Programmable aging function-related mitochondrial DNA 5-methylcytosine (m5C) modification with a TALE-directed methyltransferase Authors: Guo Li, Xinzhi Zhou, Guanglin Zhu, Yaxian Cheng, Yingjia Pan, Xiaomeng Guo, Guoya Liao, Xiangyang Li, Xiaoxiang Hu, Yuan Yao, Zhimin Gu, Wen Jiang Published: 2026 ABSTRACT AbstractMitochondrial epigenetic editing offers a potential strategy for modulating disease-causing mitochondrial genes while leaving the underlying DNA sequence unchanged. In this study, we present MEE, a mitochondrial epigenetic editor consisting of mitochondrion-targeted TALE modules fused to Dnmt3A and Dnmt3L methyltransferases. MEE efficiently directed site-specific 5mC methylation within cellular mitochondrial DNA with low detectable off-target activity under the tested conditions. MEE-mediated methylation at the C12191 (H) site exceeded 53% and was associated with an approximately 95% reduction in steady-state MT-ND5 mRNA levels in human cells. Notably, MEE increased methylation at the aging-associated C11168 (H) site by 11.76% in vivo, leading to reduced MT-ND4 expression in the targeted brain region and altered plasma levels of t-Tau and NFL in mice. These findings demonstrate that MEE provides a tool for precise epigenetic engineering of mitochondrial DNA, enabling experimental interrogation of specific 5mC modifications and exploration of the functional roles of mitochondrial DNA methylation in aging-associated diseases.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s12964-026-03030-8.Keywords: mtDNA, Mitoepigenetic editor, 5mC methylation, Aging-related diseases Introduction Dynamic covalent modifications of DNA and proteins are essential for regulating gene expression [1]. Among known DNA modifications, mitochondrial epigenetic (Termed mitoepigenetics) editing is an emerging field that focuses on modifying the epigenetic landscape of mitochondrial DNA (mtDNA) to regulate gene expression and potentially treat mitochondrial-related diseases. Alterations in mtDNA methylation can directly affect mtDNA expression or coding sequences, leading to premature aging [2] and metabolic and neurodegenerative diseases [3]. Numerous studies indicate that mtDNA methylation may influence cell senescence and aging [4–6]. Substantial evidence points to mitochondrial dysfunction as a contributor to neurodegenerative diseases, particularly Alzheimer’s, Parkinson’s, and amyotrophic lateral sclerosis, prompting researchers to explore the role of mitoepigenetic mechanisms in these diseases’ development [7–9]. Advances in mtDNA editing now allow researchers to model mitochondrial genetic disorders in cell lines [10] and animal systems [11], offering a pathway toward future treatments. Several mtDNA editing tools have been developed, including TALENs (Transcription activator-like effector nucleases) [12], which introduce targeted DNA breaks to disrupt mitochondrial genes, as well as DdCBEs (DddA-derived cytosine base editors) [13] and TALEDs (TALE-linked deaminases) [14, 15], which enable precise base editing of mtDNA. These tools are engineered to make targeted modifications to the DNA sequence, making them particularly suited for repairing or introducing disease-related mutations [16–18]. However, their dependence on endogenous DNA repair mechanisms poses challenges, as the complexity of these pathways often results in unpredictable outcomes beyond the intended edit. An alternative strategy involves reprogramming the mitochondrial epigenetic landscape to regulate gene expression without altering the DNA sequence itself, providing a complementary approach to traditional editing methods. Targeted epigenome editing is gaining recognition as a promising alternative to circumvent the limitations of existing technologies. Multiple studies indicate that precisely adding or removing epigenetic marks, such as DNA methylation or histone modifications, directly influences gene expression [19–23]. Despite this evidence, the effectiveness of epigenome editing in reliably regulating endogenous mitochondrial gene expression in human cells and animal models has yet to be conclusively established. We have developed a novel platform for achieving sustained silencing of mitochondrial target genes. By integrating the precision of TALE DNA-binding domains with multiple epigenome-editing functionalities into a single molecule, termed the mitochondrial epigenetic editor (MEE), we demonstrate robust and stable mtDNA silencing in human cells. MEEs act through simultaneous modification of DNA methylation and chromatin accessibility at 19 distinct target sites. As proof of concept, we applied this approach to three mouse mtDNA sites implicated in natural aging processes. Our results show that MEEs enable lasting, precise modulation of mitochondrial gene expression, underscoring their potential to advance mitoepigenome editing as a therapeutic strategy for human diseases. Results Rational design of mitochondrial epigenetic editor Prior research established Dnmt3A as a key tool for epigenome editing [20–25]. Indirect evidence for mtDNA methylation includes the mitochondrial localization of DNA methyltransferases, Dnmt1, which contains a mitochondrial targeting sequence (MTS), localizes to mitochondria [26], while Dnmt3A and Dnmt3B, despite lacking canonical MTSs, have been detected in mouse mitochondria [27–29]. To test whether Dnmt3A enables targeted mitoepigenome editing, we generated three cell models, Dnmt3A knockout cells (Fig. S1), cells over-expressing MTS-tagged Dnmt3A (Supplementary Table S1), and wild-type controls (Fig. S2a). Analysis of mitochondrial DNA 5mC methylation revealed that Dnmt3A knockout disrupted regional methylation patterns and altered mitochondrial gen2e expression. Strikingly, MTS-Dnmt3A overexpression markedly increased mtDNA methylation levels and introduced strand-specific methylation biases (Fig. 1a, b, S2b, Supplementary Table S2). Then we engineered a programmable mitochondrial epigenome editor (MEE) by fusing a MTS, the DNA methyltransferase domains Dnmt3A (D3A) and Dnmt3L (D3L), to a TALE DNA-binding scaffold (Fig. 1c). Three specific sites within the ND2 and ND5 genes were selected to evaluate the methylation efficiency of MEEs by constructing TALE-based expression vectors. Strand-specific analysis of mitochondrial DNA methylation and mitoepigenetic editor (MEE) development. (a) Bisulfite sequencing (BS-seq; whole-genome bisulfite sequencing) methylation profiles of linearized mitochondrial DNA (mtDNA) in HEK293T cells. Data compare three cellular models: wild-type (WT) cells, cells overexpressing mitochondria-targeted Dnmt3A (MTS-Dnmt3A), and Dnmt3A knockout (KO) cells. Annotations between heavy (H)- and light (L)-strand methylation maps indicate positions of mtDNA functional elements (n = 3). (b) Quantification of mtDNA methylation levels at CG and non-CG (CH) sites on H and L strands. Mean methylation: average methylation of cytosine sites with > 10× sequencing coverage. Total methylation: sum of methylation values across all qualifying cytosines. Analyses include all cytosines with ≥ 10× coverage (left) or a subset with > 1% methylation (right). Error bars denote standard error of the mean (SEM). (c) Schematic outlining the design and development of mitoepigenetic methylation editors (MEEs). (d) Time-dependent methylation efficiency of MEEs targeting ND2 and ND5 loci in HEK293T cells. Cells were transfected with MEE expression plasmids and harvested at 2- and 6-days post-transfection (n = 3) To assess methylation efficiency, plasmids were transfected and analyzed at 2 days and 4 days post-transfection. Mitochondrial DNA was isolated from samples and subjected to bisulfite sequencing PCR (BSP). The results revealed that the MEEs achieved 5mC methylation efficiencies of 33.7% (2 days) and 41.7% (6 days) at MT-ND2 site 1, 33.3% (2 days) and 45.8% (6 days) at MT-ND2 site 2, and 20.8% (2 days) and 70.8% (6 days) at the MT-ND5 site (Fig. 1d). These findings indicate that MEEs exhibit promising site-specific methylation activity. MEEs enable efficient mtDNA methylation To further investigate the functional capacity of the mitochondrial epigenome editor, we selected the top 20 methylation sites that exhibited significant differences in methylation efficiency across Dnmt3A knockout cells, MTS-Dnmt3A over-expressing cells, and wild-type cells (Fig. S2c, Supplementary Table S3). The 20 methylation sites were distributed as follows: 1 cytosine position C15634 (L) in the MT-RNR1 gene; 4 positions C3687 (H), C3951 (H), C3966 (H), C4152 (H) in MT-ND1; 3 positions C6464 (H), C6824 (H), C9228 (L) in MT-COX1; 4 positions C7829 (H), C8720 (L), C8642 (L), C8452 (L) in MT-COX2; 1 position C8544 (H) in MT-ATP8; 1 position C9867 (H) in MT-COX3; 2 positions C9001 (H), C11492 (H) in tRNA regions; 3 positions C12191 (H), C3795 (L), C13967 (H) in MT-ND5; and 1 position C14984 (H) in MT-CYTB. We predicted the human mitochondrial protein structure with highlighting the genomic positions of 5-methylcytosine (5mC) target sites (Fig. S4a). The architecture of MEE monomers for targeted specific mtDNA methylation site was conferred by programmable TALE DNA-binding domains, and the Dnmt3A_3L fusion protein mediated methylation editing at designated cytosine (C) residues (Fig. S4b). The strategy for MT-COX2 gene knockdown via targeted C7829 and C7850 mtDNA methylation using strand-specific MEE, MEE-H targeted C7829 (H-strand) and MEE-L targeted C7850 (L-strand), respectively (Fig. S4c, Supplementary Table S4). MEEs were designed to target these 20 sites, and the corresponding plasmids were transfected into HEK293T cells. At 72 h post-transfection, 4 × 10⁶ GFP-positive cells were isolated via fluorescence-activated cell sorting (FACS). Of these, 2 × 10⁶ cells were used for mtDNA extraction and bisulfite sequencing PCR analysis, while the remaining 2 × 10⁶ cells were utilized for RNA extraction and subsequent quantitative PCR (qPCR) detection (Fig. 2a, b). λ DNA (Takara, #3019) was selected as an unmethylated control for bisulfite assay, with all unmethylated C converted into T, indicating high reliability of bisulfite sequencing PCR analysis (Supplementary Table S5). The results demonstrated that MEEs induced 5mC methylation at these mitochondrial DNA sites, leading to down-regulation of several mitochondrial gene expressions. At C3687 (H), C3951 (H), C3966 (H), and C4152 (H), methylation levels of 48.6%, 9.8%, 52.1%, and 7.9% respectively down-regulated MT-ND1 expression by 75.1%, 24.6%, 83.2%, and 58.9%. Similarly, methylation levels of 33.3%, 42.0%, and 54.4% at C6464 (H), C6824 (H), and C9228 (L) down-regulated MT-COX1 expression by 18.9%, 25.7%, and 41.6%. At C7829 (H), C8720 (L), C8642 (L), and C8452 (L), methylation levels of 30.0%, 5.2%, 16.1%, and 69.5% down-regulated MT-COX2 expression by 34.0%, 28.2%, 60.2%, and 68.1%. MEEs mediated 42.7% 5mC methylation at C15634 (L), reducing MT-RNR1 expression by 39.3%. A 58.4% methylation level at C8544 (H) reduced MT-ATP8 expression by 83.7%, while a 17.5% level at C9867 (H) down-regulated MT-COX3 expression by 42.9%. Methylation levels of 19.0% and 21.2% at C9001 (H) and C11492 (H) resulted in 81.6% and 45.8% down-regulation of tRNA expression. Lastly, methylation levels of 53.6%, 17.1%, and 13.3% at C12191 (H), C3795 (L), and C13967 (H) down-regulated MT-ND5 expression by 94.2%, 53.5%, and 58.3%, and a 10.0% level at C14984 (H) reduced MT-CYTB expression by 65.7% (Fig. 2c-e, Fig. S5a-b, Supplementary Table S6, 7). Similar transcriptional effects were observed for MEEs targeting other mitochondrial loci. Based on these results, MT-COX1, MT-COX2, and MT-ND5 were further selected for Western blot analysis to assess whether methylation-induced repression was reflected at the protein level (Fig. 2f; Fig. S5c). Consistently, MEE-mediated methylation caused a significant reduction in protein expression compared with both the untreated control and the dMEE group. These findings indicate that MEEs are effective for mitoepigenetic editing. Functional validation of mitochondrial epigenetic editors (MEEs). (a) Schematic of the MEE (MTS-TALE-Dnmt3A_3L) backbone and experimental workflow for site-specific mtDNA methylation. (b) Confocal microscopy of MEE localization. Green: MTS-TALE-Dnmt3A_3L-GFP (MEE); Red: MitoTracker (mitochondria); Blue: DAPI (nucleus); Merge: Co-localization confirms mitochondrial targeting. (c) Genome-wide distribution of mtDNA methylation sites with predicted protein structures (AlphaFold2) of targeted mitochondrial genes. (d) Mean methylation efficiency of MEEs by targeting specific sites in HEK293T cells. (NT) control group (Data presented as mean, n = 3 biological replicates, each averaged from three technical replicates). (e) Relative expression levels of mitochondrial ND1, COX1, COX2, and ND5 genes. Statistical comparisons were performed using two-tailed Student’s t-tests against the non-targeting (NT) control group (p < 0.05; data presented as mean ± SEM, n = 3, each averaged from three technical replicates). (f) Western blot of mitochondrial ND5, COX1 and COX2 expression after MEEs or dead MEEs targeting C12191 (H), C9228 (L) and C8452 (L), respectively. COX-IV served as an internal control to monitor for equal loading. NC is referred to as Dnmt3A-KO HEK293T cells To determine whether the observed reduction in mitochondrial RNA abundance was driven by 5mC methylation or by physical interference of MEEs with mitochondrial transcription or RNA processing, we generated catalytically inactive MEEs (dMEEs) by introducing Cys-to-Ala substitutions at the key catalytic residues of DNMT3A. dMEEs and their active counterparts were designed to target 20 mitochondrial loci. After 72 h of incubation, steady-state RNA levels of mtDNA genes targeted by active MEEs were markedly reduced compared with both the untreated control and the corresponding dMEE group. In contrast, genes targeted by dMEEs exhibited only a moderate decrease relative to the control. These results indicate that MEE-induced reduction in mitochondrial RNA abundance likely arises from the combined effects of editor occupancy and DNMT3A-mediated 5mC methylation, rather than from a methylation-only mechanism (Fig. S3). The catalytically inactive dMEE controls also indicate that TALE-mediated mtDNA binding can moderately reduce steady-state mitochondrial RNA abundance, suggesting that physical occupancy of the editor may interfere with mitochondrial transcription, transcript processing or RNA stability. However, active MEEs generally induced stronger reductions in mitochondrial RNA levels than the corresponding dMEEs, supporting an additional contribution from DNMT3A-mediated cytosine methylation. Thus, the observed MEE-induced reduction in mitochondrial RNA abundance likely reflects a combined effect of editor occupancy and methylation-dependent regulation, rather than a methylation-only mechanism. MEE is highly specific To assess the specificity of the MEE, we selected three target sites: C15634 (L) in MT-RNR1, C3687 (H) in MT-ND1, and C3951 (H) in MT-ND1. Expression vectors were transfected into HEK293T cells, and at 72 h post-transfection, 4 × 10⁶ GFP-positive cells were sorted for downstream analyses. Mitochondrial DNA was isolated from 4 × 10⁶ cells for whole-genome bisulfite sequencing (WGBS), while RNA was extracted from an additional 4 × 10⁶ cells for RNA sequencing (RNA-seq). Analysis of WGBS data revealed differential 5mC methylation patterns between experimental groups. Compared to GFP transfection controls, we identified 786, 771, and 771 differential 5mC methylation sites in the MEE-based C15634 (L), C3687 (H), and C3951 (H) editing sites, respectively (Fig. 3a, Supplementary Table S8). Compared to wild type controls, we identified 748, 730, and 731 differential 5mC methylation sites in the MEE-based C15634 (L), C3687 (H), and C3951 (H) editing sites, respectively (Fig. 3b, Supplementary Table S7). Similarly, compared to GFP transfection controls, we detected 480, 479, and 507 differential 5mC methylation sites in the MEE-based C15634 (L), C3687 (H), C3951 (H) editing sites, respectively (Fig. 3c, Supplementary Table S8). Off-target 5-methylcytosine (5mC) methylation analysis of mitochondrial epigenetic editors (MEEs). a-b. Nuclear genome off-target effects. (a) 5mC levels at nuclear off-target sites in MEE-treated groups (C1: mtDNA C15634 (L)-targeting; C2: mtDNA C3687 (H)-targeting; C3: mtDNA C3951 (H)-targeting) compared to GFP-transfected controls. Data derived from whole-genome bisulfite sequencing (WGBS; n = 3). (b) Nuclear 5mC levels in MEE- or GFP-transfected groups relative to wild-type (WT) controls (WGBS; n = 3). c-d, Mitochondrial genome off-target effects. (c) Mitochondrial 5mC levels in MEE-treated groups (C1-C3) versus GFP-transfected controls (WGBS; n = 3). (d) Mitochondrial 5mC levels in MEE- or GFP-transfected groups relative to WT controls (WGBS; n = 3). C1-C3, MEEs targeting mtDNA positions C15634 (L), C3687 (H), and C3951 (H), respectively. GFP, GFP-transfected negative control group. WT, Untreated wild-type baseline. e-h, The distribution of DRGs in MEE-treated groups. DRGs: Down-regulated genes. (e) DRGs in MEE-based C15634 (L) editing sites versus wild type controls (RNA-seq; n = 3). (f) DRGs in MEE-based C3687 (L) editing sites versus wild type controls (RNA-seq; n = 3). (g) DRGs in MEE-based C3951 (L) editing sites versus wild type controls (RNA-seq; n = 3). (h) DRGs in GFP-transfected controls versus wild type controls (RNA-seq; n = 3) Compared to wild-type controls, we detected 229, 201, 192, and 4 differential 5mC methylation sites in the MEE-based C15634 (L), C3687 (H), C3951 (H) editing sites, and GFP transfection groups, respectively (Fig. 3d, Supplementary Table S8). Analysis of RNA-sequencing data showed minimal off-target gene knockdown in untransfected cells versus cells transfected with mitochondrial epigenetic editors. Compared with wild-type cells, there were 868, 547, 373, and 675 down-regulated genes in the MEE-based C15634 (L), C3687 (H), C3951 (H), and GFP-expressing control groups, respectively (Fig. 3e-h, Supplementary Table S9). Because comparable changes were also observed in the GFP-transfected control group, the observed differential 5mC methylation and gene-expression changes are likely influenced by transfection-related effects. These results suggest that MEEs exhibit limited detectable off-target activity under the tested conditions. Aged mice exhibit aberrant mtDNA 5mC methylation patterns The pace of biological aging varies between people independently of chronological age and mitochondria dysfunction is a key hallmark of biological aging [30, 31]. The role of mitochondrial DNA 5mC methylation in aging remains an emerging and debated area of research. While nuclear DNA methylation is a well-characterized epigenetic regulator of aging [32], mtDNA methylation, once considered negligible due to the absence of histones and compact genome structure, has gained attention as a potential modulator of mitochondrial function and longevity. Studies suggest that age-associated alterations in mtDNA 5mC patterns may influence the expression of mitochondrial genes critical for oxidative phosphorylation, potentially contributing to mitochondrial dysfunction, reactive oxygen species (ROS) accumulation, and cellular senescence [33]. To investigate whether 5mC methylation differences exist between young and aged individuals, we selected C57BL/6 mice representing five age groups: 1 month, 3 months, 5 months, 12 months, and 21 months. These correspond to human age ranges of 0–10 years, 10–20 years, 20–30 years, 40–50 years, and 60–70 years, respectively (Fig. 4a). Brain [34] and thymus [35] tissue samples were collected from these five age groups for mtDNA extraction and WGBS analysis (Fig. S6a). Results revealed numerous 5mC methylation sites with distinct methylation levels between young and aged mice. In brain tissues, mitochondrial genomes across age groups exhibited a strong bias toward L-strand CpG methylation, with conserved peak loci predominantly localized at gene-gene boundaries (Fig. 4b, c). This pattern, however, was not statistically significant in thymus tissues of different ages (Fig. S6b, c). Strand-specific mitochondrial DNA (mtDNA) methylation dynamics in aging mouse tissues. (a) Sample overview: Brain and thymus tissues collected from mice at ages corresponding to human developmental stages (youth to old age). (b) Methylation profiling: Bisulfite sequencing (BS-seq; whole-genome bisulfite sequencing) maps of linearized mtDNA in brain tissues. Annotations between heavy (H)- and light (L)-strand methylation profiles denote positions of mtDNA functional elements (n = 3). (c) Quantitative analysis: CG and non-CG (CH) methylation levels on H and L strands. Mean methylation: average methylation at cytosines with > 10× sequencing coverage. Total methylation: sum of methylation values across all cytosines with ≥ 10× coverage. Left: All qualifying cytosines; Right: Subset with > 1% methylation. Data shown as mean ± SEM We collected serum samples from mice across the five age groups to assess the expression of aging-related biomarkers, including IGF-1 (insulin-like growth factor) [36], NFL (neurofilament light chain) [37], sTREM-2 (soluble triggering receptor expressed on myeloid cells 2) [38], p-Tau (phosphorylated tau protein) [39], GFAP (glial fibrillary acidic protein) [40], and t-tau (total tau protein) [41, 42]. Serum biochemical analysis revealed age-dependent changes, IGF-1 expression significantly decreased in 12- and 21-month-old mice compared to 1-month-old controls (p < 0.05; Fig. S7a). NFL levels were markedly reduced in 21-month-old mice versus 1-month-old mice (p < 0.05; Fig. S7b). sTREM-2 expression showed significant reductions in 3-, 5-, 12-, and 21-month-old mice relative to 1-month-old mice (p < 0.05; Fig. S7c). p-Tau levels increased significantly in 3-, 5-, 12-, and 21-month-old mice compared to 1-month-old controls (p < 0.05; Fig. S7d). GFAP and T-tau expression trended downward in older age groups (3, 5, 12, and 21 months) compared to 1-month-old mice, but these changes were not statistically significant (Fig. S7e, f). These findings suggest that the observed age-related changes in biomarker expression may be linked to mitochondrial DNA 5mC methylation. Furthermore, MEEs represent a promising tool for investigating epigenetic mechanisms in aging. MEEs enable efficient age-related mtDNA methylation in vivo To evaluate the in vivo methylation efficiency of mitochondrial epigenetic editors (MEEs) and their therapeutic potential, we targeted three age-associated mtDNA methylation sites in mouse brain tissues: C2383 (H) in MT-RNR2, C3933 (H) in MT-ND2, and C11168 (H) in MT-ND4. We designed MEEs for these sites and packaged them into PHP.eB AAV (Fig. 5a). A cohort of 23 four-week-old mice was divided into five experimental groups: C2383 (H) methylation, C3933 (H) methylation, C11168 (H) methylation, GFP PHP.eB-virus control, and PBS-injection control. Each mouse received an intravenous injection of 10¹¹ vg PHP.eB AAV. Three weeks post-injection, brain tissues (cerebellum, cortex, hippocampus, striatum) and plasma were collected for analysis (Fig. 5b). Two GFP control mice were assessed for hippocampal GFP expression via fluorescence staining, confirming high transduction efficiency (Fig. 5c). Total RNA and mtDNA were extracted from cerebellum tissues. The mtDNA was used for mitochondrial genome methylation sequencing, site C11168 (H) revealing 11.76% methylation efficiency in cerebellum tissues (Fig. 5d). RNA sequencing and qPCR assessed off-target effects and aging-related gene expression. Compared to GFP controls, C2383-MEE significantly downregulated MT-RNR2 in the cortex, C3933-MEE reduced MT-ND2 in the hippocampus and striatum, and C11168-MEE suppressed MT-ND4 in the cerebellum (p < 0.05) (Fig. 5e). MEE-mediated age-related mtDNA 5mC methylation in vivo. (a) Schematic of the PHP.eB AAV vector design, highlighting the inverted terminal repeats (ITRs) and the MEE expression cassette. (b) Experimental timeline for PHP.eB virus injection in 4-week-old C57BL/6J mice. Groups: MEE-targeted sites (C2383, C3933, C11168), GFP-PHP.eB control, and PBS control. (c) GFP expression in the central nervous system (CNS) following intravenous PHP.eB-AAV delivery, confirming broad transduction. Scale bar: 50 μm (left); 200 μm (right). (d) Methylation efficiency at mtDNA sites C2383 (H), C3933 (H), and C11168 (H) in cerebellum, cortex, hippocampus, and striatum, quantified by mitochondrial genome bisulfite sequencing (n = 5 mice/group). (e) Relative mRNA levels of mitochondrial gene MT-RNR2 in brain regions after MEE-mediated methylation (n = 5 mice/group). (f) Relative mRNA levels of mitochondrial gene MT-ND2 in brain regions after MEE-mediated methylation (n = 5 mice/group). (g) Relative mRNA levels of mitochondrial gene MT-ND4 in brain regions after MEE-mediated methylation (n = 5 mice/group). (h) Plasma levels of aging biomarkers (t-Tau, NFL, p-Tau) in mice treated with MEEs (C2383, C3933, C11168; n = 5/group), GFP-AAV (n = 3), or PBS (n = 3) Similar trends were observed versus PBS controls (Fig. 5e). Biochemical analysis showed altered aging-associated plasma biomarkers, including p-Tau in the C2383-MEE group, t-Tau in the C3933-MEE group, and t-Tau and NFL in the C11168-MEE group (Fig. 5h). Furthermore, through mitochondrial genome methylation library construction and sequencing analysis, no additional differentially methylated off-target sites were detected in the brain tissue samples across all groups. RNA-seq revealed 372 (C2383), 127 (C3933), and 42 (C11168) down-regulated genes (DRGs) versus GFP controls (Fig. S8a), and 292, 90, and 31 DRGs versus PBS controls (Fig. S8b, Supplementary Table S10, 11). These results demonstrate that MEEs can mediate mtDNA methylation in vivo and perturb age-associated epigenetic sites, establishing MEEs as a useful tool for mitochondrial epigenome research and future mechanistic studies. Discussion The development of TALE-directed mitochondrial epigenetic editors (MEEs) for 5mC modification provides a programmable approach to perturb mtDNA methylation in cells and in vivo. By enabling targeted manipulation of a poorly understood mitochondrial epigenetic mark, this tool may help dissect the relationship between mtDNA methylation, mitochondrial gene expression and aging-associated phenotypes. Our demonstration of MEEs targeting loci such as C3951 (H) in MT-ND1 (Fig. 2d, e) with limited detectable off-target activity under the tested conditions (Fig. 3) supports their utility as a platform for probing mtDNA-environment interactions, particularly in the context of aging. The relationship between MEE-induced methylation and mitochondrial RNA reduction was not strictly linear across all target sites. This likely reflects the fact that steady-state mtRNA levels are determined by multiple layers of regulation, including local editor occupancy, methylation status, strand context, transcript processing, RNA stability and turnover. Therefore, methylation efficiency at an individual cytosine should not be expected to directly predict the magnitude of mature RNA reduction. Reconstituted in vitro transcription assays with POLRMT and defined methylated mtDNA templates will be important in future studies to determine whether 5mC directly affects transcription initiation, elongation or RNA processing. Because mammalian mtDNA is transcribed as polycistronic transcripts from heavy- and light-strand promoters, changes in mature mitochondrial RNA levels should not be interpreted solely as altered transcription initiation. Mature mtRNA abundance is shaped by transcript elongation, processing, stabilization and degradation. Recent work by Miranda et al. [43] showed that POLRMT overexpression increases mitochondrial transcription initiation but does not necessarily increase mature mitochondrial mRNA abundance, indicating that post-transcriptional regulation can buffer mitochondrial transcript output. Therefore, the qPCR and RNA-seq results in this study are best interpreted as changes in steady-state mitochondrial RNA abundance rather than direct measurements of transcriptional capacity. The age-associated changes in serum biomarkers [44], including t-Tau, NFL, and p-Tau (Fig. 5h), were observed alongside shifts in mtDNA 5mC methylation patterns, suggesting a potential association between mitochondrial epigenetic remodeling and systemic aging. Notably, the tissue-specific L-strand CpG methylation bias in brain mitochondria (Fig. 4) suggests that mtDNA methylation may be linked to mitochondrial transcription, replication fidelity, or RNA metabolism in metabolically active tissues that are disproportionately affected by aging. The absence of this pattern in thymus tissues (Fig. S6) further highlights the organ-specific nature of mitochondrial aging, possibly reflecting differences in oxidative stress or mitochondrial turnover rates. MEEs now provide a tool to test these hypotheses directly, for example, by selectively methylating sites near MT-ND1 (a subunit of Complex I) to determine whether methylation exacerbates or mitigates age-related electron transport chain (ETC) dysfunction [45]. Compared to existing mtDNA manipulation tools such as CRISPR-based systems [46] or mitoTALENs [47], MEEs offer a complementary strategy for aging studies. TALE proteins circumvent the challenge of delivering guide RNA into mitochondria, a critical limitation for CRISPR applications in mitochondrial editing [48]. The ability to modify mtDNA methylation without introducing double-strand breaks or permanent sequence mutations enables epigenetic perturbation that may better mimic naturally occurring methylation changes. Nevertheless, the reversibility, dose-dependence and long-term stability of MEE-induced mtDNA methylation remain to be systematically evaluated. The modular design of MEEs permits retargeting to other loci linked to aging, such as the D-loop regulatory region or mitochondrial tRNA genes, which accumulate mutations and methylation changes with age [49]. The observed increase in p-Tau levels in older mice (Fig. 5h), a hallmark associated with Alzheimer’s disease (AD) [50], suggests that MEEs may be useful for modeling mitochondrial contributions to neurodegeneration. For instance, targeted methylation of mtDNA regions associated with reactive oxygen species (ROS) production [51] could help test whether epigenetic modulation of ETC genes influences tau-related pathology through oxidative stress. Similarly, future studies attempting to restore youthful methylation patterns in aged mitochondria could test whether age-related biomarker changes are reversible, offering insights into potential epigenetic rejuvenation strategies [52]. Limitations and future directions While MEEs provide a programmable approach for mtDNA methylation perturbation, several key challenges remain. An important technical limitation of this study is the use of conventional bisulfite-based methylation analysis. Bisulfite sequencing cannot distinguish 5mC from 5hmC and may introduce additional biases due to DNA fragmentation, loss of converted DNA molecules, strand- or sequence-dependent coverage imbalance, and reduced mapping complexity. These issues are particularly relevant for mtDNA methylation analysis because apparent methylation signals may be affected by NUMT contamination, incomplete conversion or selective loss of unmethylated cytosine-rich fragments. Guitton et al. [53] highlighted that methodological artifacts can lead to overestimation of mtDNA non-CpG methylation. In this study, we attempted to minimize these effects by enriching mtDNA, using λ DNA as an unmethylated conversion control, applying coverage thresholds and retaining uniquely mapped mtDNA reads. Nevertheless, future studies should validate key findings using complementary approaches such as EM-seq, oxBS/TAB-based methods or long-read methylation detection to more accurately distinguish 5mC from oxidized cytosine derivatives. The current system introduces 5mC modifications, and complementary tools for demethylation or hydroxymethylation are needed to fully interrogate the dynamic nature of mitochondrial epigenetics [26]. The mechanism by which MEE perturbation reduces steady-state mitochondrial RNA abundance also remains unresolved. This effect may involve methylation-dependent changes in POLRMT progression, physical occupancy of the TALE editor, altered RNA processing, or changes in RNA stability. In addition to cellular assays, in vitro transcription assays using reconstituted POLRMT machinery and defined methylated mtDNA templates would provide direct insight into how 5mC affects mitochondrial transcription under simplified conditions. Additionally, in vivo delivery efficiency and tissue-specific targeting must be optimized to study systemic aging phenotypes. Protein-level verification was performed in the cell model for selected targets; however, mitochondrial protein expression could not be systematically evaluated in AAV-transduced mice because the available tissue samples had been exhausted. Future in vivo studies should include protein-level analysis of mitochondrial OXPHOS components to correlate methylation, RNA abundance and protein expression. This work mainly focused on molecular and biochemical phenotypes mediated by MEEs, whereas physiological, behavioral and cognitive outcomes following in vivo editing have not been well characterized. Although the in vivo experiments demonstrate that MEEs can modify age-associated mtDNA methylation sites and alter selected aging-related biomarkers, this study does not establish a causal relationship between mtDNA methylation and organismal aging. Cellular senescence assays, including proliferation analysis and SA-β-gal staining, as well as longitudinal behavioral and physiological studies, will be required to determine whether targeted mtDNA methylation changes are sufficient to induce or modify aging-related phenotypes. Future work could integrate MEEs with single-cell multi-omics to map methylation-transcription-metabolism relationships across aging trajectories or combine them with senolytic therapies to assess how mitochondrial epigenetics influences cellular senescence. Age-related methylation and its association with behavioral and psychological phenotypes require further systematic assessment, including rigorous evaluation criteria and longitudinal research methodologies. Conclusions By enabling targeted perturbation of mtDNA methylation, MEEs provide a tool for moving from correlative observations toward causal testing of mitochondrial epigenetic regulation. Their ability to model selected age-associated methylation changes, examine epigenetic perturbations and dissect tissue-specific mitochondrial responses may facilitate future studies of mitochondrial dysfunction in aging and age-related disease. As aging populations face increasing burdens of neurodegenerative and metabolic disorders, MEEs offer a useful experimental platform for exploring whether mitochondrial epigenetic modulation can contribute to healthspan-related interventions. Methods and materials Plasmid design and construction The D3A-D3L fusion sequence was sourced from the CRISPRoff-v2.1 plasmid (Addgene, #167981) [20]. The MTS signal sequence “MLGFVGRVAAAPASGALRRLTPSASLPPAQLLLRAAPTAVHPVRDYAAQTSESGGGGSPG” was synthesized and constructed in one CMV promoter plasmid. TALEs were designed and constructed. Repeat variable diresidues (RVDs) were designed to recognize each nucleotide in the target sequence using the following code: HD for cytosine (C), NG for thymine (T), NI for adenine (A), and NN for guanine (G). TALE arrays were designed using TALE-NT to ensure specificity and avoid off-target binding. A TALE expression backbone pTAL was used, containing a CMV promoter, MTS, Dnmt3A_3L and ampicillin antibiotic resistance. TALE repeat modules were cloned into the backbone vector via BsmBI restriction enzyme-mediated Golden Gate assembly. Reaction conditions were 10 cycles of 37 °C (2 min) and 16 °C (5 min), followed by 50 °C (5 min) for enzyme inactivation. Assembled plasmids were transformed into E. coli DH5α competent cells and plated on LB-agar with ampicillin (100 µg/mL). Colonies were screened via colony PCR. Plasmid DNA from positive clones was Sanger-sequenced using primers flanking the TALE array to confirm RVD order and integrity. Plasmids were digested with EcoRI/HindIII to verify expected fragment sizes. The sequences of the expression plasmid are listed in Table S1. Targeted TALE sequences are listed in Table S4. Cell culture and transfection HEK293T cells were maintained in Dulbecco’s Modified Eagle Medium (DMEM, Thermo Fisher Scientific, #11965092) supplemented with high glucose, sodium pyruvate, GlutaMAX™, 10% fetal bovine serum (FBS, Thermo Fisher Scientific, #10100147), and Penicillin-Streptomycin (PS, Thermo Fisher Scientific, #15140148). Cells were cultured in 100-mm dishes until reaching 80–90% confluency, after which they were seeded into 6-well plates for transfection experiments. For transfection, cells were seeded at a density of 70% confluence and allowed to adhere for 12–16 h. Prior to transfection, the culture medium was replaced with 500 µL (24-well plates) or 750 µL (12-well plates) of fresh medium. Transfections were performed using the EZ Cell Transfection Reagent (Shanghai Life iLab Biotech Co., Ltd., #AC04L098) according to the following protocol. For 6-well plates: 3 µg of plasmid DNA was diluted in 120 µL of serum-free DMEM. Separately, 9 µL of EZ-trans reagent was mixed with 120 µL of serum-free DMEM. The diluted plasmid DNA and transfection reagent were combined, incubated at room temperature for 15 min, and gently added to the cells. Cell harvesting and genomic DNA isolation Following aspiration of the culture medium, adherent cells were gently washed with 1× Dulbecco’s phosphate-buffered saline (PBS; Welgene). Cells were trypsinized, detached, and pelleted via centrifugation (500 × g, 4 °C, 4 min). The trypsin-containing supernatant was discarded, and the cell pellet was resuspended in PBS for an additional wash. After complete PBS aspiration, genomic DNA was extracted using the DNeasy Blood & Tissue Kit (Qiagen) in accordance with the manufacturer’s protocol. mtDNA extraction For mtDNA extraction, tissue samples were collected and washed with cold PBS, then homogenized in isolation buffer (0.3 M sucrose, 10 mM Tris-HCl, 1 mM EDTA, pH 7.4) and centrifuged at 1000 g for 10 min to pellet nuclei. The supernatant was centrifuged at 10,000 g for 10 min to collect the mitochondrial fraction. The pellet was resuspended in lysis buffer and incubated on ice for 30 min, followed by centrifugation at 16,000 g for 10 min. The supernatant was treated with RNase A and Proteinase K, then incubated at 50 °C for 1 h. DNA was precipitated with isopropanol, washed with 70% ethanol, and resuspended in TE buffer. Finally, mtDNA concentration and purity were measured using a NanoDrop spectrophotometer, and integrity was assessed via agarose gel electrophoresis. To minimize potential NUMT contamination, mtDNA was isolated from mitochondria-enriched fractions before bisulfite conversion. For targeted BSP assays, all primer pairs were evaluated by in silico PCR/BLAST against both the nuclear and mitochondrial genomes, and only primers uniquely matching mtDNA were used. For WGBS analysis, bisulfite-converted reads were aligned to a combined nuclear-plus-mitochondrial reference genome. Only uniquely mapped mtDNA reads with high mapping quality were retained for downstream methylation calling, whereas multi-mapping reads or reads with potential NUMT ambiguity were discarded. Mitochondria protein isolation For analyzing changes in mitochondria protein levels by Western blotting, we used Cell Mitochondria Isolation Kit (Beyotime, #C3601) to isolate mitochondria protein of HEK293T cells. 7 × 106-9 × 106 GFP-positive cells were harvested by FACS and resuspended 1mL lysis buffer containing 1% phenylmethylsulfonyl fluoride (PMSF), then incubating on ice for 15 min. After homogenizing, the cell suspension was centrifuged at 600 g, 4℃ for 10 min. The supernatant was transferred into a new tube, followed by centrifugation at 11,000 g, 4℃ for 10 min. The pellet was resuspended with 50µL lysis buffer and boiled with 15µL 4×LDS Loading buffer (Invitrogen, #NP0008) at 95℃ for 10 min, subjected to following SDS-PAGE. Cell DNA samples sodium bisulfite treatment Bisulfite conversion of mtDNA was performed using the EpiTect Bisulfite Kit (Qiagen). DNA (200 ng for PFC samples or 1 µg for cell line samples) was treated following the manufacturer’s “Bisulfite DNA Conversion” protocol (Sodium Bisulfite Conversion of Unmethylated Cytosines in DNA), with modifications: an additional cycle (5 min denaturation at 95 °C, 2 h incubation at 60 °C, and final hold at 20 °C) was included to enhance conversion efficiency. Post-conversion cleanup was conducted using the “Cleanup of Bisulfite-Converted DNA” protocol (Sodium Bisulfite Conversion of Unmethylated Cytosines in Small Amounts of Fragmented DNA), incorporating carrier RNA during spin column purification to maximize DNA recovery. Residual RNA was digested by incubating with 1 µL RNase A (Qiagen) at 70 °C for 10 min with agitation in a final volume of 200 µL. RNase A was eliminated via phenol-chloroform extraction, and DNA was precipitated with isopropanol. Bisulfite sequencing PCR Genomic DNA was isolated from 2 × 106 cells using the PureLink™ Genomic DNA Mini Kit (Invitrogen, #K182001) according to the manufacturer’s protocol. For bisulfite conversion, 1 µg of genomic DNA per sample was treated with the EpiTect® Bisulfite Kit (QIAGEN, #59104), followed by purification to remove unconverted DNA. Bisulfite-converted DNA was amplified using EpiMark® Hot Start Taq DNA Polymerase (New England Biolabs, #M0490S) with locus-specific primers. PCR products were resolved on a 1.5% agarose gel, excised, and purified using the Zymoclean™ Gel DNA Recovery Kit (Zymo Research, #D4007). Purified amplicons were re-amplified with EpiMark® Hot Start Taq to ensure sufficient yield for cloning. Amplicons were ligated into the pMD19-T Vector using the pMD19-T Vector Cloning Kit (Takara Bio, #6013) following the manufacturer’s guidelines. Ligation products were transformed into chemically competent E. coli DH5α cells (Takara Bio, #9057), plated onto LB agar containing 100 µg/mL carbenicillin and 40 µg/mL X-gal, and incubated overnight at 37 °C for blue-white colony screening. For each experimental condition, at least 20 white colonies were randomly selected, cultured in LB broth with carbenicillin, and plasmid DNA was isolated. Insert sequences were verified by Sanger sequencing Primer sequences for bisulfite-PCR amplification are listed in Table S2. Gene expression detection At 72 h post-transfection, 2 × 106 GFP-positive cells were isolated by fluorescence-activated cell sorting (FACS) using a BD FACSAria™ III system (BD Biosciences). Total RNA was extracted from sorted cells using RNA Isolater Total RNA Extraction Reagent (Vazyme, #R401-01) according to the manufacturer’s protocol. RNA was reverse-transcribed into cDNA using HiScript® II Q Select RT SuperMix (+ gDNA wiper) (Vazyme, #R233-01) to remove genomic DNA contamination. Quantitative PCR (qPCR) was performed on a QuantStudio™ 7 Pro Real-Time PCR System (Applied Biosystems, #A43055) using AceQ® Universal SYBR® qPCR Master Mix (Vazyme, #Q121-02). Reactions were run in triplicate for each sample, with primers listed in Table S3. Relative gene expression levels were calculated using the -ΔΔCt method. Ct values of target genes were normalized to the reference gene β-actin, and fold changes were determined relative to control groups. Statistical significance was assessed across three biological replicates. Protein expression analysis by Western Blot To validate Dnmt3A knockout (KO) and overexpression (OE) in HEK293T cells, protein lysates were analyzed by Western blot as follows: (1) Sample Preparation, HEK293T cells were harvested from three experimental groups (wild-type controls, Dnmt3A-ko cells generated via CRISPR/Cas9 editing, Dnmt3A-OE cells transfected with a Dnmt3A overexpression plasmid). Cells were lysed in RIPA buffer containing protease inhibitors, and protein concentrations were quantified using a BCA assay. (2) Equal protein amounts (20–30 µg) were resolved on 10% SDS-PAGE gels and transferred to PVDF membranes. Membranes were blocked in 5% non-fat milk/TBST and probed overnight at 4 °C with primary antibodies (Anti-Dnmt3A, 1:1,000; Abcam, #ab307503) and anti-β-actin (1:5,000; loading control, Abcam, #ab179467). HRP-conjugated secondary antibodies (1:10,000) were applied for 1 h at room temperature. Signals were visualized using enhanced chemiluminescence (ECL) and imaged on a imager (BIO-RAD Gel Doc™ XR). To assess protein expression of ND5, COX1, and COX2 following MEE editing, mitochondrial protein extracts were separated by SDS–PAGE and analyzed by Western blotting, with COXIV used as the internal loading control. The following primary antibodies were employed: anti-ND5 (1:1,000; Absin, #abs148283), anti-COX1 (1:1,000; Selleck, #F1628), anti-COX2 (1:1,000; Selleck, #F2349), and anti-COXIV (1:1,000; STARTER, #S0B0510). Whole genome bisulfite sequencing and analysis To investigate Dnmt3A’s role in mitochondrial DNA (mtDNA) methylation, whole-genome bisulfite sequencing (WGBS) was performed on the following groups, each with three biological replicates: WT (untreated) cells, Dnmt3A knockout cells, and Dnmt3A over-expressing cells. This generated WGBS data for 9 samples in total. To assess off-target methylation effects of MEE (mitochondrial editing effector), WGBS libraries were prepared for 15 mtDNA samples, including, WT (untreated) cells, GFP transfection control cells, and three MEE-targeted site experiments. Each condition was analyzed in three biological replicates. All WGBS libraries were prepared and sequenced by GENEWIZ. The raw data were processed with Cutadapt (v1.9.1) to trim adapter sequences and remove low-quality reads, generating clean data. Using Bismark (v0.22.3), the clean reads were aligned to a combined reference containing both the nuclear genome and mitochondrial genome to reduce ambiguity between mtDNA reads and NUMT-derived reads. PCR duplicates were then eliminated from the alignment files using the deduplicate_bismark module. Only uniquely mapped reads assigned to mtDNA with high mapping quality were retained for mitochondrial methylation analysis, whereas multi-mapping reads and reads with potential NUMT ambiguity were removed. Methylation levels at cytosine bases were identified across the genome via the bismark_methylation_extractor module, with parameters set to: --comprehensive, --merge_non_CpG, --CX, and --no_overlap. Default settings were retained for all other steps. MEEs’ off-target sites statistics Differential methylation analysis of mCpG sites was conducted using methylKit (v0.7.12). Comparisons included: (1) C1/C2/C3/GFP group samples versus WT controls and (2) C1/C2/C3 group samples versus GFP controls. The workflow comprised three steps: Methylation rate differences between experimental and control groups were quantified. Fisher’s exact test was applied to calculate P-values, which were adjusted to Q-values using false discovery rate (FDR) correction. Significantly differentially methylated sites were defined as those with a methylation rate difference > 25% and a Q-value < 0.01. Animal studies All mouse experiments were conducted in compliance with national and provincial regulations for laboratory animal welfare, including guidelines from the State Science and Technology Commission of China, the Ministry of Science and Technology of China, and the Zhejiang Province Administration of Laboratory Animals. Adult C57BL/6J mice were obtained from MSLT BIOTECH (Hangzhou, China) and housed in a specific pathogen-free (SPF) facility under controlled conditions: 12-hour light/dark cycle, ambient temperature 20–22 °C, and humidity 40–60%. Animals were provided ad libitum access to food and water in standard cages. Mice at designated ages (1, 3, 5, 12, and 21 months) were euthanized, followed by retro-orbital blood collection and dissection of brain and thymus tissues. Blood samples were centrifuged to isolate serum. Serum levels of IGF-1, NFL, sTREM-2, p-Tau, GFAP, and T-tau were quantified by MSLT BIOTECH using standardized assays. Statistical analysis Statistics and reproducibility Statistical analyses were conducted using GraphPad Prism v.8 (GraphPad Software). Data are presented as mean ± SEM, derived from a minimum of three biologically independent experiments or samples. Significance was assessed using unpaired two-tailed Student’s t-tests, with p < 0.05 defined as statistically significant. Significance levels are denoted as follows: * p < 0.05*, ** p < 0.01, and ns (non-significant) for p ≥ 0.05. Western blotting (WB) experiments were each independently replicated twice, with consistent results observed across replicates. Data availability The WGBS and RNA-seq data for MEEs-mediated mtDNA methylation and off-targets detection are available in the National Center for Biotechnology Information (NCBI) Sequence Read Archive (SRA) database under accession code PRJNA1196097. The hg38 reference genome (Homo sapiens GRCh38.p14) was used for aligning HEK293T cell RNA-seq data. There are no restrictions on data availability, and source data are provided with this paper. Supplementary Information Supplementary Material 1. Supplementary Material 2. Authors’ contributions W. J. and G. L. conceived the project and designed the study. Experiments were designed by W. J. and G. L., and the manuscript was written by G. L. G. L., with the assistance of Y. C., X. Z., and G. Z. conducted the experiments. X. L. designed the TALE modules and constructed the expression plasmids. Y. P., and X. G. analyzed the RNA-seq data. G. L. conducted the WB experiments. W. J., X. H. and X. H. contributed to the discussion of the manuscript. Y. Y., Z. G., W. J. and G. L. took funding acquisition for this study. Funding This work was partially supported by National Natural Science Foundation of China (grant no. 32401241 to Guo Li), and the “Pioneer” and “Leading Goose” R&D Program of Zhejiang (grant no. 2024SSYS0101 to Zhimin Gu), Chongqing Natural Science Foundation (grant no. CSTB2023NSCQ-MSX0456 to Wen Jiang) and Senior Medical Talents Program of Chongqing for Young and Middle-aged (to Wen Jiang). Data availability The WGBS and RNA-seq data for MEEs-mediated mtDNA methylation and off-targets detection are available in the National Center for Biotechnology Information (NCBI) Sequence Read Archive (SRA) database under accession code PRJNA1196097. The hg38 reference genome (Homo sapiens GRCh38.p14) was used for aligning HEK293T cell RNA-seq data. There are no restrictions on data availability, and source data are provided with this paper. Declarations Competing interests The authors declare no competing interests. Conflict of interest The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Guo Li, Xinzhi Zhou, Guanglin Zhu and Yaxian Cheng contributed equally to this work. Contributor Information Guo Li, Email: liguo@xhlab.ac.cn. Yuan Yao, Email: yyao1@zju.edu.cn. Zhimin Gu, Email: guzhimin2006@163.com. Wen Jiang, Email: jiangwen_work@126.com. References Associated Data Supplementary Materials Supplementary Material 1. Supplementary Material 2. Data Availability Statement The WGBS and RNA-seq data for MEEs-mediated mtDNA methylation and off-targets detection are available in the National Center for Biotechnology Information (NCBI) Sequence Read Archive (SRA) database under accession code PRJNA1196097. The hg38 reference genome (Homo sapiens GRCh38.p14) was used for aligning HEK293T cell RNA-seq data. There are no restrictions on data availability, and source data are provided with this paper. The WGBS and RNA-seq data for MEEs-mediated mtDNA methylation and off-targets detection are available in the National Center for Biotechnology Information (NCBI) Sequence Read Archive (SRA) database under accession code PRJNA1196097. The hg38 reference genome (Homo sapiens GRCh38.p14) was used for aligning HEK293T cell RNA-seq data. There are no restrictions on data availability, and source data are provided with this paper.