Date Published: November 01, 2019
Publisher: International Union of Crystallography
Author(s): Matthew L. Dennis, Lygie Esquirol, Tom Nebl, Janet Newman, Colin Scott, Thomas S. Peat.
The structure of AtzT, which is thought to be involved in atrazine uptake in bacteria, was solved by SAD phasing using an ethylmercury phosphate derivative. Density in the binding site was subsequently found to be guanine that was retained by the protein throughout the purification process. Replacing the guanine with 2-hydroxyatrazine allowed the manner of binding of this substrate to be determined and its comparison with what appears to be the original ligand for this protein.
Atrazine is one of the most commonly used herbicides on the planet, with more than 70 million pounds used in the US alone every year (United States Environmental Protection Agency, 2013 ▸). While the use of atrazine has been reported to enable benefits to crop yields, there are significant concerns about its impact on the environment and human health (Nicolopoulou-Stamati et al., 2016 ▸; Rickard et al., 2018 ▸). Bacteria have evolved several pathways for the catabolism of atrazine and related synthetic compounds (i.e. other s-triazines), allowing them to convert these nitrogen-rich compounds to ammonia to support growth.
As periplasmic binding proteins often contain a twin-arginine translocation (TAT) signal at the N-terminus to target the protein to the periplasm (Palmer & Berks, 2012 ▸), the TAT signal sequence was identified using TatP (http://www.cbs.dtu.dk/services/TatP/) and Protter (http://wlab.ethz.ch/protter/start/). Primers were designed to amplify the gene sequence without the sequence encoding the TAT signal in order to facilitate heterologous expression. The X-ray crystallographic analysis shows that AtzT is a two-domain protein, with each domain containing a central β-sheet of 6–7 strands, with 2–3 α-helices on either side of this central β-sheet and a C-terminal extension of about 45 residues (see Fig. 1 ▸). The N-terminal domain is made up of residues 27–133 plus an extension from the C-terminal domain of residues 257–287, which forms a single α-helix and a β-strand that is on one edge of the β-sheet through the middle of the first domain. The second (C-terminal) domain runs from residue 134 to residue 315, minus the extension 257–287, and then has an extension from 316 to 360 which adopts a predominantly random-coil structure. Between the two domains we find the binding site, in which clear density (>10σ) was seen for an unknown compound in both the original 1.87 Å resolution SAD structure and the subsequent 1.67 Å resolution native structure. Mass-spectrometric analysis allowed us to determine that the compound had a mass of 151 Da, and fragmentation analysis of this compound led us to believe that the compound was guanine. Guanine was placed into the electron density, where it refined well, and the resulting hydrogen-bond pattern to the binding-site residues satisfied every N and O atom of guanine. As can be seen in Table 1 ▸, the average B factors for the guanine are comparable to the average B factors of the protein, and this is also true of the B factors of the binding-site residues. The binding site consists of the Asn218, Asn220, Glu165 and Ser97 side chains, which make hydrogen bonds, and the Phe98 backbone N atom, which makes a hydrogen bond to the carbonyl O atom of guanine (see Fig. 2 ▸). The guanine ring is packed between (sandwiched by) Tyr45 and Trp194. Every possible heteroatom of the guanine appears to engage with the protein. We note that Asn218 is in an unconventional rotamer [180° rotation to rotamer 3 (t-20°) in Coot] and that the interaction with the guanine amine is out of plane with the atoms of this compound. Flipping the Asn218 side chain to have the O atom of the side chain presented to the guanine would require a hydrogen of the amine to move out of plane at an ∼65° angle to generate a hydrogen bond. As neither of these scenarios seemed to be likely, we decided to model the Asn218 in the unconventional rotamer state, which allows a hydrogen of the side chain to interact with the π-cloud of guanine. Additionally, the Asn218 forms a hydrogen-bond network with Asn169, which is oriented within hydrogen-bonding distance of the Glu165 backbone carbonyl. Asn169 is also positioned to hydrogen-bond to the O atom of the Thr139 side chain.
We have expressed, purified, crystallized and determined the structure of AtzT using SAD phasing from an ethylmercury derivative. We found strong density in the binding site of AtzT that represented a small aromatic compound and have determined by mass spectrometry that this small molecule is very likely to be guanine. Additionally, we have run SPR against a bank of small-molecule compounds of similar size and guanine is the tightest-binding compound tested, binding with a KD of around 110 nM. We have also determined the structure of AtzT bound to 2-hydroxyatrazine and shown that the binding affinity of this substrate for AtzT is 2.2 µM, which is significantly (>600 fold) tighter than atrazine at ∼1.5 mM.