CRISPR-Free Mitochondrial DNA Base Editing

A New Variation on Base Editing Offers a Promising Pathway Toward the Treatment of Mitochondrial Diseases, Previously Intractable to CRISPR-Based Technologies

Since the first report that mitochondrial DNA (mtDNA) base substitution mutations can cause inherited disease,^1–3 hundreds of clinically relevant mtDNA mutations have been reported and listed in our MITOMAP database. ⁴ The pressing question now is how can these pathogenic mtDNA mutations be treated given that each cell harbors hundreds to thousands of copies of the mtDNA, the majority of which will need to be corrected for the desired clinical effect. A recent report in Nature—a collaboration between the labs of two Howard Hughes Medical Institute (HHMI) investigators, Joseph Mougous and David Liu—has solved this problem and brought us significantly closer to having an mtDNA gene therapy. ⁵

The minimum prevalence of primary mtDNA disease mutations has been estimated to be at least 1 in 5000. ⁶ However, on average each person carries one heteroplasmic mtDNA mutation and one in eight individuals harbor a disease-associated mutation at ≥1% heteroplasmy (mixture of mutant and normal mtDNAs). ⁷ Moreover, the mtDNA has been accumulating sequence variants along radiating maternal lineages throughout human history, starting in Africa and followed approximately 65,000 years ago by two mtDNA lineages leaving Africa and colonizing the rest of the planet. ⁸

These radiating mtDNA lineages are punctuated by clusters of related mtDNA haplotypes, designated haplogroups. The originally geographically constrained haplogroups were each founded by functional mutations that permitted groups of human ancestors to alter their mitochondrial metabolism to adapt to regional environmental conditions. ⁹ However, these same haplogroups in a different environment can be maladaptive and predispose to a variety of common diseases. ¹⁰

Clinically relevant mtDNA variants can be either heteroplasmic or homoplasmic (all mtDNAs having the same variant). Recently arising pathogenic mutations start out as heteroplasmic; the more severe remain heteroplasmic because at high percentages of mutant they are lethal. By contrast, milder pathogenic mutations, such as those that cause Leber's hereditary optic neuropathy, can segregate to essentially homoplasmic. Furthermore, all haplogroup variants are homoplasmic having segregated to pure mutant over thousands of years.^11,12

Interventions to eliminate heteroplasmic mutations have been greatly advanced by the laboratory of Carlos Moraes (University of Miami, Miami, FL). His group has employed mitochondrially targeted restriction enzymes^13–15 or a two-part FokI nuclease directed to a specific mtDNA sequence linked to a pair of DNA sequence–specific polypeptides known as TAL(transcription activated-like) effector nucleases (TALENs).^16,17 Both of these approaches cleave the mtDNA at the mutant nucleotide, resulting in its degradation. While these approaches effectively remove mutant mtDNAs, they have two limitations: They are only applicable for heteroplasmic mtDNA mutants, and they transiently reduce the mtDNA copy number, which may be toxic to patients with high proportions of mutant mtDNAs.

The new method presented by Beverly Mok, Marcos de Moraes, and colleagues overcomes these limitations by employing a mitochondrially targeted double-strand DNA cytidine deaminase. TALENs, in some respects forgotten tools in the genome editors' arsenal, are used to bind the deaminase to specific regions of the mtDNA.

To avoid the toxicity of the cytidine deaminase, the enzyme is divided into two portions, with each portion bound to opposing strand TALENs. The structure of each half TALEN is as follows: the mtDNA sequence-specific TALEN, a 2- to 16-amino-acid linker, either the N or C terminal portion of the cytidine deaminase, and one copy of a uracil glycosylase inhibitor; the inhibitor being used to block cleavage of the U containing mtDNA by endogenous uracil glycosylases.

Co-transformation of human cells with plasmids carrying the two complementary TALENs was shown to convert a “C” located adjacent to a “T” (5′-TC-3′) to a U resulting in C · G to T · A transitions. These constructs are designated as RNA-free DddA-derived cytosine base editors (DdCBEs). What is remarkable about the DdCBEs is the high proportion of the mtDNAs within a cell in which the target “Cs” can be converted to Ts. Using the cytidine deaminase “G1397-split” DdCBE, the average percentage of cellular mtDNAs that were edited was 42%. In the case of an MT-ND5 edited by DdCBE, the resulting missense mutation was associated with a reduction in the basal and maximum respiration rates. ⁵

While these are remarkable advances, there are still some limitations to the current DdCBE system. These include that only C to T transitions can be generated, the target C must be preceded by a T, and the C-to-T editing occurs within a stretch of mtDNA sequence between the two TALENs such that more than one C may be edited. ⁵ Still, it is likely that, as we have seen with the CRISPR-Cas9 system, this system may be modified to address these limitations (see Table 1).

OPEN IN VIEWER

Table 1.

Mitochondrial DNA point mutations amenable to base editing

The DdCBE system can now be used to generate human and mouse cell lines harboring novel mtDNA mutations in designated genes. These could be used to investigate the function of potentially interesting mtDNA-coded genetic elements such as the mtDNA control region or the recently discovered intragenic peptide hormones; humanin within the 16S rRNA, ¹⁸ or MOTS-c within the 12S rRNA gene. ¹⁹ The system may also be useful in analysis of the functional variants found in the mtDNA haplogroups. Until now, this has been problematic because, without recombination, it has been difficult to disentangle the effects of linked mtDNA polymorphisms within a mtDNA haplotype.

While the DdCBE method permits generation of a subset of regional mtDNA variants, mtDNA mutations have also been generated in culture cells ²⁰ and in mice^21–23 by substituting the normal nuclear DNA-coded mtDNA polymerase with a variant that lacks the proofreading function. This mtDNA mutation system has the advantage that virtually any mtDNA mutation can be obtained. However, it has the disadvantage that the initial mutation starts as a single mutant mtDNA among hundreds of nonmutant mtDNAs, so the mutant mtDNA must be enriched either by depletion-reamplification ²⁴ or transmission through the female germline bottleneck.^25,26 Also, because the mutations are random it is necessary to screen large numbers of mutant cell lines and/or mice to obtain a specific mtDNA mutation.

In its current state, the DdCBE system could be used to correct specific pathogenic mtDNA sequence variants in which the mutation changed a T to C to generate a TC dyad. A list of potentially treatable pathogenic mutations from MITOMAP is provided in Mok et al.'s supplemental Table 10, ⁵ excerpts in Table 1. However, to deliver the two large half TALEN-DdCBE polypeptides will require vectors with a large cargo capacity, which will preclude use of current adeno-associated viral vectors.