Mutations in the nuclear genome are the root cause of many diseases, and mitochondria, as semi-autonomous organelles within cells, possess their own genome. Mutations in the mitochondrial genome are also closely associated with various genetic diseases. Mitochondrial diseases typically affect multiple tissues and organs, with the most well-known examples being Leigh syndrome and LHON (Leber's hereditary optic neuropathy). Symptoms of Leigh syndrome include developmental delay, hypotonia, motor and respiratory disorders, while LHON is characterized by vision loss, central scotoma, and optic nerve atrophy. According to MITOMAP statistics, 97 pathogenic mitochondrial mutations have been validated, with point mutations accounting for as much as 95% of these. However, due to the lack of effective mouse models for point mutation-related mitochondrial diseases, research and therapeutic development in this field have been severely hindered.

 

Early mouse models were primarily constructed through chemical induction or genetic engineering [1]; however, these methods are complex, costly, and offer poor precision in controlling mutations, resulting in the successful establishment of only a few models. In recent years, researchers have successfully developed mitochondrial base editing tools that can edit mitochondrial DNA, enabling C-to-T and A-to-G conversions, such as DdCBEs and TALEDs. These tools are based on the double-stranded DNA deaminase DddA protein [2, 3]. Although attempts have been made to apply these tools in mouse models, their editing efficiency is still insufficient to simulate the high mutation loads characteristic of human mitochondrial diseases [4, 5]. Additionally, studies have shown that DdCBEs can trigger significant off-target effects in the nuclear genome. This non-TALE-dependent off-targeting primarily arises from the self-assembly of the DddA protein and its interaction with CTCF [6]. Therefore, DddA-based mitochondrial base editing tools face the risk of nuclear genome off-target effects in their applications, making it challenging to directly establish causal links between mitochondrial mutations and disease phenotypes.

 

To address this challenge, the research group led by Prof. Wensheng Wei at Peking University previously developed mitoBEs, a novel mitochondrial base editing tool that combines nickase and single-strand DNA deaminase, enabling C-to-T and A-to-G editing in mitochondrial DNA. Compared to DdCBEs and TALEDs, mitoBEs exhibit superior strand specificity and significantly reduced off-target effects. Due to its bidirectional base editing capability, mitoBEs can precisely model approximately 87% of pathogenic mitochondrial mutations [7].

 

On January 22, 2025, the Prof. Wensheng Wei’s research group at Changping Laboratory and Peking University published a study in Nature titled "Precise modelling of mitochondrial diseases using optimized mitoBEs". The study reports the successful creation of efficient and accurate mitochondrial disease mouse models using optimized mitoBEs. By using the enhanced version of mitoBEs, the research team successfully established mouse models with high mutation frequencies, which exhibited typical disease-associated phenotypes. Additionally, through hybridization experiments, they obtained precise mouse models with a mutation load of 100% and models with only single-base mutations.

 

 

To establish a direct link between mutations and disease phenotypes, it is crucial to eliminate off-target effects of base editing tools. When using mitoBEs for modeling, RNA-encoded mitoBEs need to be injected into mouse embryos. Therefore, the study first conducted a comprehensive evaluation of the off-target effects of the RNA-encoded mitoBE system. The results showed that RNA-encoded mitoABE exhibited widespread transcriptome off-target effects, while mitoCBE demonstrated a certain degree of mitochondrial genome off-targeting dependent on the APOBEC1 protein.

 

To improve the precision of mitoBEs, the study focused on optimizing the deaminases. For mitoABE, mutation screening revealed that TadA8e-V106W-V28F could significantly reduce transcriptome off-targeting to background levels (Figure 1). For mitoCBE, several existing cytosine deaminases were screened, and it was found that the TadA-derived cytosine deaminase CBE6d exhibited off-target effects on the mitochondrial genome close to background levels. Based on these optimization results, the research team named the improved mitoBEs as mitoBEs v2, which includes mitoABE v2 and mitoCBE v2 (Figure 1).

 

Additionally, the study systematically assessed the off-target effects of mitoBEs on the nuclear genome before and after optimization. The results showed that, whether pre-optimized or optimized, mitoBEs did not induce significant off-target effects on the nuclear genome, thus confirming their safety and reliability in gene editing.

 

Figure 1 Optimizing the specificity of mitoBEs

 

By aligning 85 human pathogenic mitochondrial DNA point mutations with the mouse mitochondrial genome, the study identified 70 editable sites. Further cell-level preliminary screening successfully achieved editing of 68 of these sites. Comparison revealed that mitoBEs v2, encoded by circular RNA (circRNA), exhibited higher editing efficiency compared to the mRNA-encoded tools. Therefore, the research team injected circRNA-encoded mitoBEs v2 into mouse embryos and performed embryo transplantation. The results showed that mitoBEs v2 achieved high editing efficiency in various F0 generation mouse models, with the mutation frequency reaching 82% in the mt-Nd5 A12784G F0 mouse model (Figure 2).

 

Additionally, the study systematically assessed the off-target effects of F0 generation mice on both the mitochondrial genome and the nuclear genome. The results indicated that no off-target effects were detected across the entire genome, suggesting that mitoBEs v2 can construct mitochondrial disease mouse models with a clean genetic background. More importantly, the editing results in the mitochondrial genome were found to be widely and persistently present across various tissues in the mice (Figure 2), and the edits were stably transmitted through maternal inheritance. Through further hybridization experiments, the research successfully obtained the mt-Nd5 A12784G mouse model, which exhibited 100% editing efficiency at the target site and contained only the targeted mutation.

 

Figure 2 mitoBEs v2 enables efficient establishment of mitochondrial disease mouse models

 

The mt-Atp6 T8591C and mt-Nd5 A12784G mutations correspond to the human mitochondrial pathogenic mutations m.T9191C and m.A13379G, which respectively lead to Leigh syndrome and LHON. The research team evaluated the disease phenotypes in the F0 generation mouse models with high mutation rates. The results showed that the mt-Atp6 T8591C mice exhibited significant cardiac dysfunction, which is consistent with the clinical features of Leigh syndrome. In contrast, the mt-Nd5 A12784G mice exhibited vision impairment resembling LHON (Figure 3).

 

Additionally, the study successfully constructed a single-base mutation model for mt-Nd5 A12784G, which only contained the target site mutation, by adjusting the TALE binding sites. These findings strongly demonstrate the efficiency and precision of mitoBE v2 in creating mitochondrial disease mouse models. This provides an important tool for further exploring the pathogenic mechanisms of mitochondrial diseases and developing novel therapeutic strategies.

 

Figure 3 Mouse model shows corresponding disease phenotype

 

Dr. Zongyi Yi from Peking University is the co-corresponding author of this paper. Dr. Xiaoxue Zhang from Changping Laboratory is the first author. Xue Zhang, Jiwu Ren, Jiayi Li, Xiaoxu Wei, and Dr. Ying Yu also made contributions to the study. This research was funded by Changping Laboratory, the National Natural Science Foundation of China, the Center for Life Sciences, and the China Postdoctoral Science Foundation.

 

Paper Link: https://www.nature.com/articles/s41586-024-08469-8

 

Reference:

1.         Stewart, J.B., Current progress with mammalian models of mitochondrial DNA disease. Journal of Inherited Metabolic Disease, 2021. 44(2).

2.         Mok, B.Y., et al., A bacterial cytidine deaminase toxin enables CRISPR-free mitochondrial base editing. Nature, 2020. 583(7817).

3.         Cho, S.I., et al., Targeted A-to-G base editing in human mitochondrial DNA with programmable deaminases. Cell, 2022. 185(10).

4.         Silva-Pinheiro, P., et al., A library of base editors for the precise ablation of all protein-coding genes in the mouse mitochondrial genome. Nature Biomedical Engineering, 2023. 7(5).

5.         Cho, S.I., et al., Engineering TALE-linked deaminases to facilitate precision adenine base editing in mitochondrial DNA. Cell, 2024. 187(1).

6.         Lei, Z.X., et al., Mitochondrial base editor induces substantial nuclear off-target mutations. Nature, 2022. 606(7915).

7.         Yi, Z.Y., et al., Strand-selective base editing of human mitochondrial DNA using mitoBEs. Nature Biotechnology, 2024. 42(3).