Deep transcriptome annotation suggests that small and large proteins encoded in the same genes often cooperate

Abstract

25 26Recent studies in eukaryotes have demonstrated the translation of alternative open 27 reading frames (altORFs) in addition to annotated protein coding sequences (CDSs). We 28show that a large number of small proteins could in fact be coded by altORFs. The 29 putative alternative proteins translated from altORFs have orthologs in many species and 30 evolutionary patterns indicate that altORFs are particularly constrained in CDSs that 31 evolve slowly. Thousands of predicted alternative proteins are detected in proteomic 32 datasets by reanalysis using a database containing predicted alternative proteins. Protein 33 domains and co-conservation analyses suggest a potential functional relationship between 34 small and large proteins encoded in the same genes. This is illustrated with specific 35 examples, including altMiD51, a 70 amino acid mitochondrial fission-promoting protein 36 encoded in MiD51/Mief1/SMCR7L, a gene encoding an annotated protein promoting 37 mitochondrial fission. Our results suggest that many coding genes code for more than one 38 protein that are often functionally related.

Recognition of the polycistronic nature of human genes is critical to understanding the genotype-phenotype relationship

“…Peptide identification relies on matching mass spectra to predicted peptides from CDS (e.g., from UniProt database). If the database does not contain the relevant peptide, the associated protein will never be identified since it is not included in the scope of possibilities (Samandi et al 2017). As of today, <50% of all MS/MS spectra from a proteomics experiment are matched with high confidence (Heo et al 2010;Chick et al 2015).…”

“…5A). The phyloP score (a measure of probability to be under purifying selection) can be computed for every third base giving a triplet signal (three graphs corresponding, respectively, to the first, second, and third nucleotide for each codon) (Cooper et al 2005;Samandi et al 2017). After noise reduction, we can detect independent purifying selection signals in each of the reading frames, e.g., for the dual coding MIEF1 gene (Fig.…”

“…These unassigned peptides can correspond to peptide modifications or to proteins not in the database (Heo et al 2010). In the recently developed proteogenomics approaches, addition of more inclusive databases to the search space allows for the discovery of novel proteins thus far undetected (Oyama et al 2007;Saghatelian and Couso 2015;Samandi et al 2017;openprot. org).…”

“…Functional proteins are expected to be under purifying selection and the ratio of nonsynonymous to synonymous mutations highlights protein-coding sequences (Hughes 1999;Pál et al 2006). The first and second nucleotide of a codon experience stronger selection than the third because of the genetic code's redundancy (Pollard et al 2010;Samandi et al 2017). This selection periodicity can allow for detection of conservation signatures in each of the reading frames (Fig.…”

“…They are, in general, notably smaller than annotated CDSs, but they are not limited to small ORFs (smORFs-ORFs smaller than 100 codons) (Samandi et al 2017). Here, we define alternative ORFs as any coding sequence with an ATG start codon encoded within any reading frame of either lncRNAs or known coding mRNAs (either in UTRs or overlapping the CDS).…”