Perdeuterated GbpA Enables Neutron Scattering Experiments of a Lytic Polysaccharide Monooxygenase

Lytic polysaccharide monooxygenases (LPMOs) are surface-active redox enzymes that catalyze the degradation of recalcitrant polysaccharides, making them important tools for energy production from renewable sources. In addition, LPMOs are important virulence factors for fungi, bacteria, and viruses. However, many knowledge gaps still exist regarding their catalytic mechanism and interaction with their insoluble, crystalline substrates. Moreover, conventional structural biology techniques, such as X-ray crystallography, usually do not reveal the protonation state of catalytically important residues. In contrast, neutron crystallography is highly suited to obtain this information, albeit with significant sample volume requirements and challenges associated with hydrogen’s large incoherent scattering signal. We set out to demonstrate the feasibility of neutron-based techniques for LPMOs using N-acetylglucosamine-binding protein A (GbpA) from Vibrio cholerae as a target. GbpA is a multifunctional protein that is secreted by the bacteria to colonize and degrade chitin. We developed an efficient deuteration protocol, which yields >10 mg of pure 97% deuterated protein per liter expression media, which was scaled up further at international facilities. The deuterated protein retains its catalytic activity and structure, as demonstrated by small-angle X-ray and neutron scattering studies of full-length GbpA and X-ray crystal structures of its LPMO domain (to 1.1 Å resolution), setting the stage for neutron scattering experiments with its substrate chitin.


■ INTRODUCTION
Lytic polysaccharide monooxygenases (LPMOs) are surfaceactive enzymes that introduce breaks in chitin, cellulose, or other polysaccharide layers through a copper-dependent redox reaction. Conserved among all active LPMOs is the histidinebrace motif, where a copper-ion is coordinated by the amino group and a side-chain nitrogen of the N-terminal histidine together with a side-chain nitrogen of a second histidine residue. 1−3 The enzymatic reaction is initiated by the reduction of Cu(II) to Cu(I) by an electron donor and subsequent activation of an oxygen co-substrate. It is unclear if the oxygen co-substrate is O 2 or H 2 O 2 , with the growing consensus being that H 2 O 2 is the more likely co-substrate. 4−6 The natural electron donor likely varies significantly between organisms. 7 Interestingly, the function of LPMOs often goes beyond polysaccharide conversion. Several LPMOs are part of multidomain proteins, often with an additional carbohydrate-binding module. Some LPMOs have shown to be key virulence factors for pathogenic bacteria, with importance for colonization. One example is the AA10 class LPMO, N-acetylglucosamine-binding protein A (GbpA) of Vibrio cholerae. GbpA is a four-domain protein with chitin affinity of the first and fourth domains, and LPMO activity in the first domain. 8,9 This domain in addition exhibits affinity for human intestinal mucin, 8,10 a feature likely important for the colonization of the small intestine by V. cholerae.
While the structures of several LPMOs have been determined to atomic resolution by X-ray crystallography, including the first three domains of GbpA, 8 and all four domains of the close homologue VhLPMO10A from Vibrio campbellii, 11 the functional information has been limited, in part due to the inability to map hydrogen atoms in the structures. For redox-active enzymes like LPMOs, neutron macromolecular crystallography (NMX) is a strong complementary method to X-ray crystallography that can reveal additional information. 12 With NMX, functionally relevant hydrogens can be modeled even at relatively low (∼2.5 Å) resolution. 13 Furthermore, neutrons are non-ionizing, 13 leaving the active site of LPMOs intact throughout the diffraction experiment. Unfortunately, neutron sources have a very low flux, and hydrogen atoms have a high incoherent scattering; this diminishes the signal-to-noise ratio for the crystallographic data. Longer data collection times (e.g., multiple days) and large crystal volumes (0.1−1.0 mm 3 ) can therefore be required to determine a neutron protein crystal structure. Also, hydrogen has a negative coherent scattering length compared to the positive scattering lengths of carbon, nitrogen, sulfur, and oxygen, which may cause density cancellation effects in neutron maps at intermediate resolution. 13 These issues can be resolved with protein deuteration. Deuterium has strong positive coherent scattering and incoherent scattering that is essentially negligible compared to that of hydrogen. 13 This alleviates the sample volume requirement by as much as a factor of ten. A few neutron crystal structures of LPMOs have already been determined; 14−17 however, none of them is perdeuterated, enabling mapping of only select hydrogen atoms.
Other neutron techniques that can benefit from protein deuteration are small-angle neutron scattering (SANS) and neutron reflectometry (NR), where contrast-matching can help visualize individual components in a complex, provided that the components have sufficiently different scattering-length densities (SLD). Deuterated biomolecules have very different SLDs from non-deuterated biomolecules. Using SANS or NR on deuterated LPMOs can therefore yield information on the interactions with other biomolecules, most obviously the polysaccharide substrates. An extensive review of deuteration methodologies/protocols/applications for neutron scattering has been given by Haertlein et al. 18 Perdeuteration of proteins by expression in deuterated bacterial cultures can be a challenging and time-consuming process. Protocols usually require adaptation of the cells to increasing amounts of D 2 O; however, Cai et al. 19 recently developed a faster and simpler protocol, yielding high amounts of perdeuterated proteins. We adapted this protocol for both full-length (FL) GbpA (GbpA-FL) and for its first domain (GbpA-D1), scaled up production at international deuteration facilities, and characterized perdeuterated GbpA-FL (D-GbpA-FL) by small-angle X-ray scattering (SAXS) and SANS. The protein retained its catalytic activity. We also crystallized and determined the structure of deuterated GbpA-D1 (D-GbpA-D1) using X-ray crystallography and compared it with the X-ray structure of the hydrogenated protein (H-GbpA-D1).
These results demonstrate the feasibility of perdeuterating LPMOs for neutron-based structural biology studies, with the promise of increased knowledge for the functional mechanisms of this important class of enzymes.
Production of Hydrogenated GbpA (H-GbpA). Unless specified otherwise, all the work described in this study was carried out with expression constructs provided by the Vaaje-Kolstad laboratory (Norwegian University of Life Sciences). They are described in detail by Wong et al. 8 Briefly, the GbpA-FL sequence was cloned from the genomic DNA of V. cholerae strain N1RB3 into a pET-22b vector between the NdeI and XhoI sites. The construct for expressing GbpA-D1 was obtained by the addition of two codon stops at position 203. The natural tag for protein secretion (amino acids 1−23) is cleaved off by the E. coli expression system. The pET-22b vector contains an ampicillin resistance gene, exploited for selection in growth and expression phases.
H-GbpA-FL was expressed in BL21(DE3) STAR cells transfected with the GbpA-encoding plasmid using Terrific Broth (TB), Luria Bertani (LB), or non-deuterated M9glyc+ media (Table 1,  flask and incubated for another 15 h. Subsequently, we added the culture to 225 mL M9glyc+ media in a 2 L baffled flask. We tested induction with 1 mM IPTG at four different OD 600 levels (0.7, 3.0, 7.2 and 11.0; Figure S1B) and obtained the best yield at OD 600 = 3.0 ( Figure 1A). After addition of IPTG, the temperature was lowered to 25°C for 20 h of expression. For a scheme of the protocol, see Figure 2. 100 μM sodium ampicillin was used for selection.

Production of Deuterated Protein (D-GbpA).
When producing deuterated protein (D-GbpA-FL and D-GbpA-D1), we used the same conditions as for H-GbpA production in minimal media, except for one additional step: After growing BL21(DE3) Star cells containing GbpA-encoding plasmid for 3 h in 1 mL LB/H 2 O medium in 15 mL tubes, this culture was diluted ten times into 2.5 mL LB/D 2 O medium in 15 mL tubes, before incubation for another 5 h. Subsequently, 2.5 mL of this culture were transferred to 25 mL deuterated M9glyc+ medium (Table 1)  For the production of D-GbpA-D1 at D-lab (ILL) and DEMAX (ESS), a different construct was used (but coding for the same amino acid sequence; UniProt ID: Q9KLD5, residues 24−485). The gbpa gene was codon-optimized and cloned into pET vector pET-26b by GenScript (Leiden, The Netherlands) using the restriction sites NcoI and XhoI. The pelB leader sequence is cleaved off during post-translational translocation to the periplasmic space to instate the catalytically important His24 as N-terminal amino acid. The C-terminal His-tag was omitted by inclusion of a stop codon at the end of the insert. The pET-26b vector contains a kanamycin resistance gene, which we exploited for selection in growth and expression phases.
For D-GbpA-D1 production at the D-lab deuteration facility of Institut Laue-Langevin (ILL, Grenoble, France; proposal number DL-03-223), the protein was over-produced in E. coli strain BL21 (DE3) adapted to growth in deuterated minimal medium, 18 using the expression construct cloned in the pET-26b plasmid. A 1.9 L (final volume) deuterated high-cell-density fed-batch fermenter culture was grown at 30°C. Feeding with glycerol-d 8 was started at an OD 600 value of about 5. Expression of D-GbpA-D1 was induced at an OD 600 of about 13 by addition of 1 mM IPTG (final concentration). Cells were harvested at OD 600 = 15.8, yielding 40 g perdeuterated cell paste (wet weight), thus approximately 20 g/L media. The cell paste was flash frozen and stored at −80°C to prevent proteolysis before transport to Oslo for further processing.
D-GbpA D1 production in the DEMAX biodeuteration labs of the European Spallation Source (ESS, Lund, Sweden; proposal number 890320) followed a different protocol, without adaptation, using an approach described in Koruza et al. 21 D-GbpA D1 was expressed in BL21 (DE3) Tuner cells after transformation with GbpA-D1-encoding plasmid using selective growth conditions (i.e., in the presence of kanamycin) on LB agar plates. A 5 mL LB starter culture was grown with kanamycin (50 μg/mL final concentration) after inoculation from a single colony, and was left to grow while shaking at 180 rpm at 37°C overnight. From this overnight culture, small-scale expression tests were performed in 50 mL LB cultures and a glycerol stock was prepared. For scaling up the yield, the cells were pre-grown in LB medium, a 100 mL overnight culture was started from the glycerol stock, and kanamycin was added (50 μg/mL final concentration). The next morning, 6 × 1 L cultures were inoculated with 10 mL of the overnight culture and fresh antibiotic was added. The cultures were grown in Tunair baffled    21 with the addition of 10 mL of Botryococcus braunii deuterated algal extract and a reduction in the amount of glycerol-d 8 to 2 g/L of media. At this point, fresh antibiotic was added, and the cells were allowed to recover for 1 h at 25°C while shaking at 100 rpm. Thereafter, the shaking was increased to 200 rpm, 1 mM IPTG (final concentration) was added, and expression was left to continue overnight for up to 20 h. Finally, the cells were harvested by centrifugation at 10,000 × g for 15 min in a JLA8.1000 rotor and immediately further processed for periplasmic protein extraction as described below. Using this approach, it was possible to obtain 60 g of wet cell paste (i.e., 10 g/L of media) for further processing. The Certificate of Analysis of the material provided is deposited with DOI 10.5281/zenodo.6631673. The periplasmic fraction was frozen at −20°C and shipped on ice packs for further purification. Autolysis Procedure for Deuterated Algal Extract. The procedure for algal autolysis was adjusted and modified from literature descriptions. 22,23 Microalgae B. braunii (UTEX Showa strain, Culture Collection of Algae at the University of Texas at Austin) were continuously cultivated in perdeuterated modified Bolds 3 N medium. 24 Cells were grown in a 12 h:12 h light−dark cycle and illuminated with 240 μmol photons/m 2 s LED lights in a Multitron Pro incubator (INFORS HT). Cultures were agitated by shaking at 60 rpm, and the atmosphere was kept at 2% CO 2 . The cells were periodically harvested, typically every 14−21 days, by centrifugation in a JLA8.100 rotor (Beckman) at 5000 × g for 15 min. The pelleted cells were frozen at −80°C until further processing. To prepare the autolysate from frozen cells, 400 mL of 99.9% D 2 O was added to 100 g of frozen wet microalgae cells. The cells were thawed and resuspended to a uniform suspension and then incubated at 50°C in a water bath for 24 h. The digested cell product was centrifuged in a JA-21 rotor (Beckman) at 10,000 × g for 20 min. The supernatant was aliquoted in 10 mL tubes and frozen at −20°C until needed. For modified ModC1 medium preparation, 10 mL of d-algal extract was used per liter of culture media.
Periplasmic Lysis and Protein Purification. GbpA (both FL and D1, deuterated or hydrogenated) was harvested from the E. coli periplasm using the following periplasmic lysis protocol: First, the cell culture was pelleted by centrifugation (10,000 × g). The pellet was resuspended in a solution containing 25% sucrose, 20 mM tris(hydroxymethyl)aminomethane (Tris)-HCl pH 8.0 and 5 mM ethylenediaminetetraacetic acid (EDTA) (4− 5 mL/g cells), and incubated for 30 min. Thereafter, the cells were once again pelleted and resuspended in ∼50 mL (4−5 mL/ g cells) solution of 5 mM MgCl 2 , 1 mM phenylmethylsulfonyl fluoride (PMSF) and 0.25 mg/mL lysozyme. The suspension was incubated for 30 min on ice, pelleted, and the GbpAcontaining supernatant was subjected to purification.
GbpA (both FL and D1, deuterated or hydrogenated) was purified by anion-exchange and size-exclusion chromatography. Anion-exchange chromatography (AEX) was performed using a HiTrap Q HP 5 mL column (Cytiva) connected to an ÄKTA Start protein purification system (GE Healthcare). After loading, the protein was eluted by a salt gradient from 100 to 400 mM NaCl buffered with 20 mM Tris−HCl pH 8.0. Sizeexclusion chromatography (SEC) was performed using a HiLoad Superdex 200 prep grade column on an ÄKTA Purifier system (GE Healthcare) in a running buffer containing 20 mM Tris−HCl pH 8.0 and 100 mM NaCl. For GbpA-D1, we used a salt gradient for elution during AEX from 50 to 400 mM and performed SEC using a Superdex 75 Increase 10/300 column (Cytiva) on an ÄKTA pure system (GE Healthcare).
Small-Angle X-ray Scattering. SAXS data were acquired on a Bruker NanoStar instrument using 40.0 μM H-GbpA-FL or 32.0 μM D-GbpA-FL in 100 mM NaCl, 20 mM Tris−HCl pH 8.0, with data acquisition times of 1 h per data set. Scattering intensities were recorded as a function of the scattering vector q = (4π/λ)sinθ, where 2θ is the scattering angle and λ is the wavelength (λ = 1.54 Å). Data were collected in the q-range: 0.009 to 0.3 Å −1 .
The scattering intensities were corrected for electronic noise, empty cell scattering, and detector sensitivity. The scattering contribution from the buffer was subtracted, and intensities were calibrated to absolute units with H 2 O scattering as standard,  26 ) were calculated with PRIMUS 27 from the ATSAS 28 package. For both H-GbpA-FL and D-GbpA-FL, 20 low-resolution models were calculated by ab initio shape determination using the DAMMIF 29 software. We built average models with DAMAVER 30 and refined them with DAMMIN. 31 All three programs are from the ATSAS 28 package. SAXS data are summarized in Table 2.
Small-Angle Neutron Scattering. H-GbpA was dialyzed into a buffer containing 100 mM NaCl and 20 mM Tris−HCl pH 8.0 in 100% D 2 O. D-GbpA was dialyzed into the same buffer, but using different D 2 O/H 2 O ratios, i.e., at 100% H 2 O, 45% D 2 O, or 100% D 2 O. SANS data for the four samples were acquired at ILL beamline D11 at λ = 5.6 Å for 2 h, using a protein concentration of 39.2 μM. Subsequently, the data were processed with beamline software ( Table 2).
X-ray Crystallography. Hydrogenated and deuterated GbpA-D1 were crystallized using the same protocol and conditions. First, the proteins were saturated with Cu 2+ by addition of CuCl 2 in a molar ratio of 3:1 (CuCl 2 to GbpA) and subsequently desalted to a buffer containing 100 mM NaCl, 20 mM Tris−HCl pH 8.0 using a 5 mL HiTrap desalting column. No crystals were obtained when the copper-binding step was omitted. Initial screening yielded crystals under many conditions, but most were either very small, irreproducible, or exclusive to either H-GbpA or D-GbpA. The following paragraph describes the procedure and conditions that yielded reproducible crystals in space group P2 1 2 1 2 for both proteins, which also diffracted to high resolution.
GbpA-D1 crystals grew from a solution containing the purified protein at 6−10 mg/mL. Sitting-drop vapor diffusion experiments were set up in 96-well 3-drop PS plates (SwissCI). 0.5 μL protein were added to an equal volume of crystallization solution containing 100 mM sodium cacodylate pH 6.5, 200 mM zinc acetate and 18% w/v PEG 8000 (VWR). Crystals grew over the course of two weeks at 20°C. Crystals were subsequently cryoprotected in mother liquor supplemented with 15% glycerol and flash-cooled in liquid nitrogen before data collection.
X-ray data were processed automatically by the EDNA processing pipeline 34 for H-GbpA-D1 and XIA2_DIALS 35 for D-GbpA-D1. In all subsequent steps, the CCP4 software suite 36 was used. Structures were solved by molecular replacement (with Phaser 37 ) using domain D1 of the published GbpA structure (PDB ID: 2XWX) 8 as a model. Real-space refinement and model building were performed with Coot, 38 and subsequent refinement cycles using REFMAC5. 39 Ions and water molecules were added only after the protein chain had been modeled. Finally, occupancies were refined for protein atoms and anomalous scattering ions. Zinc and copper ions were identified with confidence based on data collected at their absorption edges for anomalous scattering (Tables S1 and S2; Figure S3). Full anisotropic refinement was carried out for the 1.1 Å D-GbpA-D1 model, whereas this was not warranted for the lower-resolution H-GbpA-D1 structure (1.6 Å). The refined structures were deposited in the Protein Data Bank (PDB, www. rcsb.org) 40 with PDB IDs: 8CC3 and 8CC5. Data collection and refinement statistics are reported in Table 3.
To unambiguously identify the metal species observed in GbpA-D1, diffraction data were collected at the K absorption edge of zinc (around 9660 eV) and copper (around 8980 eV). Three data sets were collected at the BioMAX beamline of MAX IV (Lund, Sweden; Table S1) using isomorphous crystals grown under the same conditions as the H-GbpA-D1 and D-GbpA-D1 structures described above. The data were integrated and scaled by the autoPROC automatic processing pipeline at MAX IV, 41 and subsequently scaled and truncated to 2.5 Å with XSCALE, a component of the XDS software package. 42 The phases for the highest-resolution dataset (9320 eV) were determined by molecular replacement as described above, and the structure thereafter refined in iterative cycles of maximum-likelihood refinement using REFMAC5 39 and manual real-space refinement in Coot. 38 Data collection and refinement statistics for this dataset are given in Table S2.
Phase information from the 9320 eV refined structure was used to generate anomalous difference maps for each of the datasets (D ano 10.0k , D ano 9.3k , D ano 8.5k ) using the FFT tool from the CCP4 program suite. 36 Completeness of the datasets at a resolution higher than 3 Å was limited due to the presence of water ice (see Table S1); however, the anomalous signal at lower resolution was sufficient to allow identification of the positions and identities of the metal ions. Difference density maps of anomalous difference maps were generated using the CAD and FFT tools to find peaks corresponding to copper atoms (D ano maps revealed whether the position was occupied by copper, zinc, or a combination of both. The histidine brace motif only showed a peak in the D ano 9.3k and the D ano 9.3k − D ano 8.5k maps (copper absorption edge) but not in the D ano 10.0k − D ano 9.3k map calculated around the zinc absorption edge. This confirmed the exclusive presence of copper in its catalytic position. Activity Studies. GbpA was saturated with Cu 2+ as described in the previous section, and β-chitin nanofibers from France Chitine (Orange, France) were prepared according to a protocol developed by Loose et al. 9 1 μM GbpA-FL was mixed with 5 mg/mL β-chitin nanofibers in 20 mM Tris−HCl pH 8.0, and the reaction initiated by addition of 1 mM sodium ascorbate. After incubation on an INFORS shaker for 1.5 h at 37°C, the reaction was stopped by boiling and subsequent filtering through a 0.2 μm cellulose filter. The products were measured with MALDI-TOF MS. The experiment was performed in triplicates for both H-GbpA and D-GbpA.

■ RESULTS AND DISCUSSION
High-Yield Production of Deuterated GbpA. In order to optimize expression conditions for GbpA, we first carried out experiments using non-deuterated modified M9 medium, which we refer to as M9glyc+ (Table 1; recipe modified from refs 19,43). A plateau in optical density at 600 nm (OD 600 ) was reached at 12 AU after 12 h without induction ( Figure S1A). For comparison, expression in LB medium reached a plateau much earlier (after 5 h) but only at 4 AU ( Figure S1A). Expression of GbpA-FL in M9glyc+ was induced by adding 1 mM IPTG at four different cell densities, OD 600 = 0.7, 3.0, 7.2, and 11.0 ( Figure S1B). The optimal induction point was identified to be at OD 600 = 3.0 based on band intensities over background on SDS-PAGE ( Figure 1A). The protein was subsequently purified using a periplasmic isolation protocol, followed by AEX and SEC. The protein eluted at the same retention volume as GbpA-FL expressed in TB ( Figure S1C) and exhibited equivalent purity ( Figure S1D).
We then grew cell cultures containing GbpA-FL-encoding plasmid in deuterated M9glyc+ medium (Table 1), following a stepwise adaptation protocol, from LB/H 2 O to LB/D 2 O to D 2 O and glycerol-d 8 -containing minimal medium ( Figure 2). In the final step, cells were grown to OD 600 = 3.0, and expression was induced. Both H-GbpA-FL and D-GbpA-FL eluted as a single peak from SEC and were shown to be highly pure as evaluated by SDS-PAGE ( Figure 1B−D).
We found that using deuterated glycerol instead of the more expensive deuterated glucose used by Cai et al. 19 worked well. A protein yield of 12 mg purified protein from 1 L of expression media was calculated from absorbance at 280 nm. Surprisingly, the yield was approximately two-fold higher than from optimized expression in non-deuterated TB medium, as suggested by final protein contents of the purified samples as compared to the total volume of the expression cultures.
Subsequently, we applied the same protocol to GbpA-D1, in preparation of NMX experiments. A protein yield of 3 mg pure deuterated protein from 1 L of expression media was obtained, as calculated from absorbance at 280 nm (thus 4-fold less than for D-GbpA-FL). Purity was assessed with SDS-PAGE ( Figure  1D). The protocol, with small variations, was subsequently scaled up with improved yields by D-lab at ILL and DEMAX at ESS. The procedure for producing deuterated proteins used at DEMAX (ESS) is done in traditional shaker flasks and includes supplementing the deuterated glycerol in the media with deuterated algal extracts. The method described in this work serves as a proof-of-concept, and the optimized formulation and procedure is the subject of a separate publication that is in preparation. The procedures at D-Lab (ILL) use fermenters and as such require far less D 2 O (almost 3-fold less) than the shaker method, but the batch-fed approach uses more deuterated glycerol. Despite the fundamental differences in how the deuterium labeling is achieved, both D-Lab and DEMAX produced around 30 mg of D-GbpA-D1 each, indicating that the methods are equivalent for this protein.
An additional difference that could potentially affect the protein quality and yield was the handling after the protein production stage. Usually, the cells were immediately further processed for periplasmic protein extraction, but in one case, the cell paste was frozen and shipped prior to further extraction. This step may have reduced the final yield, but the purified    where MW dE and MW hE correspond to the experimentally determined masses of the deuterated and non-deuterated GbpA, respectively. This gives a deuteration level of 97%.
The remaining 3% undesired hydrogen may result from vapor exchange with the air or contamination from the various chemicals used (Table 1). However, 97% deuteration is more than sufficient for neutron-based experiments, and the neutron SLD of 6.12 × 10 −6 Å −2 (in H 2 O) for D-GbpA (compared to 1.97 × 10 −6 Å −2 for H-GbpA) is expected to be significantly distinguishable from all non-deuterated biomolecules. Interaction studies exploiting contrast matching should therefore be possible with any interaction partner with a defined SLD, most obviously chitin or mucin, but also potential bacterial cellsurface interaction partners or enzyme complex partners yet to be identified.
To confirm that the overall structure of GbpA-FL was unaffected by deuteration, we performed SAXS experiments for both D-GbpA and H-GbpA ( Figure 4A). Indeed, the SAXS profiles of both proteins were highly similar, with a radius of gyration (R g ) of 37.4 Å for H-GbpA and 36.8 Å for D-GbpA. Both proteins are monomers in solution, with a maximal diameter of approximately 140 Å ( Figure 4B). Structural parameters are summarized in Table 2. Ab initio modeling suggests similar, elongated structures of both hydrogenated and deuterated GbpA ( Figure 4D,E). In order to test the neutron scattering of D-GbpA at different concentrations of D 2 O, we carried out SANS experiments on D-GbpA under three different conditions ( Figure 4C), comparing the scattering in buffer at 100% H 2 O with 45% D 2 O (around the match point of nondeuterated proteins and carbohydrates) and 100% D 2 O (where perdeuterated proteins should have the lowest scattering, since the match point is above 100%). We saw that the form factor of D-GbpA was fairly consistent over the three different D 2 O concentrations and similar to that of H-GbpA, with comparable R g for all conditions. We also confirmed that D-GbpA scattered well at 45% D 2 O and that the D-GbpA match point was above 100% D 2 O, i.e., that D-GbpA could not be matched out even at 100% D 2 O due to its high deuteration level. Parameters from SANS are summarized in Table 2.
In order to verify that the catalytic activity was unaffected by deuteration, LPMO activity towards β-chitin was assessed by measuring the masses of chitooligosaccharide products by MS ( Figure 5). H-GbpA and D-GbpA yielded the same products, thus activity was maintained after deuteration. We refer to Loose et al. 9 for a more in-depth study of the activity of GbpA and the components involved in catalysis.
Crystal Structures of GbpA-D1 Show no Differences upon Deuteration. The LPMO domains (D1) of both deuterated and non-deuterated GbpA were crystallized ( Figure  S2), and X-ray diffraction data were collected. Both proteins crystallized in space group P2 1 2 1 2 with very similar cell parameters. P2 1 2 1 2 is a relatively high-symmetry space group, which is preferred for NMX, as less angular data are needed to obtain complete data sets. X-ray crystal structures were determined to 1.6 Å resolution and 1.1 Å resolution for H-GbpA-D1 and D-GbpA-D1 (sample produced at ILL), respectively. The high resolution for D-GbpA is especially promising, as crystal packing and diffraction quality are important for NMX.
Both proteins were crystallized in the catalytically active, copper-bound states ( Figure 6B), differing from the crystal structure of GbpA (domains 1−3; PDB ID: 2XWX) from Wong et al., 8 which lacked the copper ion in the active site. Analysis of anomalous scattering from data sets collected at the absorption edges of copper and zinc confirmed the active site to be exclusively occupied by copper ( Figure S3 and Tables S1 and S2). The conformation of the histidine brace, which is very similar among copper-free and copper-bound LPMO structures in the Protein Data Bank (see, e.g., PDB IDs: 6IF7, 45 6RW7, 46 and 5FTZ 47 ), is conserved in our structures. Table 3 summarizes the data collection and refinement statistics. Superimposition of the deuterated and hydrogenated GbpA-D1 structures yielded r.m.s.d. values of appr. 0.2 Å for C α atoms (Figure 6), showing that the fold is not affected by deuteration. In addition, when comparing our X-ray structures with the one from Wong et al. 8 for domains 1−3 of GbpA, the r.m.s.d. values are below 0.3 Å. The crystallization condition thus serves as a good starting point for further optimization for NMX.

■ CONCLUSIONS
Neutron scattering techniques have great potential to offer deep insights into the molecular structure and function of proteins in complex environments, providing an important complement to other structural biology techniques. Here, we have demonstrated the feasibility of perdeuteration for GbpA, a bacterial colonization factor with LPMO activity, resulting in yields and deuteration levels highly compatible with NMX and SANS without the need for specialized fermentation equipment. We report a new deuteration protocol based on algal extracts and showed that the protein investigated remains structurally uncompromised and active. In addition to first SANS studies of full-length GbpA (GbpA-FL), we succeeded in obtaining well-diffracting crystals of the GbpA LPMO domain (GbpA-D1) in both the deuterated and non-deuterated states. Whereas production of 70% deuterated LPMOs has been achieved before and used for SANS interaction studies, 48 this is to our knowledge the first time that perdeuteration of an LPMO has been achieved, and a more in-depth study of the effect of deuteration of LPMOs has been performed.
This work, alongside numerous other neutron crystallographic studies, further demonstrates the feasibility of NMX and the important role that perdeuteration plays in optimizing the quality of results. Although large crystals are needed for NMX, and optimization of crystal growth is still required, the possibility of using deuterated protein can reduce the volume requirements by up to a factor of 10. With better NMX structures of LPMOs, it is anticipated that the protonation states of key amino acids in and around the active sites will be revealed and that the importance of the water network within the LPMOs can be understood. Given the recently solved crystal structure of fulllength VhLPMO10A fromV. campbellii, a close homologue of GbpA, NMX studies of FL-GbpA and homologues may also become a possibility. 11 Neutron studies with SANS or NR can reveal how LPMOs interact with the carbohydrate substrates and how the LPMO structure adapts to carbohydrate surfaces or fibers.
Our results will thus further enable neutron-based studies of perdeuterated GbpA and LPMOs in general. In addition, other techniques that benefit from isotope labeling, such as NMR spectroscopy, may also benefit from the labeling procedure presented here. ). Most of the work was carried out at the UiO Structural Biology core facilities, which are part of the Norwegian Macromolecular Crystallography Consortium (NORCRYST) and which received funding from the Norwegian INFRASTRUKTUR-program (project no. 245828) as well as from UiO (core facility funds). SAXS experiments were performed at the Norwegian Centre for X-ray Diffraction, Scattering and Imaging (RECX), funded by the Norwegian INFRASTRUKTUR-program (project no. 208896). MS experiments were carried out at the UiO Proteomics Core Facility at the Department of Biosciences, which is a member of the National Network of Advanced Proteomics Infrastructure (NAPI), funded by the Norwegian INFRASTRUKTURprogram (project no. 295910).