Modelling covalent linkages in CCP4

Abstract

In this contribution, the current protocols for modelling covalent linkages within the CCP4 suite are considered. The mechanism used for modelling covalent linkages is reviewed: the use of dictionaries for describing changes to stereochemistry as a result of the covalent linkage and the application of link-annotation records to structural models to ensure the correct treatment of individual instances of covalent linkages. Previously, linkage descriptions were lacking in quality compared with those of contemporary component dictionaries. Consequently, AceDRG has been adapted for the generation of link dictionaries of the same quality as for individual components. The approach adopted by AceDRG for the generation of link dictionaries is outlined, which includes associated modifications to the linked components. A number of tools to facilitate the practical modelling of covalent linkages available within the CCP4 suite are described, including a new restraint-dictionary accumulator, the Make Covalent Link tool and AceDRG interface in Coot, the 3D graphical editor JLigand and the mechanisms for dealing with covalent linkages in the CCP4i2 and CCP4 Cloud environments. These integrated solutions streamline and ease the covalent-linkage modelling workflow, seamlessly transferring relevant information between programs. Current recommended practice is elucidated by means of instructive practical examples. By summarizing the different approaches to modelling linkages that are available within the CCP4 suite, limitations and potential pitfalls that may be encountered are highlighted in order to raise awareness, with the intention of improving the quality of future modelled covalent linkages in macromolecular complexes.

Keywords: AceDRG; CCP4; CCP4 Monomer Library; covalent linkages; link dictionary; link records; link-restraint dictionary; mmCIF; monomer library; restraints.

Figure 1

Example LINK and LINKR records, corresponding to the covalent linkage of the NZ atom in lysine (LYS) and the C4A atom in pyridoxal phosphate (PLP). In this case, LYS-A226 is linked to PLP-A501. The two fields marked ‘1555’ correspond to symmetry operators (in this case the linked atoms are located within the same asymmetric unit). LINK and LINKR records differ only in the final field: LINK records have a link distance (for example ‘1.27’), whereas LINKR records have a link identifier (for example ‘LYS-PLP’). For further details about the format of LINK records, see Callaway et al. (1996 ▸).

Figure 2

Dataflow involved in modelling covalent linkages using the CCP4 suite, as coordinated by the graphical project-management environments CCP4i2 and CCP4 Cloud. Processes and programs are depicted as orange rectangles, data as document symbols (arrays indicate the potential presence of multiple instances), models as parallelograms and databanks as cylinders. Arrows indicate directional flow, coloured according to the matching data type: red for link dictionaries, blue for link records and green for models. Text labels are coloured black for graphical interactive processes and white for (semi-)automated processes and data. Additional representations are provided as supporting information: see Supplementary Fig. S1 for a simplified linkage information dataflow and Supplementary Fig. S2 for a GUI-centric process flow diagram.

Figure 3

Modelling haem B and haem C using monomer descriptions from the Chemical Component Dictionary (CCD; Westbrook et al., 2015 ▸). (a) Haem B (CCD identifier HEM). (b) Haem C (CCD identifier HEC) covalently bound to protein via cysteine thiols. Note that the wwPDB recommends against using the CCD component HEM for modelling haem C, which is found covalently linked to other components via thioether bridges. Also, the Fe atom should have charge +2 (unless bound to another molecule); standard representations are presented. Images were created using ChemDraw Professional 17.1.

Figure 4

Description of the covalent linkage of N-acetylglucosamine (NAG) and asparagine (ASN) using AceDRG. (a) Chemical diagrams of the individual NAG and ASN components (from the CCP4-ML) and (b) the linked composite compound, in which the covalent linkage is depicted as a dotted line (created using ChemDraw Professional 17.1). (c) Comparison of a deposited 2.4 Å resolution model (PDB entry 3kwf; Mattei et al., 2010; purple) and the model re-refined with REFMAC5 using an AceDRG link dictionary (green), focusing on NAG-A796 and ASN-A229, displayed using Coot. Interatomic distances and dotted lines corresponding to the linkage are shown for both models (note that the deposited model did not contain a corresponding link record). The 2mF _o − DF _c map corresponding to the re-refined model is shown as a grey mesh. Transparent surfaces surrounding atoms in the linked complex highlight the atoms involved in link-dictionary restraints, corresponding to (d) changes in bond/angle/chirality restraints (red surface), (e) torsion-angle restraints (green), (f) planar restraints that are removed (blue) and (g) planar restraints that are added (gold) due to the covalent linkage. H atoms were modelled in riding positions using REFMAC5.

Figure 5

Description of the covalent linkage of lysine (LYS) and pyridoxal phosphate (PLP) using AceDRG. (a) Chemical diagrams of the individual LYS and PLP components (from the CCP4-ML) and (b) the linked composite compound, in which the covalent linkage is depicted as a dotted line (created using ChemDraw Professional 17.1). (c) Comparison of a deposited 1.8 Å resolution model (PDB entry 6ndn; J. F. Scortecci, J. Brandao-Neto, H. M. Pereira & O. H. Thiemann, unpublished work; purple) and the model re-refined with REFMAC5 using an AceDRG link dictionary (green), focusing on LYS-A226 and PLP-A501, displayed using Coot. Interatomic distances and dotted lines correspond to the covalent linkage. The 2mF _o − DF _c map corresponding to the re-refined model is shown as a grey mesh. Transparent surfaces surrounding atoms in the linked complex highlight the atoms involved in link-dictionary restraints, corresponding to (d) changes to bond/angle restraints (red surface), (e) torsion-angle restraints (green), (f) planar restraints that are removed (blue) and (g) planar restraints that are added (gold) due to the covalent linkage. Note that the O4A atom deleted from PLP (and thus not shown) was involved in the removed planar restraint. H atoms were modelled in riding positions using REFMAC5.

Figure 6

Description of the covalent linkages between methionine (MET), tyrosine (TYR) and tryptophan (TRP) in a MET–TYR–TRP cross-link; examples correspond to haem-dependent catalase–peroxidase enzymes. (a) Chemical diagrams of the individual MET, TYR and TRP components (from the CCP4-ML) and (b) the linked composite compound, in which the covalent linkages are depicted as dotted lines (created using ChemDraw Professional 17.1). (c) and (d) show Coot depictions of a 2.4 Å resololution model (PDB entry 1sj2; Bertrand et al., 2004 ▸) focused on MET-A255, TYR-A229 and TRP-A107 after re-refinement with REFMAC5. Models were re-refined without modelling the covalent linkage (c) (yellow), using an AceDRG link dictionary (c, d) (green) and using a link record but no dictionary (d) (blue). (e) Coot depiction of a 1.4 Å resolution model (PDB entry 5jhy; Gasselhuber et al., 2016 ▸) focused on MET-A299, TYR-A273 and TRP-A140. The deposited model is shown (purple), as well as that after re-refinement with REFMAC5 using the AceDRG link dictionary (green). Interatomic distances between covalently linked atoms are shown and are coloured according to the corresponding model.