This page has been approved by the Responsible Curator as essentially complete. CAZypedia is a living document, so further improvement of this page is still possible. If you would like to suggest an addition or correction, please contact the page's Responsible Curator directly by e-mail.
Sequence classification methods require knowledge of at least part of the amino acid or nucleotide sequence for a protein. Algorithmic methods are then used to compare and classify sequences (e.g. the original classification of Glycoside Hydrolase Families relied largely on hydrophobic cluster analysis and multiple sequence alignment [1, 2], while sequence alignment and Hidden Markov Model methods have become dominant with the evolution of the carbohydrate-active enzymes classification [3, 4]). Each of the resulting sequence-based families contain proteins that are related by sequence, and by corollary, three-dimensional fold. An obvious shortcoming of sequence-based classifications is that they can only be applied to proteins for which sequence information is available. On the other hand sequence-based classification schemes allow classification of proteins for which no biochemical evidence has been obtained such as the thousands of uncharacterized sequences of carbohydrate-active enzymes that originate from genome sequencing efforts worldwide. Sequence-based classification methods are rather different (and in many ways complementary) to the Enzyme Commission classification scheme, which assigns proteins to groups based on the nature of the reactions that they catalyze [5].
Using a combination of comparison algorithms the glycoside hydrolases have been classified into more than 100 GH families [2]. This classification is permanently available through the Carbohydrate Active enZyme database [4]. Classification of glycoside hydrolases into families allows many useful predictions to be made since it has long been noted that the catalytic machinery and molecular mechanism is conserved for the vast majority of the GH families [6] as well as the geometry around the glycosidic bond (irrespective of naming conventions) [7]. Usually, the mechanism used (ie retaining or inverting) is conserved within a GH family. One notable exception is the glycoside hydrolases of family GH97, which contains both retaining and inverting enzymes; a glutamate acts as a general base in inverting members, whereas an aspartate likely acts as a catalytic nucleophile in retaining members [8]. Another mechanistic curiosity are the glycoside hydrolases of familes GH4 and GH109 which operate through an NAD-dependent hydrolysis mechanism that proceeds through oxidation-elimination-addition-reduction steps via anionic transition states [9]. This allows a single enzyme to hydrolyze both alpha- and beta-glycosides.
As a consequence of the evolution of the classification, several GH families have been deleted. Once deleted, family numbers are never reused in order to prevent confusion.
Classification of GH families into larger groups, termed "clans", has been proposed [10, 11]. A clan is a group of families that possess significant similarity in their tertiary structure, catalytic residues and mechanism. Thus knowledge of three-dimensional structure and the functional assignment of catalytic residues is required for classification into clans. Families within clans are thought to have a common evolutionary ancestry. Please see the CAZy Database for a current table of glycoside hydrolase clans.
Classification of glycosyltransferases
Using sequence comparison algorithms glycosyltransferases that use nucleotide diphospho-sugar, nucleotide monophospho-sugars and sugar phosphates have been grouped into over 90 GT families [12, 13]. This classification is permanently available through the Carbohydrate Active enZyme database[14]. As for the GH families above, the same three-dimensional fold is expected to occur within each of the GT families. Just as for the glycoside hydrolases, several of the families defined on the basis of sequence similarities turn out to have similar three-dimensional structures.
As a consequence of the evolving classification, GT families may be deleted; one example is GT36, which has been reclassified as GH94. Once deleted, family numbers are never reused in order to prevent confusion.
Davies, G.J. and Sinnott, M.L. (2008) Sorting the diverse: the sequence-based classifications of carbohydrate-active enzymes. The Biochemist, vol. 30, no. 4., pp. 26-32. Download PDF version. [DaviesSinnott2008]
