The ability to safely modify the DNA of living organisms may soon be within our grasp thanks to CRISPR-Cas9. Despite its immense promise, CRISPR-Cas9 gene editing has not yet been approved for widespread therapeutic use because of its potential for introducing unwanted mutations (Kosicki et al., 2018). Recent reports have proposed that homologous recombination following endonuclease cleavage is responsible for CRISPR-Cas9’s on-target mutagenicity. This is because DNA-binding proteins, such as endonucleases, interfere with double-strand break repair by competing with the repair machinery for the target site (Reijns et al., 2015; Kaiser et al., 2016; Sabarinathan et al., 2016). Moreover, it has been shown that mutation rates are higher in regions of high recombination and that indels are often associated with a high density of single nucleotide polymorphisms (SNPs) (Lercher and Hurst, 2002; Tian et al., 2008).

An excellent model for studying the relationship between endonuclease activity and on-target mutagenicity in a native genomic context is the movement of self-splicing introns in mitochondrial genomes. Self-splicing introns are a class of mobile element commonly found in the mitochondrial DNAs (mtDNAs) of plants, fungi, and protists (Lambowitz and Belfort, 1993; Burger et al., 2003; Lang et al., 2007; Brown et al., 2014; Smith and Keeling, 2015; Megarioti and Kouvelis, 2020). These elements can copy and paste themselves at specific insertion sites in the host mitochondrial genome via their ability to form a complex secondary structure, which functions as a homing endonuclease (Cech et al., 1994; Lambowitz and Zimmerly, 2004). The homing endonuclease recognizes and binds to its target site with high affinity, inducing a double-stranded break (DSB). The break is then repaired by homologous recombination using the intron-containing strand as a template, resulting in intron movement (Cech et al., 1994; Lambowitz and Zimmerly, 2004).

Although self-splicing mitochondrial introns are found in a diversity of taxa (Del Vasto et al., 2015; Mower, 2020; Wideman et al., 2020), they are particularly prevalent within the yeast genus Metschnikowia (Lee et al., 2020), many members of which are commonly found in the guts of beetles inhabiting morning glories and other flowers around the world. Metschnikowia mtDNAs harbour among the largest numbers of organelle self-splicing introns of all eukaryotes. The cox1 and cob genes from Metschnikowia mtDNAs are especially intron rich, with the number of introns varying from 0–20 within cox1 and 0–13 within cob. These introns can be found at 25 and 13 unique intron insertion sites across cox1 and cob, respectively (Supplementary Tables S1A, B), making this genus the largest known reservoir of self-splicing, mobile organelle introns (Férandon et al., 2010; Lee et al., 2020). For detailed summary statistics on the mitochondrial introns of Metschnikowia yeasts, including the variation in intron number across species/strains, the types of introns (e.g., group I or group II), and the presence/absence of intronic homing endonucleases, please refer to Supplemental Information of Lee et al. (2020).

Large variations in intron number can lead to remarkable differences in gene and genome size (Del Vasto et al., 2015; Mower, 2020; Wideman et al., 2020). Among Metschnikowia mtDNAs, there is a 30-fold size range for cox1 and a 20-fold size range for cob, which encompasses the entire known size variation for both genes across the eukaryotic domain (Lee et al., 2020). More astonishing is that even within members of the same species, the size of cox1 has been found to vary by up to 5 kb (Lee et al., 2020) — and keep in mind that the mature cox1 mRNA from Metschnikowia species is less than 1.6 kb (Figure 1). Metschnikowia mitogenomes have other notable features, including extreme differences in intergenic content and genomic structure (e.g., linear vs circular mitochondrial chromosomes), both between and within species.

The co-occurrence of mtDNA diversity and large numbers of self-splicing introns in Metschnikowia may not be coincidental (Lee et al., 2020). If the movement of self-splicing introns is mutagenic, then this could help explain why Metschnikowia mitogenomes are so diverse. To explore this idea further, we studied the relationship between SNPs and their proximity to self-splicing-intron insertion sites within cox1 and cob from 71 different species/strains of Metschnikowia. These two genes are ideal candidates for investigating this topic as they are both intron-dense and highly conserved, making sequence alignments and comparisons relatively straightforward (Chandrasekharan and Simmons, 2004; Linacre and Lee, 2005).

We predict that SNP abundance will increase close to self-splicing intron insertion sites in cox1 and cob. If true, this would support the notion that the movement of mitochondrial introns is mutagenic and that this mutagenicity may be connected to endonuclease interference of homologous recombination or gene conversion. We should stress that this study is partly inspired by the work of Repar and Warnecke (2017) who found elevated levels of SNPs close to the insertion sites of self-splicing introns in the mitochondrial genomes of Saccharomyces cerevisiae, Schizosaccharomyces pombe, and Lachancea kluyveri. These findings ultimately led the authors to conclude that intron mobility is a direct driver of host genetic diversity.

Annotated Metschnikowia mitochondrial genomes (Lee et al., 2020) were downloaded from GenBank (accessions MT421951-MT449704) and uploaded to Geneious Prime v. 2020.1.2 (Biomatter Ltd., Auckland, New Zealand). These genomes were generated as part of a large-scale collaborative effort to explore Metschnikowia diversity, natural history and phylogenetics (Lachance et al., 2016), and include Metschnikowia strains that have been collected intensively over 3 decades across a broad array of biogeographic zones (Lachance, 2011; de Oliveira Santos et al., 2015). The 71 cox1 and cob genes were extracted, their introns were removed, and their coding sequences were aligned with MUSCLE (Edgar, 2004), implemented in Geneious using default settings. An initial pre-assessment of the alignments was performed to confirm that the coding regions and intron insertion sites were properly annotated. A few minor corrections were made, and the final alignments were consistent with previous analyses: 25 and 13 intron insertion sites across the cox1 and cob alignments, respectively (Lee et al., 2020).

SNPs were evaluated in Geneious, but each polymorphic site within the alignments was also checked and counted manually. We used the chi-squared test of independence to determine if the number of SNPs within a given window was significantly greater than the number outside that window. To conduct this test, a series of 2x2 contingency tables for each window size were constructed in Excel for the cox1 and cob alignments (Table 1). From the tables, the expected values were calculated, which were then used in the CHISQ.TEST function in Excel to obtain the p-values (Table 1).

We observed a higher density of SNPs near to self-splicing intron insertion sites, which we argue is the result of mitochondrial intron movement. The mutagenicity of self-splicing introns may result from incorrect mtDNA repair following homologous recombination due to the persistent presence and high binding affinity of endonucleases. These findings provide further support for the notion that, while once considered to have no ill effects (Werren, 2011), self-splicing introns impose a mutational burden on their host genomes (Mueller et al., 1996; Repar and Warnecke, 2017). We hope that this work will encourage more research on endonuclease binding and its potential interactions with host-cell repair machinery. A better understanding of this phenomenon could be crucial to treating diseases via CRISPR-Cas9 gene editing.