In short
- Carbohydrate-active enzymes (CAZymes) build, modify and break down carbohydrates. The CAZy database groups them into families by sequence.
- The main classes are glycoside hydrolases (GH), glycosyltransferases (GT), polysaccharide lyases (PL), carbohydrate esterases (CE), auxiliary activities (AA) and carbohydrate-binding modules (CBM).
- dbCAN2 combines three search methods. Keeping only enzymes found by at least two of them improves accuracy. dbCAN3 adds prediction of the carbohydrate substrate.
- A CAZyme profile predicts which carbohydrates a strain may use. That matters for prebiotic pairing, fermentation and gut colonisation, but it needs confirming with growth tests.
What CAZymes are
Carbohydrates come in an enormous variety of structures, and the enzymes that act on them are grouped into sequence-based families in the CAZy database. The classification began about 30 years ago. The database has been freely available for more than 20 years, is updated monthly and is curated by hand, including the biochemical evidence for each family [1].
| Class | What it does |
|---|---|
| Glycoside hydrolases (GH) | Break glycosidic bonds, for example in starch, cellulose, fructans or lactose |
| Glycosyltransferases (GT) | Form glycosidic bonds, building polysaccharides and cell-surface glycans |
| Polysaccharide lyases (PL) | Cleave uronic acid-containing polysaccharides such as pectin and alginate |
| Carbohydrate esterases (CE) | Remove ester groups, such as acetyl groups, from carbohydrates |
| Auxiliary activities (AA) | Redox enzymes that work alongside other CAZymes, for example in lignocellulose breakdown |
| Carbohydrate-binding modules (CBM) | Non-catalytic modules that help enzymes bind their carbohydrate target |
How dbCAN annotates a genome
dbCAN2 runs three searches on every protein: HMMER against dbCAN's hidden Markov models, DIAMOND against sequences already classified in CAZy, and a short conserved-peptide search. Combining the three and discarding enzymes found by only one method significantly improves accuracy. dbCAN2 can also find CAZyme gene clusters, which point to putative polysaccharide utilisation loci (PULs) [2].
dbCAN3 adds substrate prediction. It uses subfamily-level models (dbCAN-sub), a comparison with experimentally characterised PULs whose glycan substrates are known (dbCAN-PUL), and a majority vote to predict what each gene cluster acts on [3]. This moves the result from "which enzymes are present" towards "which carbohydrates can this strain use".
What a CAZyme profile tells you
- Gut and probiotic strains: CAZymes are needed to digest most of the complex dietary polysaccharides in the human diet, and gut bacteria carry a large variety of them [4]. A strain's profile suggests which fibres and prebiotics it can use, which supports synbiotic design.
- Fermented foods: the profile helps explain which sugars a starter culture ferments and how it behaves in a given substrate.
- Industrial biotechnology: strains rich in GH, PL and AA families are candidates for breaking down plant biomass or producing enzymes.
- Comparisons between strains: differences in CAZyme content are often among the most informative accessory genes in a pangenome.
Limits of genome-based prediction
- A gene shows potential, not activity. It may be silent, regulated or non-functional.
- Many CAZy families contain enzymes with different substrates, so family membership alone doesn't fix what an enzyme acts on.
- Using a carbohydrate also requires transporters and regulatory genes, which CAZyme annotation doesn't cover.
- Growth tests on defined carbohydrates are the way to confirm what the genome predicts.
CAZyme analysis in BacSeq reports
CAZyme annotation with dbCAN2 is part of the Premium package, alongside secondary metabolite analysis with antiSMASH and pathway reconstruction. The report lists every CAZyme family found with counts by class, highlights CAZyme gene clusters, and interprets the profile for your research question, such as prebiotic use for a probiotic strain or enzyme potential for an industrial one.
References
- Drula E, Garron ML, Dogan S, Lombard V, Henrissat B, Terrapon N. The carbohydrate-active enzyme database: functions and literature. Nucleic Acids Res. 2022;50:D571–D577. doi:10.1093/nar/gkab1045
- Zhang H, Yohe T, Huang L, et al. dbCAN2: a meta server for automated carbohydrate-active enzyme annotation. Nucleic Acids Res. 2018;46:W95–W101. doi:10.1093/nar/gky418
- Zheng J, Ge Q, Yan Y, Zhang X, Huang L, Yin Y. dbCAN3: automated carbohydrate-active enzyme and substrate annotation. Nucleic Acids Res. 2023;51:W115–W121. doi:10.1093/nar/gkad328
- El Kaoutari A, Armougom F, Gordon JI, Raoult D, Henrissat B. The abundance and variety of carbohydrate-active enzymes in the human gut microbiota. Nat Rev Microbiol. 2013;11:497–504. doi:10.1038/nrmicro3050