Ureidoglycolate hydrolase, amidohydrolase, lyase: how errors in biological databases are incorporated in scientific papers and vice versa

Abstract

An opaque biochemical definition, an insufficient functional characterization, an interpolated database description, and a beautiful 3D structure with a wrong reaction. All these are elements of an exemplar case of misannotation in biological databases and confusion in the scientific literature concerning genes and enzymes acting on ureidoglycolate, an intermediate of purine catabolism. Here we show biochemical evidence for the relocation of genes assigned to EC 3.5.3.19 (ureidoglycolate hydrolase, releasing ammonia), such as allA of Escherichia coli or DAL3 of Saccharomyces cerevisiae, to EC 4.3.2.3 (ureidoglycolate lyase, releasing urea). The EC 3.5.3.19 should be more appropriately named ureidoglycolate amidohydrolase and include genes equivalent to UAH of Arabidopsis thaliana. The distinction between ammonia- or urea-releasing activities from ureidoglycolate is relevant for the understanding of nitrogen metabolism in various organisms and of virulence factors in certain pathogens rather than a nomenclature problem. We trace the original fault in database annotation and provide a rationale for its incorporation and persistence in the scientific literature. Notwithstanding the technological distance, yet not surprising for the constancy of human nature, error categories and mechanisms established in the study of the work of amanuensis monks still apply to the modern curation of biological databases.

Curation accuracy of model organism databases

“…Previous discussions of errors in manual extraction of information from publications include discussions of errors that occur in chemical databases and the importance of manual curation and quality-control procedures for ensuring accuracy in such databases ( 11 , 12 ), and an in-depth analysis ( 13 ) of the scientific confusion and curation errors that led to the incorrect attributions of enzyme function in a number of bioinformatics databases. A study of curated protein-interaction data examined variation in curation of the same papers by different database projects, and found that differences in curated facts reflected both divergent curation policies for those databases, as well as curation errors ( 14 ).…”

Automated Detection of Records in Biological Sequence Databases that are Inconsistent with the Literature

“…Even in very small bacterial genomes, many misannotations may arise [31]. As for high-throughput functional annotations, errors may occur due to a variety of factors [34,39], but the most common errors are over-annotations, in which a gene is given a specific but incorrect function [49,31,36]. Once made, functional annotation errors can be difficult to correct in large scale sequence databases and as functional annotations are often inferred 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 from sequence similarity to other annotated sequences, errors may "propagate" to newly sequenced genomes through "(mis)annotation transfer" [18,24,43].…”

Coping with inevitable accidents in metabolism

“…. Remembering that annotations in databases are often unreliable (Percudani et al ., ), it is essential to try and follow the chemical constraints in the postulated pathways and reappraise current annotations. In aerobic cells, urate may be oxidized to 5‐hydroxyisourate by two distinct enzymes: either a coenzyme‐independent urate oxidase (uricase, EC 1.7.3.3) found in eukaryotes and bacteria like B. subtilis (PucL) or a prokaryotic flavoprotein urate hydroxylase (HpxO) originally found in some Klebsiella species (Michiel et al ., ).…”