Characterization of cysteine thiol modifications based on protein microenvironments and local secondary structures

Abstract

We have demonstrated earlier that protein microenvironments were conserved around disulfide-bridged cystine motifs with similar functions, irrespective of diversity in protein sequences. Here, cysteine thiol modifications were characterized based on protein microenvironments, secondary structures and specific protein functions. Protein microenvironment around an amino acid was defined as the summation of hydrophobic contributions from the surrounding protein fragments and the solvent molecules present within its first contact shell. Cysteine functions (modifications) were grouped into enzymatic and non-enzymatic classes. Modifications studied were-disulfide formation, thio-ether formation, metal-binding, nitrosylation, acylation, selenylation, glutathionylation, sulfenylation, and ribosylation. 1079 enzymatic proteins were reported from high-resolution crystal structures. Protein microenvironments around cysteine thiol, derived from above crystal structures, were clustered into 3 groups-buried-hydrophobic, intermediate and exposed-hydrophilic clusters. Characterization of cysteine functions were statistically meaningful for 4 modifications (disulfide formation, thioether formation, sulfenylation, and iron/zinc binding) those have sufficient amount of data in the current dataset. Results showed that protein microenvironment, secondary structure and protein functions were conserved for enzymatic cysteine functions, in contrast to the same function from non-enzymatic cysteines. Disulfide forming enzymatic cysteines were tightly packed within intermediate protein microenvironment cluster, have alpha-helical conformation and mostly belonged to CxxC motif of electron transport proteins. Disulfide forming non-enzymatic cysteines did not belong to conserved motif and have variable secondary structures. Similarly, enzymatic thioether forming cysteines have conserved microenvironment compared to non-enzymatic cystienes. Based on the compatibility between protein microenvironment and cysteine modifications, more efficient drug molecules could be designed against cysteine-related diseases.

Computational Structural Biology of S-nitrosylation of Cancer Targets

“…These different pathways do not necessarily take place in the same cellular and protein environment. Thus, it is especially difficult to describe the microenvironment properties driving the specificity of S -nitrosylation sites ( 4 , 14 , 119 , 120 ).…”

Crystal structure of human CRM1, covalently modified by 2-mercaptoethanol on Cys528, in complex with RanGTP

“…However, the pK a value can be significantly affected by the microenvironment of the cysteine (Bhatnagar & Bandyopadhyay, 2018;Klomsiri et al, 2011). For example, metal-binding enzymatic cysteines were shown to exhibit a lower range of pK a values (8.1 AE 2.2) when buried in a hydrophobic cluster (Bhatnagar & Bandyopadhyay, 2018). Furthermore, BME exhibits decreased stability as the pH increases, which can lead to the formation of covalent adducts with surface cysteines (Wingfield, 1995).…”

DeepCys: Structure‐based multiple cysteine function prediction method trained on deep neural network: Case study on domains of unknown functions belonging to COX2 domains

“…63 The keyword sea- The feature, pKa, has been identified as one of the important parameters for cysteine function predictions. 51,54 However, pKa computed using PROPKA has a predefined value of 99.99 for disulphide connectivity. This fixed pKa value for disuphide from PROPKA makes the deep learning model circular in nature.…”

“…Earlier we have annotated four cysteine functions, namely, disulphide, thioether, metal-binding, and sulphenylation, based on protein structural properties, like, buried fraction, quantitative microenvironment descriptor (rHpy), secondary structure, and pKa values. [54][55][56] Only these four modifications were chosen because of their abundance in PDB crystal structures. Inspired by these functional annotations of cysteine, here we propose a deep neural network-based model, DeepCys, that exploits six different protein features, and predict any one of these four different cysteine modifications.…”