Reconstructing NOD-like receptor alleles with high internal conservation in Podospora anserina using long-read sequencing

Abstract

NOD-like receptors (NLRs) are intracellular immune receptors that detect pathogen-associated cues and trigger defence mechanisms, including regulated cell death. In filamentous fungi, some NLRs mediate heterokaryon incompatibility, a self-/non-self-recognition process that prevents the vegetative fusion of genetically distinct individuals, reducing the risk of parasitism. The het-d and het-e NLRs in Podospora anserina are highly polymorphic incompatibility genes (het genes) whose products recognize different allelic variants of the HET-C protein via a sensor domain composed of WD40 repeats. These repeats display unusually high sequence identity maintained by concerted evolution. However, some sites within individual repeats are hypervariable and under diversifying selection. Despite extensive genetic studies, inconsistencies in the reported WD40 domain sequence have hindered functional and evolutionary analyses. Here, we confirm that the WD40 domain can be accurately reconstructed from long-read sequencing (Oxford Nanopore and PacBio) data, but not from Illumina-based assemblies. Functional alleles are usually formed by 11 highly conserved repeats, with different repeat combinations underlying the same phenotypic het-d and het-e incompatibility reactions. AlphaFold 3 structure models suggest that their WD40 domain folds into two 7-blade β-propellers composed of the highly conserved repeats, as well as three cryptic divergent repeats at the C-terminus. We additionally show that one particular het-e allele does not have an incompatibility reaction with common het-c alleles, despite being 11-repeats long. Finally, we present evidence that the recognition phenotypes of het-e and het-d arose through convergent evolution. Our findings provide a robust foundation for future research into the molecular mechanisms and evolutionary dynamics of het NLRs, while also highlighting both the fragility and the flexibility of β-propellers as immune sensor domains.

Keywords: WD40 domain; allorecognition; fungi; heterokaryon incompatibility.

Fig. 1.. Primer on the het-d and het-e genes. (a) The domain structure of an HNWD NLR. (b) Incompatibility interactions between the most common het-c alleles and those of het-e and het-d. Shaded squares indicate a vegetative incompatibility reaction, while white squares indicate compatibility, following Saupe et al. [38]. (c) A typical PCR result when amplifying the WD40 domain of an HNWD gene from genomic DNA, in this case het-e (1% agarose gel). Three strains with known het-e alleles are shown.

Fig. 2.. Colour classification of WD40 repeats with HIC using het-e as an example. The individual repeats were categorized based on the unique combination of aa present at seven positions, highlighted in a logo diagram of het-e on the top left corner. A matrix of functional dissimilarity of aa (upper right) was then used to calculate pairwise distances between repeat variants (lower left), where the dissimilarity of two variants is the sum of the dissimilarities of their aa at each of the seven selected positions. These distances were then used to project the repeat variants into the 3D L*a*b colour space, such that repeats with functionally similar aa are assigned perceptually similar colours (lower right). Each repeat variant is plotted in L*a*b space as a coloured square, along with a dashed line connecting it to a point at its projection in the a*b plane. Three repeats are highlighted to illustrate the case of one distant and two similar repeats (1, 17 and 24).

Fig. 3.. Assembly of the WD40 domain from different sequencing technologies. Only repeats with HIC are shown. Each repeat was arbitrarily classified based on unique aa combinations (see Table S2), but the colours reflect their physicochemical similarity (each gene has an independent palette). Repeats with a track of missing data (Ns) are coloured black. Black lines linking the repeats symbolize the containing scaffold.

Fig. 4.. Long-read assemblies of het-d (a) and het-e (b) WD40 domain from different WT strains. Only repeats with HIC are shown, arbitrarily classified based on unique aa combinations (see Table S2), but coloured based on their physicochemical similarity (each gene has an independent palette). Repeats containing stop codons or frameshifts are coloured black. The red arrow highlights the single repeat distinguishing the reactive D1^A allele (in strain ChEhDa+) from the non-reactive d3^Z allele (Z+). The black arrow marks the repeat with a deletion in the T_G+ sequence that is likely a misassembly. Blue arrows point to inferred alleles based on sequence or the number of repeats. A specific allele of het-e was selected for phenotypic testing. The beginning of the first repeat is missing in the E1^A sequence (GenBank accession number FJ897789), but the missing aa happen to be perfectly conserved in all sequences, and hence, we inferred it to be identical to the first repeat of E1^Y.

Fig. 5.. The het-e allele from Wa63 is compatible with the four common het-c alleles. Barrage assay of C1 and C2 strains transformed with the het-e allele of Wa63 cloned on a plasmid and tested with the four common het-c alleles. The C2 recipient strain (upper plate) allows for testing against C1, C3 and C4. The C1 recipient (bottom plate) allows for testing against C2, C3 and C4. The E1 and E2 alleles (on the upper left on the upper and bottom plates, respectively) are used for positive controls for the barrage (incompatibility) reaction. Note the barrage formation between the E2 and C4 testers in the bottom plate. In the strain designation, the het-c, het-d and het-e genotypes are omitted for clarity when strains carry inactive alleles. In other words, a C1 strain has d3 and e4 alleles, an E1 strain has a d3 and a null het-c allele, etc. The source strain of each active allele is marked as a superscript (e.g. E1^A indicates that E1 comes from the A strain).

Fig. 6.. Ribbon diagrams of the WD40-domain structure from the HET-E1 protein (E1^H allele) produced by AlphaFold 3. The first 817 sites containing the HET and NACHT domains were removed for clarity (pTM=0.69 for the full model). The first propeller is coloured with a rainbow palette to illustrate the direction of the individual β-sheets. The cryptic d β-sheet in the N-terminus of the WD40 domain that forms the molecular velcro with the C-terminus is also highlighted (forest green). The second propeller is coloured based on HIC (red) and cryptic (salmon) repeats. Individual blades are numbered with Latin (HIC) or Roman (cryptic) numerals.

Fig. 7.. Relationship between different domains of HNWD genes and related NLRs. (a) Maximum likelihood phylogenies along the HET, NACHT and WD40 domains. Branch support values correspond to standard non-parametric bootstrap (values <70 omitted). Branches are proportional to the scale bar (aa substitutions per site). The beginning of the WD40 domain contains a cryptic (divergent) repeat, used to make a phylogeny, followed by repeats with HIC. (b) The individual nt sequences of the HIC WD40 repeats were extracted from all P. anserina long-read assemblies and analysed in a PCA. (c) PCA of HIC WD40 repeats in genes closely related to het-e. The two repeats identical between het-e and nwd6 are easiest to see when comparing the principal components 2 and 3 (arrows).