Mitonuclear interactions may contribute to fitness of fungal hybrids

Hybridization between species is being recognized as a major force in the rapid adaptive evolution of fungal plant pathogens. The first stages of interspecific hybridization necessarily involve nuclear-mitochondrial chimeras. In their 2001 publication, Olson and Stenlid reported that mitochondria control the virulence of first generation hybrids between the North American fungal pathogen Heterobasidion irregulare and its congeneric H. occidentale. By assessing saprobic ability and gene expression of H. irregulare × H. annosum sensu stricto hybrids and of their parental genotypes, we demonstrate that mitochondria also influence saprobic growth of hybrids. Moreover, gene expression data suggest that fungal fitness is modulated by an intimate interplay between nuclear genes and mitochondrial type, and is dependent on the specific mitonuclear combination.

Fungal hybridization is a process recently acknowledged to occur in nature more frequently than originally thought [1][2][3] . In basidiomycetous fungi, first generation (F1) hybrids are generated through plasmogamy of two haploid mycelia (n) belonging to different interfertile species carrying compatible mating types. As in all Basidiomycetes, karyogamy is delayed, and the extensive life phase in between plasmogamy and karyogamy represents the main vegetative or growth phase of hybrid mycelia. Such phase is characterized by the presence of nuclear haploid genomes from both parental species and is known as heterokaryotic (n + n). Once the heterokaryotic mycelium is well established and has sufficiently colonized its substrate, fruit bodies are produced in which karyogamy and meiosis occur resulting in the production of haploid meiospores, responsible for spread of the fungus 2 ( Fig. 1). In the majority of Basidiomycetes, including the genus Heterobasidion, nuclei migrate between mating individuals but mitochondria do not 4 . As a result, mitochondria are inherited from only one of the parental species 4 (Fig. 1). As in plants and animals, the uniparental inheritance of mitochondria in fungi is a genetically regulated process, often leading to the selective removal of one of the two parental mitochondria in natural populations 4,5 .
Hybrid breakdown has been observed in several studies on F1 hybrids between species that were once living in allopatry 6 . Fitness reduction in hybrids is being recognized as the consequence of incompatibility between mitochondrial and nuclear alleles, plausibly according to the classical biological model of evolution of genetic incompatibility known as the Bateson-Dobzhansky-Muller incompatibility (BDMI) model 7 . For example, mitonuclear interactions in yeasts have been reported to affect growth 8 , to reinforce reproductive isolation 9,10 , and to result in phenotypes that are more sensitive to environmental extremes compared to their parental strains 11,12 . Nonetheless, the current knowledge on the role and the effect of mitonuclear interactions in fungal hybrids is still limited 13 .
Olson and Stenlid 14 generated interspecific F1 heterokaryotic hybrids between the North American pine pathogen Heterobasidion irregulare and the North American non-pine pathogen H. occidentale and performed inoculation studies on pine germlings to show that hybrids with the H. irregulare mitochondria were more virulent than hybrids with the H. occidentale mitochondria. That report was one of the first to highlight the importance of mitochondria in the infection process of a host by a pathogen, and to provide timely evidence on the fitness of fungal hybrids. Since both heteroplasmy of distinct mitochondria in the same individual and mitochondrial recombination occur rarely in fungi 15 , reduced fitness cannot be ascribed to the co-occurrence of different mitochondria in the same cell or to the emergence of new mitochondrial genomes. Rather, fitness may be the result of conflicting, albeit specific, interactions between nuclear and mitochondrial genomes. Notwithstanding that H. irregulare × H. occidentale F1 hybrids appear to be extremely rare in nature 16 , the experimental design of Olson and Stenlid was adequate to uncover a likely mitochondrial control of virulence, although without providing any explanation of the mechanisms involved.
It has recently been reported that H. irregulare is massively hybridizing with the European species H. annosum sensu stricto (s.s.) across multiple sites in Central Italy 17 . Sympatry between the two species can be dated to 1944, when the North American species was introduced in Europe. Interestingly, H. irregulare is clearly dominant and outcompetes the native H. annosum s.s., in spite of its relatively recent history in Italy 18 . Its dominance in Central Italy is associated not with higher pathogenicity on the main Italian host, Pinus pinea, but rather with its higher saprobic and sporulation potential, i.e. its greater ability to grow on wood substrates and to produce fruit bodies 19,20 . These two traits ensure high establishment rates in forest stands because several Heterobasidion annosum sensu lato (s.l.) species become established through the infection of stumps by airborne spores, followed by saprobic colonization of infected stumps and stumps' roots by vegetative mycelium 16 . A significant proportion of genotypes across the entire invasive range of H. irregulare are actually admixtures of H. irregulare and H. annosum s.s. nuclear genomes, but the overwhelming majority of them are characterized by a H. irregulare mitochondrial genome 17 . This observation prompted us to confirm that fitness of admixed genotypes may be correlated with the mitochondrial genome.  Table 1). These results confirmed that fitness of hybrids is associated with the mitochondrial genome, and indicated that a depression of fitness is to be expected when the H. annosum s.s. mitochondrion is combined with the H. irregulare nucleus. Conversely, the opposite is not true, as evidenced by the high fitness recorded for the IA,i hybrid. Expression levels for each of 9 nuclear genes, for which gene products were reported to be putatively involved in wood and host colonization in previous studies [21][22][23] , were measured and compared among all four genotypes, between each hybrid and the pure genotype carrying the same mitochondrial genome, and between the two hybrids. Hierarchical clustering analysis on gene expression data (Fig. 3) showed that the overall pattern of gene expression during saprobic growth was strongly correlated with mitochondrial type, rather than with nuclear composition. Consequently, overall gene expression pattern of the IA,a hybrid clustered with that of the A,a genotype, while that of IA,i clustered with that of I,i. However, Principal Component Analysis (PCA) on the same dataset showed that I,i and IA,i were closest in terms of overall gene expression, while A,a and IA,a were much less similar to one another (Fig. 3).

Results and Discussion
In order to investigate differences in interactions between the H. annosum s.s. mitochondrion (a) and the H. irregulare nucleus (I), as opposed to those between the H. irregulare mitochondrion (i) and the H. annosum s.s. nucleus (A), we compared gene expression of the pure genotype A,a with that of the hybrid IA,a, and gene expression of I,i with that of IA,i. Results showed that 7 out of 9 genes were differentially expressed in the IA,a hybrid compared to the pure A,a genotype ( Fig. 4 and Supplementary Table 2), while only 2 genes were differentially expressed in the IA,i hybrid compared to the pure I,i genotype ( Fig. 4 and Supplementary Table 3). We hypothesize that the significant changes in gene expression levels (both under-and over-expression) detected in hybrids with the H. annosum s.s. mitochondrion may be the result of a costly lack of compatibility between the mitochondrion of that species and the H. irregulare nucleus. Therefore, the negative effect on fitness may be the result of a significantly imbalanced presence of transcripts when compared to transcriptomic profile of the parental genotype.
In contrast, the hybrid carrying the H. irregulare mitochondrion more closely matches the pure H. irregulare genotype in gene expression.
Finally, we also compared gene expression data of IA,a versus IA,i. Interestingly, 3 out of 9 genes belonging to three different categories, i.e. a gene putatively encoding an hydrophobin-like protein (46054), a gene putatively encoding a fatty acid desaturase with oxidoreductase activity (148283), and one putatively encoding a mitochondrial carrier (33584), were significantly (p < 0.05) differentially regulated (Supplementary Table 4). This result suggested that different mitochondrial types may affect different gene categories in F1 hybrids.  It should be noted that in order to compare gene expression of all four genotypes, we opted to focus on transcripts of regions that are conserved between the two Heterobasidion species for the 9 selected genes. Thus, the role played by each of the two species in the generation of the analyzed transcriptome cannot be determined. However, while we cannot exclude that gene expression of heterokaryotic F1 hybrids IA,i and IA,a may be biased by different nuclear ratios of the two parental species as observed in other Heterobasidion species 24,25 , there is yet no evidence that the mitochondrial genome may affect nuclear ratios 24 . Hence, we conclude that differences in gene expression may be due to differential interactions between nuclei and the mitochondria.
Our results showed that mitochondria influence saprobic growth in H. irregulare × H. annosum s.s. hybrids, and that expression of a set of nuclear genes involved in wood and host colonization varies in association with different mitonuclear combinations. It should be noted that nuclear genomes of H. irregulare and H. annosum s.s. have been reported to be very similar, i.e. 98% of interspecific nucleotide homology 23 . Although no information on divergence between mitochondrial genomes is available, it could be assumed that the mitochondrial genomes may be similar as well. Based on this assumption, it is unlikely that mitochondrial genomes themselves may be responsible for fitness reduction in hybrids, though we cannot exclude that they may encode factors directly involved in saprobic growth. Our results strongly support the hypothesis that the interplay between mitochondria and nuclei may be crucial to determine the phenotype of hybrids. Moreover, the asymmetry of the results in both phenotypic and transcriptomic experiments indicates that it is the particular mitonuclear combination that determines the results and not just a general mismatch between mitochondrion and nucleus. Such finding supports previous observations by Olson and Stenlid 14 .
We acknowledge that our findings, although convincing, will require to be corroborated by further studies because they are based on a single combination of two genotypes of the two species, and on gene expression analysis of a selected set of representative genes. Nonetheless, this work is one of the first to highlight a possible and plausible role of mitonuclear interactions in determining the fitness of fungal hybrids in broader terms than previously done. This study, however, not only expands the notion of the association between fitness and mitochondria introduced by Olson and Stenlid using virulence, but also provides an explanation of the reason why hybrids carrying the H. annosum s.s. mitochondrion may be at a significant disadvantage in Italian admixed populations.
Previous phylogenetic studies suggested that H. annosum s.s. is more similar to the common ancestor of the two species than H. irregulare 26 . Furthermore, it has been suggested that, in the absence of recombination, mitochondrial evolution may predate nuclear evolution 27 . In North America, the newly evolved H. irregulare mitochondrion may have interacted with a nuclear genome resembling to H. annosum s.s. Conversely, in Eurasia, the H. annosum s.s. mitochondrion never interacted with the allopatrically evolved H. irregulare nuclear genome, until the recent introduction of H. irregulare in Italy 17 . As a result, natural selection may have already 'tested' and selected fit genotypes with H. irregulare mitochondria and nuclear genomes similar to H. annosum s.s., but the same is not true with genotypes bearing H. annosum s.s. mitochondria and H. irregulare nuclear genomes. This scenario is consistent with the BDMI model of mitonuclear interactions proposed by Sloan and colleagues 28 .
In summary, this study shows that hybrids carrying a nucleus of H. irregulare and mitochondria of H. annosum s.s. have a reduced saprobic ability, thus limiting their establishment in the environment. Highly differential genic expression, i.e. 'expression bias' 29 , in IA,a compared to the pure parent with the same mitochondrion may be regarded as the result of an energetic cost for F1 hybrids carrying the H. annosum s.s. mitochondria. This hypothesis fits our results and the recently proposed model of speciation based on mitonuclear epistatic interactions 28 .

Methods
Generation of artificial hybrids and saprobic assay. Hybrids F1 were obtained by mating in vitro one pure H. irregulare and one pure H. annosum s.s. genotype. In detail, mycelial plugs 6 mm in diameter were removed with a cork borer from actively growing cultures (7 days old) and were paired by placing the plugs about 10 mm apart and 30 mm from the edge of a 90-mm Petri dish containing Malt Extract Agar (2% w/v). Petri dishes were incubated at room temperature (RT) (25 °C) under light. To verify that mating between genotypes occurred, presence of clamp connections in the contact zone was assessed under a dissecting microscope (20X magnification). In addition, molecular analysis as previously described 18 was also performed to confirm the presence of either H. irregulare or H. annosum s.s. mitochondria in heterokaryotic hybrids. Two hybrids with the same nuclear genome but with mitochondrial genomes of either species were obtained by transferring pieces of mycelia taken from sectors behind the inocula into Malt Extract Agar in Petri dishes as previously described 30,31 . All fungal genotypes were deposited at the Mycotheca Universitatis Taurinensis (MUT) (see Supplementary Table 5). Saprobic assays were performed in 90-mm Petri dishes filled with P. pinea sawdust and water agar (15 g of P. pinea sawdust and 12 g of agar per liter). Parents (I,i and A,a) and F1 hybrids (IA,i and IA,a) were inoculated on the center of 90-mm Petri dishes. Ten replicates for each pair were used and Petri dishes were incubated at RT (25 °C) at dark, to limit any possible bias due to light conditions which can affect fungal phenotype 32 . Mycelial growth expressed in mm of colonization (two measures of radial growth for each replicate) was measured every 48 hours for 15 days, and significant (p < 0.05) differences were assessed through Kruskal Wallis and Mann-Whitney U tests at the end of the experiment.

Gene expression analyses.
Nine primer pairs were designed by using Primer3Plus (http://www.bioinformatics.nl/cgi-bin/primer3plus/primer3plus.cgi/) to target genes chosen for their role in wood and host colonization processes. In particular, all selected genes are reported to be expressed during growth in lignin and cellulose-based media, in wood and in infected bark, according to microarray data 21 . Five out of 9 (147699, 148283, 143314, 108968, and 59167) putatively encode for catalytic proteins with oxidoreductase activity; one (33584) for a mitochondrial substrate carrier; one (46054) for an hydrophobin-like protein; one (109183) for a factor putatively involved in the secondary metabolite pathway; one (436250) for a transcription factor. The genes 147699, 148283, 143314 and 109183 were also selected because their transcripts were detected in large scale transcriptomic analysis both in H. irregulare and H. annosum s.s. 21,22 . In addition, all selected genes showed signs of positive selection (high dN/dS ratio) according to a whole-genome analysis of several genotypes of the two species 23 . Full-length sequences of candidate genes were obtained from the available H. irregulare genome (http:// genome.jgi-psf.org/Hetan2/Hetan2.home.html). Primers were designed on conserved genic regions between H. irregulare and H. annosum s.s. on the basis of whole-genome alignments previously performed 23 . RNA was extracted from mycelia of three biological replicates per genotype at the end of the saprobic assay (15 th day). RNA extraction was carried out using Spectrum ™ Plant Total RNA Kit (Sigma-Aldrich, Saint Louis, Missouri, USA) following the manufacturer's protocol. RNA was cleaned of DNA using RQ1 RNase-Free DNase (Promega Corp., Madison, WI, USA), and then quantified using a NanoDrop (Thermo Scientific, Hudson, NH, USA). The absence of genomic DNA was verified through PCR with primers for gene 46054. The absence of signal after PCR amplification without retrotranscription was regarded as absence of DNA. Four hundred ng of total RNA were used for each sample to synthesize the cDNA, according to the SuperScript II Reverse Transcriptase (Invitrogen, Carlsbad, CA, USA) protocol. Quantitative RT-PCRs (RT-qPCRs) were carried out with the Connect ™ Real-Time PCR Detection System (Bio-Rad Laboratories, Hercules, CA, USA). Each RT-qPCR reaction was carried out on a total volume of 20 μl, containing 2 μl diluted cDNA (dilution 1:3), 15 μl SsoAdvanced ™ Universal SYBR ® Green Supermix (Bio-Rad Laboratories) and 1.5 μl of each primer (3 μM), using a 96-well plate. Primer sequences with their optimal annealing temperature are listed in Supplementary Table 6. The following PCR program, which includes the calculation of the melting curve, was used: 95 °C for 30 seconds, 40 cycles of 95 °C for 10 seconds, the optimal annealing temperature for 30 seconds, ramp from 65 °C to 93 °C with a temperature increment of 0.5 °C and a read plate every 2 seconds. All reactions were performed for three biological and three technical replicates. The baseline range and Ct values were automatically calculated using the Bio-Rad CFX Manager software. The candidate gene expression was normalized to that of the reference gene by subtracting the Ct value of the reference gene Tryp Metab (Protein ID: 43087) 33 from the Ct value of the candidate gene with efficiency correction, SCIenTIFIC RePoRtS | (2018) 8:1706 | DOI:10.1038/s41598-018-19922-w from equation 2 −ΔΔCT , where ΔΔCt represents the ΔCt sample − ΔCt control 34 . The gene Tryp Metab was selected as reference given its high expression stability in different tissues and samples based on previous experiments 33  Data availability. All data generated or analysed during this study are included in this article (and its Supplementary Information files).