Date Published: April 01, 2018
Publisher: International Union of Crystallography
Author(s): Paul Emsley, Max Crispin.
Newly improved tools for carbohydrate modelling in Coot are presented.
Cell-surface and secreted proteins are often modified by numerous asparagine (N)-linked glycans. In addition to their role in lectin-mediated protein folding, glycans often play a structural role by forming intramolecular interactions with the protein surface which can stabilize protein domains and influence dynamics (Petrescu et al., 2006 ▸). Although glycans have the capacity to be highly dynamic and therefore conformationally heterogenous, they are increasingly being observed by both X-ray crystallography and cryo-electron microscopy (cryo-EM; see, for example, Bai et al., 2015 ▸). This trend includes a growing number of examples of glycans that are braced against protein surfaces, including by antibody binding (Pejchal et al., 2011 ▸), and by the advent of methods to manufacture chemically homogenous glycoforms for structural analysis (Chang et al., 2007 ▸).
We wanted to provide a tool in Coot that was interactive and could provide the user with a knowledge-based model-building guide through glycan space. The carbohydrate-building tool was designed to have three modes.(i) ‘Expert User’ mode: monomer-by-monomer addition of the ‘next’ monosaccharide. The user chooses the link type and the monosaccharide type. The different hypotheses for the position, orientation and conformation of the ‘next’ monosaccharide are assessed and the best one is added and refined; control is then returned back to the user.(ii) Linked Monosaccharide Addition (LMA): as above, with the modification that Coot uses glycan comprehension. Given a (user-selected) glycan type, only certain monosaccharide types with certain link types are available for any given position on the tree [for example, only N-acetyl-β-d-glucosamine (NAG) linked by ‘NAG-ASN’ is available for the first position].(iii) Whole Tree Addition (WTA), where the user need only identify the starting asparagine and the glycosylation tree type to be added. The options are ‘High Mannose’, ‘Hybrid (Mammal)’, ‘Complex (Mammal)’ and ‘Complex (Plant)’. This mode automatically (i.e. without user intervention) applies built-in knowledge of residues and link types for particular glycosylation trees, and uses density fit for branch termination.
The work described here is motivated to help to tackle the challenge of accurately interpreting both crystallographic and cryo-EM maps of glycans and in part to address the concerns raised by Agirre et al. (2017 ▸). The LMA mode is the mode that we imagine that users will find most useful.