Simplified heavy-atom derivatization of protein structures via co-crystallization with the MAD tetragon tetrabromoterephthalic acid

Tetrabromoterephthalic acid (B4C) provides sufficient anomalous signal to solve the crystal structure of hen egg-white lysozyme using MAD phasing. The use of B4C as an experimental phasing compound offers several advantages over existing compounds, including its compatibility with common crystallization reagents and its amenability to co-crystallization.

Abstract

The phase problem is a persistent bottleneck that impedes the structure-determination pipeline and must be solved to obtain atomic resolution crystal structures of macromolecules. Although molecular replacement has become the predominant method of solving the phase problem, many scenarios still exist in which experimental phasing is needed. Here, a proof-of-concept study is presented that shows the efficacy of using tetrabromoterephthalic acid (B4C) as an experimental phasing compound. Incorporating B4C into the crystal lattice using co-crystallization, the crystal structure of hen egg-white lysozyme was solved using MAD phasing. The strong anomalous signal generated by its four Br atoms coupled with its compatibility with commonly used crystallization reagents render B4C an effective experimental phasing compound that can be used to overcome the phase problem.

1. Introduction  

X-ray crystallography is the gold-standard technique used to determine the three-dimensional structures of proteins and other macromolecules. At the time of writing, 88% of the total number of deposited structures in the Protein Data Bank (PDB) have been solved using X-ray crystallography (Burley et al., 2019 ▸). Although the use of this technique is ubiquitous in the structural biology community, there are several challenges associated with its implementation that must be addressed in order to elucidate a protein structure from a diffraction pattern. One is that the conversion of an X-ray diffraction pattern requires the solution of the phase problem. Molecular replacement has quickly become the preferred method to overcome the phase problem due to its accessibility, speed and affordability. This technique approximates the phases of the target structure from structurally similar proteins that have previously been solved (Evans & McCoy, 2008 ▸; Rossmann, 1990 ▸). Despite the many benefits of using molecular replacement and the growing number of deposited structures in the PDB that are available for use as models, there are still limitations to this technique. Many proteins lack a suitable search model and so experimental phasing must be used to solve the phase problem.

To acquire initial phases experimentally, heavy atoms are incorporated into the crystal lattice at defined locations and phases are traditionally obtained using one of two methods: isomorphous replacement or anomalous dispersion (Cowtan, 2003 ▸). If isomorphous crystals (crystals with very similar unit-cell dimensions with protein molecules arranged similarly within the unit cell) can be obtained with and without the heavy atom, phases can be obtained using the isomorphous replacement method. In contrast, the anomalous dispersion method only requires a protein crystal to have a heavy atom incorporated. In this case, data collection includes measuring diffraction at an X-ray wavelength close to the absorption edge of the heavy atom. At these wavelengths, Friedel’s law breaks down, resulting in anomalous scattering. The theories for phase determination using isomorphous replacement and anomalous scattering have been described previously (Taylor, 2010 ▸).

Methods of incorporating heavy atoms into crystals include labelling the protein (incorporating selenomethionine or selenocysteine amino acids; Strub et al., 2003 ▸), chemical modification (such as 5-bromouracil labelling of RNA; Baugh et al., 2000 ▸; Kieft et al., 2002 ▸), and soaking and co-crystallization of the crystal with a heavy-atom compound. The last method, although a general method to derivatize crystals, often suffers from nonspecific binding, resulting in multiple binding sites with low occupancy in the unit cell and ultimately, low anomalous signal.

To address this problem, Tobias Beck, in his PhD dissertation Sticky Triangles: New Tools for Experimental Phasing of Biological Molecules, proposed a set of five new phasing compounds that could be used to derivatize protein crystals and incorporate heavy atoms into the lattice for phasing (Beck, 2010 ▸). These compounds have anomalous scattering halogen atoms, either bromine or iodine, attached to a benzene scaffold. In addition, the compounds were designed to have functional groups such as carboxyl, amine, methoxyl and hydroxyl groups surrounded by the heavy atoms on the benzene scaffold. These functional groups and the aromatic ring scaffold allow more specific interactions with the protein to improve occupancy and yield higher anomalous signals.

Two of these compounds, the magic triangle and the MAD triangle (Fig. 1 ▸), were later published in a series of journal articles (Beck et al., 2008 ▸, 2009 ▸, 2010 ▸). The MAD triangle, 5-amino-2,4,6-tribromobenzene-1,3-dicarboxylic acid (B3C), has only been used to phase two structures in the PDB, whereas the magic triangle, 5-amino-2,4,6-triiodobenzene-1,3-dicarboxylic acid (I3C), has gained significant traction and has been incorporated into 21 structures in the PDB. Two suppliers of protein crystallography reagents, Hampton Research and Molecular Dimensions, also sell kits for I3C phasing, demonstrating the accessibility and feasibility of using I3C as a compound for experimental phasing.

Figure 1.

Chemical structure of the phasing compounds I3C (5-amino-2,4,6-triiodobenzene-1,3-dicarboxylic acid), B3C (5-amino-2,4,6-tribromobenzene-1,3-dicarboxylic acid) and B4C (tetrabromoterephthalic acid).

There could be advantages to exploring phasing compounds outside of I3C and B3C. For larger proteins and protein complexes, the limited number of I3C or B3C sites on the protein surface may provide insufficient phasing power to successfully solve the phase problem. This limit could be overcome by employing phasing molecules with increased phasing power per molecule, such as tetrabromoterephthalic acid (B4C; Fig. 1 ▸). This compound, dubbed ‘the MAD tetragon’, has four anomalous scattering Br atoms and should provide increased anomalous signal over I3C and B3C. B4C has successfully been incorporated into crystals via soaking and has been used to phase the structures of thaumatin and thermolysin (Beck, 2010 ▸). This work has yet to be published in a peer-reviewed journal and we would like to explicitly credit Beck and the Sheldrick group as the first users of B4C as a phasing tool.

Recently, we have demonstrated that I3C can be combined with random microseed matrix screening (rMMS) to efficiently generate derivatized protein crystals (Truong et al., 2019 ▸, 2021 ▸). To expand this screening technique to other phasing compounds, we have searched for other specific phasing ligands that could be effectively co-crystallized with different proteins. In this study, we have successfully co-crystallized B4C with hen egg-white lysozyme (HEWL) and confirm that B4C provides sufficient anomalous signal for structure solution using MAD phasing. This work expands the arsenal of existing phasing compounds and provides a proof of concept for the experimental phasing of other proteins.

2. Materials and methods  

2.1. Crystallization  

Tetrabromoterephthalic acid (B4C, catalogue No. 524441) and hen egg-white lysozyme (HEWL; catalogue No. L6876) were commercially acquired from Sigma–Aldrich. Lyophilized HEWL powder was dissolved in TBS (50 mM Tris–HCl pH 7.6, 150 mM NaCl) to a final concentration of 30 mg ml⁻¹ as determined by UV absorbance at 280 nm. B4C was dissolved in 2 M LiOH solution to a final concentration of 0.5 M. To prepare HEWL samples for co-crystallization experiments, 0.5 M B4C was added directly to the HEWL solution to a final concentration of 40 mM. HEWL crystals were grown via sitting-drop vapour diffusion in 96-well Intelli-Plates (Art Robbins) by mixing 1 µl HEWL solution with 1 µl reservoir solution [20%(w/v) polyethylene glycol 3350, 0.2 M potassium sulfate] and equilibrating against 75 µl reservoir solution (Table 1 ▸). The crystals were harvested onto a cryoloop, passed through Paratone-N (Hampton Research) for cryoprotection and flash-cooled in liquid nitrogen for data collection (Teng, 1990 ▸).

Table 1. Statistics of data processing for HEWL co-crystallized with B4C.

Method	Sitting-drop vapour diffusion
Plate type	Art Robbins 96-well Intelli-Plate
Temperature (K)	289.15
Protein concentration (mg ml−1)	30
Buffer composition of protein solution	50 mM Tris–HCl pH 7.6, 150 mM NaCl
Composition of reservoir solution	20%(w/v) polyethylene glycol 3350, 0.2 M potassium sulfate
Volume and ratio of drop	1 µl:1 µl
Volume of reservoir (µl)	75

2.2. Data collection and processing  

A 1.3 Å resolution MAD data set was collected at a wavelength of 0.92 Å using an EIGER X 16M detector on the macromolecular beamline MX2 at the Australian Synchrotron, which is part of ANSTO (Aragão et al., 2018 ▸). This wavelength produced the maximal anomalous signal close to the Br K edge, as determined by a fluorescence scan (Fig. 2 ▸). A total of 3600 diffraction images were collected with a 0.1° oscillation width and a crystal-to-detector distance of 170 mm. The diffraction data were processed using XDS (Kabsch, 2010 ▸) and were combined and scaled with AIMLESS (Evans & Murshudov, 2013 ▸) (Table 2 ▸).

Figure 2.

Fluorescence scan of HEWL co-crystallized with B4C. The peak wavelength (13.4837 keV) and inflection points (13.4767 keV) were assigned with CHOOCH (Evans & Pettifer, 2001 ▸)

Table 2. Statistics of data processing for HEWL co-crystallized with B4C.

Values in parentheses are for the highest resolution shell.

	Peak	Inflection
Diffraction source	MX2, Australian Synchrotron	MX2, Australian Synchrotron
Wavelength (Å)	0.9106	0.9200
Temperature (K)	100	100
Detector	EIGER X 16M	EIGER X 16M
Crystal-to-detector distance (mm)	170	170
Rotation range per image (°)	0.1	0.1
Total rotation range (°)	360	360
Space group	P43212	P43212
a, b, c (Å)	77.99, 77.99, 37.76	77.98, 77.98, 37.75
α, β, γ (°)	90, 90, 90	90, 90, 90
Mosaicity (°)	0.09	0.09
Resolution range (Å)	38.99–1.30	38.99–1.28
Total No. of reflections	734500 (32208)	753940 (31166)
No. of unique reflections	29530 (1477)	30377 (1428)
Completeness (%)	99.8 (97.1)	99.8 (96.2)
Multiplicity	24.9	24.8 (21.8)
〈I/σ(I)〉	31.0 (2.4)	30.9 (2.5)
R meas	0.052 (1.52)	0.050 (1.51)
Overall B factor from Wilson plot (Å2)	16.54	16.97

2.3. Structure solution and refinement  

The structure was solved using the two-wavelength multiple-wavelength anomalous diffraction (MAD) protocol of Auto-Rickshaw, the EMBL-Hamburg automated crystal structure-determination platform (Panjikar et al., 2005 ▸). The diffraction data processed using AIMLESS (Evans & Murshudov, 2013 ▸) were used as input. Marker-atom structure-factor amplitude (F _A) values were calculated using SHELXC (Sheldrick et al., 2001 ▸). Based on an initial analysis of the data, the maximum resolution for substructure determination and initial phase calculation was set to 1.75 Å. SHELXD found 27 potential heavy-atom sites (Schneider & Sheldrick, 2002 ▸). The correct hand for the substructure was determined using ABS (Hao, 2004 ▸) and SHELXE (Sheldrick, 2002 ▸). Initial phases were calculated after density modification using SHELXE (Sheldrick, 2002 ▸). Using ARP/wARP, 94% (121 of 129 residues) of the model was built (Perrakis et al., 1999 ▸; Morris et al., 2004 ▸) and correctly docked into electron density. At this stage, the structure had an R factor of 27.4%. The structure was then iteratively rebuilt and refined using Coot (Emsley et al., 2010 ▸) and phenix.refine (Afonine et al., 2012 ▸) to an R and R _free of 19.38% and 22.35%, respectively. Structure-solution statistics are summarized in Table 3 ▸.

Table 3. Statistics for structure solution of HEWL co-crystallized with B4C.

Values in parentheses are for the highest resolution shell.

Resolution range (Å)	38.99–1.30
Completeness (%)	99.8 (97.1)
No. of reflections, working set	29473 (2840)
No. of reflections, test set	978 (102)
Final R cryst	0.19 (0.30)
Final R free	0.22 (0.35)
No. of non-H atoms
Total	1169
Protein	991
Ligand	16
Water	162
R.m.s. deviations
Bonds (Å)	0.003
Angles (°)	0.60
Average B factors (Å2)
Overall	25.31
Protein	23.59
Ligand	31.14
Water	35.22
Ramachandran plot
Most favoured (%)	98.43
Allowed (%)	1.57

2.4. Accession numbers  

The coordinates and structure factors for hen egg-white lysozyme co-crystallized with B4C have been deposited in the Protein Data Bank under accession number 7kh5.

3. Results  

3.1. B4C is an effective compound for the experimental phasing of proteins  

The anomalous signal from a single B4C molecule was sufficient to solve the crystal structure of HEWL to a resolution of 1.3 Å using MAD phasing. HEWL forms a compact and globular structure, consisting of four α-helices, three β-strands and three 3₁₀-helices (Fig. 3 ▸). Four disulfide bonds are formed between Cys6 and Cys127, between Cys30 and Cys115, between Cys64 and Cys80 and between Cys76 and Cys94. The overall fold and structure of HEWL solved by experimental phasing with B4C is consistent with previously solved structures (data not shown).

Figure 3.

Hen egg-white lysozyme co-crystallized with B4C (PDB entry 7kh5). The crystal structure of HEWL is displayed in cartoon representation and B4C is represented in magenta as sticks.

3.2. B4C forms a characteristic arrangement of anomalous scatterers, and functional groups on the benzene scaffold of B4C facilitate protein binding  

A single molecule of B4C was bound to one monomer of HEWL (with an occupancy of 27%), positioned between the α3 helix and the β3/η1 loop. When bound, B4C is positioned within an overall positively charged surface-exposed cleft with two ortho-arranged Br atoms protruding into a deeper pocket (Fig. 4 ▸ a). Due to the arrangement of Br atoms on the benzene scaffold of B4C, the anomalous signal generated forms a characteristic 5.7 × 3.37 Å tetragon (Fig. 4 ▸ b). Key hydrogen-bonding and hydrophobic interactions facilitate the binding of the compound to HEWL. When bound, the carboxylic acid functional group on the benzene scaffold of B4C forms a 2.9 Å hydrogen bond between the hydroxyl group and the side chain of Asn93 (Fig. 4 ▸ c). Additional hydrophobic interactions formed between B4C and Cys76, Ile78, Ala90, Asn93 and Cys94 further stabilize binding. Analysis of the crystal contacts reveals two molecules of B4C, one from each symmetry mate, interlocking and likely aiding crystallization (Fig. 4 ▸ d).

Figure 4.

Characteristics of B4C binding to hen egg-white lysozyme. (a) Electrostatic surface representation of HEWL with B4C bound to a positively charged surface-exposed cleft. (b) The binding site of B4C in the lysozyme crystal overlaid with the substructure density of B4C (anomalous difference map contour of 5σ). This map was generated using phenix.maps. The four Br atoms arranged on the benzene scaffold form a characteristic tetragon with dimensions of 5.7 × 3.3–3.4 Å. (c) Binding of B4C to HEWL occurs through hydrogen-bond interactions (black dashed line) with the Asn93 side chain. All distances are reported in Å. (d) Symmetry mates of HEWL (depicted in grey and magenta) show B4C molecules interlocking at the interface of the crystal contact.

4. Discussion  

Although the number of structures deposited in the PDB is steadily growing, a significant proportion of proteins remain that lack a suitable template for molecular replacement and therefore require experimental phasing. Commonly, phasing compounds containing heavy atoms are soaked into pre-existing crystals or co-crystallized with the protein to enable experimental phasing. However, many of these compounds have inherent limitations that hinder the structure-determination pipeline. Some phasing compounds, such as heavy-metal salts, show poor solubility, which can reduce their availability in both soaking and co-crystallization experiments. Halide-containing compounds may bind with low specificity, leading to low occupancy and poor anomalous scattering (Pike et al., 2016 ▸). Furthermore, many of these phasing compounds are incompatible with common crystallization reagents, including phosphate, sulfate, citrate, acetate, tris(2-amino-2-hydroxymethyl-propane-1,3-diol) and HEPES [4-(2-hydroxy­ethyl)-1-piperazineethanesulfonic acid] buffers. Since there is no singular phasing compound that is universally compatible with all proteins and crystallization conditions, it is necessary to generate a library of available compounds that can be utilized in different situations.

In this study, we have demonstrated that B4C is an effective phasing compound by successfully solving the structure of the model protein HEWL. Due to the characteristic arrangement of Br atoms in B4C, the position of the B4C ligand was readily identified in the anomalous density map. In B4C, the Br atoms form a characteristic rectangle (5.7 × 3.37 Å) which can be used to easily validate that the substructure identified is correct (Fig. 4 ▸ b).

The chemical composition of B4C offers several advantages that improve its qualities as a phasing compound. The unique arrangement of functional groups on the benzene scaffold affords superior binding specificity to the protein. In turn, this improves the occupancy, generates greater anomalous signal and thus reduces noise. As shown, a single molecule of B4C bound to HEWL provides sufficient anomalous signal for experimental phasing (Fig. 3 ▸).

B4C has a para arrangement of carboxyl groups that are surrounded by four Br atoms on the aromatic scaffold. This arrangement of functional groups differs from the arrangement of I atoms in I3C and Br atoms in B3C. Thus, B4C can recognize and bind different sites on the protein compared with I3C and B3C. The two available crystal structures of I3C-bound HEWL (PDB entries 6pbb and 3e3d) show six positions at which I3C binds on the HEWL protein surface (Truong et al., 2019 ▸; Beck et al., 2008 ▸). The position where B4C binds on the protein surface differs from the previously occupied positions of I3C and does not clash spatially (Fig. 5 ▸). This result indicates that proteins that fail to be derivatized with I3C (and likely with B3C) could potentially be derivatized with B4C. B4C also has four anomalous scattering Br atoms, compared with three Br atoms in B3C, and thus in theory could provide greater anomalous signal. The presence of Br atoms also means that it can be phased by MAD phasing, instead of being restricted to single-wavelength anomalous dispersion (SAD) phasing as is the case for I3C. MAD can provide additional phasing information in comparison to the SAD method (Rice et al., 2000 ▸).

Figure 5.

B4C can occupy different sites on a protein surface compared with I3C. Structural superimposition of HEWL bound to I3C (PDB entries 6pbb, yellow, and 3e3d, cyan) and HEWL bound to B4C (PDB entry 7kh5, magenta) reveals that B4C binds to a different surface patch on HEWL to I3C.

As B4C is a common starting reagent for the synthesis of polyesters, it is commercially available. It can be acquired easily in large quantities and is relatively inexpensive. It is also highly soluble in lithium hydroxide and other basic solutions, undergoing an acid–base reaction upon dissolving to produce a B4C solution that has an approximately neutral pH. This is beneficial for both co-crystallization and soaking experiments, since the process of crystallization and the crystals themselves are sensitive to changes in pH.

Additionally, B4C is compatible with co-crystallization experiments and thus derivatization using this compound is not entirely reliant on soaking. Co-crystallization reduces the number of potentially damaging crystal-handling steps. B4C also could be added to commercial sparse-matrix screens to identify suitable co-crystallization conditions whilst simultaneously growing derivatized protein crystals. A further application may involve combining B4C co-crystallization with rMMS to expand the number of potential conditions for derivatization of crystals. The efficiency of this technique has been demonstrated previously using the phasing compound I3C (Truong et al., 2019 ▸).

In this study, we have co-crystallized HEWL with B4C and solved the structure using MAD phasing. The several advantageous properties of B4C, including its high phasing power and its compatibility with common crystallization reagents, highlight the effectiveness of B4C as an experimental phasing molecule. Due to its accessibility and affordability, B4C is an excellent addition to the arsenal of phasing compounds suitable for scenarios in which molecular replacement is not possible.

Supplementary Material

PDB reference: hen egg-white lysozyme, complex with tetrabromoterephthalic acid, 7kh5

Funding Statement

This work was funded by Australian Research Council grants DP150103009 and DP160101450. University of Adelaide grant .