Joint X-ray/neutron structure of Lentinus similis AA9_A at room temperature

Lytic polysaccharide monooxygenases (LPMOs) are copper metalloenzymes which cleave polysaccharides oxidatively and are important in pathogen biology, carbon cycling and biotechnology. The Lentinus similis family AA9 isoform A (LsAA9_A) has been extensively studied as a model system because its activity towards smaller soluble saccharide substrates has allowed detailed structural characterization of its interaction with a variety of substrates by X-ray crystallography at high resolution. Here, the joint X-ray/neutron room-temperature crystallographic structure of carbohydrate-free LsAA9_A in the copper(II) resting state refined against X-ray and neutron data at 2.1 and 2.8 Å resolution, respectively, is presented. The results provide an experimental determination of the protonation states of the copper(II)-coordinating residues and second-shell residues in LsAA9_A, paving the way for future neutron crystallographic studies of LPMO–carbohydrate complexes.

1. Introduction

Copper-containing lytic polysaccharide monooxygenases (LPMOs) are a class of metalloenzymes that have recently been brought to the forefront of research advancing renewable energy and the conversion of biomass to value-added chemicals (Johansen, 2016), as well as having increasingly recognized biological functions in pathogenesis (Vandhana et al., 2022). They are classified as auxiliary activities (AA) families AA9–AA11 and AA13–AA17 in the Carbohydrate Active enZymes (CAZy) database (Levasseur et al., 2013; Drula et al., 2022). Fungal LPMOs are some of many enzymes secreted by fungi to break down plant matter, namely cellulose, and release glucose. LPMOs randomly oxidize recalcitrant crystalline cellulose, disrupting local crystallinity and creating new cellodextrin chain ends that serve as additional starting points for the endo- and exo-activities of glycoside hydrolases (GHs; Agger et al., 2014). While LPMOs were first described as monooxygenases, binding molecular dioxygen as a co-substrate, it is now established that LPMOs also have peroxidase/peroxygenase activity (Bissaro et al., 2017). The monooxygenase reaction requires molecular oxygen, two protons and two single-electron transfers from either small-molecule reductants or electron-donating proteins such as cellobiose dehydrogenases that are co-secreted by fungi (Fig. 1). Mechanistic details, including the source of protons and the chemical nature of oxygen-activated reaction intermediates, for both the O₂-based and H₂O₂-based mechanisms remain to be elucidated in order to understand how these enzymes carry out carbohydrate oxidation.

Figure 1 LPMO reaction scheme. Depending on the specific LPMO and the available conditions, one of several reactions may take place. Initial reduction (a) and substrate binding (b) is followed by the formation of oxidized or non-oxidized chain ends depending on the specfic LPMO, using H2O2 (c) or O2 (d) as a co-substrate. Oxidizing reactions may also occur in the presence of small molecues (e). In the presence of O2 but without substrate, the LPMO may produce H2O2 (f).

Neutron protein crystallography is a powerful tool for investigating protein chemistry because it directly locates H-atom positions in a protein structure (Schröder et al., 2018). Early X-ray and neutron crystallographic studies of LPMOs from the AA9 family have focused on understanding the activation of O₂ by Neurospora crassa LPMO 9D (NcAA9_D) in the absence of substrate (Bodenheimer et al., 2017; O'Dell, Swartz et al., 2017; O'Dell, Agarwal et al., 2017; Schröder et al., 2021, 2022). In addition to the direct determination of protonation states, these structures have revealed the geometry and the chemical nature of the initial Cu—O₂⁻ and Cu—O₂H intermediates. These results are important to understand the activation of O₂ (Fig. 1f). However, elucidation of the mechanism of carbohydrate oxidation requires the trapping of early intermediates in an LPMO–carbohydrate complex (Fig. 1b). The insolubility of the substrates of NcAA9_D has so far prevented the investigation of LPMO–O₂–carbohydrate complexes. Later, LsAA9_A was demonstrated to be active towards soluble oligosaccharides and the first structures of an LPMO enzyme in complex with short soluble carbohydrates were reported (Frandsen et al., 2016; Simmons et al., 2017). LsAA9_A has now been used extensively as a model LPMO, including high-resolution crystallo­graphic studies to determine the protonation state of key histidine residues (Banerjee et al., 2022) and detailed photoreduction studies (Tandrup et al., 2022). Furthermore, the dependence of LsAA9_A on H₂O₂ has been established and a novel twist on the LPMO mechanism has been suggested, at least for this specific enzyme (Brander et al., 2021). Therefore, LsAA9_A opens the opportunity to analyze substrate-bound enzyme complexes and the activation of H₂O₂ using neutron crystallography.

Here, we report the neutron structure of carbohydrate-free LsAA9_A in the copper(II) resting state. The protonation states of catalytic residues at and around the copper center are determined.

2. Materials and methods

2.1. Protein purification and crystal growth

The fungal enzyme LsAA9_A was expressed in Aspergillus oryzae, purified and deglycosylated with endoglycosidase H as described previously (Frandsen et al., 2016; Simmons et al., 2017; Tandrup et al., 2020). The protein sample was pre-incubated with an equimolar amount of copper(II) acetate for 1 h at 4°C prior to crystallization setup. Crystals of LsAA9_A were grown via sitting-drop vapor diffusion by adapting previously reported protocols (Frandsen et al., 2016). For neutron crystallography, crystals were grown in a nine-well glass-plate and sandwich-box setup (Hampton Research) at 10°C. The sitting drops were equilibrated against 50 ml reservoir solution consisting of 3.0 M NaCl, 0.1 M citric acid pH 3.5. Large crystals grew from 100 µl sitting drops prepared at a protein concentration ranging from 16 to 20 mg ml⁻¹ in 1.3 M NaCl, 0.1 M citric acid pH 3.5 (Fig. 2).

Figure 2 Octahedral crystal of hydrogenated LsAA9_A grown from 100 µl sitting drops.

2.2. Neutron and X-ray data collection at room temperature

Crystals were mounted in thin-walled quartz capillaries (Hampton Research) using hydrogenated crystallization buffer at pH 3.5 from the sandwich box. Excess buffer was removed and plugs of 100 mM deuterated citric acid buffer at pD 5.5 (pH electrode reading 5.1) containing 3 M NaCl were placed on both sides of the crystal prior to sealing the capillaries to vapor-exchange the crystal solvent water molecules and labile protein H atoms to D₂O molecules and D atoms, respectively. The pH was increased from 3.5 to 5.5 to prevent disorder of the histidine brace (Frandsen et al., 2017). The exchange occurred over two weeks prior to neutron data collection.

Neutron time-of-flight diffraction data were collected at room temperature on the MaNDi instrument at the Spallation Neutron Source (Coates & Sullivan, 2020; Meilleur et al., 2018). An incident neutron wavelength bandpass of 3–5 Å was used. A total of six diffraction patterns with a Δφ of 10° between frames were collected with an exposure of 48 h per frame. Following neutron diffraction data collection, an X-ray data set was collected from the same crystal at room temperature on a microfocus rotating-anode X-ray diffracto­meter (MicroMax-007 HF, Rigaku). A total of 57 diffraction patterns were collected with a Δφ of 1.0° and an exposure of 45 s per frame.

The neutron data set was reduced using the Mantid package (Arnold et al., 2014) and integrated using three-dimensional profile fitting (Sullivan et al., 2018). The data were wavelength-normalized using LAUENORM from the LAUEGEN suite (Helliwell et al., 1989; Campbell et al., 1998; Arzt et al., 1996). The X-ray data were indexed and integrated using CrysAlis^Pro (Rigaku, Woodlands, Texas, USA) and scaled and merged with AIMLESS in the CCP4 suite (Evans & Murshudov, 2013; Winn et al., 2011). Data-collection and processing statistics are summarized in Table 1. The data and structure have been deposited in the Protein Data Bank (PDB entry 8e1w; Berman et al., 2000).

Table 1
Neutron and X-ray data-collection and joint refinement statistics

Values in parentheses are for the highest resolution shell.

	X-ray	Neutron
Wavelength (Å)	1.54	3.0–5.0
Resolution range (Å)	30.67–2.10 (2.18–2.10)	14.8–2.80 (2.90–2.80)
Space group	P4132	P4132
a, b, c (Å)	126.469, 126.469, 126.469	126.469, 126.469, 126.469
α, β, γ (°)	90, 90, 90	90, 90, 90
Total reflections	212841 (21125)	91012 (7436)
Unique reflections	20749 (2019)	8942 (869)
Multiplicity	10.3 (10.5)	10.2 (8.6)
Completeness (%)	99.87 (99.95)	99.3 (99.9)
Mean I/σ(I)	14.16 (2.09)	10.1 (4.8)
Rmerge	0.1758 (1.482)	0.262 (0.290)
Rmeas	0.1851 (1.559)	0.275 (0.308)
Rp.i.m.	0.0571 (0.4802)	0.081 (0.102)
CC1/2	0.988 (0.746)	0.920 (0.348)
Reflections used in refinement	20738	8850
Reflections used for Rfree	1033	886
Rwork	0.1484	0.2251
Rfree	0.1702	0.2598
No. of non-H atoms
Total	1931
Macromolecule	1798
Cu	1
Cl	1
N-Acetylglucosamine	14
Solvent	117
Protein residues	235
R.m.s.d., bond lengths (Å)	0.015
R.m.s.d., angles (°)	1.27
Ramachandran favored (%)	91.81
Ramachandran allowed (%)	8.19
Ramachandran outliers (%)	0.00
Rotamer outliers (%)	1.53
Clashscore	1.42
Average B factor (Å^2)
Overall	33.56
Macromolecule	32.92
Cu	35.33
Cl	33.0
N-Acetylglucosamine	50.54
Solvent	40.84

2.3. Structure refinement

Joint X-ray/neutron refinement was performed using the Phenix software suite (Liebschner et al., 2019) with manual model building performed in Coot (Emsley et al., 2010).

The ligand-free model of LsAA9_A at pH 3.5 (PDB entry 5n04; Frandsen et al., 2017) stripped of the Cu²⁺ ion and water molecules was used as the starting model. Iterative rounds of refinement and model building were conducted. When the model refined against the X-ray data alone was complete, H and D atoms were generated using the Phenix ReadySet! tool as described by Schröder & Meilleur (2020). The model was further refined against both the X-ray and neutron data sets. Refinement statistics are listed in Table 1.

3. Results and discussion

Neutron and X-ray data sets were collected from the same crystal to 2.8 Å and 2.1 Å resolution, respectively. The LsAA9_A model refined jointly against the X-ray and neutron data includes key LPMO structural features at the active site.

3.1. Vapor exchange

Neutron diffraction data collection requires hydrogen to be exchanged for deuterium to increase the signal-to-noise ratio of the data and the visibility of the hydrogen/deuterium positions (Meilleur, 2020). Vapor exchange was performed over a period of two weeks, which is typical of the time used in a neutron protein crystallography experiment to exchange hydrogen for deuterium (O'Dell et al., 2016). The hydrogen–deuterium exchange pattern for the backbone amide groups is presented in Fig. 3. As expected, the outer loops show a high level of exchange, while the inner β-sheets undergo limited exchange. However, the overall exchange is low (35%), indicating that at this low pH longer times are required to achieve higher overall exchange. The pH dependence of the amide hydrogen-exchange rate has previously been examined using mass-spectrometry experiments. These studies suggest that the exchange time required at pH 5.5 to achieve the same level of exchange as at pH 7.5 is 100 times longer (Li et al., 2014). This time scale is challenging to achieve when planning for neutron data collection.

Figure 3 Hydrogen/deuterium exchange of the backbone amide groups. Full (100% D, 0% H) exchange is plotted in red. No exchange (0% D, 100% H) is plotted in blue. Copper(II) is represented by an orange sphere.

3.2. The histidine brace

As previously described for LsAA9_A, the Cu²⁺ ion in the joint X-ray/neutron structure presented here is coordinated by the N-terminal amino group and N^δ of His1, N^ɛ of His78 and a water molecule in the equatorial position (H₂O-eq), while the OH group of Tyr164 and a water molecule are located close to the axial coordination sites. The distances of the copper to the ligands are listed in Table 2.

Table 2
Distances within the LsAA9_A copper site

Ligand	Copper–ligand distance (Å)
His1 N	2.2
His1 N^δ	2.0
His78 N^ɛ (0.75)/Cl^− (0.25)†	2.1/2.5
H2O-eq	2.1
H2O-ax	2.7
Tyr164 O^η	2.8

†The occupancies of His78 and the Cl⁻ ligand are shown in parentheses.

The crystals were grown at pH 3.5. At this pH the histidine brace is disordered, with His78 adopting two conformations as reported previously for LsAA9_A (Frandsen et al., 2017). Ordering of the histidine brace requires the pH to be increased. The pH increase is typically achieved by soaking crystals directly in buffer at pH 5.5. Here, to avoid compromising the diffraction quality of the large crystals, the pH was increased by vapor diffusion.

Clear electron and neutron scattering-length densities were observed for His78 coordinated to Cu²⁺, confirming that vapor pH exchange drove ordering of the histidine brace (Figs. 4a and 4b). No residual F_o − F_c electron density was observed for an alternate conformation of His78 flipped away from the active-site Cu²⁺. However, residual F_o − F_c electron density indicated that a Cl⁻ ion from the crystallization condition alternatively coordinates the Cu²⁺ ion instead of His78 as observed in the structure of LsAA9_A previously solved at pH 3.5 (Frandsen et al., 2017; Fig. 4c). The occupancies of His78 and Cl⁻ refined to 0.75 and 0.25, respectively. The N-terminal amino group is neutral (Fig. 4d).

Figure 4 The LsAA9_A active site modeled in the joint X-ray/neutron structure at room temperature. (a) 2Fo − Fc electron-density map in light blue contoured at 1.0σ. (b) Fo − Fc neutron scattering-length density omit map for His78 in purple contoured at 3.0σ. His78 is modeled in a single conformation with 75% occupancy. (c) Fo − Fc electron-density omit map for Cl^− in dark blue contoured at 3.0σ. The Cl^− ion is modeled with 25% occupancy. (d) Fo − Fc neutron scattering-length density omit map for the N-terminal amino group in purple contoured at 3.0σ, indicating that the N-terminal amino group is neutral (NH2).

3.3. Active-site waters

The active sites of LPMOs are readily prone to X-ray-induced photochemistry, as shown by photoreduction studies on families AA9, AA10 and AA13 (Gudmundsson et al., 2014; Muderspach et al., 2019; Banerjee et al., 2022). At the active site of LPMO, photoreduction causes disorder (and ultimately the disappearance) of the copper-coordinating water molecules in the copper axial and equatorial positions and other geometrical changes. An advantage of neutron diffraction in the structural characterization of metalloenzymes is the lack of radiation-induced in-beam chemistry (Bodenheimer et al., 2017; Schröder & Meilleur, 2021). While joint refinement against neutron and X-ray data is a standard approach (O'Dell et al., 2016), the structure refined against the X-ray data alone must be carefully examined when planning the joint X-ray/neutron refinement of a metalloenzyme. For the joint refinement performed here, we collected an X-ray data set on a home source. The X-ray data show clear F_o − F_c electron-density peaks for the copper equatorial and axial water molecules, as well as for the pocket water (Fig. 5). The distances from H₂O-eq and H₂O-ax to Cu²⁺ are 2.1 and 2.7 Å, respectively, confirming that the copper ion is in the resting oxidation state, copper(II), and validating the use of the X-ray data for joint refinement (Tandrup et al., 2022).

Figure 5 Active-site water molecules. Fo − Fc electron-density omit map for H2O-eq, H2O-ax and H2O-pocket in dark blue contoured at 7.0σ.

3.4. Residues near the active site: Tyr164 and His147

The LPMO family AA9 active site includes a conserved tyrosine residue. The tyrosine hydroxy group is in the second axial coordination site of the copper, but the Cu–O_Tyr distance is too long to form a Cu—O_Tyr bond. Recent analysis of X-ray structures at low X-ray dose have confirmed that on binding oligosaccharide the distance to Tyr164 is shortened (Tandrup et al., 2022). In the structure presented here, Tyr164 is positioned in the axial position of the copper with a Cu–O_Tyr distance of 2.80 Å. The Tyr164 OH group forms a hydrogen bond to the side-chain carbonyl group of Gln162 with an O_Tyr164–O_Gln162 distance of 2.72 Å. An F_o − F_c nuclear map calculated omitting the D atom of the Tyr164 hydroxy group show a clear positive peak at 3.0σ (Fig. 6a). It will be interesting to observe the effect of saccharide binding on this hydrogen bond in future neutron structures.

Figure 6 Conserved residues near the copper center. (a) Fo − Fc neutron scattering-length density omit map for the hydroxy group of Tyr164 in purple contoured at 3.0σ. The hydrogen bond between Tyr164 and Gln162 OE1 is represented by a black dashed line. (b) Fo − Fc neutron scattering-length density omit map for His147 in purple contoured at 3.0σ, indicating that His147 is singly protonated at the NE2 position.

His147, a conserved second-shell residue, is singly proton­ated at the NE2 position (Fig. 6b). This confirms that in this conformation His147 remains neutral at low pH (Schröder et al., 2022), as also shown by analysis of high-resolution X-ray structures (Banerjee et al., 2022).

4. Conclusion

The protonation state of key amino acids in the active site of LsAA9_A could be determined despite the modest resolution of the neutron data. For future studies, longer pH exchange times are advised, as a small amount of disorder remained at the histidine brace. Perdeuteration allows diffraction from similar-sized crystals to extend to higher resolution by eliminating the incoherent scattering of hydrogen which contributes to the high background. The heterologous expression of LsAA9_A was recently optimized in Escherichia coli, which opens the opportunity to perdeuterate the protein (Hernández-Rollán et al., 2021). LsAA9_A produced in E. coli was crystallized and high-quality X-ray structures were obtained, including a complex with cellotriose (Banerjee et al., 2022; Tandrup et al., 2022). Future neutron structural investigation of LsAA9_A in complex with carbohydrate ligands will be greatly enhanced by the use of perdeuterated protein.

Footnotes

‡Current affiliation: Department of Biotechnology and Biomedicine, Technical University of Denmark, Søltofts Plads 224, 2800 Kongens Lyngby, Denmark.