Mitochondria are functionally and genetically dynamic organelles essential to the eukaryotic cell (Burger et al., 2003; Henze and Martin, 2003; McBride et al., 2006). They are the center of aerobic production of energy, but also play a role in signaling between cellular components, cellular growth and differentiation, apoptosis, control of the cell cycle, disease, and aging (McBride et al., 2006; Bernhardt et al., 2014; Urbani and Babu, 2019). Although the majority of mitochondrial proteins are encoded by nuclear genes, mitochondria have their own self-replicating genome bearing phylogenetic evidence of prokaryotic ancestry (Gray et al., 2001). According to the endosymbiotic theory, mitochondria originate from the endocytosis of an alphaproteobacterium that took place about two billions years ago as an adaptation to an atmosphere richer in oxygen (Gray et al., 1999). Mitochondrial and nuclear genomes have subsequently coevolved throughout the evolutionary history of Eukaryota, following two major trends: one of reduction of the number of mitochondrial genes non-essential to the eukaryotic cell, the other of transfer to the nucleus of unique essential genes whose products can be imported from the cytosol to the mitochondrion without fitness cost (Kurland and Andersson, 2000). Mitochondrial genes encoding proteins involved in cellular respiration are highly conserved while other loci are more variable depending on the species, and some contribute to mitochondrial genome evolution and plasticity, like highly polymorphic, repetitive elements found within introns and intergenic regions (Burger et al., 2003). Consequently, conformation of the mitochondrial genome, its size, its degree of relatedness to its bacterial ancestor, its genes number, their order and their expression vary by species (Gray et al., 1999; Medina et al., 2020 for a recent review in fungi). Maintaining mitochondrial activity with a reduced genome implies multiple mitonuclear genetic interactions (Stojković and Dordević, 2017), but also requires nuclear and mitochondrial DNA permanently to coadapt despite their frequently different mutation rates (Allio et al., 2017; Sandor et al., 2018; Medina et al., 2020). Population-wise divergence of mitochondrial and nuclear coadapted alleles is predicted to lead to speciation due to Bateson-Dobzhansky-Muller incompatibility (Nguyen et al., 2020 and references therein).

The goal of this article is to show that harnessing natural intraspecific variation of coadapted mitochondrial and nuclear alleles in order to study genetics of mitonuclear interactions is possible. Natural variation in the mitochondrial genome of two species of the tree pathogen Heterobasidion is associated with the emergence of interspecific heterokaryotic hybrids with virulence that correlates with mitochondrial inheritance (Olson and Stenlid, 2001; Giordano et al., 2018). This observation suggests that, in self-incompatible Basidiomycetes species such as Heterobasidion spp., new combinations of mitochondrial and nuclear DNA formed at the heterokaryotic phase could drastically alter phenotype. Isolates of Heterobasidion spp. spend most of their life cycle as heterokaryotic mycelium (n + n) in host wood (Johannesson and Stenlid, 2004; Garbelotto and Gonthier, 2013). The homokaryotic (haploid) phase is short and exposed mainly to purifying and neutral selections: Basidiospores dispersal by the wind, their landing on an exposed area of a potential host, and the encounter of homokaryotic mycelia originating from different basidiocarps on host wood are largely stochastic. Mating is under the control of a unifactorial multiallelic genetic system (Holt et al., 1983), which, preceded by basidiospore dispersal, is supposed to favor outbreeding (Nieuwenhuis et al., 2013). It is therefore when mitochondrial and nuclear DNA of different homokaryotic isolates meet at the heterokaryotic phase that new mitonuclear combinations might emerge. If so, some degree of mitochondrial DNA-dependent phenotypic variation is expected at each generation of intraspecific heterokaryons as well.

Our recent work in Heterobasdion parviporum demonstrated and quantified mitochondrial genetic effects on two fitness proxies in heterokaryons and their parents (Clergeot et al., 2019). Our experimental design was based on heterokaryon synthesis (Olson and Stenlid, 2001), during which mitochondrial DNA is unparentally inherited while nuclei undergo bidirectional exchange, resulting in a pair of heterokaryons with the same haploid nuclei but a different mitochondrial genome (Xu and Wang, 2015). Heterokaryon synthesis was implemented with thirty homokaryotic parental isolates sampled at various locations in Eurasia and showing no population structure. We observed that variance of growth rate does not correlate with genetic distance in heterokaryons and that mitochondrial genetic effects account for 35% of phenotypic variance among homokaryons and 3% among heterokaryons (Clergeot et al., 2019). Heterokaryons are more fit than their parents in average, but not more than the fittest parents, suggesting heterozygosity and the masking of deleterious alleles at some of the nuclear loci involved in mitonuclear interactions in the offspring, rather than an intrinsic difference of fitness between the heterokaryons and their parents.

The present study is an investigation of the genetic basis of mitonuclear interactions in Heterobasidion parviporum: As phenotyping the heterokaryotic offspring of pairwise crossings of homokaryotic parents in controlled conditions brings a more reliable assessment of the variation resulting from combining different nuclear and mitochondrial alleles (Nguyen et al., 2020), we have revisited our data of heterokaryon variance for growth rate and parental genome sequence polymorphism (Clergeot et al., 2019) in order to identify candidate loci acting in trans across the two genomes. We have been able to single out this way two possible candidates for a nuclear locus encoding proteins involved in molecular trafficking to the mitochondrion (a mitochondrial carrier protein or a mitochondrial calcium uniporter), and a non-conserved mitochondrial locus of unknown function and whose distribution is limited to the Russulales order of Agaricomycetes. We believe that our results on mitonuclear genetics of growth rate in H. parviporum could pave the way to similar investigations of mitochondrial DNA-dependent variation of important traits e.g., virulence in fungal pathogens.

From the thirty homokaryotic isolates of H. parviporum in our previous study from 2019, reciprocal exchange of nuclei during mating was successful with thirteen of them crossed pairwise, leading to 73 pairs of heterokaryons having the same parental haploid nuclei, but mitochondria from either of the two parental isolates. In this study, we have used phenotypic data from these thirteen homokaryotic parents and their heterokaryotic offspring to investigate mitonuclear interactions.

Our phenotypic data show that two major competing trends are at work in heterokaryons, with opposite effects on growth rate. The first trend is “heterokaryon vigor” (Clergeot et al., 2019), which results from the reciprocal masking of deleterious alleles brought by one or both parents (Clark and Anderson, 2004). Our growth rate data show that dominance is frequent, but overdominance is rare and limited to the offspring of parents that are both less fit. As the fittest parental alleles are overwhelmingly dominant, the center of the distribution of all heterokaryons data is shifted toward the position it would have in a hypothetic homokaryon having the largest combination of the fittest parental alleles. Homokaryons growth rate data suggest that the fastest parents in our sample are close to such an optimal allelic combination, as none of the heterokaryotic offspring can grow at a higher rate than they do.

The second trend is high heterokaryon-specific phenotypic variation, a phenomenon that we have observed in several heterokaryons (26%) after carrying out different growth rate assays over time. These heterokaryons can nevertheless grow as fast as the others. Their genetic potential for this specific trait is therefore not different from the rest of the heterokaryons. The fact that their phenotype does vary more between assays is due to other factors, unidentified so far, although probably linked to the heterokaryotic state, during which parental haploid nuclei coexist in the same cytoplasm, but do not fuse (Raper, 1966). The heterokaryon is an intermediary state between haploidy and true diploidy, more plastic than the latter due to the possibilities of remating (Hansen et al., 1993; Johannesson and Stenlid, 2004; James et al., 2009), secondary formation of haploid sectors (Stenlid and Rayner, 1991; Hansen et al., 1993), and production of haploid conidia (Hansen et al., 1993). Although considered as functionally equivalent to a diploid (Day and Roberts, 1969), such a dynamic and possibly conflicting association of haploid nuclei is likely to lead to stochastic variation of gene expression (Kilfoil et al., 2009), and therefore, of heterokaryon phenotype. Degree of phenotypic variation of each heterokaryon might be genetically determined by specific combination of parental alleles, as for nuclear ratio in heterokaryons of H. parviporum (James et al., 2008).

Analysis of heterokaryons growth rate data showed that higher values reflect their genetic potential for this trait better, which made possible the identification of pairs of heterokaryons issued from the same parents, but whose phenotypes are different. It confirms our previous detection of mitochondrial genetic effects in the same cohort of heterokaryons of H. parviporum (Clergeot et al., 2019). In one occurrence, owing to the facts that two non-kin-related parental isolates (Ho9 and Ho18) have nearly identical mitochondrial haplotypes and that their crossings with a third isolate, Ho32, gave heterokaryons with clear-cut phenotypes, we have identified alleles of a mitochondrial locus (mt2) and a genomic locus (nuc2) involved in a genetic interaction in trans across the two genomes. Different combination of alleles at these two loci contribute to phenotypic variation in the offspring of the crossings with these three isolates.

Candidates for genomic locus nuc2 were searched using phenotypic information of heterokaryons from all other crossings involving parent Ho9 and SNPs markers called from the genome sequences of all homokaryotic isolates (Clergeot et al., 2019). In this work, SNP alleles associated with a non-synonymous mutation in two H. parviporum ORFs were investigated in priority as more likely candidates for nuc2, although other causes of phenotypic variation cannot be excluded at this stage. Similar bioinformatics methods have been used to track polymorphism in gene regulatory sequences or introns too, and led to the identification of two additional candidates for nuc2. The two genes whose sequences in Ho9 and Ho32 are bearing mutations altering the structure of their translation products (although it cannot be ascertained that the function of these products is modified), but also carrying mutations that might affect their expression level are considered the best possible candidates at this stage. Both are involved in molecular trafficking to the mitochondrion (a carrier protein and a calcium uniporter). Although the gene encoding a calcium uniporter cannot be ruled out at this stage, the other one encoding a mitochondrial carrier protein is a more likely candidate for nuc2 for several reasons: Mutations in this gene result from multiple transversions (3) and one indel in the same exon, whereas the mutation in the gene encoding the calcium uniporter results from a single transition; A multiple protein alignment with orthologs from other species shows that substitution R to K in the sequence of the carrier is in a conserved motif, and this substitution is present in Ho9 and Ho32 only; Amino-acid substitutions/deletion observed in the carrier are more likely to lead to a change of balance between structure and function.

In contrast, it is not a mitochondrial gene encoding a key protein in oxidative metabolism that has been uncovered as the potential candidate involved in a mitonuclear interaction with nuc2, but a non-conserved gene instead, with orthologs in the Russulales order only. The ortholog of this gene in the mitochondrial genome of H. irregulare has been previously identified as nc-ORF5 (non-conserved Open Reading Frame 5; Himmelstrand et al., 2014). The presence of nc-genes in mitochondrial DNA is a typical feature of Agaricomycetes fungi. Nc-genes are supposed to share the same evolutionary origin as the mitochondrial plasmid DNA frequently found in the mitochondrial genome of species in this class of fungi and contribute to explain its unusually large size. This non-conserved and exchangeable part of Agaricomycetes mitochondrial DNA might originate from the uptake of plasmids by mitochondria and its subsequent integration into mitochondrial DNA early in the evolution of this class of fungi (Himmelstrand et al., 2014; Medina et al., 2020). Protein blast searches made with the translation of the ortholog of nc-ORF5 in H. parviporum (Hparv-nc-ORF5p) and by excluding proteins from the Russulales retrieved exclusively a hypothetical protein from soilborne gammaproteobacteria (Diamond et al., 2019). Phenotypic variation associated with mitochondrial DNA sequence polyporphism at this locus seems to originate from a change in the regulation of nc-ORF5 expression rather than from its loss of function, as the SNP is located upstream of nc-ORF5 coding sequence, either in its promoter or its 5' UTR sequence.

Further investigation would imply (1) studying the expression of all mitochondrial and nuclear candidate genes to confirm or infirm hypotheses made about them, (2) providing molecular genetics evidence of mitonuclear interaction by carrying complementation experiments of homokaryon Ho9 with wild-type copies of candidate nuclear loci for nuc2. Genetic transformation of Heterobasidion spp. using Agrobacterium tumefaciens has been described previously (Samils et al., 2006), including in H. parviporum (Ihrmark, unpubl.), and could be used for that purpose. Another option would be to shuffle mitochondria into various homokaryotic nuclear backgrounds by reisolating homokaryotic conidia from heterokaryons cultures (Ramsdale and Rayner, 1994) and assessing their phenotype. Finally, although our phenotypic analysis points at a very simple genetic explanation for the phenotypic variation observed in this sample of isolates (homogeneization of the heterokaryons phenotypes to the level observed in the best homokaryotic parents is the signature of the complementation of a few recessive and defective alleles unique to each slower parent by dominant wild-type alleles borne by any other parent), mitonuclear interactions involving more than one locus in the mitochondrial and/or the nuclear genome cannot be excluded so far. It could be determined by carrying out a segregation analysis (Lind et al., 2005).

Mitonuclear interactions are suspected to be the cause of some breakthrough increase of virulence or change of host range in rare interspecific hybrids (Olson and Stenlid, 2001; Giordano et al., 2018; Hu et al., 2020), but they are also part of the ordinary evolutionary process of each fungal pathogenic species and could hence contribute to mold the interactions with its host. Coadaptation between mitochondrial and nuclear DNA might be an important driver in the evolution of host-pathogen interactions. Several examples have been documented in other fungal pathogens of a change of host range or of a variation in virulence associated either with polymorphism of mitochondrial DNA, with a loss of function, or with a differential expression of nuclear genes encoding mitochondrial proteins (Monteiro-Vitorello et al., 1995; Lorenz and Fink, 2001; Inoue et al., 2002; Zhan et al., 2004; Ma et al., 2009; Mahlert et al., 2009; Patkar et al., 2012; Zhang et al., 2012; Khan et al., 2015; van de Vossenberg et al., 2018). In the context of this work, it is worth noticing that a nuclear-encoded mitochondrial carrier protein has already been mentioned in Heterobasidion spp. as possibly involved in virulence variation related to mitochondrial background (Giordano et al., 2018; Hu et al., 2020). Furthermore, in pathogenic Basidiomycetes fungi like Heterobasidion spp., the capacity of heterokaryon remating (Johannesson and Stenlid, 2004) gives mitonuclear interactions the opportunity to add to the selection process for fitter heterokaryotic associations while genetically diverse heterokaryons grow, meet and exchange nuclei on host wood. Our work brings forth evidence that the investigation of mitonuclear interactions is possible for a trait like hyphal growth rate, but the same could be considered for virulence.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/Supplementary Material.

P-HC and ÅO: conceived and designed the experiments, performed the experiments, analyzed the data, and revised the paper. P-HC: exploratory data analysis, statistics, and bioinformatics and wrote the paper. ÅO: additional growth rate assays. All authors contributed to the article and approved the submitted version.

This work was supported by the Carl Tryggers Foundation for Scientific Research.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.