Protein networks identify novel symbiogenetic genes resulting from plastid endosymbiosis

Abstract

The integration of foreign genetic information is central to the evolution of eukaryotes, as has been demonstrated for the origin of the Calvin cycle and of the heme and carotenoid biosynthesis pathways in algae and plants. For photosynthetic lineages, this coordination involved three genomes of divergent phylogenetic origins (the nucleus, plastid, and mitochondrion). Major hurdles overcome by the ancestor of these lineages were harnessing the oxygen-evolving organelle, optimizing the use of light, and stabilizing the partnership between the plastid endosymbiont and host through retargeting of proteins to the nascent organelle. Here we used protein similarity networks that can disentangle reticulate gene histories to explore how these significant challenges were met. We discovered a previously hidden component of algal and plant nuclear genomes that originated from the plastid endosymbiont: symbiogenetic genes (S genes). These composite proteins, exclusive to photosynthetic eukaryotes, encode a cyanobacterium-derived domain fused to one of cyanobacterial or another prokaryotic origin and have emerged multiple, independent times during evolution. Transcriptome data demonstrate the existence and expression of S genes across a wide swath of algae and plants, and functional data indicate their involvement in tolerance to oxidative stress, phototropism, and adaptation to nitrogen limitation. Our research demonstrates the "recycling" of genetic information by photosynthetic eukaryotes to generate novel composite genes, many of which function in plastid maintenance.

Keywords: endosymbiosis; eukaryote evolution; gene fusion; novel gene origin; photosynthesis.

Fig. 1.

Origin of composite genes in algae and plants. (A) The role of plastid endosymbiosis in providing the genetic toolkit for S-gene origin. (B) Network analysis of the AtGRXS16 (family 14) S gene in A. thaliana. The red nodes identify the S genes; green and blue nodes are the components from GIY–YIG and GRXS domains, respectively, that gave rise to S genes through gene fusion. (C) Domain structure of AtGRXS16. An intramolecular disulfide bond can be formed between the two domains. TP, transit peptide.

Fig. 2.

Sixty-seven S-gene families identified in our study. Domains in bold originated from Cyanobacteria. Plastid-localized protein families (i.e., families with at least one protein predicted to be plastid-targeted according to ChloroP and ASAFind) are shaded in gray. *, domain of cyanobacterial origin occurring more than once per S gene. ^†, highly confident domain of cyanobacterial or prokaryotic (noncyanobacterial) origin (Table S4).

Fig. 3.

Putative nuclear gene-based phylogeny of photosynthetic eukaryotes, showing the distribution of the 67 S-gene families we report. SAR, Stramenopiles-Alveolates-Rhizaria.

Fig. S1.

Taxonomic distribution of the 67 S genes discovered in our study. The taxonomic distribution of the data is shown with black boxes indicating a presence and white boxes indicating presumed absence in the genome or transcriptome data from the taxon.

Fig. S2.

Mapping data for S genes in Picochlorum and A. thaliana. (A) S-gene family 31, which encodes RIBR + DUF1768 (PyrR) and is involved in riboflavin biosynthesis. The CDS derives from the MMETSP database and encodes this 1,848-nt S gene from Picochlorum oklahomensis CCMP2329. The 6,609 RNA-seq reads mapped to it are from the closely related species Picochlorum SE3 (55). The homologous SE3 gene is encoded on Picochlorum contig 185.g609.t1. (B) S-gene family 23, which encodes a TPR repeat/RING and an ATP-dependent protease domain. This CDS also derives from the MMETSP database, and encodes the 1,554-nt S gene from P. oklahomensis CCMP2329. The 9,253 RNA-seq reads mapped to it are from Picochlorum SE3. The homologous SE3 gene is encoded on Picochlorum contig 43.g1 98.t1. (C) S-gene family 14, which encodes GIY–YIF superfamily and thioredoxin superfamily domains. This genic region is from A. thaliana (ArGrxS16), and the 16,906 unique reads mapped to the exons are also from this species [see Table S2 for Sequence Read Archive (SRA) run accession numbers]. Thin blue lines indicate a spliceosomal intron. The mapping in all cases is colored in green, red, and blue for forward, reverse, and paired-end reads, respectively. The fused domains and their putative annotations are shown. These data are typical for all of the Picochlorum and Arabidopsis mappings and for many of the other plant and algal S genes when sufficient RNA-seq data are available (Table S2). This unambiguous, “deep” mapping across the region that spans the domain fusion argues strongly against misassembly of this (and other) S genes.

Fig. S3.

Genomic PCRs that targeted S genes identified in Picochlorum species. The contig number in the Picochlorum SE3 assembly is given for each S gene, as is the S-gene family number (see Table S2 for details). The PCR primers were complementary to regions at the 5′ and 3′ termini of the S genes to span the domain-fusion region. The sizes of these S-gene CDS fragments matched the fragment sizes resulting from PCR amplification as follows: S gene 11 (1,015 nt), S gene 23 (1,419 nt), S gene 4 (1,214 nt), S gene 34 (924 nt), and S gene 12 (1,119 nt). Sanger sequencing of these PCR fragments showed identity to the genomic region in Picochlorum SE3, and BLASTX analysis using the fragments showed that each spanned the domain-fusion region in the respective S gene. The match of the CDS size to the genomic region is explained by the paucity of spliceosomal introns in Picochlorum SE3. These data demonstrate that the tested S genes exist as intact fragments in this green alga.

Fig. S4.

Maximum-likelihood (RAxML) (57) tree of species encoding S-gene family 31. This composite gene is limited to the Viridiplantae (Fig. S1) and encodes fused RIBR + DUF1768 (PyrR) domains that are involved in riboflavin biosynthesis. This manually trimmed alignment includes a selection of taxonomically diverse green lineage species and is of length 524 amino acids. The intact gene (see also mapping evidence in Fig. S2) was analyzed using the LG + Γ + I evolutionary model with the results of 100 bootstrap replicates, when ≥50% are shown at the branches. The topology of this tree is consistent with the expected phylogeny of Viridiplantae (e.g., 56), indicating an ancient origin of this S gene. The NCBI gi numbers are shown after each species name, when available.