Conformational Changes in Wild Type KRAS Induced by Two Novel Variants p.E31K and p.G138V

Ras proteins (H, K ras4A, K ras4B, and N-ras) have 85% homology and amino acids 1-165 (G domain) are highly conserved between these four Ras proteins. Several motifs present in conserved domain are important for protein function including GTP binding, effector binding etc. (Wittinghofer and Vetter, 2011; Hunter et al., 2014). The human K ras contains an alternative fourth coding exon. Alternative RNA splicing specifies either of two isomorphic proteins differing by 25 amino acid residues at their carboxy-terminus (Schubbert et al., 2007). The first 85 amino acids have role in binding to GDP and GTP, this region also includes the phosphate binding loop (P loop, comprising of amino acids 10-16), Switch I region (amino acids 32-38) and Switch II region (amino acids 59-67). P loop binds to the γ phosphate of GTP and Switch I, II regulate binding to Ras regulators and effectors. At the carboxy terminal 25 amino acids show considerable variation, it is named as hypervariable domain (amino acids 165-168/169). In this region terminal-CAAX farnesylation motif is present which specify membrane localization of Ras (Fig. 1). Key cysteine residues responsible for palmitoylation in H ras, N ras and K ras4A are also present in this region (C in Fig. 1). A stretch of lysines, proximal to CVIM motif is responsible for localization of K ras4B (KKKKKK in Fig. 1) (Schubbert et al., 2007). Deletion of the CAAX motif leads to the interruption of the post translational modification, thereby preventing the trafficking of Ras to plasma membrane (Wright and Philips, 2006). Fluctuations in the switch I and II regions or P-loop regions can promote instability in these protein regions leading to the hampering of the GTP binding (Chen et al., 2013; Goitre et al., 2014). As a consequence of mutations, the KRAS remains in an active GTP binding state and results in the continuous promotion of pro-proliferative signals downstream and tumorogenesis (Forbes et al., 2006; Buhrman et al., 2011). More than one third of all human cancers are found associated with Kras gene mutations resulting in one million deaths per year (Hutchins et al., 2011; Inoue et al., 2012; Thompson, 2013; McCormick, 2016). High frequency of Krasgene mutations are found in colorectal, lung and pancreatic cancers which are three of the four leading death cause by cancer in the world (Siegel et al., 2014; Wang et al., 2016; McCormick, 2016).

The most common Kras mutations are found in codon 12 and 13, which contribute about 95% of all mutation types, about 80% in codon 12 and 15% found in codon 13. The most common mutations in these two codons include G12C, G12V, G12D, G12A and G13D, which are found among 12% to 40% cases of cancers. Some other point mutations have been identified at codon 11 (Hongyo et al., 1995), 15, 18 (Wang et al., 2003), 19, 20 (Naguib et al., 2011), 27, 30 (Wang et al., 2003) 31 (Murtaza et al., 2012), 61 (Enomoto et al., 1992), 117, 146 (Neumann et al., 2009; Dogan et al., 2012), 154 (Neumann et al., 2009; Janakiraman et al., 2010; Dogan et al., 2012) but with less frequency. Certain conformational changes in mutant protein can result in its constitutive activation. Tumors harbouring mutated isoforms may have achilles heel (Habeck, 2002). In advance CRC, positive mutational status of K ras is associated with worse prognosis of the disease. KRAS mutation status should be considered as an important variable at the time of selection of therapy (Lièvre et al., 2008). To target the mutant K RAS pharmacologically, many in vitro and in vivo trials along with the computational metrics and MD simulation data analysis approach are being used (Ostrem et al., 2013; Patricelli et al., 2016; Jamal-Hanjani et al., 2017; Janes et al., 2018; Pantsar et al., 2018; Misale et al., 2019) and few of them are found promising. By using recent bioinformatics tools, we have analysed the possible functional impacts of two novel heterozygous mutations E31K and G138V in two colorectal cancer patients in silico.

Two novel variants E31K and G138V of K RAS were selected for analysis. An heterozygous mutation at codon 31, substituting glutamic acid (GAA) to lysine AAA, was previously identified (Murtaza et al., 2012) in a male patient (45 years) with moderately differentiated adenocarcinoma of mucinous type in T3 N0 M0 stage, invaded to the muscularis propria of colon with metastatic involvement in four or more regional lymph nodes. G138V was identified in a male patient (40 years) diagnosed with adenocarcinoma. Located in transverse colon, tumor was moderately differentiated infiltrative with vascular invasion classified in stage B. No family history of cancer was present in any of the studied subject.

To predict the deleterious nature of the identified genetic variants (c.91G>A and c.413G>T corresponding to p.E31K and p.G138V) in KRAS gene, we employed MutationTaster as in-silico genomic prediction tools (Schwarz et al., 2010). MutationTaster is conservation/ evolutionary based algorithm, which demands the query in the form of Ensembl gene ID and nucleotide change with few flanking nucleotides of the variation (Schwarz et al., 2010; Adzhubei et al., 2010).

The 3D structure of wild type KRAS protein, developed by X-Diffraction method, was obtained from Protein Data Bank. The crystal structure ‘4DSN’ (Maurer et al., 2012) was selected as a template to build further mutant protein models using Swiss-Model (Guex et al., 2009), as it covered both mutation positions (4DSN residue coverage: 2-164). The ligands, water molecules and other Het atoms present in the crystal structure were removed manually to avoid errors in building mutant models (DeLano, 2002).

The structural stability of the mutant KRAS proteins was analysed by I-Mutant Suite and DUET server (Capriotti et al., 2008; Pires et al., 2014). I-Mutant is a support vector machine (SVM)-based tool for the prediction of protein stability changes, upon single point mutations. After processing the protein sequence and the amino acid change, I-Mutant produces the prediction in the form of change in the Gibbs free energy (ΔG) value. The PDB structure 4-letter code, chain identifier as well as the mutation information such as residue position, wild-type and mutant residues codes in one-letter format were provided as an input for this server. The output of DUET is also in the form of ΔΔG values, wherein negative values denote destabilizing mutations.

The three-dimensional structures of wild-type and mutant proteins were subjected to molecular docking with GTP by Hex docking server (Macindoe et al., 2010). Hex is the only docking and superposition program to use spherical polar Fourier (SPF) correlations to accelerate the calculations. Binding sites of the GTP was also analysed in the wild-type and mutant KRAS protein using PyMol, displaying all the interacting residues around 8Å.

The output of the MutationTaster reveals one of the four predictions: ‘disease causing’ (probably deleterious), ‘disease causing automatic’ (known to be deleterious by database records) ‘polymorphism’ (probably harmless), and ‘polymorphism automatic’ (known to be harmless by database records). MutationTaster reported both p.E31K and p.G138V to be ‘disease causing’.

When genetic variants were structurally aligned with the wildtype protein for structural deviation analysis using PyMol-molecular graphic system, wild-type KRAS was superimposed with mutant models and the Root Mean Square Deviation (RMSD) of 0.01 Å was noticed in p.E31K and p.G138V mutant models (Fig. 2). As DUET is an integrated computational web server; it calculated the combined/consensus predictions of mutation Cutoff Scanning Matrix (mCSM) and Site Directed Mutator (SDM) methods in a non-linear regression fashion using SVMs. I-Mutant Suite and DUET scores suggest that both the genetic variants are deleterious in nature. Particularly p.G138V mutant model is predicted to be of extremely low stability when compared to p.E31K protein mutant model (Table I).

Table I.- Protein stability analysis score of different tools for p.E31K and p.G138V mutant KRAS proteins. Units for all scores are Kcal/mole, negative values denotes destabilized protein.

Molecular docking is extensively employed computation tool to analyse the molecular recognition that aims to predict the binding affinity and mode of a protein. Figure 2 depicts the docked complexes of A) Wt-KRAS-GTP B) Mt-E31K-KRAS-GTP and C) Mt-G138V-KRAS-GTP. The docking results of wild-type KRAS and its mutants (i.e., p.E31K and p.G138V) to GTP have revealed the binding energies -277.60, -278.27 and -279.67 Kcal/mole. The mutated KRAS protein may have low binding affinity to the GTP when compared to the wild-type KRAS protein. Moreover, the mutant and wild-type KRAS-GTP complex revealed different binding sites as exhibited by the visualization software (Table I). Herein, conformational changes induced in p.E31K and p.G138V proteins may result in its constitutive activation of Kras which resulted in worse prognosis of the disease; however the detailed functional activity and association of these variants to CRC needs be clarified by further studies.

KRAS has three sensitive sites which participate in the GTP hydrolysis; γ-phosphate of GTP binds to P-loop i.e., phosphate binding loop (10-16aa) of KRAS. After the KRAS-GTP complex is formed, a conformational change occurs in switch-I (30-38aa) and switch-II (59-67aa) regions. Any mutation in these sites might affect its regulatory function (Xu et al., 2017; Janes et al., 2018). Structural implications caused p.G12D and p.G13D were analysed by Chen et al. (2013) by calculating free energy profiles of binding processes of GTP, interacting with mutant and wild type and it was observed that GTP-binding pocket in mutant with p.G12D is more open than that of wild type and p.G13D proteins. By using a new integrated MD simulation data analysis approach Vatansever et al. (2017) depicted the induction of negative correlations between the fluctuations of SII and those of the P-loop, Switch I (SI) and α3 regions in K ras G12D and it was postulated that the deviation of active site residues impairs the GTP hydrolysis and GAP binding.SII fluctuations display increased level of fluctuations and negative correlations (Chen et al., 2013). New close-range salt bridges observed in G12D variant were absent in wild type Kras. It was assumed that, negative charge of aspartate triggers several conformational and dynamical changes in Kras G12D which forms an electrostatic interactions with K16 and K88, furthermore, a salt bridge between K16 with D57 will be formed.

Herein, it was observed that, E31K is predictable to disrupt the formation of effector loop thus affecting the downstream transducers and G138 residue is among the residues involved in the interprotein crystal interaction comprising the λ9 loop of KRAS. G138V may also destabilize the protein as it was predicted by the Duet stability analysis. Present results bring further acquiesce to the notion that heterogeneity of clinical outcomes in patients with mutant K ras may depends on variability in copy number and nature of diversity of mutant isoforms.

The authors declare no conflict of interest.