Pangenome-based trajectories of intracellular gene transfers in Poaceae unveil high cumulation in Triticeae

Abstract

Intracellular gene transfers (IGTs) between the nucleus and organelles, including plastids and mitochondria, constantly reshape the nuclear genome during evolution. Despite the substantial contribution of IGTs to genome variation, the dynamic trajectories of IGTs at the pangenomic level remain elusive. Here, we developed an approach, IGTminer, that maps the evolutionary trajectories of IGTs using collinearity and gene reannotation across multiple genome assemblies. We applied IGTminer to create a nuclear organellar gene (NOG) map across 67 genomes covering 15 Poaceae species, including important crops. The resulting NOGs were verified by experiments and sequencing data sets. Our analysis revealed that most NOGs were recently transferred and lineage specific and that Triticeae species tended to have more NOGs than other Poaceae species. Wheat (Triticum aestivum) had a higher retention rate of NOGs than maize (Zea mays) and rice (Oryza sativa), and the retained NOGs were likely involved in photosynthesis and translation pathways. Large numbers of NOG clusters were aggregated in hexaploid wheat during 2 rounds of polyploidization, contributing to the genetic diversity among modern wheat accessions. We implemented an interactive web server to facilitate the exploration of NOGs in Poaceae. In summary, this study provides resources and insights into the roles of IGTs in shaping interspecies and intraspecies genome variation and driving plant genome evolution.

Figure 1.

The IGTminer pipeline for connecting NOGs across assemblies and recording their dynamic trajectories. A) A schematic diagram of organelle-to-nucleus gene transfer. B) The length distribution of NUPTs and NUMTs in the genomes of several major cereal crops. NUPTs and NMPTs that contain full homologous sequences of organellar genes are highlighted in different colors. TaA, TaB, and TaD indicate the A, B, and D subgenomes of bread wheat, respectively. Hv represents H. vulgare. Os represents O. sativa. Zm represents Z. mays. C) The design of the IGTminer pipeline for uncovering the evolutionary trajectories of the IGT (NOG) using gene reannotation and collinearity information. The raw NOG annotations in the reference genome are removed and reannotated, and the reannotated genes are clustered into a single transfer event. The collinearity block search pipeline is optimized and used to search collinear NOGs. Only NOGC networks containing at least 1 coding gene are used. Finally, a NOG map across multiple genomes is generated to record the dynamics of NOGs. D) A typical example showing the polymorphism and collinearity of NPGs between 4 genomes. Genomes of 2 bread wheat accessions (Chinese Spring and Jagger), a wild emmer wheat accession (Zavitan), and a spelt wheat accession are presented. An unfilled pentagon represents a pseudogene, and a filled pentagon represents an annotated coding gene. Lines represent orthology relationships. The direction of the spelt wheat track was reversed manually. LSC represents a large single-copy region. SSC represents a small single-copy region. IRA and IRB represent a pair of IRs. E) Confirmation of the presence of the NPGC in the nuclear genome with PCR amplification experiments. The PCR products span plastid DNA-nucleus junction sites on both sides of the NPGC.

Figure 2.

Overview and diversity of NPGs in Poaceae. A) The cladogram depicting evolutionary relationships among 12 Poaceae species/lineages. The numbers in parenthesis show the counts of the genomes used in each lineage. The left tree is supported by previous studies (Wang et al. 2020; Zhou et al. 2021; Li et al. 2022) and the TimeTree website (http://timetree.org/). The pies show the proportion of lineage-specific NPGCs in each lineage. A total of 93 diploid assemblies were used. The tribe Triticeae was highlighted. B and C) Average counts of NPGs B) and NPGCs C) in each lineage. D) Comparison of plastid-to-nucleus gene transfer counts between Triticeae and non-Triticeae in Poaceae. The Wilcoxon test was performed using 69 Triticeae subassemblies and 24 non-Triticeae assemblies. P values <0.05 were statistically significant. E) Estimated transfer times of NPGs, as investigated by the Ks method. The dashed line indicates the differentiation time of the wheat A and D lineages. The tribe Triticeae evolved ∼25 million years ago in the Poaceae (Feldman and Levy 2015). F) An example shows the wheat D lineage-specific NPGC network. Cluster P65 is highly conserved in the wheat D lineage. In the left network, each node represents an NPGC, and edges indicate collinearity relationships. The right microcollinearity plot shows the polymorphism and collinearity of NPGCs among the 4 genomes. An unfilled pentagon represents a pseudogene, and a filled pentagon represents an annotated coding gene. Lines represent orthology relationships. The direction of the Zavitan track was reversed manually. The genomes shown in the microcollinearity plot are highlighted with rings in the network.

Figure 3.

The distribution and transfer preference of NOGs in Triticeae. A) The positions of all NOGCs in the nuclear genome of bread wheat. Dark and light colors indicate number of events per megabase. NOGCs in 11 hexaploid wheat genomes were anchored to the Chinese Spring reference genome with collinearity. B) Colocalization of NPGCs and NMGCs in the Chinese Spring wheat genome. The adjacent NOGCs were determined by an interval of 1 Mb. A hypergeometric test was performed to evaluate the significance of the colocalization. C) The density of NOGCs on different chromosome zones, as defined by a previous study (Choulet et al. 2014; International Wheat Genome Sequencing Consortium 2018). The Wilcoxon test was performed. P values <0.05 were statistically significant. All 21 chromosomes were used. R1 and R3 represent distal regions of the chromosomal long arm and short arm, respectively. C represents the centromeric/pericentromeric region. R2a and R2b represent the rest. D) A Circos plot shows the distributions of original plastid sequences of NPGCs. The tracks show chromosomal position (a), gene position (b), density of mapped NPGC sequences (c), and mapping position of NPGC sequences (d). In track a, the numbers indicate the length in kilobases. In track b, medial blocks represent genes on the reverse strand; lateral blocks represent genes on the forward strand. In track c, the mapping positions of NPGC sequences are normalized with repeat sequence abundance. E) The NPGC coverages on original plastid genes in 3 regions, LSC, SSC (small single-copy), and IR regions. Each point represents 1 gene. The symbol names of the top 10 genes with the highest coverage are noted. For an easier comparison of plastid and mitochondrion genomes, the coverage of each site was normalized by dividing by the coverage of the site with the highest transfer frequency. F) Comparison of coverage of integration sequences in different bins of the original plastid (pt) and mitochondrial (mt) genomes. For NPGCs and NMGCs, the coverage of each site is normalized by referring to the method in E). The coefficient of variation, defined as the Sd of coverage divided by the mean coverage, is calculated.

Figure 4.

Dynamic trajectories of NOGs during wheat evolution. A) Overview of the retained NPGCs in wheat during 2 rounds of polyploidization. Numbers in circles indicate the number of retained NPGCs. The percentages next to the numbers represent the retention ratio of NPGCs during polyploidization. Two examples show representative transfer events during the evolution of the wheat A genome. Microcollinearity plots show the polymorphism and collinearity of NPGCs among multiple wheat A genomes. An unfilled pentagon represents a pseudogene, and a filled pentagon represents a coding gene. Lines represent orthology relationships. B) NOGCs-based cladograms. The left tree is supported by previous studies (Wang et al. 2020; Zhou et al. 2021; Li et al. 2022) and the TimeTree website (http://timetree.org/). The trees constructed by the PAV matrices of NPGCs and NMGCs from multiple genomes are highlighted in different colors. C) The counts of the NPGCs and NMGCs shared by 11 wheat genomes. The classification of core, dispensable, and private sets was performed according to a previous study (Qin et al. 2021). D) The Ks distributions of NOGs in private and core sets. The Wilcoxon test was performed. *P < 0.05. **P < 0.01. ***P < 0.001. E) The cumulative distributions show NOGC occurrence frequencies across wheat accessions. The Wilcoxon test was performed. Resequencing data sets of 418 bread wheat accessions were used for analysis.

Figure 5.

A high rate of selective retention of NOGs in wheat. A) NOGC sizes decrease over time. The X-axis shows the median of Ks values for each NOGC. The Y-axis shows the median counts of genes in each NOGC. B) The proportions of retained NPGs and NMGs in main crops such as wheat, rice, and maize. The Wilcoxon test was performed on 2 groups: the “wheat subassemblies” group (n = 33), which contains 33 subassemblies from 11 wheat accessions, and the “rice and maize assemblies” group (n = 20), which contains 10 rice genomes and 10 maize genomes. P values <0.05 were statistically significant. Mean and standard error are shown. C) The shared ratio of NPGCs and NMGCs revealed by wheat, rice, and maize genome assemblies. The shared ratio was calculated by dividing the counts of genomes containing the NOGCs by the total number of genomes. The colors of the lines correspond to the colors of the bars in B). D) NPG retention in the bread wheat A, B, and D subgenomes. NPGs were classified as retained when Ks > 0.04. The genes were grouped according to their biological function, as defined by a previous study (Green 2011). NPGs were clustered using hierarchical clustering. E) The microcollinearity plot shows the polymorphism and collinearity of an NPGC between 4 genomes. An unfilled pentagon represents a pseudogene, and a filled pentagon represents a coding gene. Lines represent orthology relationships. The right panel represents the sequence alignment of the NPGC Cluster P3-located region between the 3D chromosome of the CS genome (x-axis) and the plastid genome (y-axis). The lines parallel to the coordinate axis indicate the location of the genes.

Figure 6.

Schematic representation of the Poaceae NOG platform and the radiocarbon-like model of IGTs. A) Overview of the pNOGmap database. The “NOG viewer” module shows detailed information about NOGs and the distribution of NOGs in the genome. The “microcollinearity” module shows the gene context of NOGC on a local scale across multiple genomes. With this tool, users can trace the origin, elimination, and retention dynamics of NOGs. Detailed gene information can be viewed by clicking on the gene. B) A schematic summary showing the radiocarbon-like model illustrating the evolutionary dynamics of organelle-to-nucleus gene transfer and their transition from stability to dynamism. The upper panel represents radiocarbon decay, and the bottom panel represents the transfer and evolution of NOGs. Before transfer, the genetic material in organelles had a low mutation rate and was relatively stable. These organellar sequences can be likened to carbon-14. After transfer, the organelle-to-nucleus genetic material transfer could undergo elimination and mutation. We liken the evolutionary dynamic of NOGs in the nuclear genome to radiocarbon decay in the remains of once-living organisms. The observed IGTs in organisms now aggregate many nonrandom retentions and recent transfers over evolutionary time.