Accuracy of AlphaFold models: Comparison with short N O contacts in atomic resolution protein crystal structures.

Artificial intelligence (AI) has revolutionized structural biology by predicting protein 3D structures with near-experimental accuracy. Here, short backbone N-O distances in high-resolution crystal structures were compared to those in three-dimensional models based on AI AlphaFold/ColabFold, specifically considering their estimated standard errors. Experimental and computationally modeled distances very often differ significantly, showing that these models' precision is inadequate to reproduce experimental results at high resolution. T-tests and normal probability plots showed that these computational methods predict atomic position standard errors 3.5-6 times bigger than experimental errors. SYNOPSIS: Positional standard errors in AI-based protein 3D models are 3.5-6 times larger than in atomic resolution crystal structures.

Location of S-nitrosylated cysteines in protein three-dimensional structures.

Although S-nitrosylation of cysteines is a common protein posttranslational modification, little is known about its three-dimensional structural features. This paper describes a systematic survey of the data available in the Protein Data Bank. Several interesting observations could be made. (1) As a result of radiation damage, S-nitrosylated cysteines (Snc) are frequently reduced, at least partially. (2) S-nitrosylation may be a protection against irreversible thiol oxidation; because the NO group of Snc is relatively accessible to the solvent, it may act as a cork to protect the sulfur atoms of cysteines from oxidation by molecular oxygen to sulfenic, sulfinic, and sulfonic acid; moreover, Snc are frequently found at the start or end of helices and strands and this might shield secondary structural elements from unfolding.

Automated identification of chalcogen bonds in AlphaFold protein structure database files: is it possible?

Protein structure prediction and structural biology have entered a new era with an artificial intelligence-based approach encoded in the AlphaFold2 and the analogous RoseTTAfold methods. More than 200 million structures have been predicted by AlphaFold2 from their primary sequences and the models as well as the approach itself have naturally been examined from different points of view by experimentalists and bioinformaticians. Here, we assessed the degree to which these computational models can provide information on subtle structural details with potential implications for diverse applications in protein engineering and chemical biology and focused the attention on chalcogen bonds formed by disulphide bridges. We found that only 43% of the chalcogen bonds observed in the experimental structures are present in the computational models, suggesting that the accuracy of the computational models is, in the majority of the cases, insufficient to allow the detection of chalcogen bonds, according to the usual stereochemical criteria. High-resolution experimentally derived structures are therefore still necessary when the structure must be investigated in depth based on fine structural aspects.

Structural biology: A golden era.

In the past 2 decades, structural biology has transformed from a single technique used on single proteins to a multimodal integrative approach. Recently, protein structure prediction algorithms have opened new avenues to address challenging biological questions.

Interplay between hydrogen and chalcogen bonds in cysteine.

Protein structures are stabilized by several types of chemical interactions between amino acids, which can compete with each other. This is the case of chalcogen and hydrogen bonds formed by the thiol group of cysteine, which can form three hydrogen bonds with one hydrogen acceptor and two hydrogen donors and a chalcogen bond with a nucleophile along the extension of the CS bond. A survey of the Protein Data Bank shows that hydrogen bonds are about 40-50 more common than chalcogen bonds, suggesting that they are stronger and, consequently, prevail, though not always. It is also observed that frequently a thiol group that forms a chalcogen bond is also involved, as a hydrogen donor, in a hydrogen bond.

Chalcogen bonds formed by protein sulfur atoms in proteins. A survey of high-resolution structures deposited in the protein data bank.

The presence of chalcogen bonds in native proteins was investigated on a non-redundant and high-resolution (≤ 1 Angstrom) set of protein crystal structures deposited in the Protein Data Bank. It was observed that about one half of the sulfur atoms of methionines and disulfide bridges from chalcogen bonds with nucleophiles (oxygen and sulfur atoms, and aromatic rings). This suggests that chalcogen bonds are a non-bonding interaction important for protein stability. Quite numerous chalcogen bonds involve water molecules. Interestingly, in the case of disulfide bridges, chalcogen bonds have a marked tendency to occur along the S-S bond extension rather than along the C-S bond extension. Additionally, it has been observed that closer residues have a higher probability of being connected by a chalcogen bonds, while the secondary structure of the two residues connected by a chalcogen bond do not correlate with its formation.Communicated by Ramaswamy H. Sarma.

Survey of the Intermolecular Disulfide Bonds Observed in Protein Crystal Structures Deposited in the Protein Data Bank.

About 5% of the disulfide bonds (DBs) observed in the Protein Data Bank bridge two protein chains. Several of their features were comprehensively analyzed, resulting in a structural atlas of the intermolecular DBs. The analysis was performed on a very large set of data extracted from the Protein Data Bank, according to the RaSPDB procedure. It was observed that the two chains tend to have different sequences and belong to the same structural class. Intermolecular DBs tend to be more solvent accessible and less distorted from the most stable conformation than intermolecular DBs while showing similar B-factors. They tend to occur in beta strands and in mainly-beta structures. These and other data should prove useful in protein modelling and design.

Uses and Abuses of the Atomic Displacement Parameters in Structural Biology.

B-factors determined with X-ray crystallographic analyses are commonly used to estimate the flexibility degree of atoms, residues, and molecular moieties in biological macromolecules. In this chapter, the most recent studies and applications of B-factors in protein engineering and structural biology are briefly summarized. Particular emphasis is given to the limitations in using B-factors, in order to prevent inappropriate applications. It is eventually predicted that future applications will involve anisotropically refined B-factors, deep learning, and data produced by cryo-EM.

B-factor accuracy in protein crystal structures.

The accuracy of B factors in protein crystal structures has been determined by comparing the same atoms in numerous, independent crystal structures of Gallus gallus lysozyme. Both B-factor absolute differences and normal probability plots indicate that the estimated B-factor errors are quite large, close to 9 Å2 in ambient-temperature structures and to 6 Å2 in low-temperature structures, and surprisingly are comparable to values estimated two decades ago. It is well known that B factors are not due to local movements only but reflect several, additional factors from crystal defects, large-scale disorder, diffraction data quality etc. It therefore remains essential to normalize B factors when comparing different crystal structures, although it has clearly been shown that they provide useful information about protein dynamics. Improved, quantitative analyses of raw B factors require novel experimental and computational tools that are able to disaggregate local movements from other features and properties that affect B factors.

Random sampling of the Protein Data Bank: RaSPDB.

A novel and simple procedure (RaSPDB) for Protein Data Bank mining is described. 10 PDB subsets, each containing 7000 randomly selected protein chains, are built and used to make 10 estimations of the average value of a generic feature F-the length of the protein chain, the amino acid composition, the crystallographic resolution, and the secondary structure composition. These 10 estimations are then used to compute an average estimation of F together with its standard error. It is heuristically verified that the dimension of these 10 subsets-7000 protein chains-is sufficiently small to avoid redundancy within each subset and sufficiently large to guarantee stable estimations amongst different subsets. RaSPDB has two major advantages over classical procedures aimed to build a single, non-redundant PDB subset: a larger fraction of the information stored in the PDB is used and an estimation of the standard error of F is possible.

Decline of protein structure rigidity with interatomic distance.

BACKGROUND: Protein structural rigidity was analyzed in a non-redundant ensemble of high-resolution protein crystal structures by means of the Hirshfeld test, according to which the components (uX and uY) of the B-factors of two atoms (X and Y) along the interatomic direction is related to their degree of rigidity: the atoms may move as a rigid body if uX = uY and they cannot if uX ≠ uY. RESULTS: It was observed that the rigidity degree diminishes if the number of covalent bonds intercalated between the two atoms (d_seq) increases, while it is rather independent on the Euclidean distance between the two atoms (d): for a given value of d_seq, the difference between uX and uY does not depend on d. No additional rigidity decline is observed when d_seq ≥ ~ 30 and this upper limit is very modest, close to 0.015 Å. CONCLUSIONS: This suggests that protein flexibility is not fully described by B-factors that capture only partially the wide range of distortions that proteins can afford.

Chalcogen Bonds Involving Selenium in Protein Structures.

Chalcogen bonds are the specific interactions involving group 16 elements as electrophilic sites. The role of chalcogen atoms as sticky sites in biomolecules is underappreciated, and the few available studies have mostly focused on S. Here, we carried out a statistical analysis over 3562 protein structures in the Protein Data Bank (PDB) containing 18 266 selenomethionines and found that Se···O chalcogen bonds are commonplace. These findings may help the future design of functional peptides and contribute to understanding the role of Se in nature.

How anisotropic and isotropic atomic displacement parameters monitor protein covalent bonds rigidity: isotropic B-factors underestimate bond rigidity.

Under the assumption that covalent bonds are rigid, it is possible to compare the estimations of rigidity based on anisotropic and isotropic B-factors. This is done by computing the difference of the mean-square displacements (Delta-u) of atoms A and Z along the covalent bond A-Z, which must be close to zero for a rigid bond. The analysis of a high-quality set of protein structures, refined at a resolution better than (or equal to) 0.8 Angstroms, showed that Delta-u is significantly close to zero when anisotropic B-factors are used, with an average 60% Delta-u reduction. This reduction is larger for larger B-factors and this suggests that care should be taken in data-mining procedures that involve isotropic B-factors, especially at lower resolution, when anisotropic B-factors cannot be determined and when the average B-factor increases.

Lithium-Protein Interactions: Analysis of Lithium-Containing Protein Crystal Structures Deposited in the Protein Data Bank.

BACKGROUND: Despite the fact that lithium is not a biologically essential metallic element, its pharmacological properties are well known and human exposure to lithium is increasingly possible because of its used in aerospace industry and in batteries. OBJECTIVE: Lithium-protein interactions are therefore interesting and the surveys of the structures of lithium-protein complexes is described in this paper. METHODS: A high quality non-redundant set of lithium containing protein crystal structures was extracted from the Protein Data Bank and the stereochemistry of the lithium first coordination sphere was examined in detail. RESULTS: Four main observations were reported: (i) lithium interacts preferably with oxygen atoms; (ii) preferably with side-chain atoms; (iii) preferably with Asp or Glu carboxylates; (iv) the coordination number tends to be four with stereochemical parameters similar to those observed in small molecules containing lithium. CONCLUSION: Although structural information on lithium-protein, available from the Protein Data Bank, is relatively scarce, these trends appears to be so clear that one may suppose that they will be confirmed by further data that will join the Protein Data Bank in the future.

Mobility of water and of protein atoms at the protein-water interface, monitored by anisotropic atomic displacement parameters, are largely uncorrelated.

A non-redundant set of 231 protein crystal structures refined at a resolution better than (or equal to) 1 Å was extracted from the Protein Data Bank and the degree of conformational rigidity at the protein-water interface was examined by means of the Hirshfeld test and by comparing the orientations of the anisotropic Us for contacting protein and water atoms. Contacts between protein and water atoms are more rigid that contacts between water atoms and the degree of rigidity increases for shorter contacts and for more hydrogen-bonded atoms. Nevertheless, water and protein atoms are not rigidly held together. On the contrary, they seem to have little influence on their mobility to such an extent that hydration water, different from the protein atoms, cannot be considered to be properly in the solid state.

Anisotropic waters in atomic resolution protein crystal structures.

In protein crystal structures at extremely high resolution, B-factors of water oxygen atoms can be refined anisotropically. Here, some properties and trends associated with the anisotropy of the water oxygen B-factors are presented and commented. Anisotropy, defined as the ratio of the highest and the smallest eigenvalue of the anisotropic B-factor, is very variable and its distribution can be described by a Gumpel function, which slightly depends on the equivalent isotropic B-factor. Moreover, water oxygen atom anisotropies are very similar in the first and in the second hydration layers; are nearly independent of the number of hydrogen bonds that are formed by the water molecule; and are weekly correlated with protein atom anisotropy. This suggests that hydration water molecules might be in a physicochemical state intermediate between the liquid state, which is present and abundant in the crystal, and the solid state, which is assumed by protein. Further studies and analyses are apparently necessary to investigate this hypothesis.

Globular protein backbone conformational disorder in crystal structures.

Proteins are not static molecules but dynamic entities able to modify their structure for several reasons, from the necessity to recognize partners to the regulation of their thermodynamic stability. Conformational disorder is frequent in protein structures and atoms can have, in protein crystal structures, two or more alternative, equilibrium positions close to each other. Here, a set of protein crystal structures refined at very high resolution (1 Å or better) is examined to characterize the conformational disorder of the backbone atoms, which is not infrequent: about 15% of the protein backbone atoms are conformationally disordered and three quarters of them have been deposited with two or more equilibrium positions (most of the others were not detected in the electron density maps). Several structural features have been examined and it was observed that Cα atoms tend to be disordered more frequently than the other backbone atoms, likely because their disorder is induced by disordered side chains: side-chain disorder is two times more frequent than backbone disorder. Surprisingly, backbone disorder is only slightly more frequent in loops than in helices and strands and this is in agreement with the observation that backbone disorder is a localized phenomenon: in about 80% of the cases, it is observed in one amino acid and not in its neighbors. However, although backbone disorder does not cluster along the polypeptide sequence, it tends to cluster in 3D, since backbone-disordered amino acids distant in sequence are close in the 3D space.

Structural features of uranium-protein complexes.

Uranium toxicity depends on its chemical properties rather than on its radioactivity and involves its interaction with macromolecules. Here, a systematic survey of the structural features of the uranyl sites observed in protein crystal structures deposited in the Protein Data Bank is reported. Beside the two uranyl oxygens, which occupy the axial positions, uranium tends to be coordinated by five other oxygen atoms, which occupy the equatorial vertices of a pentagonal bipyramid. Even if one or more of these equatorial positions are sometime empty, they can be occupied only by oxygen atoms that belong to the carboxylate groups of Glu and Asp side-chains, usually acting as monodentate ligands, to water molecules, or to acetate anions. Although several uranium sites appear undefined or unrefined, with a single uranium atom that lacks the two uranyl oxygen atoms, this problem seems to become less frequent in recent years. However, it is clear that the crystallographic refinements of the uranyl sites are not always well restrained and a better parametrization of these restraints seems to be necessary.

Atomic displacement parameters in structural biology.

Atomic displacement parameters (ADPs, also known as B-factors), which depend on structural heterogeneity, provide a wide spectrum of information on protein structure and dynamics and find several applications, from protein conformational disorder prediction to protein thermostabilization, and from protein folding kinetics prediction to protein binding sites prediction. A crucial aspect is the standardization of the ADPs when comparisons between two or more protein crystal structures are made, since ADPs are differently affected by several factors, from crystallographic resolution to refinement protocols. A potential limitation to ADP analysis is the modern tendency to let ADPs to inflate up to extremely large values that have little physico-chemical meaning.

How large B-factors can be in protein crystal structures.

BACKGROUND: Protein crystal structures are potentially over-interpreted since they are routinely refined without any restraint on the upper limit of atomic B-factors. Consequently, some of their atoms, undetected in the electron density maps, are allowed to reach extremely large B-factors, even above 100 square Angstroms, and their final positions are purely speculative and not based on any experimental evidence. RESULTS: A strategy to define B-factors upper limits is described here, based on the analysis of protein crystal structures deposited in the Protein Data Bank prior 2008, when the tendency to allow B-factor to arbitrary inflate was limited. This B-factor upper limit (B_max) is determined by extrapolating the relationship between crystal structure average B-factor and percentage of crystal volume occupied by solvent (pcVol) to pcVol =100%, when, ab absurdo, the crystal contains only liquid solvent, the structure of which is, by definition, undetectable in electron density maps. CONCLUSIONS: It is thus possible to highlight structures with average B-factors larger than B_max, which should be considered with caution by the users of the information deposited in the Protein Data Bank, in order to avoid scientifically deleterious over-interpretations.