In this study, we found that the model AMF, R. irregularis, harbours diversity at the MAT-locus that is higher than that expected for a sexually reproducing fungus. We also tested whether R. irregularis is likely a sexually reproducing species, based on the fundamental evolutionary concept of decoupling between the evolutionary history of a MAT-locus, another locus independent of mating and genome-wide loci [28]. In R. irregularis, the congruence between the evolutionary history of these three datasets, coupled with newly uncovered diversity at the MAT-locus, allows us to reject the hypothesis of sexual reproduction in this important fungal species (Figure 1c). Taken together with the information from a population genomics study, indicating identical or very similar, R. irregularis genotypes existing in geographically distant locations, it is highly unlikely that this species is sexually reproducing. Our findings have a number of important consequences that we discuss in more detail.
Variation in the R. irregularis MAT-locus is not consistent with the MAT-locus being involved in mating
Seven MAT-types had been reported in R. irregularis which suggested possible cryptic, but active sex and recombination [26]. In that study, the approach to define different MAT-types was based on phylogenetic analyses of MAT-locus sequence similarity with a defined node-support value or threshold for each cluster. However, the posterior probability values for the nodes depend on the relative similarity between MAT-locus sequences and do not reflect the true sequence difference. We found that nucleotide sequence variation at the MAT-locus affects the HD2 amino acid sequence; a homeodomain transcription factor that plays an important role in self/non-self recognition in other fungi [15]. If the sequence at the R. irregularis MAT-locus defines mating types in this species, the sequences should be highly conserved as is the case for other fungi [18]. Relying on an arbitrary threshold of node support to define MAT-type clusters overlooks the actual diversity of existing MAT-types. Previous studies targeted a less variable region of the locus. and this likely hindered finding true MAT-locus variation. We found 15 MAT types out of 50 unambiguous R. irregularis isolates. The frequency (approx. 0.3) is almost five time higher than that previously reported (0.061; 7 MAT-types in 114 isolates). However, several of the 50 isolates in our study are undistinguishable from each other with ddRAD-seq data. Thus, it is likely that the 15 MAT-types occur in considerably less than 50 genetically different R. irregularis isolates. Furthermore, our study and previous studies only considered partial sequences at the MAT-locus. Full length sequences could contain more nucleotide sequence variation. We conclude that the higher-than-expected level of MAT-type variation is inconsistent with sexuality or this MAT-locus being involved in mating.
The co-evolution of genome-wide variation with PTG, and the MAT-locus variation points strongly to asexuality
Mating, followed by recombination, decouple the evolution of different loci in the population. In contrast, loci of asexual organisms share the same genealogical history [28]. If R. irregularis is sexual and recombination takes place, it is highly unlikely that the intraspecific diversification of two non-related loci is linked to the evolutionary history of the entire genome. The PTG is an important AMF gene, as the symbiosis between plants and AMF is constituted by the nutrient exchange, especially translocation of soil phosphate from the fungus to the plant [1]. The genes of MAT-locus and PTG were known to have clearly different functions [15, 38]. The two loci are located on two different chromosomes (chromosome 11 for HD2 and 18 for PTG in R. irregularis DAOM197198) [39]. Strikingly, we found a clear positive linear correlation in the divergence between these two functionally and distantly unrelated loci, which is expected in the absence of recombination. Moreover, we found that the sequence divergence, based on multiple genome-wide coding and non-coding regions across the R. irregularis genome, and in which both the MAT-locus and PTG were excluded, was congruent with the MAT-locus or PTG phylogenies. The fact that the MAT-locus, PTG and R. irregularis genome (excluding the MAT-locus and PTG) are correlated strongly suggests a lack of sexuality.
Should we be searching for alternative MAT-loci in AMF and what are the possible roles of the locus?
The conserved putative MAT-locus reported in Rhizophagus spp. is the homolog of a fungal mating type locus that is conserved in Basidiomycete fungi. From the evidence presented here, it seems highly unlikely that this putative MAT-locus is involved in mating in AMF. Some other known genes of the Dikarya or Mucoromycota (ancient and recent fungal lineages) that are involved in mating and recognition are present in AMF genomes [10] and are expressed during co-inoculation of roots with two genetically different R. irregularis isolates [40, 41]. However, those genes are not all fully conserved, most have other known functions and none of them have two different alleles in dikaryons. For these reasons, they are unlikely candidates as MAT-loci in R. irregularis. Thus, if a true MAT-locus exists in AMF then it would be a previously undescribed locus in the fungal kingdom.
So, what is the role of the MAT-locus studied here? One possibility is that R. irregularis was very active sexually in the past, giving rise to high MAT-type diversity, and at a certain point in the evolution of the lineage, the fungi lost the ability to sexually reproduce. However, this locus still contains a conserved HD1-like and HD2 protein. MAT-loci identified in other asexual fungi were found to be involved in asexual functions, such as asexual sporulation [42]. It is well known that R. irregularis readily anastomoses with hyphae of the same genotype and that four different stages of recognition and compatibility between pairs of genetically different R. irregularis isolates have been described [11]. In the case of successful fusion allows cytoplasm of the two individuals to flow rapidly in both directions [11]. It is conceivable that the locus could be involved in some of these recognition mechanisms allowing or not allowing the fusion of hyphae of compatible individuals. This could also allow the formation of dikaryons, even if no recombination takes place between them.
The asexual Rhizophagus irregularis lineage does not mean Glomeromycota are ancient asexuals
Explaining the existence of ancient asexual lineages is problematic in evolutionary biology because a fundamental role of recombination is to purge deleterious mutations [43]. The Glomeromycota are thought to be an ancient lineage that formed symbioses with plants since the colonisation of land. Coupled with their seemingly low morphological diversification, they were suggested to be ancient asexuals. The fossil record for Glomeromycota is extremely poor and there could have been great diversification in the Glomeromycota in the past, as seen for the species diversity of major plant lineages that preceded angiosperm radiation. There is a danger that our results on the asexuality of R. irregularis will be interpreted as evidence for the long-term asexuality of the Glomeromycota lineage. While our results strongly support asexuality in R. irregularis, we are not claiming the whole Glomeromycota lineage to be asexual. While there are hardly any confirmed examples of long-lived asexual lineages, there are many examples in nature where an order or genus contains sexual and asexual species [44]. This may be the case in the Glomeromycota. To answer the separate question of sexuality versus asexuality in the Glomeromycota lineage, we urge researchers to carry out similar studies on other Glomeromycota species spread widely across the phylogeny.
Exciting consequences for AMF applications in agriculture and the environment
The question of sexuality in R. irregularis will greatly affect how AMF can be applied to improve agricultural production and ecosystem functions. Our findings may disappoint researchers intending to develop a breeding program relying on crossing to improve AMF. Our results show that this will likely not be possible with R. irregularis. However, other genetic mechanisms in R. irregularis allow the development of new strains of this fungus that have been shown to greatly alter productivity of globally important crops [5, 8, 45].
However, there are positive consequences of our findings. First, R. irregularis is a safe AMF species to develop for agricultural applications because it can be produced readily in vitro without other unwanted microorganisms. That introduced an R. irregularis strain will not recombine with local AMF populations is of great benefit for applications because there should be no introgression of introduced genes into the local population and the introduced fungus should retain its functional characteristics. Secondly, there is no current method to track in introduced R. irregularis in soil where Rhizopagus spp already occur (which is usually the case in agricultural soils). Each nucleus of R. irregularis is haploid and, thus, represents the genome of the individual and all nuclei in a homokaryon individual are identical. Each nucleus carries one MAT-type. Because MAT-type variation is positively correlated with variation in the R. irregularis genome, MAT-type variation represents an excellent proxy for studying Rhizophagus variation in populations, which was previously not possible. This will allow direct tracking of introduced Rhizophagus to finally allow the study of AMF invasiveness. Furthermore, it will be the first time researchers have a tool for directly studying the population biology of this important fungus to measure diversity, and to allow the study AMF competition and co-existence.