Allorecognition genes drive reproductive isolation in Podospora anserina

Abstract

Allorecognition, the capacity to discriminate self from conspecific non-self, is a ubiquitous organismal feature typically governed by genes evolving under balancing selection. Here, we show that in the fungus Podospora anserina, allorecognition loci controlling vegetative incompatibility (het genes), define two reproductively isolated groups through pleiotropic effects on sexual compatibility. These two groups emerge from the antagonistic interactions of the unlinked loci het-r (encoding a NOD-like receptor) and het-v (encoding a methyltransferase and an MLKL/HeLo domain protein). Using a combination of genetic and ecological data, supported by simulations, we provide a concrete and molecularly defined example whereby the origin and coexistence of reproductively isolated groups in sympatry is driven by pleiotropic genes under balancing selection.

Fig. 1. Genetic diversity and balancing selection in the Wageningen collection of P. anserina.

a, Sliding window analysis (10 kb long with steps of 1 kb) of representative chromosomes 1 (top) and 3 (bottom) with values of genetic diversity (measured as either the pairwise nucleotide diversity π in red or as Watterson’s theta θ_W in blue) and the Tajima’s D statistic (orange). The location of relevant loci is marked when present: het genes, the centromere, the mating type locus (MAT) and the location of meiotic drivers of the Spok family. Loci with highly divergent alleles (for example, MAT) and repetitive regions (for example, WD40 repeats of HNWD genes) were filtered out by the variant-calling pipeline but linked variants can still show signals of balancing selection. Note that Spok3 and/or Spok4 can be found at different locations in the genome depending on the strain. b, Changes of allele frequencies through time for three known het loci (data shown from years with more than five samples).

Fig. 2. Mating success correlates with genetic differentiation in chromosome 5.

a, Left, a PCoA of mating success data reveals two groups (indicated by lilac and green ovals); right, a PCA of the SNP data from chromosome 5 returns clusters of the same two groups (divided by the grey line). Samples with both mating and genomic data are coloured on the basis of the mating success clustering (lilac and green points). Samples without mating success data are in grey. b, Genetic differentiation between the two groups (as defined by the PC1 axis of the SNP data) shown as F_st and D_XY statistics estimated from sliding windows (10 kb long with steps of 1 kb).

Fig. 3. Genetic and molecular dissection of the het-v locus.

a, When two individuals with incompatible alleles fuse during vegetative growth, an incompatibility reaction occurs producing a divisive line of dead cells known as barrage. During sex, when the female organ (protoperithecium) is fertilized by microconidia, different genotypes of the two het genes confer various degrees of infertility depending on the crossing partner. The RV genotype is lethal. b, The location of het-v was identified through nested deletions, marked in orange if they lost the barrage formation in confrontations with a V1 strain (that is, became compatible) or in grey if they did not. The black bar represents a cluster of genes absent in the strain Y, which is of incompatibility type V1. c, Deletion Δ113 (Δ12915-123170) from a V strain leads to the V1 phenotype. A barrage test with a V (top row) and V1 tester (bottom row) is given for three transformants obtained with the Δ113 deletion cassette using an rV recipient strain. The central transformant t2 produces a barrage reaction to V but not to V1 (that is, it acquired the V1 phenotype). Transformants t1 and t3 retain the V phenotype and are presumably V + V1 heterokaryons. d, Genetic structure of the het-r and het-v loci is represented by the domain architecture of the encoded proteins. Arrows represent incompatible interactions. e, To the left, a cross between rV (light colour) and RV1 (dark colour) produces very few fruiting bodies. However, fertility is recovered by deleting the het-v locus and adjacent genes (Δ12810-12690) as shown in the cross in the middle. Reintroduction of the het-Va and het-Vb genes fully restores the sterile phenotype characteristic of rV × RV1 in the cross to the right.

Fig. 4. The effects of the het-r/v system on vegetative incompatibility and reproductive isolation.

a, Vegetative cell fusion between members of the two RI groups is prevented by the allelic interaction of the het-v alleles (V/V1) and the non-allelic interaction between het-v and het-r (V/R), which deters the transfer of viruses and other deleterious cytoplasmatic elements. b, During sex, the het-r/v system remains active, resulting in failed fertilization. c, In most cases the zygote formation does not occur and female organs (protoperithecia) do not develop into perithecia. The few successful perithecia produce spores but a fraction of these spores are self-incompatible and die 15 h after germination^,–. While female R organs are irreversibly damaged after abortive fertilization with V microconidia, female V organs are still viable after fertilization with R (or V1) microconidia^,.

Fig. 5. The RI groups maintain isolation despite coexistence.

a, Genotypes of het-v and het-r in the Wageningen population through time. The rV1 genotype is very rare, confirming that the RI groups (RV1 and rV) are not mixing freely. RV is lethal. b, There is no significant difference (NS) in abundance between RI groups based on the source herbivore (Pearson’s Χ²_{1, 82} = 0.46365, P = 0.4959; genotyped samples from 1991 to 2016 with substrate data, n = 82). Sheep dung was not tested due to low sample sizes. c, The strains collected in 2017 from horse dung (plus one sample from sheep) show that the RI groups can be found in the same dung piece (n = 63). Insert, fruiting body of P. anserina in the wild (photo by S.L.A.V.).

Fig. 6. Summary of individual-based simulations of rV invading an RV1 population with intensity of V/R balancing selection fixed to 0.5.

For each parameter combination, the distribution of genotype frequencies of 100 replicated simulations is given. Each simulation is represented by a dot and their distribution by a boxplot, which shows the median as a diamond, the 25th to 75th percentiles as the box bounds and 1.5× interquartile ranges as whiskers. If dots and boxplots are not visible, the distribution is concentrated behind the diamond of the median.