n = 10–19), AF (n = 9–19). *P < 0.05, ***P < 0.001, comparisons made using unpaired Student’s t-test. (D) Protein expression of methyltransferases METTL3, METTL14, WTAP, and demethylase FTO, ALKBH5 in HL-1 cells at different stimulation frequencies. Data are presented as mean ± SD. 0 Hz (n = 3), 5 Hz (n = 3), and 10 Hz (n = 3). *P < 0.05, ***P < 0.001, comparisons made using ordinary one-way ANOVA. (E) METTL3 mRNA levels were detected by RT–qPCR in patients with AF. Data are presented as mean ± SD. Sinus rhythm (n = 19), AF (n = 19), comparison made using unpaired Student’s t-test; (F) METTL3 mRNA levels in HL-1 cells at certain stimulation frequencies. Data are presented as mean ± SD. 0 Hz (n = 3), 5 Hz (n = 3), and 10 Hz (n = 3), comparison made using ordinary one-way ANOVA; (G) Endogenous immunoprecipitation (IP) performed on 0 or 10 Hz paced HL-1 cells using METTL3 antibody or negative control rabbit IgG. Western blotting was then performed using the corresponding antibodies; Co-IP/western blotting experiments revealed that 10 Hz stimulation significantly increased the ubiquitination level of METTL3. AF, atrial fibrillation; ANOVA, analysis of variance; LAA, left atrial appendages; METTL3, methyltransferase like 3; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; ns, not significant; RT–qPCR, reverse transcriptase–quantitative polymerase chain reaction; SD, standard deviation; SR, sinus rhythm.

Figure 1

M6a modification and METTL3 were down-regulated in AF. (A) Dot blot analysis was used to assess the m⁶A modification levels in total RNA (500 ng) from LAA tissues of patients with SR (n = 15) and AF (n = 18). Data are presented as mean ± SD. ***P < 0.001, comparison made using an unpaired Student’s t-test. m⁶A, anti-m⁶A antibody; MB, methylene blue staining. (B) Dot blot analysis of m⁶A modification levels in total RNA from HL-1 cells at varying stimulation frequencies, with loading amounts of 250 ng (up) and 125 ng (down). The graph below shows the statistical analysis results at RNA loading amount of 250 ng. Data are presented as mean ± SD. 0 Hz (n = 3), 5 Hz (n = 3), and 10 Hz (n = 3). *P < 0.05, **P < 0.01, comparisons made using ordinary one-way ANOVA. (C) Protein expression of methyltransferases METTL3, METTL14, WTAP, and demethylase FTO, ALKBH5 in LAA tissue of patients with AF. Data are presented as mean ± SD. Sinus rhythm (n = 10–19), AF (n = 9–19). *P < 0.05, ***P < 0.001, comparisons made using unpaired Student’s t-test. (D) Protein expression of methyltransferases METTL3, METTL14, WTAP, and demethylase FTO, ALKBH5 in HL-1 cells at different stimulation frequencies. Data are presented as mean ± SD. 0 Hz (n = 3), 5 Hz (n = 3), and 10 Hz (n = 3). *P < 0.05, ***P < 0.001, comparisons made using ordinary one-way ANOVA. (E) METTL3 mRNA levels were detected by RT–qPCR in patients with AF. Data are presented as mean ± SD. Sinus rhythm (n = 19), AF (n = 19), comparison made using unpaired Student’s t-test; (F) METTL3 mRNA levels in HL-1 cells at certain stimulation frequencies. Data are presented as mean ± SD. 0 Hz (n = 3), 5 Hz (n = 3), and 10 Hz (n = 3), comparison made using ordinary one-way ANOVA; (G) Endogenous immunoprecipitation (IP) performed on 0 or 10 Hz paced HL-1 cells using METTL3 antibody or negative control rabbit IgG. Western blotting was then performed using the corresponding antibodies; Co-IP/western blotting experiments revealed that 10 Hz stimulation significantly increased the ubiquitination level of METTL3. AF, atrial fibrillation; ANOVA, analysis of variance; LAA, left atrial appendages; METTL3, methyltransferase like 3; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; ns, not significant; RT–qPCR, reverse transcriptase–quantitative polymerase chain reaction; SD, standard deviation; SR, sinus rhythm.

Given that m⁶A modification is catalysed by METTL3–METTL14–WTAP complex and demethylated by FTO or ALKBH5, we assessed the protein levels of these enzymes in the LAAs of patients with AF and in cell models via western blotting. The results revealed that only the protein levels of METTL3 and METTL14 were significantly down-regulated in the LAAs of patients with AF (Figure 1C). In AF cell models, only the METTL3 protein level was reduced (Figure 1D). These findings were confirmed by immunofluorescence analyses (see Supplementary material online, Figure S1C and D). In the left atria of mice with CaCl₂-acetylcholine (ACh)-induced AF, METTL3 expression was also significantly decreased (see Supplementary material online, Figure S1E).

Subsequently, we focused on the role of METTL3 in atrial myocytes in the context of AF. Reverse transcriptase–qPCR analysis showed no significant change in METTL3 mRNA levels in AF patient samples or cell models (Figure 1E and F), leading us to hypothesize that METTL3 protein degradation is accelerated during AF progression. To determine whether METTL3 degradation in atrial cardiomyocytes primarily relies on the proteasomal or lysosomal pathway, we treated HL-1 cells with the proteasome inhibitor MG132 or the lysosomal inhibitor chloroquine. The results demonstrated that METTL3 accumulated in cells treated with MG132 but not in those treated with chloroquine (see Supplementary material online, Figure S1F), suggesting that METTL3 in atrial myocytes was predominantly degraded via the proteasomal pathway. Additionally, Co-immunoprecipitation/western blotting experiments showed that the ubiquitination level of METTL3 was significantly increased in AF cell models (Figure 1G). These findings suggest that the reduction of METTL3 and its mediated m⁶A modification in atrial myocytes is closely linked to the onset of AF.

METTL3 deletion promoted CaCl₂-acetylcholine-induced atrial fibrillation and spontaneous atrial fibrillation in CREM mice

To investigate the impact of METTL3 on AF, we induced AF by injecting CaCl₂-ACh into the tail vein of wild-type (WT) and Mettl3^+/− mice for 2 weeks (see Supplementary material online, Figure S2A).^12–14 Genotype identification of WT and Mettl3^+/− mice was confirmed via PCR analysis (see Supplementary material online, Figure S2B). Compared with WT mice, both the mRNA and protein levels of METTL3 in the atria of Mettl3^+/− mice were significantly reduced (Figure 2A and B). Electrocardiogram (ECG) results showed that heterozygous deletion of METTL3 significantly increased the duration of CaCl₂-ACh-induced AF (Figure 2C).

METTL3 deletion promoted CaCl2-ACh-induced AF and spontaneous AF in CREM mice. (A) RT–qPCR was used to detect the expression of METTL3 mRNA in the atrium of Mettl3+/− mice. Data are presented as mean ± SD. WT (n = 4), Mettl3+/− (n = 6). **P < 0.01, comparison made using unpaired Student’s t-test. (B) Western blotting was used to detect expression of METTL3 protein in the atrium of Mettl3+/− mice. Data are presented as mean ± SD. WT (n = 3); Mettl3+/− (n = 3). ***P < 0.001, comparison made using unpaired Student’s t-test. (C) Duration of induced AF in each group of mice following CaCl2-ACh injection (left). The representative ECG of the limb leads from each group of mice (right). WT; control (n = 5): WT mice injected with saline, Mettl3+/−; control (n = 5): Mettl3+/− mice injected with saline, WT; AF (n = 6): WT mice injected with CaCl2-ACh, Mettl3+/−; AF (n = 3): Mettl3+/− mice injected with CaCl2-ACh. Data are presented as mean ± SD. **P < 0.01, ****P < 0.0001, comparison made using two-way ANOVA. (D) Schematic diagram illustrating the process of specific METTL3 knockout in atrial myocytes of CREM mice. Parental Mettl3flox/flox mice were crossed with CREM mice to generate Mettl3flox/+:CREM mice, which were then bred with Mettl3flox/flox mice to obtain Mettl3flox/flox:CREM mice. These mice were then injected with AAV9-Anf-Cre virus to specifically knock out METTL3 in atrial myocytes. The Mettl3flox/flox:CREM mice injected with AAV9-Anf-Zsgreen virus served as controls. (E) Expression of METTL3 protein in the atrium, ventricle, and liver of Mettl3flox/flox:CREM mice was evaluated by western blotting assay 4 weeks after AAV9-Anf virus injection. Data are represent as mean ± SD. AAV9-Anf-Zsgreen (n = 3), AAV9-Anf-Cre (n = 3). ***P < 0.001, comparisons made using unpaired Student’s t-test. (F) Statistics of spontaneous AF in Mettl3flox/flox:CREM mice 4 weeks after AAV9-Anf virus injection (left). Schematic diagram of limb lead ECG (right). Data represent the number of mice with or without sAF. AAV9-Anf-Zsgreen (n = 10), AAV9-Anf-Cre (n = 9). *P < 0.05, comparison made using the Fisher’s test. AF, atrial fibrillation; ANOVA, analysis of variance; CREM, cAMP-responsive element modulator; ECG, electrocardiogram; METTL3, methyltransferase like 3; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; ns, not significant; sAF, spontaneous atrial fibrillation; SD, standard deviation; WT, wild-type.

Figure 2

METTL3 deletion promoted CaCl₂-ACh-induced AF and spontaneous AF in CREM mice. (A) RT–qPCR was used to detect the expression of METTL3 mRNA in the atrium of Mettl3^+/− mice. Data are presented as mean ± SD. WT (n = 4), Mettl3^+/− (n = 6). **P < 0.01, comparison made using unpaired Student’s t-test. (B) Western blotting was used to detect expression of METTL3 protein in the atrium of Mettl3^+/− mice. Data are presented as mean ± SD. WT (n = 3); Mettl3^+/− (n = 3). ***P < 0.001, comparison made using unpaired Student’s t-test. (C) Duration of induced AF in each group of mice following CaCl₂-ACh injection (left). The representative ECG of the limb leads from each group of mice (right). WT; control (n = 5): WT mice injected with saline, Mettl3^+/−; control (n = 5): Mettl3^+/− mice injected with saline, WT; AF (n = 6): WT mice injected with CaCl₂-ACh, Mettl3^+/−; AF (n = 3): Mettl3^+/− mice injected with CaCl₂-ACh. Data are presented as mean ± SD. **P < 0.01, ****P < 0.0001, comparison made using two-way ANOVA. (D) Schematic diagram illustrating the process of specific METTL3 knockout in atrial myocytes of CREM mice. Parental Mettl3^flox/flox mice were crossed with CREM mice to generate Mettl3^flox/+:CREM mice, which were then bred with Mettl3^flox/flox mice to obtain Mettl3^flox/flox:CREM mice. These mice were then injected with AAV9-Anf-Cre virus to specifically knock out METTL3 in atrial myocytes. The Mettl3^flox/flox:CREM mice injected with AAV9-Anf-Zsgreen virus served as controls. (E) Expression of METTL3 protein in the atrium, ventricle, and liver of Mettl3^flox/flox:CREM mice was evaluated by western blotting assay 4 weeks after AAV9-Anf virus injection. Data are represent as mean ± SD. AAV9-Anf-Zsgreen (n = 3), AAV9-Anf-Cre (n = 3). ***P < 0.001, comparisons made using unpaired Student’s t-test. (F) Statistics of spontaneous AF in Mettl3^flox/flox:CREM mice 4 weeks after AAV9-Anf virus injection (left). Schematic diagram of limb lead ECG (right). Data represent the number of mice with or without sAF. AAV9-Anf-Zsgreen (n = 10), AAV9-Anf-Cre (n = 9). *P < 0.05, comparison made using the Fisher’s test. AF, atrial fibrillation; ANOVA, analysis of variance; CREM, cAMP-responsive element modulator; ECG, electrocardiogram; METTL3, methyltransferase like 3; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; ns, not significant; sAF, spontaneous atrial fibrillation; SD, standard deviation; WT, wild-type.

With cardiomyocyte-specific expression of CREM-IbΔC-X (an isoform of CREM in human), CREM mice are a widely used transgenic mouse model of AF.^15–17 These mice progressively develop atrial dilation, atrial fibrosis, and spontaneous AF with age (see Supplementary material online, Figure S3). To test whether loss of METTL3 in atrial myocytes promotes the development of spontaneous AF, we generated Mettl3^flox/flox:CREM double-mutant (DM) mice and injected them with AAV9-Anf-Zsgreen or AAV9-Anf-Cre virus¹⁸ (Figure 2D). Genotyping of DM mice was confirmed via PCR analysis (see Supplementary material online, Figure S4A and B). Atrial-specific expression of the virus was validated by the presence of fluorescence signals in the atrium but not the ventricle of Tg-tdTomato mice 4 weeks after AAV9-Anf-Cre virus injection (see Supplementary material online, Figure S4C). Four weeks post-injection in DM mice, METTL3 protein levels were significantly reduced in the atrium but not in the ventricle and liver (Figure 2E). Electrocardiogram monitoring revealed that atrial-specific deletion of METTL3 in CREM mice increased the occurrence of spontaneous AF within 8 weeks (Figure 2F). These findings suggested that both global and atrial myocyte-specific deletion of METTL3 promotes AF development.

Neither atrial myocyte-specific deletion nor global heterozygous deletion of METTL3 caused atrial structural remodelling such as atrial enlargement or fibrosis (see Supplementary material online, Figure S5). Since ACh activates the ACh-sensitive potassium channels (I_K,ACh) of atrial myocytes, shortening action potential duration and promoting AF,^19,20 we examined the mRNA expression of KCNJ3 and KCNJ5 (potassium inwardly rectifying channel Subfamily J Members 3 and 5, the subunits of potassium channels) in the atrial tissues of Mettl3^+/− mice via RT–qPCR. The results showed that KCNJ5 mRNA was significantly increased, while KCNJ3 mRNA was unchanged (Figure 3A). Western blotting confirmed that KCNJ5 protein was significantly elevated in the right atrial appendage (Figure 3B), a finding consistent with results from AF patient samples (Figure 3C). Additionally, KCNJ5 expression was also significantly up-regulated in AF cell models (Figure 3D and E), consistent with the results of METTL3 knockdown (see Supplementary material online, Figure S6A). In vitro electrophysiological experiments showed that knockdown of METTL3 increased ACh-sensitive potassium current (I_K,ACh) (see Supplementary material online, Figure S6B and C).

Figure 3

METTL3 deficiency increased KCNJ5 expression in the right atrium. (A) RT–qPCR was used to evaluate expression of KCNJ3 and KCNJ5 mRNA in bilateral atria of Mettl3^+/− mice. Data are represented as mean ± SD. WT (n = 4), Mettl3^+/− (n = 6). **P < 0.01, comparisons made using unpaired Student’s t-test. (B) Western blotting was used to evaluate KCNJ5 protein expression in bilateral atria of Mettl3^+/− mice. Data are represented as mean ± SD. WT left atrium (LA) (n = 5), Mettl3^+/− LA (n = 5), WT right atrium (RA) (n = 8), Mettl3^+/− RA (n = 8). **P < 0.01, comparisons made using unpaired Student’s t-test. (C) KCNJ5 protein expression levels were detected by western blotting in bilateral atrial appendages from patients with AF and SR. Data are represented as mean ± SD. SR (n = 6), AF (n = 6). **P < 0.01, comparisons made using unpaired Student’s t-test; (D) KCNJ5 mRNA expression levels were detected by RT–qPCR at certain stimulation frequencies in AF cell models. Data are represented as mean ± SD. 0 Hz (n = 5), 5 Hz (n = 5), and 10 Hz (n = 5). *P < 0.05, ****P < 0.0001, comparison made using ordinary one-way ANOVA. (E) KCNJ5 protein expression levels were detected at certain stimulation frequencies in AF cell models. Data are represented as mean ± SD. 0 Hz (n = 3), 5 Hz (n = 3), and 10 Hz (n = 3). *P < 0.05, **P < 0.01, comparisons made using ordinary one-way ANOVA. AF, atrial fibrillation; ANOVA, analysis of variance; LAA, left atrial appendages; METTL3, methyltransferase like 3; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; ns, not significant; RAA, right atrial appendage; RT–qPCR, reverse transcriptase–quantitative polymerase chain reaction; SD, standard deviation; SR, sinus rhythm; WT, wild type.

Knockdown of METTL3 resulted in abnormal calcium handling in atrial myocytes

In addition to changes in ion channels expression, calcium handling in atrial myocytes is also involved in the pathogenesis of AF.¹ To investigate whether METTL3 affects calcium handling, we knocked down METTL3 in primary neonatal mice atrial myocytes (NMAMs) (Figure 4A and B). This resulted in a reduction of m⁶A modification levels in the NMAMs (Figure 4C). Calcium transient measurement in these atrial myocytes, paced at 1 Hz, showed that METTL3 knockdown increased amplitude and shortened the decay time of the calcium transients (Figure 4D).

Figure 4

METTL3 deficiency resulted in accelerated decay of calcium transients in atrial myocytes. (A) Expression of METTL3 mRNA was determined by RT–qPCR in NMAMs infected with knockdown lentivirus. Data are represent as mean ± SD. Sh-Ctrl (n = 6), sh-Mettl3 (n = 4). **P < 0.01, comparisons made using unpaired Student’s t-test. (B) Expression of METTL3 protein was determined by western blotting in NMAMs infected with knockdown lentivirus. Data are represented as mean ± SD. Sh-Ctrl (n = 3), sh-Mettl3 (n = 3). **P < 0.01, comparisons made using unpaired Student’s t-test. (C) N6-methyladenosine modification levels were detected by dot blot assay in METTL3 knockdown NMAMs. (D) Calcium transients were measured in METTL3 knockdown NMAMs using confocal microscopy. Data are represent as mean ± SD. Sh-Ctrl (n = 42 cells from three technical replicates), sh-Mettl3 (n = 48 cells from three technical replicates). **P < 0.01, ****P < 0.0001, comparisons made using unpaired Student’s t-test. (E) Calcium transients in AMAMs were detected by the confocal microscopy. Data are represented as mean ± SD. WT (n = 15 cells from three mice), Mettl3^+/− (n = 11 cells from three mice). **P < 0.01, comparisons made using non-parametric Mann–Whitney test (for amplitude) and unpaired Student’s t-test (for decay time). AMAMs, adult mice atrial myocytes; m⁶A, N6-methyladenosine; MB, methylene blue staining; METTL3, methyltransferase like 3; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; NMAMs, neonatal mice atrial myocytes; RT–qPCR, reverse transcriptase–quantitative polymerase chain reaction; SD, standard deviation; WT, wild type.

Next, we isolated primary adult mice atrial myocytes (AMAMs) from WT and Mettl3^+/− mice using Langendorff perfusion and performed calcium transient analysis. In AMAMs, the decay of calcium transients was also significantly accelerated (Figure 4E). However, key calcium-handling proteins, including NCX1, SR Ca²⁺-ATPase Type 2a (SERCA2a), phospholamban (PLN), and the phosphorylation of PLN at pS16 and pT17, did not exhibit significant changes in METTL3 knockdown HL-1 cells or the atrial tissues of Mettl3^+/− mice (see Supplementary material online, Figure S7).

METTL3 affects calcium handling in atrial myocytes through protocadherin gamma subfamily A, 10 via an m⁶A-HuR-mediated pathway

METTL3 catalyses the m⁶A modification of target mRNA, affecting their splicing, maturation, nuclear export, translation, and stability, thereby influencing target genes expression.²¹ To identify the target genes regulated by METTL3 during AF, we performed methylated RNA immunoprecipitation sequencing (MeRIP-Seq) in the atrial myocyte model of AF. We focused on genes that showed reduced m⁶A modification and altered transcription levels. Cross-analysis of m⁶A modification and transcription levels revealed that m⁶A modification levels of Gm6842, 9330104G04Rik, and protocadherin gamma subfamily A, 10 (PCDHGA10) were significantly decreased, while their transcription levels were significantly increased (Figure 5A and Supplementary material online, Appendix S2). Since Gm6842 is a pseudogene and 9330104G04Rik is a non-coding RNA with no human homologue, we focused on PCDHGA10 for further investigation.

PCDHGA10 was regulated by METTL3. (A) MeRIP-Seq in the AF cell model showed reduced m6A modification and increased transcript levels of PCDHGA10. Ten Hertz paced HL-1 cells served as the experimental group (n = 3), and HL-1 cells without pacing served as the control group (n = 3). (B) MeRIP-qPCR was used to validate that m6A modifications on PCDHGA10 mRNA were regulated by METTL3. Fragments of RNA from sh-Ctrl or sh-Mettl3 HL-1 cells were enriched with m6A or rabbit IgG antibody, followed by reverse transcription and qPCR assay. Data are represented as mean ± SD. Sh-Ctrl (n = 4), sh-Mettl3 (n = 3). *P < 0.05, comparison made using two-way ANOVA. (C) PCDHGA10 mRNA expression was evaluated by RT–qPCR in sh-Mettl3 or OE-Mettl3 HL-1 cells. Data are represented as mean ± SD. Sh-Ctrl (n = 4), sh-Mettl3 (n = 4), OE-Ctrl (n = 4), and OE-Mettl3 (n = 4). **P < 0.01, ***P < 0.001, comparisons made using unpaired Student’s t-test. (D) Protocadherin gamma subfamily A, 10 protein expression was evaluated by western blotting in sh-Mettl3 or OE-Mettl3 HL-1 cells. Data are represented as mean ± SD. Sh-Ctrl (n = 3), sh-Mettl3 (n = 3), OE-Ctrl (n = 3), and OE-Mettl3 (n = 3). **P < 0.01, comparisons made using unpaired Student’s t-test. (E) Protein expression of PCDHGA10 was evaluated by western blotting in LAA tissue of patients with AF. Data are represented as mean ± SD. SR (n = 17), AF (n = 18). *P < 0.05, comparison made using non-parametric Mann–Whitney test. (F) Ythdf2, Ythdf3, and Pcdhga10 mRNA were evaluated by RT–qPCR in sh-Ythdf2 or sh-Ythdf3 HL-1 cells. Data are represented as mean ± SD. Sh-Ctrl (n = 3), sh-Ythdf2 (n = 3), and sh-Ythdf3 (n = 3). ***P < 0.001, ****P < 0.0001, comparisons made using unpaired Student’s t-test. (G) HuR and Pcdhga10 mRNA were evaluated by RT–qPCR in si-HuR HL-1 cells. si-Ctrl (n = 3), si1-HuR (n = 3), and si2-HuR (n = 3). *P < 0.05, ***P < 0.001, comparisons made using ordinary one-way ANOVA. (H) RIP-PCR was used to detect the enrichment of PCDHGA10 mRNA by HuR protein under the condition of METTL3 knockdown in HL-1 cells. AF, atrial fibrillation; ANOVA, analysis of variance; FC, fold change; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; LAA, left atrial appendage; m6A, N6-methyladenosine; MeRIP-Seq, methylated RNA immunoprecipitation sequencing; METTL3, methyltransferase like 3; ns, not significant; PCDHGA10, protocadherin gamma subfamily A, 10; RT–qPCR, reverse transcriptase–quantitative polymerase chain reaction; SD, standard deviation; SR, sinus rhythm; Ythdf3, YTH N6-methyladenosine RNA-binding protein F2, 3.

Figure 5

PCDHGA10 was regulated by METTL3. (A) MeRIP-Seq in the AF cell model showed reduced m⁶A modification and increased transcript levels of PCDHGA10. Ten Hertz paced HL-1 cells served as the experimental group (n = 3), and HL-1 cells without pacing served as the control group (n = 3). (B) MeRIP-qPCR was used to validate that m⁶A modifications on PCDHGA10 mRNA were regulated by METTL3. Fragments of RNA from sh-Ctrl or sh-Mettl3 HL-1 cells were enriched with m⁶A or rabbit IgG antibody, followed by reverse transcription and qPCR assay. Data are represented as mean ± SD. Sh-Ctrl (n = 4), sh-Mettl3 (n = 3). *P < 0.05, comparison made using two-way ANOVA. (C) PCDHGA10 mRNA expression was evaluated by RT–qPCR in sh-Mettl3 or OE-Mettl3 HL-1 cells. Data are represented as mean ± SD. Sh-Ctrl (n = 4), sh-Mettl3 (n = 4), OE-Ctrl (n = 4), and OE-Mettl3 (n = 4). **P < 0.01, ***P < 0.001, comparisons made using unpaired Student’s t-test. (D) Protocadherin gamma subfamily A, 10 protein expression was evaluated by western blotting in sh-Mettl3 or OE-Mettl3 HL-1 cells. Data are represented as mean ± SD. Sh-Ctrl (n = 3), sh-Mettl3 (n = 3), OE-Ctrl (n = 3), and OE-Mettl3 (n = 3). **P < 0.01, comparisons made using unpaired Student’s t-test. (E) Protein expression of PCDHGA10 was evaluated by western blotting in LAA tissue of patients with AF. Data are represented as mean ± SD. SR (n = 17), AF (n = 18). *P < 0.05, comparison made using non-parametric Mann–Whitney test. (F) Ythdf2, Ythdf3, and Pcdhga10 mRNA were evaluated by RT–qPCR in sh-Ythdf2 or sh-Ythdf3 HL-1 cells. Data are represented as mean ± SD. Sh-Ctrl (n = 3), sh-Ythdf2 (n = 3), and sh-Ythdf3 (n = 3). ***P < 0.001, ****P < 0.0001, comparisons made using unpaired Student’s t-test. (G) HuR and Pcdhga10 mRNA were evaluated by RT–qPCR in si-HuR HL-1 cells. si-Ctrl (n = 3), si1-HuR (n = 3), and si2-HuR (n = 3). *P < 0.05, ***P < 0.001, comparisons made using ordinary one-way ANOVA. (H) RIP-PCR was used to detect the enrichment of PCDHGA10 mRNA by HuR protein under the condition of METTL3 knockdown in HL-1 cells. AF, atrial fibrillation; ANOVA, analysis of variance; FC, fold change; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; LAA, left atrial appendage; m⁶A, N6-methyladenosine; MeRIP-Seq, methylated RNA immunoprecipitation sequencing; METTL3, methyltransferase like 3; ns, not significant; PCDHGA10, protocadherin gamma subfamily A, 10; RT–qPCR, reverse transcriptase–quantitative polymerase chain reaction; SD, standard deviation; SR, sinus rhythm; Ythdf3, YTH N6-methyladenosine RNA-binding protein F2, 3.

We first knocked down METTL3 in the HL-1 cells and performed MeRIP-qPCR. The results showed a significant reduction in the m⁶A modification level of PCDHGA10 and a significant increase in its transcription levels, consistent with findings in the AF cell model (Figure 5B and C). Overexpression of METTL3 led to a decrease in PCDHGA10 transcript levels (Figure 5C), and a corresponding reduction in PCDHGA10 protein levels (Figure 5D). Furthermore, PCDHGA10 protein level was significantly elevated in LAAs of patients with AF (Figure 5E). The m⁶A peak map combined with the SRAMP database identified multiple m⁶A modification sites in the PCDHGA10 mRNA in both mouse and human (see Supplementary material online, Figure S8A and B). Combining the BLAST results on the NCBI database, eight conserved m⁶A modification sites on the mouse and human PCDHGA10 genes were found (see Supplementary Appendix 3 and see Supplementary material online, Figure S8C).

While YTHDF2 and YTHDF3 are well-established m⁶A readers that promote mRNA degradation,²² their knockdown in HL-1 cells did not result in an increase in PCDHGA10 (Figure 5F). Additionally, HuR was found to bind to and stabilize mRNA with less m⁶A modification and knocking it down caused PCDHGA10 to decrease (Figure 5G). RNA immunoprecipitation combined with PCR (RIP-PCR) revealed increased binding of HuR to PCDHGA10 mRNA following METTL3 knockdown (Figure 5H). Together, these results suggested that METTL3 repressed PCDHGA10 in an m⁶A-HuR-mediated pathway.

Next, to determine whether PCDHGA10 regulates calcium handling in atrial myocytes, we overexpressed PCDHGA10 in NMAMs. Calcium-handling assays showed that PCDHGA10 overexpression significantly enhanced amplitude and accelerated decay of calcium transients (Figure 6A), replicating the effects seen with METTL3 knockdown. Furthermore, knockdown of PCDHGA10 in NMAMs alleviated the abnormal calcium handling caused by METTL3 knockdown (Figure 6B).

Figure 6

METTL3 affected calcium transients in atrial cardiomyocytes via PCDHGA10. (A) Calcium transients were detected by the confocal microscopy in PCDHGA10-mcherry and mcherry (control) NMAMs. Data are represented as mean ± SD. mcherry (n = 14 cells from three technical replicates), PCDHGA10-mcherry (n = 18 cells from three technical replicates). ***P < 0.001, comparisons made using unpaired Student’s t-test. (B) Calcium transients were detected by the confocal microscopy in sh-Ctrl, sh-Mettl3, sh-Pcdhga10, and sh-Mettl3 + sh-Pcdhga10 NMAMs. Data are represented as mean ± SD. Sh-Ctrl (n = 32 cells from three technical replicates), sh-Mettl3 (n = 39 cells from three technical replicates), sh-Pcdhga10 (n = 37 cells from three technical replicates), and sh-Mettl3 + sh-Pcdhga10 (n = 41 cells from three technical replicates). *P < 0.05, ****P < 0.0001, comparisons made using two-way ANOVA. (C) Immunofluorescence staining with SERCA2a antibody was performed in NMAMs overexpressing PCDHGA10-GFP. The sarcoplasmic reticulum was demonstrated by the ER tracker-red. The white arrows show the sites where PCDHGA10 colocalizes with SERCA2a and the sarcoplasmic reticulum. (D) Co-IP experiment of HEK293T cells overexpressing PCDHGA10-GFP and Flag-SERCA2a was performed with GFP and Flag antibodies, followed by western blotting experiments with GFP and Flag antibodies. (E) SERCA2a-ATPase activity measurement in HL-1 cells overexpressing PCDHGA10-GFP and GFP (Control). Data are represented as mean ± SD. GFP (n = 4), PCDHGA10-GFP (n = 4), comparison made using unpaired Student’s t-test. (F) SERCA2a ATPase activity measurement in sh-Ctrl, sh-Mettl3, sh-Pcdhga10, and sh-Mettl3 + sh-Pcdhga10 HL-1 cells. Data are represented as mean ± SD. Sh-Ctrl (n = 3), sh-Mettl3 (n = 3), sh-Pcdhga10 (n = 3), and sh-Mettl3 + sh-Pcdhga10 (n = 3). *P < 0.05, **P < 0.01, comparisons made using two-way ANOVA. ANOVA, analysis of variance; GFP, green fluorescent protein; METTL3, methyltransferase like 3; NMAMs, neonatal mice atrial myocytes; PCDHGA10, protocadherin gamma subfamily A, 10; SERCA2a, SR Ca²⁺-ATPase Type 2a; SD, standard deviation.

Protocadherin gamma subfamily A, 10 is a member of the protocadherin γ gene cluster (Pcdhg-γ), which encodes cadherin-like transmembrane proteins^23,24 and may not directly act as a calcium ion channel. Literature suggests that Pcdhg-γ proteins interact with calcium-handling proteins, such as SERCA2a.²⁵ We used lentivirus to overexpress PCDHGA10-GFP in HL-1 cells and performed immunofluorescence experiments. The result confirmed that PCDHGA10 co-localized with SERCA2a and the sarcoplasmic reticulum (Figure 6C). Co-immunoprecipitation/western blotting experiments in HEK293T cells transfected with PCDHGA10-GFP and Flag-SERCA2a plasmids further confirmed the interaction between PCDHGA10 and SERCA2a (Figure 6D). To explore whether PCDHGA10 affects the SERCA2a activity, we measured SERCA2a-ATPase activity and found that it was significantly increased in PCDHGA10-overexpressing HL-1 cells (Figure 6E). Moreover, knockdown of PCDHGA10 prevented the increase in SERCA2a-ATPase activity induced by METTL3 knockdown (Figure 6F). During diastole of atrial myocytes, Ca²⁺ is taken back up into the sarcoplasmic reticulum via SERCA2a, or extruded from cell via NCX1.^26,27 To explore whether NCX1 activity was altered, caffeine was applied, which maintains the ryanodine receptor 2 (RyR2) open.^28,29 All calcium pumped back into the SR will be immediately released by the open RyR2 channels, making SERCA pumping inefficient. Therefore, the decay rate of caffeine-induced calcium transients depends only on extrusion by NCX1, which remained unchanged in METTL3 or PCDHGA10 knockdown NMAMs. Additionally, PCDHGA10 knockdown restored the increased amplitude (reflecting sarcoplasmic reticulum calcium load) caused by METTL3 deficiency (see Supplementary material online, Figure S9). The above results indicated that decreased METTL3 led to increased PCDHGA10 expression, which in turn enhanced SERCA2a activity and increased sarcoplasmic reticulum calcium load in atrial myocytes.

CREB1/USP9X regulates protein level of METTL3

We next explored the mechanisms underlying the down-regulation of METTL3 protein during AF. Studies have suggested that reduced expression and transcriptional activity of CREB1 are associated with AF.^30,31 CREM mice, which overexpress the transcriptional repressor CREM-IbΔC-X, also show impaired CREB1 function.^15,32 To investigate this, we examined CREB1 and its phosphorylated form (p-CREB1) in LAA samples from patients with AF. Both CREB1 protein expression and p-CREB1 (S133) were significantly reduced in AF (Figure 7A). We also found that CREB1 expression was decreased in AF cell models (Figure 7B). Knockdown of CREB1 in HL-1 cells resulted in a significant decrease in METTL3 protein levels, with no change in METTL3 mRNA levels, consistent with observations in AF patient samples (Figure 7C and D).

Figure 7

Reduced CREB1/USP9X axis in AF leads to decreased METTL3 protein. (A) Protein expression of p-CREB1 (S133) and CREB1 were detected by western blotting in LAA tissue of patients with AF. Data are represented as mean ± SD. SR (n = 11), AF (n = 11). *P < 0.05, comparisons made using unpaired Student’s t-test. (B) Protein expression of p-CREB1 (S133) and CREB1 in HL-1 cells at certain stimulation frequencies. Data are represented as mean ± SD. 0 Hz (n = 6), 5 Hz (n = 6), and 10 Hz (n = 6). ***P < 0.001, comparison made using ordinary one-way ANOVA. (C) CREB1 and METTL3 mRNA levels were detected by RT–qPCR in sh-Ctrl and sh-CREB1 HL-1 cells. Data are represented as mean ± SD. Sh-Ctrl (n = 3), sh-CREB1 (n = 3). **P < 0.01, comparisons made using unpaired Student’s t-test. (D) p-CREB1 (S133), CREB1, and METTL3 protein levels were detected by western blotting in sh-Ctrl and sh-CREB1 HL-1 cells. Data are represented as mean ± SD. Sh-Ctrl (n = 3), sh-CREB1 (n = 3). *P < 0.05, **P < 0.01, comparisons made using unpaired Student’s t-test. (E) Prediction of CREB1 transcriptionally regulated deubiquitinases through ChEA, ENCODE, and JASPAR transcription factor databases. (F) USP15, USP2, USP30, USP5, and USP9X mRNA levels were evaluated by RT–qPCR in sh-Ctrl and sh-CREB1 HL-1 cells. Data are represented as mean ± SD. Sh-Ctrl (n = 3), sh-CREB1 (n = 3). **P < 0.01, comparisons made using unpaired Student’s t-test. (G) USP9X protein levels were detected by western blotting in sh-Ctrl and sh-CREB1 HL-1 cells. Data are represented as mean ± SD. Sh-Ctrl (n = 3), sh-CREB1 (n = 3). *P < 0.05, comparisons made using unpaired Student’s t-test. (H) USP9X, METTL3 protein levels were detected in si-Ctrl, si1-USP9X, and si2-USP9X HL-1 cells. Data are represented as mean ± SD. Si-Ctrl (n = 6), si1-USP9X (n = 8), si2-USP9X (n = 8). **P < 0.01, ****P < 0.0001, comparisons made using ordinary one-way ANOVA. (I) The ubiquitination level of METTL3 in si-Ctrl and si1-USP9X HL-1 cells was detected. Endogenous IP was performed using METTL3 antibody or negative control rabbit IgG antibody, followed by western blotting. AF, atrial fibrillation; ANOVA, analysis of variance; CREB1, cAMP-responsive element-binding protein 1; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; IgG, immunoglobulin G; IP, immunoprecipitation; LAA, left atrial appendage; METTL3, methyltransferase like 3; ns, not significant; RT–qPCR, reverse transcriptase–quantitative polymerase chain reaction; SD, standard deviation; SR, sinus rhythm.

Since METTL3 was mainly degraded via the ubiquitin-proteasome pathway (see Supplementary material online, Figure S1H), we hypothesized that CREB1 loss might reduce the expression of a deubiquitinating enzyme, thus increasing the ubiquitination of METTL3. Analysis of transcription factor target genes across three databases (JASPAR,³³ ChEA,³⁴ and ENCODE³⁵) revealed that CREB1 regulates several deubiquitinating enzymes, including USP15, USP2, USP20, USP5, and USP9X (Figure 7E). Among these, only USP9X showed significant down-regulation at both the mRNA and protein levels following CREB1 knockdown in HL-1 cells (Figure 7F and G). Furthermore, dual luciferase reporter assays showed that CREB1 regulates USP9X transcription, as evidenced by increased USP9X promoter activity in the presence of Forskolin (a direct activator of cAMP) and a decrease in activity following treatment with KG-501 (a CREB1 inhibitor; see Supplementary material online, Figure S10A and B). Consistently, in AF patient samples, both USP9X mRNA and protein levels were significantly reduced (see Supplementary material online, Figure S10C and D). Knockdown of USP9X in HL-1 cells resulted in a significant decrease in METTL3 protein levels (Figure 7H) and an increased in METTL3 ubiquitination (Figure 7I). In summary, our findings suggested that the reduction of CREB1 and USP9X in AF contributes to increased ubiquitination and degradation of METTL3, thereby lowering its protein levels without affecting its mRNA expression.

Discussion

Main findings of this study

This present study, to the best of our knowledge, is the first to explore the relationship between METTL3 in atrial myocytes and AF. METTL3 protein was found to decrease during AF, and METTL3 deficiency was shown to promote AF development in experimental models. Mechanically, METTL3 deficiency led to enhanced calcium handling through increased PCDHGA10 expression in atrial myocytes.

METTL3 and cardiovascular diseases

As the predominant m⁶A modification catalytic enzyme, METTL3 has been found to play critical roles in various cardiovascular diseases. For instance, METTL3 and m⁶A modification are up-regulated in atherosclerosis, and METTL3 knockdown effectively prevents the atherogenic process by targeting NLRP3 and KLF4 mRNA.³⁶ METTL3 also plays a protective role in cardiac hypertrophy through regulation of PARP10 mRNA.³⁷ Additionally, METTL3 and m⁶A modification are implicated in pulmonary hypertension,³⁸ aortic aneurysm, and dissection.³⁹ However, studies on m⁶A modification and arrhythmias are limited. Carnevali et al.⁴⁰ reported that m⁶A demethylase FTO knockout mice exhibited increased heart rhythm variability and susceptibility to premature ventricular contractions under stress. It has been reported that the up-regulation of METTL3 promotes cardiac fibroblast proliferation, thereby promoting cardiac fibrosis,⁴¹ which is an important substrate for AF. Interestingly, our findings suggest that during AF, METTL3 is down-regulated in atrial myocytes. This contradictory regulation of METTL3 between atrial fibroblasts and myocytes may be due to the distinct metabolic remodelling observed in these cell types during AF. Atrial fibroblasts undergo mitochondrial biogenesis and enhanced cellular respiration,^42,43 while atrial myocytes experience impaired electron transport chain function, increased reactive oxygen species, and mitochondrial dysfunction.⁴⁴ METTL3’s association with mitochondrial function warrants further investigation.^45,46

Atrial cardiomyopathy and atrial fibrillation

Atrial cardiomyopathy (AtCM) refers to structural, architectural, contractile, or electrophysiological changes in the atria with potential clinical relevance. Atrial cardiomyopathy typically precedes AF, and early identification is key for prevention.⁴⁷ Our study revealed that increased KCNJ5 and disordered calcium handling in atrial myocytes, features of AtCM, are mechanisms by which METTL3 deficiency promotes AF.

I_K,ACh, an atrial-specific current, is primarily composed of KCNJ3 and KCNJ5 with a stoichiometric 2:2.⁴⁸ KCNJ5 alone can also form homotetramers,^49,50 which are activated by vagally released ACh. Structurally activated I_K,ACh is implicated in chronic AF,^51,52 and I_K,ACh blockers like tertiapin effectively inhibit the re-entry waves formation and terminate AF.⁵³

By studying the calcium activity of atrial myocytes, we found that deficiency of METTL3 resulted in abnormal calcium handling through PCDHGA10 and SERCA2a in atrial myocytes. Increased expression of SERCA2a leads to sarcoplasmic reticulum calcium overload, heightening AF susceptibility.⁵⁴ While previous studies identified SERCA2a as a binding protein of PCDHGA10,²⁵ our study is the first to link PCDHGA10 to AF and its role in calcium handling.

In the past few decades, despite significant progress in AF research, mechanism-based treatments remain scarce.⁵ Our findings highlight METTL3 deficiency as a pathogenic factor in AtCM and AF. Combined with advancements in m⁶A modification interventions,⁵⁵ our research will bridge basic science and clinical translation, offering novel therapeutic targets for AF management.

CREB1/USP9X axis in heart disease

cAMP-responsive element-binding protein 1 (CREB1) is phosphorylated by cAMP-dependent protein kinase, enabling its nuclear entry and transcriptional activation.⁵⁶ CREB1 expression and transcriptional activity are reduced in AF patient LAAs,^30,31,57 consistent with our findings. We demonstrated that CREB1 regulates METTL3 protein levels indirectly via USP9X, a deubiquitinating enzyme. CREB1’s role in regulating ubiquitination has been reported, with studies showing CREB1 increasing RCAN1 ubiquitination to promote degradation, a process mitigated by proteasome inhibitors.⁵⁸

USP9X plays a protective role in heart diseases, as evidenced by its association with reduced congenital heart disease–related pulmonary arterial hypertension⁵⁹ and inhibition of aortic calcification.⁶⁰ Our findings suggest reduced USP9X expression in patients with AF aligns with its protective functions, proposing the CREB1/USP9X axis as a potential therapeutic target for AF.

Limitation

Our study still had some limitations. First, the precise mechanism by which PCDHGA10 affects SERCA2a activity remains unclear. Secondly, it is unknown whether METTL3 influences phosphorylation or oxidative modifications of SERCA2a or other calcium ion channels. Thirdly, the potential effects of METTL3 on other ion channels besides KCNJ5 require further investigation.

Conclusions

In summary, our findings suggest that METTL3 plays a crucial role in AF development by influencing calcium handling in atrial myocytes. Additionally, the CREB1/USP9X axis is implicated in the down-regulation of METTL3 protein during AF. These findings provide insights into potential mechanism-based therapeutic targets for AF treatment.

Supplementary material

Supplementary material is available at Europace online.

Authors’ contributions

D.-J.W. and Qin.Z. supervised and designed this study. Qia.Z. handled the revision comments, designed the experiment, and replied. J.S., X.-Y.Z., R.-H.Y., and W.-X.L. performed the cellular and molecular experiments. J.Y., L.T., and C.-Y.K. analysed the data. J.S. and X.-Y.Z. wrote the manuscript. H.-Q.L., F.C., and J.-J.W. helped with the human sample collection and animal studies. W.-S.X. and J.-L.F. performed additional experiments to reply to the comments.

Experimental ethics and consent

The use of human samples in this experiment was in compliance with the Declaration of Helsinki and was approved by the Institutional Review Board of Nanjing Drum Tower Hospital (2020-185-01). In addition, written consent was obtained from each included patient to remove and obtain the atrial appendages. All animal experiments were performed in accordance with the ARRIVE Guidelines and were approved by the Animal Care and Use Committee of Nanjing Drum Tower Hospital.

Funding

This work was supported by the Nanjing Medical Science and Technique Development Foundation (grant number ZKX21021), the National Natural Science Foundation of China (grant numbers 82241212, 82270346, and 82200356), and the China Postdoctoral Science Foundation (grant number 2024M761412).

Data availability

Data underlying this article will be shared upon reasonable request with the corresponding author.

References

Brundel

Hills

Kuipers

Lip

GYH

de Groot

NMS

Atrial fibrillation

Nat Rev Dis Primers

2022

;

Benz

Hohnloser

Eikelboom

Carnicelli

Giugliano

Granger

et al.

Outcomes of patients with atrial fibrillation and ischemic stroke while on oral anticoagulation

Eur Heart J

2023

;

1807

–

Escudero-Martinez

Morales-Caba

Segura

Atrial fibrillation and stroke: a review and new insights

Trends Cardiovasc Med

2023

;

–

Schotten

Verheule

Kirchhof

Goette

Pathophysiological mechanisms of atrial fibrillation: a translational appraisal

Physiol Rev

2011

;

265

–

325

Remme

Heijman

Gomez

Zaza

Odening

25 Years of basic and translational science in EP Europace: novel insights into arrhythmia mechanisms and therapeutic strategies

Europace

2023

;

–

Vinciguerra

Dobrev

Nattel

Atrial fibrillation: pathophysiology, genetic and epigenetic mechanisms

Lancet Reg Health Eur

2024

;

100785

Song

Hou

Wang

Role of N(6)-methyladenosine RNA modification in cardiovascular disease

Front Cardiovasc Med

2021

;

659628

Dominissini

Rechavi

Gene expression regulation mediated through reversible m(6)A RNA methylation

Nat Rev Genet

2014

;

293

–

306

Song

Zhou

Sun

Liu

Lin

et al.

METTL3 boosts mitochondrial fission and induces cardiac fibrosis by enhancing LncRNA GAS5 methylation

Pharmacol Res

2023

;

194

106840

Adili

Zhu

Cao

Tang

Wang

et al.

Atrial fibrillation underlies cardiomyocyte senescence and contributes to deleterious atrial remodeling during disease progression

Aging Dis

2022

;

298

–

312

Zhang

Liu

Baks-Te Bulte

Wiersma

et al.

DNA damage-induced PARP1 activation confers cardiomyocyte dysfunction through NAD(+) depletion in experimental atrial fibrillation

Nat Commun

2019

;

1307

Zhao

Guo

Cacl2–ACh induced atrial fibrillation (flutter) in mice

Zhongguo Yao Li Xue Bao

1982

;

185

–

PubMed

Wei

Ren

Zhang

Fan

miR-425-5p is negatively associated with atrial fibrosis and promotes atrial remodeling by targeting CREB1 in atrial fibrillation

J Cardiol

2022

;

202

–

Jiang

Sun

Gao

Fan

et al.

Necroptosis is required for atrial fibrillation and involved in aerobic exercise-conferred cardioprotection

J Cell Mol Med

2021

;

8363

–

Muller

Lewin

Baba

Boknik

Fabritz

Kirchhefer

et al.

Heart-directed expression of a human cardiac isoform of cAMP-response element modulator in transgenic mice

J Biol Chem

2005

;

280

6906

–

Lahiri

Nie

Pan

Abu-Taha

Reynolds

et al.

Genetic inhibition of nuclear factor of activated T-cell c2 prevents atrial fibrillation in CREM transgenic mice

Cardiovasc Res

2022

;

118

2805

–

Yao

Veleva

Scott

Jr,

Cao

Chen

et al.

Enhanced cardiomyocyte NLRP3 inflammasome signaling promotes atrial fibrillation

Circulation

2018

;

138

2227

–

Scott

Jr,

Campbell

Pan

Alsina

Reynolds

et al.

Atrial-specific gene delivery using an adeno-associated viral vector

Circ Res

2019

;

124

256

–

Verkerk

Geuzebroek

Veldkamp

Wilders

Effects of acetylcholine and noradrenalin on action potentials of isolated rabbit sinoatrial and atrial myocytes

Front Physiol

2012

;

174

PubMed

Butova

Myachina

Simonova

Kochurova

Bozhko

Arkhipov

et al.

Peculiarities of the acetylcholine action on the contractile function of cardiomyocytes from the left and right atria in rats

Cells

2022

;

–

Boulias

Greer

Biological roles of adenine methylation in RNA

Nat Rev Genet

2023

;

143

–

Zou

The YTHDF proteins display distinct cellular functions on m(6)A-modified RNA

Trends Biochem Sci

2024

;

611

–

Chen

Alvarez

Lefebvre

Friedman

Nwakeze

Geiman

et al.

Functional significance of isoform diversification in the protocadherin gamma gene cluster

Neuron

2012

;

402

–

Kiefer

Chiosso

Langen

Buckley

Gaudin

Rajkumar

et al.

WAPL functions as a rheostat of protocadherin isoform diversity that controls neural wiring

Science

2023

;

380

eadf8440

Han

Lin

Meng

Wang

Proteomics analysis reveals overlapping functions of clustered protocadherins

Mol Cell Proteomics

2010

;

–

Hamilton

Terentyev

Proarrhythmic remodeling of calcium homeostasis in cardiac disease; implications for diabetes and obesity

Front Physiol

2018

;

1517

Shah

Akhtar

Hassan

Akhtar

Paudel

Najmi

EGFR tyrosine kinase inhibition decreases cardiac remodeling and SERCA2a/NCX1 depletion in streptozotocin induced cardiomyopathy in C57/BL6 mice

Life Sci

2018

;

210

–

Callewaert

Cleemann

Morad

Caffeine-induced Ca²⁺ release activates Ca²⁺ extrusion via Na⁺-Ca²⁺ exchanger in cardiac myocytes

Am J Physiol

1989

;

257

C147

–

Sitsapesan

Williams

Mechanisms of caffeine activation of single calcium-release channels of sheep cardiac sarcoplasmic reticulum

J Physiol

1990

;

423

425

–

Deshmukh

Barnard

Sun

Newton

Castel

Pettersson

et al.

Left atrial transcriptional changes associated with atrial fibrillation susceptibility and persistence

Circ Arrhythm Electrophysiol

2015

;

–

Moreira

Takawale

Hulsurkar

Menassa

Antanaviciute

Lahiri

et al.

Paracrine signalling by cardiac calcitonin controls atrial fibrogenesis and arrhythmia

Nature

2020

;

587

460

–

Chen

Qin

Hao

Wang

Chen

Yang

et al.

Cardiac-specific overexpression of CREM-IbDeltaC-X via CRISPR/Cas9 in mice presents a new model of atrial cardiomyopathy with spontaneous atrial fibrillation

Transl Res

2024

;

267

–

Castro-Mondragon

Riudavets-Puig

Rauluseviciute

Lemma

Turchi

Blanc-Mathieu

et al.

JASPAR 2022: the 9th release of the open-access database of transcription factor binding profiles

Nucleic Acids Res

2022

;

D165

–

Lachmann

Krishnan

Berger

Mazloom

Ma'ayan

ChEA: transcription factor regulation inferred from integrating genome-wide ChIP-X experiments

Bioinformatics

2010

;

2438

–

Miao

Jia

Zhang

Guo

AnimalTFDB 3.0: a comprehensive resource for annotation and prediction of animal transcription factors

Nucleic Acids Res

2019

;

D33

–

Chien

Wang

Yarmishyn

Tsai

et al.

METTL3-dependent N(6)-methyladenosine RNA modification mediates the atherogenic inflammatory cascades in vascular endothelium

Proc Natl Acad Sci U S A

2021

;

118

–

Gao

Zhang

Liu

Ponnusamy

Zhao

Zhou

et al.

The piRNA CHAPIR regulates cardiac hypertrophy by controlling METTL3-dependent N(6)-methyladenosine methylation of Parp10 mRNA

Nat Cell Biol

2020

;

1319

–

Zhang

Yan

Zhang

The role of RNA m(6)A methylation in the regulation of postnatal hypoxia-induced pulmonary hypertension

Respir Res

2021

;

121

Wang

Jing

Sun

Expression pattern and clinical value of key m6A RNA modification regulators in abdominal aortic aneurysm

J Inflamm Res

2021

;

4245

–

Carnevali

Graiani

Rossi

Al Banchaabouchi

Macchi

Quaini

et al.

Signs of cardiac autonomic imbalance and proarrhythmic remodeling in FTO deficient mice

PLoS One

2014

;

e95499

Zhou

Song

Sun

Ding

Luo

et al.

METTL3 boosts glycolysis and cardiac fibroblast proliferation by increasing AR methylation

Int J Biol Macromol

2022

;

223

899

–

915

Bernard

Logsdon

Ravi

Xie

Persons

Rangarajan

et al.

Metabolic reprogramming is required for myofibroblast contractility and differentiation

J Biol Chem

2015

;

290

25427

–

Negmadjanov

Godic

Rizvi

Emelyanova

Ross

Richards

et al.

TGF-beta1-mediated differentiation of fibroblasts is associated with increased mitochondrial content and cellular respiration

PLoS One

2015

;

e0123046

Bode

Pronto

JRD

Schiattarella

Voigt

Metabolic remodelling in atrial fibrillation: manifestations, mechanisms and clinical implications

Nat Rev Cardiol

2024

;

682

–

700

Yang

Pang

Guo

Zhu

Song

et al.

An exploration of the coherent effects between METTL3 and NDUFA10 on Alzheimer's disease

Int J Mol Sci

2023

;

–

Hao

Song

Zhang

Zhou

Zhang

et al.

METTL family in healthy and disease

Mol Biomed

2024

;

Goette

Corradi

Dobrev

Aguinaga

Cabrera

Chugh

et al.

Atrial cardiomyopathy revisited-evolution of a concept: a clinical consensus statement of the European Heart Rhythm Association (EHRA) of the ESC, the Heart Rhythm Society (HRS), the Asian Pacific Heart Rhythm Society (APHRS), and the Latin American Heart Rhythm Society (LAHRS)

Europace

2024

;

–

Cui

Cantwell

Zorn

Logothetis

Kir channel molecular physiology, pharmacology, and therapeutic implications

Handb Exp Pharmacol

2021

;

267

277

–

356

Wickman

Nemec

Gendler

Clapham

Abnormal heart rate regulation in GIRK4 knockout mice

Neuron

1998

;

103

–

Cui

Gada

Shalomov

Ban

Eptaminitaki

et al.

A novel small-molecule selective activator of homomeric GIRK4 channels

J Biol Chem

2022

;

298

102009

Dobrev

Friedrich

Voigt

Jost

Wettwer

Christ

et al.

The G protein-gated potassium current I(K,ACh) is constitutively active in patients with chronic atrial fibrillation

Circulation

2005

;

112

3697

–

706

Voigt

Friedrich

Bock

Wettwer

Christ

Knaut

et al.

Differential phosphorylation-dependent regulation of constitutively active and muscarinic receptor-activated IK,ACh channels in patients with chronic atrial fibrillation

Cardiovasc Res

2007

;

426

–

Bingen

Neshati

Askar

Kazbanov

Ypey

Panfilov

et al.

Atrium-specific Kir3.x determines inducibility, dynamics, and termination of fibrillation by regulating restitution-driven alternans

Circulation

2013

;

128

2732

–

Herraiz-Martinez

Llach

Tarifa

Gandia

Jimenez-Sabado

Lozano-Velasco

et al.

The 4q25 variant rs13143308T links risk of atrial fibrillation to defective calcium homoeostasis

Cardiovasc Res

2019

;

115

578

–

Selberg

Blokhina

Aatonen

Koivisto

Siltanen

Mervaala

et al.

Discovery of small molecules that activate RNA methylation through cooperative binding to the METTL3-14-WTAP complex active site

Cell Rep

2019

;

3762

–

Mayr

Montminy

Transcriptional regulation by the phosphorylation-dependent factor CREB

Nat Rev Mol Cell Biol

2001

;

599

–

609

Wang

Cai

Shen

Comprehensive analysis of pertinent genes and pathways in atrial fibrillation

Comput Math Methods Med

2021

;

2021

4530180

PubMed

Seo

Chung

CREB activates proteasomal degradation of DSCR1/RCAN1

FEBS Lett

2008

;

582

1889

–

Reijnders

Zachariadis

Latour

Jolly

Mancini

Pfundt

et al.

De novo loss-of-function mutations in USP9X cause a female-specific recognizable syndrome with developmental delay and congenital malformations

Am J Hum Genet

2016

;

373

–

Majumdar

Manivannan

Basu

Ueyama

Blaser

Cameron

et al.

Nitric oxide prevents aortic valve calcification by S-nitrosylation of USP9X to activate NOTCH signaling

Sci Adv

2021

;

–