Identification of the genes at S and Z reveals the molecular basis and evolution of grass self-incompatibility

Abstract

Self-incompatibility (SI) is a feature of many flowering plants, whereby self-pollen is recognized and rejected by the stigma. In grasses (Poaceae), the genes controlling this phenomenon have not been fully elucidated. Grasses have a unique two-locus system, in which two independent genetic loci (S and Z) control self-recognition. S and Z are thought to have arisen from an ancient duplication, common to all grasses. With new chromosome-scale genome data, we examined the genes present at S- and Z-loci, firstly in ryegrass (Lolium perenne), and subsequently in ~20 other grass species. We found that two DUF247 genes and a short unstructured protein (SP/ZP) were present at both S- and Z- in all SI species, while in self-compatible species these genes were often lost or mutated. Expression data suggested that DUF247 genes acted as the male components and SP/ZP were the female components. Consistent with their role in distinguishing self- from non-self, all genes were hypervariable, although key secondary structure features were conserved, including the predicted N-terminal cleavage site of SP/ZP. The evolutionary history of these genes was probed, revealing that specificity groups at the Z-locus arose before the advent of various grass subfamilies/species, while specificity groups at the S-locus arose after the split of Panicoideae, Chloridoideae, Oryzoideae and Pooideae. Finally, we propose a model explaining how the proteins encoded at the S and Z loci might function to specify self-incompatibility.

Keywords: DUF247; Poaceae; grass; pollen; reproduction; self-incompatibility; stigma.

Figure 1

Synteny between S- and Z- loci in L. perenne. (A) Structure of the DUF247/SP/ZP at S- and Z-loci in the Kyuss genome (Frei et al., 2021). (B) Annotated genes in a 1.3 and 1.7 Mb region around SDUF247 and ZDUF247 were compared by phylogenetic analysis (see Supplementary Figure 1 ), BLAST searching and comparing domains in common. Only the DUF247 genes passed all three tests, indicated by solid lines. SP had a second tBLASTn hit at the Z-locus, indicated by a dotted line. (C) Number of unique domains identified at the S- and Z-loci and the overlap between the two. (D) Number of annotated genes in each region, and the number with secondary tBLASTn hits at the opposing (S/Z) locus.

Figure 2

Expression of DUF247 and SP/ZP in reproductive tissues from grasses. (A, B) RNAseq coverage at the S- (A) and Z-locus (B) in anther and stigma tissue from two ONE50 individuals. One plant (ONE50a) was a clone of the individual used to create the reference sequence, and had the corresponding genotype at S- and Z-, while the other (ONE50b) only had the corresponding Z-locus genotype. Numbers in the top left of each sub-panel indicate the y-axis scale in units of coverage in reads per million over 25bp windows. (C, D) Expression of DUF247 (C) and SP/ZP (D) genes from rice, wheat, sorghum and brachypodium in anther and stigma (or pistil) from expression databases.

Figure 3

Structure of the S-locus in different grass subfamilies. Structure of S-loci from different species. Grasses of the Pooideae subfamily (top) are flanked by a Peridoxial phosphate homeostasis gene (PLP; left) and SWI/SNF chromatin remodeler (right). Oryzoideae grass S-loci are flanked by a gene encoding a calmodulin binding protein (left) and a zinc permease (right). Chloridoideae grasses (bottom) are flanked by a Sterol-response element binding protein (SRE-BP; left, light grey) and Flavin reductase (left, dark grey) and a Mitogen activated protein kinase (MAPK, right). Panicoideae grasses (S. bicolor, P. virgatum, P. vaginatum) did not show obvious synteny, except for SRE-BP in S. bicolor. Gene identifiers of flanking genes are shown where available. See Supplementary Dataset S4 for details of genomes/versions.

Figure 4

Structure of the Z-locus in different grass subfamilies. Structure of Z-loci identified in other species. In all cases, a Glycerol kinase (GK) is on the left, and Ubiquitin Conjugating Enzyme (UBC) is on the right. Gene identifiers are shown for GK and UBC for each genome where available. See Supplementary Dataset S4 for details of genomes/versions.

Figure 5

Phylogeny of grass species represented by Glycerol Kinase protein similarity. A phylogenetic tree constructed using the protein sequence of a Glycerol Kinase adjacent to all Z-loci. To produce the tree a global alignment with a BLOSUM62 cost matrix was performed, followed by tree building using a Jukes Cantor genetic distance model and nearest neighbor tree-building method. The phylogenetic tree resembles the expected relationship between species. Scale bar indicates substitutions per site.

Figure 6

Phylogeny of S/Z-DUF247 genes from various grass species. A phylogenetic tree constructed using the sequences of all S/Z-DUF247 protein sequences. To produce the tree a global alignment with a BLOSUM62 cost matrix was performed, followed by tree building using a Jukes Cantor genetic distance model and nearest neighbor tree-building method. Subclades of SDUF2 and SDUF3 based on subfamily are shown. ZDUF1/ZDUF4 did not produce subclades. Self-incompatible species are highlighted with blue text. Scale bar indicates substitutions per site.

Figure 7

Phylogeny of SP/ZP genes from various grass species. A phylogenetic tree constructed using the sequences of all SP/ZP protein sequences. To produce the tree a global alignment with a BLOSUM45 cost matrix was performed, followed by tree building using a Jukes Cantor genetic distance model and nearest neighbor tree-building method. Representative structures, cleavage and cysteine location of each sub-clade shown at right. Scale bar indicates substitutions per site.

Figure 8

Model for SI in grasses. DUF247 proteins from S- and Z-loci form dimers at the pollen tube surface, and are anchored to the membrane by a conserved transmembrane domain. SP/ZP peptides are free-floating in the stigmatic exudate, allowing pollen tube arrest to occur immediately upon pollen germination. Signals from S- and Z- must be incorporated, therefore we propose a tetramerization of DUF247 proteins, resulting in a downstream signaling cascade that results in pollen tube arrest. Absence of any component, or non-self components would result in a lack of signaling, thus allowing pollen tube growth.