Discovery and revision of Arabidopsis genes by proteogenomics

Abstract

Gene annotation underpins genome science. Most often protein coding sequence is inferred from the genome based on transcript evidence and computational predictions. While generally correct, gene models suffer from errors in reading frame, exon border definition, and exon identification. To ascertain the error rate of Arabidopsis thaliana gene models, we isolated proteins from a sample of Arabidopsis tissues and determined the amino acid sequences of 144,079 distinct peptides by tandem mass spectrometry. The peptides corresponded to 1 or more of 3 different translations of the genome: a 6-frame translation, an exon splice-graph, and the currently annotated proteome. The majority of the peptides (126,055) resided in existing gene models (12,769 confirmed proteins), comprising 40% of annotated genes. Surprisingly, 18,024 novel peptides were found that do not correspond to annotated genes. Using the gene finding program AUGUSTUS and 5,426 novel peptides that occurred in clusters, we discovered 778 new protein-coding genes and refined the annotation of an additional 695 gene models. The remaining 13,449 novel peptides provide high quality annotation (>99% correct) for thousands of additional genes. Our observation that 18,024 of 144,079 peptides did not match current gene models suggests that 13% of the Arabidopsis proteome was incomplete due to approximately equal numbers of missing and incorrect gene models.

Fig. 1.

Work flow. All mass spectra are compared with three databases using Inspect. Spectra are filtered to a 1% false discovery rate and grouped into peptides. Novel peptides are separated from those that appear in TAIR7 and clustered. It is important to note that only a subset of the novel peptides appear in a peptide cluster. Novel peptide clusters are then segregated based on genome location. Those that overlap a current gene model (intragenic) are further classified by how they refine the model. Peptides that do not overlap a gene model (intergenic) are classified by whether they overlap a pseudogene. The peptide clusters, along with evidence from cDNA and current gene annotations, are given to the gene predictor AUGUSTUS to produce new gene models. Not all peptides in the peptide clusters are included in the final AUGUSTUS models.

Fig. 2.

Novel gene supported discovered by proteogenomics. (A) A cluster of 13 uniquely located peptides that do not overlap a current gene model (Chr3). The prediction track shows the single exon gene model produced by AUGUSTUS. (B) The predicted sequence shows strong homology to a Thylakoid lumen family protein (sp|P82658|TL19_ARATH). It also shows strong similarity to proteins in both grapevine (emb|CAO40861.1 a hypothetical gene) and rice (Os08g0504500 a cDNA derived gene).

Fig. 3.

Peptides overlapping a predicted transposable element gene. (A) 5 peptides, 4 unique, overlap locus AT4G07947, which is annotated as a transposable element gene. (B) Sequence alignment to an Arabidopsis Ulp1 (ubiquitin like protease) showing strong conservation (56% identity, e value 0.0). Observed peptides highlighted.

Fig. 4.

Refined Gene Model. TAIR locus AT1G63500 encodes a protein kinase. (A) Four novel peptides map within the 5′ UTR and the first exon. (B) Zoom of the region shows that the current first exon (frame 3) is out of frame with the peptides (frame 2). (C) Sequence alignment with Arabidopsis and grapevine proteins supports translation in the frame supported by peptides (observed peptides highlighted in alignment).