Notes on nitrogen geometries
- Pavan Behara
@David Mobley
DLM: In the context of the work we're doing on geometries of nitrogens to applications in crystal structure refinement, etc. do you have any literature examples handy from your work or other work you're familiar with that you could share?
TF: the following articles may serve as starting points:
Keep it together: restraints in crystallographic refinement of macromolecule–ligand complexes Roberto A. Steiner and Julie A. Tucker doi: 10.1107/S2059798316017964
Protein–Ligand Cocrystal Structures: We Can Do Better, C.H. Reynolds, doi: 10.1021/ml500220a
Intriguing role of water in protein-ligand binding studied by neutron crystallography on trypsin complexes Johannes Schiebel et al., DOI: 10.1038/s41467-018-05769-2
Another starting point is the work of Lance Westerhoff from QuantumBio – they make the point that using dictionaries in XRay refinement is not a good idea and propose doing a combined e-density and QM refinement. Lance might have some examples where the sp2 / sp3 switch is important. I haven’t seen anything particular in their papers, maybe the one where the do a large scale comparison of different refinement procedures has sth in it:
High-throughput quantum-mechanics/molecular-mechanics (ONIOM) macromolecular crystallographic refinement with PHENIX/DivCon: the impact of mixed Hamiltonian methods on ligand and protein structure
Oleg Borbulevych Roger I. Martina and Lance M. Westerhoff High-throughput quantum-mechanics/molecular-mechanics (ONIOM) macromolecular crystallographic refinement with PHENIX/DivCon: the impact of mixed Hamiltonian methods on ligand and protein structure
When it comes to actual examples, you might want to search the PDB for phenyl-piperidines, -pyrrolidines or – piperazines. There sometimes they are planar, sometimes sp3. One example I checked is the ligand 3WC in the structure 3ZPR: planar in the pdb (sum of angles around N: 359.7), a Gaussian calculation gives 353.6°, and the very similar compound 3C6 (pdb code 4U41) sums up to 331.8 degrees. One has to be careful though, its always also a matter of resolution.
Another example is 1s3v: in the pdb the aniline is strictly planar (360 degrees), in Gaussian I get 355 degrees. Admittedly, the difference is not big – but I'm sure if one searches for examples there will be others and better.
One striking example DR discovered is the structure 4dmx. From the density the nitrogen probably is much more planar than the modelled ligand, which has the sp3 nitrogen. Let us know if you need more – DR offered to search for some problematic cases if you need some.
DR: Please note that almost all big pharma companies use GRADE from Global Phasing for generating high quality restraints. There is no publication as far as I know but a web server where some information resources are linked (http://grade.globalphasing.org/cgi-bin/grade/server.cgi ).
In addition to that, geometries are usually monitored by using MOGUL. This is also the PDBe standard by now (https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind1906&L=CCP4BB&O=D&P=71908).
This does not mean, however, that the ‘nitrogen problems’ are solved … these are particularly tricky. But usually, restraints are carefully prepared and monitored.
You can also use existing forcefields instead of prepared ligand restraints (e.g. https://www.globalphasing.com/buster/wiki/index.cgi?AutobusterLigandForcefield ) or even QM (https://www.globalphasing.com/buster/wiki/index.cgi?AutobusterLigandQMIntroduction ). More work on this would be highly appreciated.