research papers\(\def\hfill{\hskip 5em}\def\hfil{\hskip 3em}\def\eqno#1{\hfil {#1}}\)

Journal logoSTRUCTURAL
BIOLOGY
ISSN: 2059-7983

The evolving story of AtzT, a periplasmic binding protein

CROSSMARK_Color_square_no_text.svg

aBiomedical Manufacturing Program, CSIRO, 343 Royal Parade, Parkville, VIC 3052, Australia, and bCSIRO Synthetic Biology Future Science Platform, GPO Box 1700, Acton, Canberra, ACT 2601, Australia
*Correspondence e-mail: tom.peat@csiro.au

Edited by B. Kobe, University of Queensland, Australia (Received 10 August 2019; accepted 11 October 2019; online 31 October 2019)

Atrazine is an s-triazine-based herbicide that is used in many countries around the world in many millions of tons per year. A small number of organisms, such as Pseudomonas sp. strain ADP, have evolved to use this modified s-triazine as a food source, and the various genes required to metabolize atrazine can be found on a single plasmid. The atomic structures of seven of the eight proteins involved in the breakdown of atrazine by Pseudomonas sp. strain ADP have been determined by X-ray crystallography, but the structures of the proteins required by the cell to import atrazine for use as an energy source are still lacking. The structure of AtzT, a periplasmic binding protein that may be involved in the transport of a derivative of atrazine, 2-hydroxyatrazine, into the cell for mineralization, has now been determined. The structure was determined by SAD phasing using an ethylmercury phosphate derivative that diffracted X-rays to beyond 1.9 Å resolution. `Native' (guanine-bound) and 2-hydroxyatrazine-bound structures were also determined to high resolution (1.67 and 1.65 Å, respectively), showing that 2-hydroxyatrazine binds in a similar way to the purportedly native ligand. Structural similarities led to the belief that it may be possible to evolve AtzT from a purine-binding protein to a protein that can bind and detect atrazine in the environment.

1. Introduction

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[United States Environmental Protection Agency (2013). Atrazine Preliminary Work Plan. Washington DC: United States Environmental Protection Agency. https://www.regulations.gov/document?D=EPA-HQ-OPP-2013-0266-0008.]). 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[Nicolopoulou-Stamati, P., Maipas, S., Kotampasi, C., Stamatis, P. & Hens, L. (2016). Front. Public Health, 4, 148.]; Rickard et al., 2018[Rickard, K., Drew, D., Liccione, J. & Donovan, W. (2018). Triazine Cumulative Human Health Risk. Washington DC: United States Environmental Protection Agency. https://www.regulations.gov/document?D=EPA-HQ-OPP-2013-0266-1160.]). 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.

The best understood of the atrazine catabolism pathways is that from Pseuodomonas sp. strain ADP, in which a series of hydrolases (AtzA, AtzB, AtzC, AtzD, AtzEG, AtzH and AtzF) convert the heterocycle to ammonia and carbon dioxide via the metabolic intermediate cyanuric acid. The genes encoding AtzD, AtzEG, AtzH and AtzF, which are responsible for degrading cyanuric acid, are co-located in an operon on a self-transmissible plasmid (Udiković-Kolić et al., 2012[Udiković-Kolić, N., Scott, C. & Martin-Laurent, F. (2012). Appl. Microbiol. Biotechnol. 96, 1175-1189.]). A second operon, transcribed divergently from the atzDGEHF operon, contains genes that encode an ABC transporter (AtzS, AtzT, AtzU, AtzV and AtzW) and a cyanuric acid-responsive transcriptional regulator (AtzR) that regulates the transcription of both operons. Both operons are transcribed from σ54-dependent promoters and are induced under nitrogen starvation (Platero et al., 2012[Platero, A. I., García-Jaramillo, M., Santero, E. & Govantes, F. (2012). J. Bacteriol. 194, 6560-6573.]). The atzDGEHF operon is further up-regulated by AtzR in the presence of cyanuric acid, while the atzRSTUVW operon is repressed by AtzR when cyanuric acid is present (Platero et al., 2012[Platero, A. I., García-Jaramillo, M., Santero, E. & Govantes, F. (2012). J. Bacteriol. 194, 6560-6573.]).

The enzymes of the atrazine catabolic pathway are now biochemically and structurally well characterized (Peat et al., 2013[Peat, T. S., Balotra, S., Wilding, M., French, N. G., Briggs, L. J., Panjikar, S., Cowieson, N., Newman, J. & Scott, C. (2013). Mol. Microbiol. 88, 1149-1163.], 2015[Peat, T. S., Newman, J., Balotra, S., Lucent, D., Warden, A. C. & Scott, C. (2015). Acta Cryst. D71, 710-720.]; Balotra, Warden et al., 2015[Balotra, S., Warden, A. C., Newman, J., Briggs, L. J., Scott, C. & Peat, T. S. (2015). PLoS One, 10, e0137700.]; Esquirol, Peat, Wilding, Liu et al., 2018[Esquirol, L., Peat, T. S., Wilding, M., Liu, J.-W., French, N. G., Hartley, C. J., Onagi, H., Nebl, T., Easton, C. J., Newman, J. & Scott, C. (2018). J. Biol. Chem. 293, 7880-7891.]; Balotra, Newman et al., 2015[Balotra, S., Newman, J., Cowieson, N. P., French, N. G., Campbell, P. M., Briggs, L. J., Warden, A. C., Easton, C. J., Peat, T. S. & Scott, C. (2015). Appl. Environ. Microbiol. 81, 470-480.]; Esquirol, Peat, Wilding, Hartley et al., 2018[Esquirol, L., Peat, T. S., Wilding, M., Hartley, C. J., Newman, J. & Scott, C. (2018). PLoS One, 13, e0206949.]), while the transport proteins encoded by the atzRSTUVW operon are not; however, there is some genetic evidence that these gene products are involved in the facilitated uptake of atrazine (Liu & Parales, 2009[Liu, X. & Parales, R. E. (2009). Appl. Environ. Microbiol. 75, 5481-5488.]). For some bacteria, such as Arthrobacter aurescens TC1, mass transfer has been demonstrated to be the rate-limiting step in atrazine catabolism (Kundu et al., 2019[Kundu, K., Marozava, S., Ehrl, B., Merl-Pham, J., Griebler, C. & Elsner, M. (2019). ISME J. 13, 2236-2251.]). Gaining an understanding of the mechanisms by which atrazine may be actively imported is therefore of high importance in understanding the dynamics of the process.

Periplasmic binding proteins (a subclass of solute-binding proteins, SBPs; Scheepers et al., 2016[Scheepers, G. H., Lycklama a Nijeholt, J. A. & Poolman, B. (2016). FEBS Lett. 590, 4393-4401.]) bind to a wide variety of small molecules and metals, and serve as a mechanism for sequestering these ligands for use by the organism. These proteins are part of a family of receptors that serve a variety of functions, for example transport, quorum sensing and chemotaxis. Additionally, some of these proteins have been engineered to be selective sensors for the identification of specific compounds in complex mixtures (Dwyer & Hellinga, 2004[Dwyer, M. A. & Hellinga, H. W. (2004). Curr. Opin. Struct. Biol. 14, 495-504.]; Donaldson et al., 2017[Donaldson, T., Iozzino, L., Deacon, L. J., Billones, H., Ausili, A., D'Auria, S. & Dattelbaum, J. D. (2017). Anal. Biochem. 525, 60-66.]; Newman et al., 2019[Newman, J., Caron, K., Nebl, T. & Peat, T. S. (2019). Acta Cryst. D75, 639-646.]). The protein structures are generally conserved and have two α/β domains that bind the ligand of interest between the domains (often referred to as a `Venus fly-trap' mechanism). These proteins can be found in both open and closed forms, where ligand binding shifts the equilibrium to the closed form, i.e. closing the fly trap (Felder et al., 1999[Felder, C. B., Graul, R. C., Lee, A. Y., Merkle, H.-P. & Sadee, W. (1999). AAPS PharmSci, 1, 7-26.]).

Here, we describe the characterization of the periplasmic binding protein AtzT, including high-resolution `closed' structures with bound ligands. We also report the ligand preference of AtzT as determined by surface plasmon reson­ance (SPR).

2. Materials and methods

2.1. Cloning, expression and purification

The atzT gene, assigned as orf97 (NC_004956.1, 1440631) encoding protein Q936X6 in UniProt, was amplified from Pseudomonas sp. strain ADP genomic DNA. The restriction sites SacI and NdeI were inserted in the 5′ primer (5′→3′) catcatGAGCTCCATATGcaggaaccgctgaaagtcgcc and the restriction sites AvrII and BamHI were inserted in the 3′ region (5′→3′) catcatGGATCCTAGGtcacttcgggatcgtgccttcgac. Following amplification, the amplified gene portion was sub­cloned into pETcc2 vector using NdeI and BamHI (Supplementary Fig. S1).

The expression vector pETcc2-atzT was used to transform Escherichia coli BL21 (λDE3) cells (Invitrogen). The bacteria were grown on Luria–Bertani medium containing ampicillin (100 µg ml−1) for the pETcc2 constructs. The cells were grown with shaking at 200 rev min−1 at 301 K. Protein expression was induced at an OD600 of 0.8 by the addition of isopropyl β-D-1-thiogalactopyranoside (1 mM final concentration). A 1 l culture produced around 100 mg of AtzT.

The cells were harvested after 24 h of expression by centrifugation at 5000g for 15 min with an Avanti J-E centrifuge (Beckman Coulter, Indianapolis, USA), resuspended in lysis buffer (25 mM potassium phosphate, 5 mM imidazole pH 7.5) and lysed by passage through a Microfluidics M-110P homogenizer (Massachusetts, USA) five times at 103 MPa. The lysis was followed by centrifugation at 18 000g for 45 min to pellet the cellular debris, and the soluble fraction was used for further purification.

The soluble fraction was syringe-filtered through a 0.22 µm filter. The filtrate was purified using a 5 ml Ni–NTA Superflow cartridge (Qiagen, Maryland, USA) with a gradient from 5 to 500 mM imidazole over 20 column volumes. SDS–PAGE gel analysis was performed to assess the purity of the fractions. AtzT was found to elute at an imidazole concentration of between 130 and 300 mM (Supplementary Fig. S2). The fractions containing AtzT were pooled and concentrated to 12 ml using an Amicon Ultra-15 centrifugal filter unit. Finally, size-exclusion chromatography was performed using a 130 ml column packed with Superdex 75 preparation-grade resin (GE Healthcare Life Sciences) equilibrated with 50 mM HEPES pH 7.5, 200 mM NaCl over 1.5 column volumes (Supplementary Fig. S1). All chromatographic steps were performed using an ÄKTApurifier UPC 10 (GE Healthcare Life Sciences).

2.2. Crystallography and structure determination

Crystallization was performed with the protein sample at 20 mg ml−1 (in 50 mM HEPES pH 7.5, 200 mM NaCl) in SD2 sitting-drop plates (Molecular Dimensions, UK) incubated at 293 K. The experimental droplets consisted of 150 nl protein sample plus 150 nl crystallization condition and were equilibrated against 50 µl crystallization condition. The crystals grew in conditions that consisted of 6–8%(v/v) Jeffamine M600 with 2–7%(v/v) MPD and trisodium citrate at 1.1–1.2 M. The crystals grew as flat plates with varying degrees of thickness, and would appear occasionally after 5–7 days without microseeding or reliably overnight with microseeding (see Supplementary Fig. S3). The crystals were harvested into MiTeGen Mylar loops and cryoprotected by the addition of MPD to give a 20%(v/v) final concentration. Diffraction data were collected at 100 K on the MX2 microfocus beamline at the Australian Synchrotron using a Dectris EIGER 16M detector (Aragão et al., 2018[Aragão, D., Aishima, J., Cherukuvada, H., Clarken, R., Clift, M., Cowieson, N. P., Ericsson, D. J., Gee, C. L., Macedo, S., Mudie, N., Panjikar, S., Price, J. R., Riboldi-Tunnicliffe, A., Rostan, R., Williamson, R. & Caradoc-Davies, T. T. (2018). J. Synchrotron Rad. 25, 885-891.]). 360° of data were taken in 36 s to obtain full data sets. For the ethylmercury phosphate derivative, the wavelength was moved to 12 300 eV (1.0082 Å) and four full data sets were obtained from a single crystal by moving down the length of the crystal. Three of these data sets were shown to be of good quality and high resolution and were merged to obtain a highly redundant data set for SAD phasing (see Table 1[link] for X-ray data-set statistics). Data were indexed using XDS (Kabsch, 2010[Kabsch, W. (2010). Acta Cryst. D66, 125-132.]) and scaled using AIMLESS (Evans & Murshudov, 2013[Evans, P. R. & Murshudov, G. N. (2013). Acta Cryst. D69, 1204-1214.]), and the structure was phased using CRANK2 (Skubák & Pannu, 2013[Skubák, P. & Pannu, N. S. (2013). Nature Commun. 4, 2777.]), which gave an initial model with four protomers in the asymmetric unit and four Hg atoms. Owing to a mix-up in the clones (two different proteins of approximately equal size were being investigated at the same time), an incorrect sequence was originally input into CRANK2. A sequence of eight amino acids in clear electron density was used to search for the correct sequence in UniProt. The most likely sequence containing these eight residues was found to be UniProt ID Q936X6. This sequence was defined as being from Pseudomonas sp. strain ADP and was identified as AtzT, which was one of the two proteins under investigation. A model with the correct sequence was built manually using Coot (Emsley et al., 2010[Emsley, P., Lohkamp, B., Scott, W. G. & Cowtan, K. (2010). Acta Cryst. D66, 486-501.]) and refined using REFMAC (Murshudov et al., 2011[Murshudov, G. N., Skubák, P., Lebedev, A. A., Pannu, N. S., Steiner, R. A., Nicholls, R. A., Winn, M. D., Long, F. & Vagin, A. A. (2011). Acta Cryst. D67, 355-367.]). Standard (default) REFMAC parameters were used for refinement with jelly-body refinement, isotropic B factors and automatic NCS restraints applied. Subsequent structures were phased using Phaser (McCoy et al., 2007[McCoy, A. J., Grosse-Kunstleve, R. W., Adams, P. D., Winn, M. D., Storoni, L. C. & Read, R. J. (2007). J. Appl. Cryst. 40, 658-674.]) and the initial SAD-based model, manually rebuilt using Coot and refined using REFMAC. phenix.elbow (Moriarty et al., 2009[Moriarty, N. W., Grosse-Kunstleve, R. W. & Adams, P. D. (2009). Acta Cryst. D65, 1074-1080.]) was used to obtain the dictionary file for the 2-hydroxyatrazine moiety; the guanine and atrazine compounds already had appropriate dictionary files available.

Table 1
X-ray data statistics

Values in parentheses are for the highest resolution bin.

Ligand Ethylmercury phosphate Guanine 2-Hydroxyatrazine
PDB code 6pii 6pi5 6pi6
Data collection
 Space group P21 P21 P21
a, b, c (Å) 98.9, 63.1, 120.4 99.0, 63.1, 120.4 98.8, 62.8, 120.2
α, β, γ (°) 90, 93.5, 90 90, 93.6, 90 90, 93.4, 90
 Wavelength (Å) 1.0082 0.9537 0.9537
 Resolution (Å) 49.4–1.87 (1.90–1.87) 49.3–1.67 (1.70–1.67) 49.0–1.65 (1.68–1.65)
 Completeness (%) 97.8 (80.1) 99.3 (93.2) 100.0 (100.0)
Rmerge (%) 0.140 (0.669) 0.083 (0.643) 0.086 (1.265)
Rp.i.m. (%) 0.032 (0.246) 0.034 (0.285) 0.035 (0.515)
 Mean I/σ(I) 13.5 (3.1) 11.9 (2.7) 10.8 (1.5)
 No. of unique reflections 119947 (4833) 171863 (7958) 176707 (8668)
 Multiplicity 19.6 (7.4) 6.8 (5.8) 6.8 (7.0)
 CC1/2 0.998 (0.809) 0.998 (0.769) 0.998 (0.653)
 Anomalous completeness (%) 97.1 (70.5)    
 Anomalous multiplicity 9.8 (4.1)    
 No. of Hg atoms 4    
Refinement
 Resolution (Å) 49.4–1.87 (1.92–1.87) 49.3–1.67 (1.71–1.67) 49.0–1.65 (1.69–1.65)
 No. of reflections 114140 (7356) 163254 (11606) 167790 (12348)
Rwork (%) 15.5 (22.8) 15.6 (22.9) 16.2 (28.1)
Rfree (%) 18.3 (27.4) 18.4 (26.1) 18.2 (30.7)
 Total No. of atoms 10867 11222 11186
 No. of ligand atoms 44 44 112
 No. of waters 648 917 798
 Wilson B value (Å2) 32.7 24.4 27.5
 Mean B value (Å2)
  Protein 34.7 25.5 28.8
  Waters 39.3 33.6 36.9
  Ligand 37.1 23.6 24.4
 R.m.s.d., bond lengths (Å) 0.011 0.012 0.012
 R.m.s.d., bond angles (°) 1.544 1.650 1.602
 Ramachandran analysis (%)
  Preferred 97.7 98.1 97.8
  Allowed 2.3 1.9 2.2
  Outliers 0 0 0

In order to remove the bound guanine and allow atrazine (2-hydroxyatrazine) to bind, the HEPES-based buffer in which the protein was formulated (50 mM HEPES pH 7.5 with 200 mM NaCl) with 5 mM atrazine added (from a 100 mM atrazine stock in DMSO) was added to the protein and allowed to rest for 1 h at room temperature before concentration in a spin concentrator. This process was repeated five times before the protein was concentrated to 16.9 mg ml−1 and drops were set up with microseeding in the same crystallization conditions as used for the native protein (the drops consisted of 150 nl protein, 130 nl crystallant and 20 nl seed stock). Crystals were obtained and handled as described above.

2.3. Surface plasmon resonance (SPR)

All SPR experiments were performed using a Biacore S200 (GE Healthcare) biosensor with a CM5 chip. AtzT was coupled to the chip surface at 298 K in HBS-P+ buffer [10 mM HEPES pH 7.4, 150 mM NaCl, 0.05%(v/v) Tween 20] after activation with a 1:1 mixture of NHS/EDC [N-hydroxy­succinimide/N-ethyl-N′-(3-diethylaminopropyl)carbo­diimide]. The protein was diluted to 15 µg ml−1 at pH 7 and injected to immobilization levels of ∼7000 RU (∼5 min, 10 µl min−1). The surface was then blocked with 1 M ethan­olamine pH 8.0. Prior to immobilization the AtzT protein is bound to guanine (as demonstrated by crystallography and mass spectrometry); however, the SPR data reveal that the majority of the guanine dissociates in as little as 20 s (Supplementary Fig. S4). The relatively long time between protein immobilization and the first analyte injection (≥1 h) thus provides a surface that is free from guanine. All SPR binding experiments were performed at 293 K in the above immobilization buffer but with the addition of 2%(v/v) DMSO. Analytes were serially diluted (twofold) in SPR binding buffer, injected for 30 s contact time at 60 µl min−1 and then allowed to dissociate back to baseline. Each analyte titration was performed in triplicate. Binding sensorgrams were processed, solvent-corrected and double-referenced using Scrubber (BioLogic Software, Australia). To determine the binding affinities, responses at equilibrium for each analyte were fitted to a 1:1 steady-state affinity model. Where the top concentration (typically 200 µM) for a compound was less than twofold above its KD, the responses were fitted using a global Rmax determined with guanine (with sensorgrams scaled to the molecular weight). Adjusting the response by molecular weight is a potential source of error, although this is generally no more than twofold (Davis & Wilson, 2000[Davis, T. M. & Wilson, W. D. (2000). Anal. Biochem. 284, 348-353.]). The compounds that were fitted to a local Rmax (ammeline, guanine and adenine) all displayed equivalent Rmax values (when responses were scaled to molecular weight); however, the potential that the other globally fitted compounds possess deviant refractive-index increments is a relevant caveat when interpreting compound affinities. The compounds melamine and ammelide appeared to contain trace amounts of the tightly binding ammeline (based on the heterogeneous curve shape and the observed off-rates); although a large discrepancy in affinity between these compounds was reliably demonstrated, an exact KD value was not determined. Instead, a range is given for melamine and ammelide (KD > 400 µM and KD > 100 µM, respectively) to avoid overinterpretation. Sensorgrams and binding isotherms are displayed in Supplementary Fig. S4.

2.4. Mass spectrometry

The accurate mass of recombinant proteins was confirmed by LC-ESI-MS. Samples were spiked with formic acid to a final concentration of 0.1%(v/v) and separated by reverse-phase liquid chromatography using a linear gradient from 0 to 80% mobile phase B [Dionex UltiMate 3000 RSLCnano System, Thermo Scientific; mobile phase A, 0.1%(v/v) formic acid; mobile phase B, 90%(v/v) acetonitrile/0.1%(v/v) formic acid] and ionized using an Apollo II electrospray source (Bruker) with a nebuliser pressure of 1 bar and a drying gas flow of 8 l min−1 at 473 K. A maXis II mass spectrometer (Bruker) was calibrated in positive-ion mode using ESI-TOF Low Concentration Tuning Mix (Agilent) and high-resolution MS raw data were processed with Bruker Compass Data­Analysis v.4.3.

To identify the unknown compound which was seen in AtzT by X-ray crystallography, we tested the original protein sample for the presence of an ∼150 Da small molecule (estimate based on the electron density at 1.67 Å resolution). Using ESI-MS in direct infusion mode, we discovered a monoisotopic mass of 151.0772 Da that was exclusively present in the spectra of the AtzT sample but not in spectra obtained from the protein-free crystallization solution. We performed a molecular-weight search of the E. coli Metabolome Database (https://ecmdb.ca/). This identified guanine (monoisotopic mass = 151.0997) as the closest match. In order to validate this possible hit we also fragmented the corresponding MH+ precursor ion at collision energies of 10–40 eV and searched the resulting MSMS spectrum against the METLIN metabolite database (https://metlin.scripps.edu). This yielded a unique, confident MSMS spectrum match for guanine (METLIN score 15, Δp.p.m. = 15, MSMS tolerance = 0.01 Da, collision energy 40 eV).

3. Results

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[Palmer, T. & Berks, B. C. (2012). Nature Rev. Microbiol. 10, 483-496.]), the TAT signal sequence was identified using TatP (https://www.cbs.dtu.dk/services/TatP/) and Protter (https://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[link]). 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[link], 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[link]). 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.

[Figure 1]
Figure 1
The image on the left is a cartoon representation coloured by Jones' rainbow with the N-terminus in dark blue and the C-terminus in dark red. The green mesh represents the unknown difference density found after solving the protein structure sitting between the two domains. The image on the right is a cartoon representation that is coloured by secondary structure with α-helices in cyan and β-sheets in magenta. It is orientated with an approximate 90° rotation to the figure on the left.
[Figure 2]
Figure 2
An enlarged image of the binding site between the two domains. The image on the left shows the difference density (mFoDFc, 3σ) before guanine was placed into the model. The relevant amino acids in the binding site are shown and the backbone is coloured by Jones' rainbow as in Fig. 1[link](a). The middle and right figures are rotated plus and minus approximately 90° from the figure on the left and are coloured by atom type, with C in cyan, O in red and N in dark blue. Potential hydrogen bonds are represented as black lines; approximate distances between potential hydrogen-bond acceptors and donors are shown in Å and the binding residues are labelled. Note that the Asn218 is out of plane to the guanine, so the hydrogen on the Asn218 N atom is likely to interact with the π-cloud perpendicular to the plane of the atoms.

The structure of AtzT allows us to classify this protein into cluster B of the solute-binding proteins (Berntsson et al., 2010[Berntsson, R. P.-A., Smits, S. H. J., Schmitt, L., Slotboom, D.-J. & Poolman, B. (2010). FEBS Lett. 584, 2606-2617.]). Cluster B has three connecting strands between the two domains and there are no α-helices interrupting these connections, placing AtzT in the B-I cluster. According to PDBeFold, AtzT is most similar to the entry with PDB code 3s99 (Seattle Structural Genomics Center for Infectious Disease, unpublished work), which is also classified as a B-I cluster protein and had electron density in the binding site for a purine. Post hoc analysis of the structure and sequence showed that PDB entry 3s99 has 54% sequence identity and an r.m.s.d. of 1.2 Å (over ∼330 residues) to AtzT. Using Phaser with PDB entry 3s99 showed that the AtzT structure could have been solved by molecular replacement using this model. PDB entry 3s99 is annotated as a `basic membrane lipoprotein from Brucella melitensis', although the deposited model has an adenine moiety modelled into the binding site, suggesting that it is another purine-binding SBP.

As AtzT was thought to be a protein that bound atrazine, we used surface plasmon resonance (SPR) to test the binding activity of the protein to several different nucleobases and to atrazine and atrazine metabolites and analogues (see Table 2[link]). The shape of the SPR curve for atrazine suggested a heterogeneous analyte; liquid-chromatography and mass-spectrometric analysis of the atrazine starting material showed that 5% of the starting material was 2-hydroxyatrazine (data not shown). Fresh atrazine and 2-hydroxyatrazine were reordered and tested, and clean sensorgrams were obtained for each (see Supplementary Fig. S4), with KD values of ∼1.5 mM and 2.2 µM, respectively. 2-Hydroxyatrazine has a similar affinity as adenine (KD ≃ 2.5 µM) but binds more weakly than guanine (KD ≃ 110 nM). The other nucleobases display limited binding (KD ≥ 620 µM). Guanine is the tightest-binding compound tested, with a KD of ∼110 nM. It is interesting to note that ammeline and ammelide bind very differently to AtzT, ammeline with a KD of 360 nM and ammelide with a KD of >100 µM (>250-fold difference), despite these compounds differing by a single nitrogen-to-oxygen substitution on the same triazine ring. Similarly, substituting the O atom of ammeline with an N atom (i.e. melamine; KD > 400 µM) also reduces binding.

Table 2
SPR kinetic values

Compound Chemical structure KDM)
Atrazine
[Scheme 1]
1500 ± 200
2-Hydroxyatrazine
[Scheme 2]
2.2 ± 0.2
Guanine
[Scheme 3]
0.11 ± 0.02
Ammeline
[Scheme 4]
0.36 ± 0.02
Adenine
[Scheme 5]
2.5 ± 0.3
Cyanuric acid
[Scheme 6]
340 ± 30
Thymine
[Scheme 7]
620 ± 40
Cytosine
[Scheme 8]
800 ± 90
Uracil
[Scheme 9]
1600 ± 300
Ammelide
[Scheme 10]
>100
Melamine
[Scheme 11]
>400
4-Iodophenylboronic acid
[Scheme 12]
5000 ± 1000
Barbaturic acid
[Scheme 13]
>8000
Barbitone
[Scheme 14]
>8000
Tetrabromoterephthalic acid
[Scheme 15]
>8000
GTP
[Scheme 16]
>8000
†Mean ± standard deviation (n = 3).
‡The sample contains trace contamination with ammeline.

The structure with 2-hydroxyatrazine diffracted to a nominal resolution of 1.65 Å; it belongs to the same space group (P21) and has the same fold and closed form as the other structures that we have determined. This structure was obtained prior to the SPR data and used the aforementioned 2-hydroxyatrazine-contaminated sample of atrazine for co-crystallization. Modelling atrazine into the structure resulted in distinct negative difference density at the Cl atom, which was alleviated completely upon replacement with 2-hydroxy­atrazine. The occupancy of the binding site is thus suggested to be effectively entirely 2-hydroxyatrazine, which agrees with the affinity data presented. The crystallography results thus verify the SPR results: a contaminant found in low abundance (about 5% of the total compound sample, but ∼700-fold more tightly binding) is the entity that is found bound in the structure. The same residues contribute to the binding of 2-hydroxyatrazine as to the binding of guanine in the binding site: Asn220, Glu165 and Ser97, with the ring packed on either side by Tyr45 and Trp194. 2-Hydroxyatrazine does not fit as well into the binding site as guanine and has some steric clashes on the side with the isopropyl moiety (see Fig. 3[link]). It has well defined density except around the isopropyl moiety, which shows close contacts with Asn218 (2.4–3.2 Å depending on the atom and the chain) and with Asn220 (3.0–3.2 Å to the N atom of the side chain). The Asn218 residue has weaker density and can be found in multiple conformations. The ethyl moiety on the other side fits the density much better and generally does not make extensive clashes with the protein residues (3.1–3.2 Å from the O atom of the Asn220 side chain). The N atoms of 2-hydroxyatrazine are well within hydrogen-bonding distances of Asn220, Glu165 and Ser97 and the hydroxyl moiety makes a bond (2.9 Å on average) to the backbone N atom of Phe98, analogous to the carbonyl of guanine.

[Figure 3]
Figure 3
An enlarged image of 2-hydroxyatrazine in the binding site. The difference map (mFoDFc) is set to 3σ and the compound has good density, with the exception of the isopropyl group, which can be found in more than one orientation but has clashes with the protein (for example Asn218) in all possible orientations. The same residues that make hydrogen bonds to guanine make very similar interactions with the 2-hydroxyatrazine heteroatoms and the compound is sandwiched between Tyr45 and Trp194 in a similar fashion to guanine. C atoms are in cyan, O atoms are in red and N atoms are in dark blue.

4. Discussion

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.

The Pseudomonas sp. strain ADP genome contains 112 genes annotated as `substrate-binding proteins' (Supplementary Table S1), none of which have similarity to AtzT. It also contains genes that code for enzymes for de novo purine synthesis (KSW22506.1, KSW23190.1, KSW23286.1, KSW23824.1, KSW23941.1, KSW24282.1, KSW25158.1, KSW25646.1 and KSW25647.1). Additionally, a purine permease (KSW22651.1) was identified on the genome, suggesting that Pseudomonas sp. strain ADP can also take up purine via at least one other pathway. The atzT gene is found with the other atz genes on the 97 kb plasmid pADP-1 in Pseudomonas sp. strain ADP. To date, no obvious gene involved in either chemotaxis or transport had been identified on the pADP-1 plasmid. Our finding suggests that atzT codes for a periplasmic transport protein. It has a TAT signal at the N-terminus and has sequence and structural homology to many other periplasmic binding proteins. Sequence comparison of AtzT with other proteins found in UniProt gives a high number of hits for proteins that are classified as `nucleoside-binding protein' or `purine-binding protein' from such organisms as Massilia (84% identity; A0A1G6IR71), Rugamonas rubra (76% identity; A0A1I4KBL4_9GAMM) and Duganella (71–76% identity; A0A1E7WQA5_9BURK). Atrazine is a modified s-triazine compound that has some similarity to purines, so it seems reasonable to hypothesize that AtzT is a purine-binding protein that can promiscuously bind atrazine and atrazine derivatives. After identifying the unknown compound binding to AtzT as guanine (using a combination of X-ray crystallography, mass-spectrometric analysis and SPR) and searching sequence databases for the sequences of homologous proteins, we believe that AtzT is a purine-binding protein that can accept 2-hydroxyatrazine (and possibly atrazine) from the environment when it is present and guanine is absent.

We speculate that the promiscuous uptake of 2-hydroxy­atrazine (which is a likely contaminant in most atrazine samples) could prove to be an evolutionary benefit for those organisms living in some of the more intensively farmed lands where atrazine has been used in abundance as a herbicide (Europe prior to 2004, and the US and Australia currently, as well as Asia). Organisms such as Pseudomonas sp. strain ADP have been shown to survive on atrazine as a nitrogen/carbon source and have evolved the necessary pathways to do so effectively (Govantes et al., 2009[Govantes, F., Porrúa, O., García-González, V. & Santero, E. (2009). Microb. Biotechnol. 2, 178-185.]). Several other organisms have been shown to possess similar pathways for the use of atrazine (and other triazine-based products) as nutrients/energy sources (Mongodin et al., 2006[Mongodin, E. F., Shapir, N., Daugherty, S. C., DeBoy, R. T., Emerson, J. B., Shvartzbeyn, A., Radune, D., Vamathevan, J., Riggs, F., Grinberg, V., Khouri, H., Wackett, L. P., Nelson, K. E. & Sadowsky, M. J. (2006). PLoS Genet. 2, e214.]; Young et al., 2006[Young, J. P. W., Crossman, L. C., Johnston, A. W., Thomson, N. R., Ghazoui, Z. F., Hull, K. H., Wexler, M., Curson, A. R., Todd, J. D., Poole, P. S., Mauchline, T. H., East, A. K., Quail, M. A., Churcher, C., Arrowsmith, C., Cherevach, I., Chillingworth, T., Clarke, K., Cronin, A., Davis, P., Fraser, A., Hance, Z., Hauser, H., Jagels, K., Moule, S., Mungall, K., Norbertczak, H., Rabbinowitsch, E., Sanders, M., Simmonds, M., Whitehead, S. & Parkhill, J. (2006). Genome Biol. 7, R34.]; Kapley et al., 2013[Kapley, A., Sagarkar, S., Tanksale, H., Sharma, N., Qureshi, A., Khardenavis, A. & Purohit, H. J. (2013). Genome Announc. 1, e00235-12.]; Nakamoto et al., 2016[Nakamoto, M., Nonaka, T., Shea, K. J., Miura, Y. & Hoshino, Y. (2016). J. Am. Chem. Soc. 138, 4282-4285.]; Devers-Lamrani et al., 2016[Devers-Lamrani, M., Spor, A., Mounier, A. & Martin-Laurent, F. (2016). Genome Announc. 4, e01733-15.]).

Periplasmic binding proteins have been engineered to serve as biosensors, exploiting the large conformation change induced by ligand binding to affect changes in fluorescence resonance energy transfer (FRET) and bio­luminescence resonance energy transfer (BRET) signals from fused fluorescent and luminescent tags (Ye & Schultz, 2003[Ye, K. & Schultz, J. S. (2003). Anal. Chem. 75, 3451-3459.]; Dacres et al., 2013[Dacres, H., Michie, M., Anderson, A. & Trowell, S. C. (2013). Biosens. Bioelectron. 41, 459-464.]). The high-resolution structures presented here provide a depth of information that will allow the re-engineering of AtzT for greater specificity for contaminants such as atrazine and melamine, and potentially for medically relevant compounds such as barbituric acid or related barbiturates.

Acknowledgements

The protein was crystallized in CSIRO's Collaborative Crystallization Facility. We thank the Australian Synchrotron for beam time and the beamline scientists for their help. This research was undertaken in part using the MX2 beamline at the Australian Synchrotron and made use of the ACRF detector. We also thank Ben Morrow for running the original atrazine sample on LC and MS to determine that there was 5% 2-hydroxyatrazine in the original sample.

References

First citationAragão, D., Aishima, J., Cherukuvada, H., Clarken, R., Clift, M., Cowieson, N. P., Ericsson, D. J., Gee, C. L., Macedo, S., Mudie, N., Panjikar, S., Price, J. R., Riboldi-Tunnicliffe, A., Rostan, R., Williamson, R. & Caradoc-Davies, T. T. (2018). J. Synchrotron Rad. 25, 885–891.  Web of Science CrossRef IUCr Journals Google Scholar
First citationBalotra, S., Newman, J., Cowieson, N. P., French, N. G., Campbell, P. M., Briggs, L. J., Warden, A. C., Easton, C. J., Peat, T. S. & Scott, C. (2015). Appl. Environ. Microbiol. 81, 470–480.  CrossRef PubMed Google Scholar
First citationBalotra, S., Warden, A. C., Newman, J., Briggs, L. J., Scott, C. & Peat, T. S. (2015). PLoS One, 10, e0137700.  CrossRef PubMed Google Scholar
First citationBerntsson, R. P.-A., Smits, S. H. J., Schmitt, L., Slotboom, D.-J. & Poolman, B. (2010). FEBS Lett. 584, 2606–2617.  Web of Science CrossRef CAS PubMed Google Scholar
First citationDacres, H., Michie, M., Anderson, A. & Trowell, S. C. (2013). Biosens. Bioelectron. 41, 459–464.  CrossRef CAS PubMed Google Scholar
First citationDavis, T. M. & Wilson, W. D. (2000). Anal. Biochem. 284, 348–353.  CrossRef PubMed CAS Google Scholar
First citationDevers-Lamrani, M., Spor, A., Mounier, A. & Martin-Laurent, F. (2016). Genome Announc. 4, e01733-15.  PubMed Google Scholar
First citationDonaldson, T., Iozzino, L., Deacon, L. J., Billones, H., Ausili, A., D'Auria, S. & Dattelbaum, J. D. (2017). Anal. Biochem. 525, 60–66.  CrossRef CAS PubMed Google Scholar
First citationDwyer, M. A. & Hellinga, H. W. (2004). Curr. Opin. Struct. Biol. 14, 495–504.  Web of Science CrossRef PubMed CAS Google Scholar
First citationEmsley, P., Lohkamp, B., Scott, W. G. & Cowtan, K. (2010). Acta Cryst. D66, 486–501.  Web of Science CrossRef CAS IUCr Journals Google Scholar
First citationEsquirol, L., Peat, T. S., Wilding, M., Hartley, C. J., Newman, J. & Scott, C. (2018). PLoS One, 13, e0206949.  CrossRef PubMed Google Scholar
First citationEsquirol, L., Peat, T. S., Wilding, M., Liu, J.-W., French, N. G., Hartley, C. J., Onagi, H., Nebl, T., Easton, C. J., Newman, J. & Scott, C. (2018). J. Biol. Chem. 293, 7880–7891.  Web of Science CrossRef CAS PubMed Google Scholar
First citationEvans, P. R. & Murshudov, G. N. (2013). Acta Cryst. D69, 1204–1214.  Web of Science CrossRef CAS IUCr Journals Google Scholar
First citationFelder, C. B., Graul, R. C., Lee, A. Y., Merkle, H.-P. & Sadee, W. (1999). AAPS PharmSci, 1, 7–26.  CrossRef Google Scholar
First citationGovantes, F., Porrúa, O., García-González, V. & Santero, E. (2009). Microb. Biotechnol. 2, 178–185.  CrossRef CAS PubMed Google Scholar
First citationKabsch, W. (2010). Acta Cryst. D66, 125–132.  Web of Science CrossRef CAS IUCr Journals Google Scholar
First citationKapley, A., Sagarkar, S., Tanksale, H., Sharma, N., Qureshi, A., Khardenavis, A. & Purohit, H. J. (2013). Genome Announc. 1, e00235-12.  CrossRef PubMed Google Scholar
First citationKundu, K., Marozava, S., Ehrl, B., Merl-Pham, J., Griebler, C. & Elsner, M. (2019). ISME J. 13, 2236–2251.  CrossRef CAS PubMed Google Scholar
First citationLiu, X. & Parales, R. E. (2009). Appl. Environ. Microbiol. 75, 5481–5488.  CrossRef PubMed CAS Google Scholar
First citationMcCoy, A. J., Grosse-Kunstleve, R. W., Adams, P. D., Winn, M. D., Storoni, L. C. & Read, R. J. (2007). J. Appl. Cryst. 40, 658–674.  Web of Science CrossRef CAS IUCr Journals Google Scholar
First citationMongodin, E. F., Shapir, N., Daugherty, S. C., DeBoy, R. T., Emerson, J. B., Shvartzbeyn, A., Radune, D., Vamathevan, J., Riggs, F., Grinberg, V., Khouri, H., Wackett, L. P., Nelson, K. E. & Sadowsky, M. J. (2006). PLoS Genet. 2, e214.  CrossRef PubMed Google Scholar
First citationMoriarty, N. W., Grosse-Kunstleve, R. W. & Adams, P. D. (2009). Acta Cryst. D65, 1074–1080.  Web of Science CrossRef CAS IUCr Journals Google Scholar
First citationMurshudov, G. N., Skubák, P., Lebedev, A. A., Pannu, N. S., Steiner, R. A., Nicholls, R. A., Winn, M. D., Long, F. & Vagin, A. A. (2011). Acta Cryst. D67, 355–367.  Web of Science CrossRef CAS IUCr Journals Google Scholar
First citationNakamoto, M., Nonaka, T., Shea, K. J., Miura, Y. & Hoshino, Y. (2016). J. Am. Chem. Soc. 138, 4282–4285.  CrossRef CAS PubMed Google Scholar
First citationNewman, J., Caron, K., Nebl, T. & Peat, T. S. (2019). Acta Cryst. D75, 639–646.  CrossRef IUCr Journals Google Scholar
First citationNicolopoulou-Stamati, P., Maipas, S., Kotampasi, C., Stamatis, P. & Hens, L. (2016). Front. Public Health, 4, 148.  PubMed Google Scholar
First citationPalmer, T. & Berks, B. C. (2012). Nature Rev. Microbiol. 10, 483–496.  CrossRef CAS Google Scholar
First citationPeat, T. S., Balotra, S., Wilding, M., French, N. G., Briggs, L. J., Panjikar, S., Cowieson, N., Newman, J. & Scott, C. (2013). Mol. Microbiol. 88, 1149–1163.  Web of Science CrossRef CAS PubMed Google Scholar
First citationPeat, T. S., Newman, J., Balotra, S., Lucent, D., Warden, A. C. & Scott, C. (2015). Acta Cryst. D71, 710–720.  CrossRef IUCr Journals Google Scholar
First citationPlatero, A. I., García-Jaramillo, M., Santero, E. & Govantes, F. (2012). J. Bacteriol. 194, 6560–6573.  CrossRef CAS PubMed Google Scholar
First citationRickard, K., Drew, D., Liccione, J. & Donovan, W. (2018). Triazine Cumulative Human Health Risk. Washington DC: United States Environmental Protection Agency. https://www.regulations.gov/document?D=EPA-HQ-OPP-2013-0266-1160Google Scholar
First citationScheepers, G. H., Lycklama a Nijeholt, J. A. & Poolman, B. (2016). FEBS Lett. 590, 4393–4401.  Google Scholar
First citationSkubák, P. & Pannu, N. S. (2013). Nature Commun. 4, 2777.  Google Scholar
First citationUdiković-Kolić, N., Scott, C. & Martin-Laurent, F. (2012). Appl. Microbiol. Biotechnol. 96, 1175–1189.  Web of Science PubMed Google Scholar
First citationUnited States Environmental Protection Agency (2013). Atrazine Preliminary Work Plan. Washington DC: United States Environmental Protection Agency. https://www.regulations.gov/document?D=EPA-HQ-OPP-2013-0266-0008Google Scholar
First citationYe, K. & Schultz, J. S. (2003). Anal. Chem. 75, 3451–3459.  CrossRef PubMed CAS Google Scholar
First citationYoung, J. P. W., Crossman, L. C., Johnston, A. W., Thomson, N. R., Ghazoui, Z. F., Hull, K. H., Wexler, M., Curson, A. R., Todd, J. D., Poole, P. S., Mauchline, T. H., East, A. K., Quail, M. A., Churcher, C., Arrowsmith, C., Cherevach, I., Chillingworth, T., Clarke, K., Cronin, A., Davis, P., Fraser, A., Hance, Z., Hauser, H., Jagels, K., Moule, S., Mungall, K., Norbertczak, H., Rabbinowitsch, E., Sanders, M., Simmonds, M., Whitehead, S. & Parkhill, J. (2006). Genome Biol. 7, R34.  CrossRef PubMed Google Scholar

This is an open-access article distributed under the terms of the Creative Commons Attribution (CC-BY) Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original authors and source are cited.

Journal logoSTRUCTURAL
BIOLOGY
ISSN: 2059-7983
Follow Acta Cryst. D
Sign up for e-alerts
Follow Acta Cryst. on Twitter
Follow us on facebook
Sign up for RSS feeds