SENJU of BL18 at MLF is a single-crystal neutron diffractometer with wide-area detectors, and it uses the time-of-flight Laue technique for structural analysis (Fig. 1) [1]. The integrated intensities measured for crystal structure and magnetic structure analyses are corrected for the Lorentz factor, incident spectrum and detector sensitivity difference. In addition, absorption correction is required for materials with large neutron absorption cross sections. Therefore, in order to perform numerical absorption correction based on the crystal shape, we developed a procedure to determine the Miller indices of polyhedral crystal faces using the photographs of crystals taken with a CCD camera and the UB matrix at SENJU [2]. When applied to DyNi3Al9 [3], structural analysis demonstrated that the absorption effect could be corrected.Figure 2 shows a single crystal taken with a CCD camera, and the outline of the crystal is defined by the vertices.

In this presentation, we will report the details of the definition of crystal planes and the results of crystal structure analysis.

SENJU at J-PARC is a time-of-flight (TOF) single-crystal neutron diffractometer designed for precise crystal and magnetic structure analyses under multiple extreme environments, such as low-temperature, high-pressure and high-magnetic field, as well as for taking diffraction intensities of small single crystals with a volume of less than 1.0 mm³ down to 0.1 mm³ [1]. We have recently upgraded some SENJU components, such as sample environment devices, the detector system, and data processing software. These upgrades of SENJU enhance the possibility and accessibility of SENJU, in other words, improvement of versatility. In this presentation, we will introduce the recent upgrades of SENJU for the improvement of its versatility.

A major advance in the sample environment is installing a radial oscillating collimator (ROC). The ROC can cut the neutron scattering from vacuum vessels of extreme sample environment devices, making low-background measurements with extreme conditions possible. By using the ROC, we can obtain low-background diffraction data with a dilution cryostat (T > 50 mK), a liquid-He free cryostat (T > 300 mK), a cryo-furnace (700 K > T > 10 K), a niobium-furnace (T < 1300 K), and a superconducting magnet (B < 6.8 T, T > 50 mK).

As an upgrade of the detector system, we added four area-detectors to SENJU in the obliquely downward direction of the sample position. The additional detectors can cover the blind region in the reciprocal space when measuring a low-symmetry sample and improve the measurement efficiency of low-symmetric molecular crystals.

A significant part of software upgrades is an improvement of accessibility. We have developed remote-access data processing software installed on a cloud computing system and works on various web browsers. This software will make remotely access to the measured data from users’ laboratories easy even in the COVID-19 situation.

Reference

[1] T. Ohhara, R. Kiyanagi, K. Oikawa, et al., J. Appl. Cryst., 49, 120-127 (2016).

Neutron single crystal diffraction provides an experimental method for the direct location of hydrogen and deuterium atoms in biological macromolecules, thus providing important complementary information to that gained by X-ray crystallography. At the FRM II neutron source in Garching near Munich the neutron single crystal diffractometer BIODIFF, a joint project of the Forschungszentrum Jülich and the FRM II, is dedicated to the structure determination of proteins. Typical scientific questions address the determination of protonation states of amino acid side chains, the orientation of individual water molecules and the characterization of the hydrogen bonding network between the protein active centre and an inhibitor or substrate. This knowledge is often crucial towards understanding the specific function and behaviour of an enzyme. BIODIFF is designed as a monochromatic diffractometer and is able to operate in the wavelength range of 2.4 Å to about 5.6 Å. This allows to adapt the wavelength to the size of the unit cell of the sample crystal. Data collection at cryogenic temperatures is possible, allowing studies of cryo-trapped enzymatic intermediates. Some recent examples will be presented to illustrate the potential of neutron macromolecular crystallography.

The DIALS project[1] provides an open-source, extensible framework to analyse X-ray diffraction data and is now used widely in the X-ray diffraction community. Much of this framework is in principle agnostic to the method used to obtain diffraction patterns. In recent years this has been expanded for stills shot serial crystallography and continuous-rotation electron diffraction experiments[2], for example, highlighting how DIALS can be adapted to cope with challenges from electron sources such as low diffraction angles and lens distortion. Continuing with this push towards generalised diffraction integration software, here we present preliminary results for processing time-of-flight neutron diffraction patterns obtained from the Single Crystal Diffractometer (SXD) at ISIS[3].

Neutron diffraction provides a complimentary technique to X-ray diffraction, exploiting nuclei interactions to render isotopes and light atoms that are generally absent from X-ray data. A lack of sample radiation damage further allows for the same crystal to be used for both X-ray and neutron diffraction, and such joint neutron/X-ray approaches are increasingly common for macromolecular crystallography[4]. By incorporating neutron diffraction into DIALS, we provide a common interface for users working with different diffraction sources to draw on the expanding toolbox of DIALS algorithms. This is further emphasised by providing the ability to convert DIALS output to formats readable by other related software, in this case Mantid[5], greatly increasing the transferability of different analysis tools. Here we show how DIALS can be used for the first time for time-of-flight neutron diffraction data, allowing Bragg peaks at a variety of wavelengths to be identified and visualised in the DIALS image (Fig. 1) and reciprocal lattice viewers, indexed, and refined to identify the sample space group. We also outline how DIALS refinement and integration is being adapted to cope with polychromatic data from pulsed neutron sources.

[1] Winter G., Waterman D. G., Parkhurst J. M., Brewster A. S., Gildea R. J., Gerstel M., Fuentes-Montero L., Vollmar M, Michels-Clark T., Young I. D., Sauter N. K., Evans G. (2017). Acta Crystallogr. D. 74, 85-97

[2] Clabbers T. B., Gruene T., Parkhurst J. M., Abrahams J. P., & Waterman D. G. (2018). Acta Crystallogr. D. 74, 506-518

[3] Keen D. A, Gutmann M. J., & Wilson C. C. (2006). J. Appl. Cryst. 39, 714-722

[4] Blakeley M. P. & Podjarny A. D. (2018). Emerg. Top. Life Sci. 2, 39-55

[5] Arnold O., et al. (2014). Nucl. Instrum. Methods Phys. Res. 764, 156-166

Lytic polysaccharide monooxygenases (LPMOs) are copper-center enzymes involved in the oxidative cleavage of the glycosidic bond. LPMOs are responsible for chain disruption of crystalline cellulose, thereby increasing the accessibility of the carbohydrate substrate to cellulases for hydrolytic depolymerization. The enhanced cellulose conversion of biomass due to addition of LPMOs makes them valuable for the generation of biofuels. The LPMO active site is located on the planar enzyme–cellulose binding surface in which a single copper ion is coordinated in a ‘histidine-brace’ motif composed of a N-terminal histidine and a second conserved histidine residue in the equatorial plane, with a coordinating tyrosine residue in the axial position. The LPMO reaction is initiated by the addition of a reductant and oxygen to ultimately form an unknown activated copper–oxygen species responsible for polysaccharide substrate hydrogen atom abstraction. Previous work in our group on LPMO9D from Neurospora crassa has provided insight into the binding and activation of oxygen at the LPMO active site as well as the role of the protonation state of a second-shell residue His 157 in oxygen-prebinding (O’Dell et al., 2017). The metallocenter of LPMO makes it highly susceptible to radiation damage, particularly photoreduction and radiolysis due to X-ray beam exposure. Neutron protein crystallography provides a non-destructive technique for structural characterization while also allowing the determination of the positions of light atoms such as hydrogen and deuterium which are central to understanding protein chemistry. Neutron cryo-crystallography permits trapping of catalytic intermediates, thereby providing insight into protonation states and chemical nature of otherwise short-lived species in the reaction mechanism. To this end, we collected a cryo-neutron diffraction dataset on an ascorbate-reduced LPMO9D crystal to characterize the reaction mechanism intermediates (Schröder et al., 2021). A second neutron diffraction dataset was collected at room temperature on a LPMO9D crystal exposed to low pH conditions to probe protonation states under acidic conditions.

References:

O’Dell, W. B., Agarwal, P. K. & Meilleur, F. (2017). Angew. Chemie - Int. Ed. 56, 767–770.

Schröder, G. C., O’Dell, W. B., Swartz, P. D. & Meilleur, F. (2021). Acta Crystallogr. Sect. F Struct. Biol. Commun. 77, 128–133.

“SENJU” is a TOF-Laue neutron single crystal diffractometer installed at J-PARC/MLF in Japan. This instrument is designed to study structures of inorganic materials and organic materials with relatively small cell sizes as well as magnetic structures. SENJU has 41 2D-detectors installed cylindrically surrounding the sample position. With the benefit of high intensity pulsed white neutrons generated at J-PARC, SENJU can measure a quite large 3D reciprocal space at once, which makes a quite efficient measurements possible. In order to make a good use of the efficient measurement, as a natural consequence, efficient data treatment is also demanded.

The experiments conducted at SENJU can be categorized into two types, one is for a standard structure analyses where a large number of Bragg reflections are collected, and the other is for a search for superlattice reflections including magnetic reflections where new reflections, which typically are weak, are searched within the observed 3D reciprocal space. For these two types of data, some mathematical methods are applied in order to efficiently treat the data.

1) Application of machine learning to Bragg reflection selection

In a measurement for a standard structure analysis, typically some hundred, or often more than ten thousands, of Bragg reflections are measured owing to the 2D-detectors and the white neutrons. The issue is that the measured data have to be checked before fed into a structure analysis software, because there can be ill reflections such as ones overlapped with a powder ring coming from, e.g., radiation shields or ones very close to each other. In order to eliminate such reflections, machine learning was adapted.

Several sets of “good” and “bad” data, including simulated ones, were prepared and used as a training data. With each training data, a model was constructed based on the convolutional neural network. Most of the trainings reached models with high accuracy, namely higher than 90%, and the models, indeed, were able to distinguish “good” and “bad” data from actual data with high probability. Further improvement is envisaged with increased number of the training data, especially “bad” data.

2) Application of kernel density estimation to reciprocal map

In a measurement for a search for superlattice reflections, the expected reflections are typically very weak. Hence longer exposure is needed and, often, even after the long exposure, the reflections could be blurry. In order to enhance the chance to find new reflections, improvement of the visibility of the data is one option. Therefore, the kernel density estimation (KDE) method was applied to a measured reciprocal map data.

It was found that the visibility of the reflections is greatly enhanced by the application of KDE. As shown in Figure 1, the reflections that are obscure in the raw data map became much clearer in the KDE map. In the KDE map, even the splitting of the peak can be recognized. The visibility was estimated to be increased by five times by KDE, which means that the application of KDE can be equal to five times longer exposure.

In the presentation, the details of the application of the machine learning and KDE will be shown, including the information about the codes used in the calculations.

The dynamic nuclear polarization method in neutron diffraction can increase the detection sensitivity of hydrogen. It is expected that the scattering length of hydrogen becomes about 8 times larger at maximum and high S/N ratio data can be obtained even with hydrogenated samples, not deuterated ones [1]. In order to realize the method, some radical molecules as an unpaired electron should be introduced into the sample, and high magnetic field (several T) and very low temperature (about 1 K) should be applied. In the previous study, the polarization ratio of 22.3% was obtained with lysozyme protein polycrystal in TEMPOL (4 - Hydroxy - 2, 2, 6, 6 - tetramethylpiperidine - 1 - oxyl) 50 mM with a normal-conducting magnet of 2.5 T under off-beam condition [2]. This time, to achieve a higher polarization rate and to obtain a diffraction image using a polarized neutron beam, a nuclear polarization experiment of protein polycrystal was conducted using a super-conducting 7 T magnet installed at BL20 in MLF in J-PARC [3].

Both lysozyme and TEMPOL were purchased from Merck. Lysozyme polycrystal was made from several 1.5 mL-solutions of 60 mg/mL lysozyme, 100 mM TEMPOL and 9 % (wt/vol) NaCl in 50 mM sodium acetate buffer of pH 4.5 by batch method. About 100 mg polycrystal was mixed with 30 % (wt/vol) glycerol, then it was sealed within a cell made from Teflon and quartz windows. Incident neutron was polarized to 93 % negatively. And sample was polarized at 1.2 K to 68 % positively, to 59 % negatively and to 0 % under equilibrium at 4.2K. The total exposure times were 4 hr, 2.5 hr and 1hr, respectively. The proton power was 600 kW, the applied magnetic field was 7 T, and the microwave frequency was 188 GHz.

According to I (Q) graph integrated from 4 to 9 Å, several powder diffraction peaks were observed at around 0.1-0.2 Å^-1 (Fig.1). Depending upon sample polarization rates, the different background levels and different peak intensities were observed clearly. If the lysozyme crystal space group is tetragonal form (P4₃2₁2), the maximum peaks around at q=0.1 Å^-1 were (110) reflections whose d-spacings were about 60 Å.

[1] Niimura, N. & Pojarny, A. (2011). Neutron Protein Crystallography. New York: Oxford University Press.

[2] Tanaka, I., Komatsuzaki, N., Yue, W.-X., Chatake, T., Kusaka, K., Niimura, N., Miura, D., Iwata, T., Miyachi, Y., Nukazuka, G. & Matsuda, H. (2018). Acta Cryst. D 74, 787.

[3] Noda, Y. & Koizumi, S. (2019). Nucl. Inst. and Meth. A 923, 127.

Keywords: dynamic nuclear polarization; protein crystal; neutron diffraction

Anharmonic movement is highly expected for the lighter atoms, as they are more influenced by the motion of their heavier parent atoms. However, because of low X-ray diffracting power of H-atoms, their motion is assumed as harmonic isotropic during routine structural studies or harmonic anisotropic during charge density studies with ADPs calculated in most cases. In this study we try to go one step further and check if it is possible to successfully refine H-atoms anharmonically.

For our study we performed neutron diffraction experiments on α-glycine (P2₁/n) high quality single crystals in 90 K and 200 K in ILL (Grenoble, France). With the high resolution of the data that we collected we tried to investigate whether anharmonic motion of H-atoms can be observed and modelled. Using Jana2020 we refined all H-atoms with Gram-Charlier coefficients up to the third level and then checked their 3D probability density function maps dependence on resolution. We performed Prince-Spiegelman test [1] to compare models with harmonic and anharmonic treatment of H-atoms to decide which model better fits the collected data, apart from simple comparison of refinement parameters like R₁, wR₂ and GooF.

At the end, to put H-atoms anharmonic motion refinements and observing changes in the model with varying resolution in perspective, we calculated lowest necessary resolution of the data collection according to Kuhs’ formula. It has been shown by Kuhs [2], that most information of the anharmonic motion of atoms is contained within higher angle reflections intensities. Our results agree with this assumption and the provided formula.