The fine details of the electron density distribution, typically far beyond the possibilities of standard X-ray diffraction analysis, can be approached when (ultra)high-resolution diffraction data are available, to allow abandoning the standard model of independent, spherically symmetrical atoms. This more complicated, and much more demanding (both experimentally and computationally) method allows, for instance, to analyze the redistribution of electron density into bonds, intermolecular interactions, etc. Moreover, the Atoms-In-Molecules approach, which is based on the analysis of topological features of high-quality electron density distribution, may offer an insight into the hierarchy of interactions, energetic features, etc. Even though such an approach is well-developed for small molecules, its application in macromolecular crystallography is still under development. Ultrahigh resolution in this case means resolution of at least 0.7 – 0.65 Å. Such data are extremely rare for macromolecular crystals. In addition, modelling problems, disorder, high solvent content, etc., severely limit the number of successful studies of experimental electron density distribution in macromolecules, which so far have been reported for proteins only (e.g. crambin [1], aldose reductase [2] and the high-potential iron–sulfur protein [3]).

For the present study, ultrahigh-resolution diffraction data (0.55 A) were collected for a Z-DNA hexamer with the sequence d(CGCGCG)₂. The results of the high-quality standard refinement [4] suggested that bonding and other features are visible in the difference electron density map (Fig). The quality of these data indeed allows the application of the more sophisticated multipolar model.

Fig. 1. The Watson–Crick base pair C3×G10 with the corresponding Fo map (blue, 3s) and difference Fo-Fc map (green, 2s) calculated without the contribution of H atoms.

The multipolar model has been successfully constructed and the drop in the R factor (and R free) seems to show that real experimental features have been included in this model. The topology of the electron density distribution has been analyzed and the intra- and intermolecular interactions characterized. A number of problems related to the multipolar approach, together with some expected and some unexpected results (e.g. unusual disorder of one of the C bases) will be presented. We will also consider the practical question concerning this kind of research: is the additional burden and investment of effort justified by the results?

[1] C. Jelsch, M. M. Teeter, V. Lamzin, V. Pichon-Pesme, R. H. Blessing, C. Lecomte. PNAS 97, 3171-3176 (2000).

[2] B. Guillot, C. Jelsch, A. Podjarny, C. Lecomte. Acta Cryst. D64, 567-588 (2008).

[3] Y. Hirano Y, K. Takeda, K. Miki. Nature 534, 281–284 (2016).

[4]K. Brzezinski, A. Brzuszkiewicz, M. Dauter, M. Kubicki, M. Jaskolski, Z. Dauter. Nucleic Acids Res. 39, 6238-6248 (2011).

One of the kinds of information gained from high resolution (sub-atomic) structures is the observation that electron density parameters are transferable between atoms having similar chemical topology. This stimulated creation of databases of multipolar pseudoatoms (Invariom [1], ELMAM2 [2], MATTS – successor of UBDB [3], etc.) and their applications in (a) structure refinements on standard (atomic) resolution data for small-molecule crystals, and (b) electrostatic properties and non-covalent bonding characterisations for macromolecules.

Transferable Aspherical Atom Model (TAAM) of scattering built from a pseudoatom database proved to be advantageous in refining the structure on X-ray diffraction (XRD) data compared to the Independent Atom model (IAM) [4], leading to better fit of the model to the data and improved localization of hydrogen atoms. We have recently showed [5] that also for small-molecule 3D electron diffraction (3D ED) data, a better model-to-data fit and more accurate structures should be expected from TAAM.

To improve its usability, we further extended the MATTS bank to cover 98% of atoms found in all the structures deposited in the Cambridge Structural Database [6] composed of chemical elements like C, H, N, O, P, S, F, Cl and/or Br. It is planned that the remaining 1% will be covered by the more general atom types resulting from multidimensional cluster analysis.

Some benefits of TAAM over IAM refinements were also reported for macromolecular XRD data of 0.9 Å resolution and better [7]. As most macromolecular crystals diffract to lower resolutions, we recently moved our investigation towards near-atomic resolutions. We quantified the differences between the macromolecular electron density Fourier maps obtained with TAAM and IAM, calculated with a resolution of 1.8 Å. We did the same for electrostatic potential maps, a key property in the context of 3D ED.

TAAM refinements affect not only the positions of the atoms, but also the atomic displacement parameters (ADPs) [8]. ADPs appears to be less resolution dependent with TAAM than with IAM. With IAM, ADPs increased for XRD and decreased for 3D ED by about 30%, when the resolution was reduced from 0.6 Å to 0.8 Å [5]. From modified Wilson plots we recently predicted, and then verified by TAAM refinements on macromolecular XRD or 3D ED data, what will happen with ADPs (B-factors) with a further resolution worsening, up to 1.8 Å.

All the above helps to understand if there will be any benefits of TAAM refinements on lower than atomic resolutions.

[1] Dittrich, B., Hübschle, C. B., Pröpper, K., Dietrich, F., Stolper, T. & Holstein, J. (2013). Acta Crystallogr. B 69, 91. [2] Domagała, S., Fournier, B., Liebschner, D., Guillot, B. & Jelsch, C. (2012). Acta Crystallogr. A 68, 337. [3] Kumar, P., Gruza, B., Bojarowski, S. A. & Dominiak, P. M. (2019). Acta Crystallogr. A 75, 398. [4] Jha, K. K., Gruza, B., Kumar, P., Chodkiewicz, M. L. & Dominiak, P. M. (2020). Acta Crystallogr. B 76, 296. [5] Gruza, B., Chodkiewicz, M., Krzeszczakowska, J. & Dominiak, P. M. (2020). Acta Crystallogr. A 76, 92. [6] Groom, C. R., Bruno, I. J., Lightfoot, M. P. & Ward, S. C. (2016). Acta Crystallogr. B 72, 171. [7]. Malinska, M. & Dauter, Z. (2016). Acta Crystallogr. D 72, 770. [8]. Sanjuan-Szklarz, F. W., Woińska, M., Domagała, S., Dominiak, P. M., Grabowsky, S., Jayatilaka, D., Gutmann, M., Woźniak, K. (2020). IUCrJ, 7, 920.

Keywords: quantum crystallography; structure refinement; X-ray diffraction, electron diffraction; 3D ED, microED, aspherical scattering factors; multipolar model; TAAM; MATTS; UBDB; ELMAM2

Support of this work by the National Centre of Science (Poland) through grant OPUS No.UMO-2017/27/B/ST4/02721 and PL-Grid through grant plgubdb2020 is acknowledged.

Lipases (E.C. 3.1.1.3) are ubiquitous hydrolases for the carboxyl ester bond of water-insoluble substrates such as triacylglycerols, phospholipids, and other insoluble substrates, acting in aqueous as well as in low-water media, thus being of considerable physiological significance with high interest also for their industrial applications. The hydrolysis reaction follows a two-step mechanism, or ‘interfacial activation’, with adsorption of the enzyme to a heterogeneous interface and subsequent enhancement of the lipolytic activity. Among lipases, Candida antarctica Lipase B (CALB) has never shown any significant interfacial activation, and a closed conformation of CALB has never been reported leading to the conclusion that its behaviour was due to the absence of a lid regulating the access to the active site. The lid open and closed conformations and their protonation states are observed in the crystal structure of CALB at 0.91 Å resolution [1]. Having the open and closed states at atomic resolution allows relating protonation to the conformation, indicating the role of Asp145 and Lys290 in the conformation alteration. Once positioned within the catalytic triad, substrates are then hydrolysed, and products released. However, the intermediate steps of substrate transfer from the lipidic-aqueous phase to the enzyme surface and then down to the catalytic site are still unclear. By inhibiting CALB with ethyl phosphonate and incubating with glyceryl tributyrate (2,3-di(butanoyloxy)propyl butanoate), the crystal structure of the lipid-enzyme complex, at 1.55 Å resolution, shows the tributyrin in the limbus region of active site [2]. The substrate is found above the catalytic Ser, with the glycerol backbone readily pre-aligned for further processing by key interactions via an extended water network with α-helix10 and α-helix5. These findings explain the lack of ‘interfacial activation’ of CALB and offer new elements to elucidate the mechanism of substrate recognition, transfer and catalysis of Candida antarctica Lipase B (CALB) and lipases in general.

[1] Stauch, B., Fisher, S. J., Cianci, M. (2015). Journal of Lipid Research, 56, 2348-2358.

[2] Silvestrini, L. & Cianci, M. (2020). International Journal of Biological Macromolecules, 158, 358-363.

Tryparedoxins are critical regulators of the redox metabolism in parasitic protozoa such as trypanosomones and leishmania, which cause the neglected tropical diseases sleeping sickness and leishmaniosis, respectively. Although tryparedoxins belong to the thioredoxin superfamily, they differ in their substrate specificity for the low molecular weight redox carrier, utilizing trypanothione, a spermidine-linked di-glutathione instead of glutathione. The unique nature of the redox carrier opens avenues for the targeted interference in the protozoan redox metabolism which hold potential for future therapeutic intervention.

We were able to obtain crystals of a tryparedoxin which have the potential to diffract to ultra-high resolution. Crystals of the oxidized protein diffract to well below 1 Å with our current highest resolution data set extending to 0.75 Å/0.85 Å resolution in the best/worst direction. Preliminary refinement indicated a mixture of oxidized and reduced states in the Cys-Pro-Pro-Cys active site with photoreduction of the disulfide bond being the reason for the appearance of the reduced state. Subsequent efforts resulted in crystal structures of the oxidized and reduced states, which at present extend to a more limited resolution in the 1 to 1.1 Å range. An analysis of the two redox states defines the redox linked conformational changes in the tryparedoxin family.

The maps of electrostatic potential from cryo-electron microscopy and micro-electron diffraction are now being obtained at atomic resolution. This extends the possibility of investigating the electrostatic potential beyond determining the non-hydrogen atom positions, taking into account also the negative regions of the maps. However, accurate tools to calculate this potential for macromolecules, without reaching to the expensive quantum calculations, are lacking. Simple point charges or spherical models do not provide enough accuracy. Here, we apply the multipolar electron scattering factors and investigate the theoretically-obtained potential maps.

The multipolar electron scattering factors are derived from the aspherical atom types from Multipolar Atom Types from Theory and Statistical clustering (MATTS) databank (successor of UBDB2018 [1]). MATTS has been created since electron densities of atom types are transferable between different molecules in similar chemical environment. These atom types can be used to recreate the electron density distribution of macromolecules via structure factors [2] and to calculate the accurate electrostatic potential maps for small molecules [3]. MATTS reproduces the molecular electrostatic potential of molecules within their entire volume better than the simple point charge models used in molecular mechanics or neutral spherical models used in electron crystallography. In this study, we calculate electrostatic potential maps for several chosen macromolecules using aspherical atom databank and compare them with experimental maps from cryo-electron microscopy and micro-electron diffraction at high resolution. Calculations at different resolutions reveal at which spatial frequencies different elements become discernible. We also consider the influence of atomic displacement parameters on the theoretical maps as their physical meaning in cryo-electron microscopy is not as well established as in X-ray crystallography.

This study could potentially pave the way for distinguishing between different ions/water molecules in the active sites of macromolecules in high resolution structures, which is of interest for drug design purposes. It could also facilitate the interpretation of the less-resolved regions of the maps and also advise in simple yet questionable issue of resolution definition in cryo-electron microscopy.

The authors acknowledge NCN UMO-2017/27/B/ST4/02721 grant.

References:

[1] Kumar et al. (2019). Acta Cryst. A75, 398-408

[2] Chodkiewicz et al. (2018). J. Appl. Cryst. 51, 193-199

[3] Gruza et al. (2019), Acta Cryst. A76, 92-109