Comparative Genome Mining of the Extremely Thermophilic Genus Caldicellulosiruptor Reveals a Conserved Architecture for Hemicellulose Deconstruction with Variable Enzyme Inventories

Hemicellulose deconstruction by the extremely thermophilic genus Caldicellulosiruptor has been studied mainly in single species. How far its enzymatic systems are conserved across the genus is unknown. Here, 19 non-redundant wild-type genomes and the cellulosomal outgroup Acetivibrio thermocellus DSM 1313 were annotated by one uniform pipeline. They were mined for carbohydrate-active enzymes (CAZymes), their predicted destinations, gene clustering, and genome relatedness. CAZyme gene numbers varied from 93 to 153 per genome, while the organisation of the system did not. In every genome, xylan degradation was predicted to be divided between compartments. All seven GH11 and 23 of 30 multi-modular GH10 xylanases were exported, that is, secreted or attached to the cell surface, whereas none of the 21 GH39 β-xylosidases, 26 GH51 α-L-arabinofuranosidases, or 17 GH67 α-glucuronidases were exported. A 20.5 kb locus joining the GH67 α-glucuronidase to an ABC importer, a two-component system, and uronate catabolism occurred in 14 of 15 closed genomes. ABC importers formed 86.1% of the transporters in hemicellulolytic clusters. Enzyme content followed relatedness only weakly (Mantel r = 0.378, p = 0.0002). One conserved architecture thus carries strain-level enzyme variation. The exported 11.4% of CAZymes, probably a minimum, is the pool relevant to thermostable biocatalysts. All assignments are computational predictions awaiting experimental confirmation.