RiSLnet: Rapid identification of smart mutant libraries using protein structure network. Application to thermal stability enhancement.

A key point of protein stability engineering is to identify specific target residues whose mutations can stabilize the protein structure without negatively affecting the function or activity of the protein. Here, we propose a method called RiSLnet (Rapid identification of Smart mutant Library using residue network) to identify such residues by combining network analysis for protein residue interactions, identification of conserved residues, and evaluation of relative solvent accessibility. To validate its performance, the method was applied to four proteins, that is, T4 lysozyme, ribonuclease H, barnase, and cold shock protein B. Our method predicted beneficial mutations in thermal stability with ~62% average accuracy when the thermal stability of the mutants was compared with the ones in the Protherm database. It was further applied to lysine decarboxylase (CadA) to experimentally confirm its accuracy and effectiveness. RiSLnet identified mutations increasing the thermal stability of CadA with the accuracy of ~60% and significantly reduced the number of candidate residues (~99%) for mutation. Finally, combinatorial mutations designed by RiSLnet and in silico saturation mutagenesis yielded a thermally stable triple mutant with the half-life (T 1/2 ) of 114.9 min at 58°C, which is approximately twofold higher than that of the wild-type.

Rational engineering of ornithine decarboxylase with greater selectivity for ornithine over lysine through protein network analysis.

Ornithine decarboxylase (ODC) converts C5 ornithine into C4 putrescine, a monomer for polyamide synthesis. However, ODC also has minor activity towards cell metabolite C6 lysine and yields C5 cadaverine. The accumulation of cadaverine in the reaction solution causes increase in the operational cost of subsequent distillation process for putrescine purification. Here, to increase ODC substrate specificity toward ornithine over lysine, molecular modelling and protein network analysis, specifically k-clique community analysis, around the substrate tunnel were performed. This resulted in a mutant with two-fold increase in substrate specificity (ornithine versus lysine) without losing its original activity towards ornithine (kcat/KM  = 61.5 s-1  mM-1), compared to the native enzyme. When this mutant was used for putrescine synthesis, 31.6 g/L putrescine (based on 51.5 g/L ornithine) titer was achieved, while 0.007 g/L (based on 2.57 g/L lysine) cadaverine was produced. This corresponds to four-fold decrease in cadaverine yield compared to the native ODC.

Interaction signatures stabilizing the NAD(P)-binding Rossmann fold: a structure network approach.

The fidelity of the folding pathways being encoded in the amino acid sequence is met with challenge in instances where proteins with no sequence homology, performing different functions and no apparent evolutionary linkage, adopt a similar fold. The problem stated otherwise is that a limited fold space is available to a repertoire of diverse sequences. The key question is what factors lead to the formation of a fold from diverse sequences. Here, with the NAD(P)-binding Rossmann fold domains as a case study and using the concepts of network theory, we have unveiled the consensus structural features that drive the formation of this fold. We have proposed a graph theoretic formalism to capture the structural details in terms of the conserved atomic interactions in global milieu, and hence extract the essential topological features from diverse sequences. A unified mathematical representation of the different structures together with a judicious concoction of several network parameters enabled us to probe into the structural features driving the adoption of the NAD(P)-binding Rossmann fold. The atomic interactions at key positions seem to be better conserved in proteins, as compared to the residues participating in these interactions. We propose a "spatial motif" and several "fold specific hot spots" that form the signature structural blueprints of the NAD(P)-binding Rossmann fold domain. Excellent agreement of our data with previous experimental and theoretical studies validates the robustness and validity of the approach. Additionally, comparison of our results with statistical coupling analysis (SCA) provides further support. The methodology proposed here is general and can be applied to similar problems of interest.