Structural basis of constitutive c-Src kinase activity due to R175L and W118A mutations

Abstract Cellular Src (c-Src) belongs to a non-receptor membrane-associated tyrosine kinase family that plays essential roles in cellular processes. Growing evidence suggests that R175L and W118A mutations in SH2/SH3 domains of c-Src functionally inactivate these domains leading to constitutive activation of kinase domain (KD). Here we modeled c-SrcR175L, c-SrcW118A and c-SrcW118A+R175L structures by inducing phosphorylation at Y416 or Y527, respectively to characterize the comparative dynamics in the active versus inactive states through molecular dynamics simulation assay. We observed more conformational readjustments in c-Srcopen than its close variants. In particular, C-terminal tail residues of c-SrcW118A-open and c-SrcW118A+R175L-open demonstrate significantly higher transitions. The cross-correlation analysis revealed an anticorrelation behavior in the motion of KD with respect to SH2, SH3 and the linker region of SrcW118A+R175L-open, while in c-SrcWT-open, SH2 and SH3 domains were anticorrelated, while KD and C-terminal tail motions were correlated. Due to these conformational differences, c-Src open forms exhibited lower interaction between pY527 and SH2 domain. Through detailed structural analysis, we observed a uniform myristate binding cavity in c-SrcWT-open, while the myristoyl pockets of mutant forms were deformed. We propose that constitutive activation of mutant Src forms may presumably be achieved by the prolonged membrane binding due to unusual conformations of C-terminal and myristoyl switch residues that may result in a higher dephosphorylation rate at pY527 in the myristoylated c-Src. Thus, our study establishes novel clues to decipher the constitutive activation status of c-Src in response to known mutations that may help in devising novel therapeutic strategies for cancer metastasis treatment. Communicated by Ramaswamy H. Sarma

Src family kinases (SFKs) belong to the membrane-associated, non-receptor tyrosine kinase family that comprises Src, Fyn, Frk Yes, Blk, Yrk, Fgr, Hck, Lck, and Lyn (Amata et al., 2014; Mart ın-P erez & Garc ıa-Mart ınez, 2015; Thomas & Brugge, 1997). All SFK members exhibit a similar domain organization with a unique N-terminal region, followed by Src homology 3 (SH3) and SH2 domains, a linker region, and a tyrosine kinase domain (KD). KD encompasses an N-terminal lobe, a C-lobe, and the regulatory C-terminal tail (Boczek et al., 2019;Young et al., 2001;Y. Meng & Roux, 2014;Xu et al., 1999). Src gene encodes proto-oncogene c-Src, which is considered the normal cellular homologue of the avian sarcoma viral transforming oncogene v-Src (Yoon et al., 2018). The tyrosine c-Src is the most oncogenic, and it is ubiquitously expressed in all tissues. It plays an imperative role in many signal transduction pathways ranging from cell proliferation to growth and survival (Amata et al., 2014;Summy & Gallick, 2006;Yeatman, 2004). Multiple lines of evidence suggest that any imbalance in the c-Src activity results in cancer ( Boczek et al., 2019;Griffiths et al., 2004;Roskoski, 2015 ). For example, elevated c-Src signaling has been implicated in colon, breast, lung, esophagus, skin, parotid, cervix, ovary and stomach cancers (Frame, 2002). The crystal structures of c-Src indicate that its catalytic activity is tightly controlled by auto-inhibition through various intramolecular interactions. Generally, c-Src remains inactive in the normal cells. In compact and inactive conformation, the SH3 domain mediates interaction with the proline-rich region localized in the linker, whereas the SH2 domain binds to phosphorylated Tyr527 (pY527) at the regulatory C-terminal tail (Boerner et al., 1996;Mart ın-P erez & Garc ıa-Mart ınez, 2015;Y. Meng & Roux, 2014;Young et al., 2001). Deletion of regulatory C-terminal tail or mutation of pY57F causes the oncogenic form of SFKs, while W118A and R175L substitutions result in the functional inactivation of both SH3 and SH2 domains and provoke the constitutive activation of KD (Griffiths et al., 2004;Hunter, 1987;Mart ın-P erez & Garc ıa-Mart ınez, 2015;Reynolds et al., 1987 ). Recently, a double mutant (W118A/ R175L) of c-Src has been reported to have inhibited activities of both SH2 and SH3 domains (Mayoral-Varo et al., 2021). In contrast, other reported mutants boost the transforming and kinase activities of pY527F c-Src [F172P, R175L, D(198-205), D(206-226), and D(176-226)] and confer the transformation competence of normal c-Src gene, indicating that mutations in SH2 may activate c-Src (Hirai & Varmus, 1990). Thus, two regions of SH2, the FLVRES sequence (residues 172 to 177) and the NVKHYKIRKLDSGGFYITSRTQFNSLQQL sequence (198-226) may be considered as mutation hotspots (DeClue & Martin, 1989;Hirai & Varmus, 1990;Sadowski et al., 1986). Evidently, pY527 must be dephosphorylated for c-Src activation; however, dephosphorylation of Y527 is not sufficient for kinase activity. c-Src KD contains another phosphorylation site at Y416 within the activation loop (A-loop), which is situated at the interface between N-and C-lobes (Banavali & Roux, 2005;Cowan-Jacob et al., 2005;Yang et al., 2009). Due to the autophosphorylation of Y416, kinase activity is enhanced that is another important event in this process. For full kinase activity of c-Src, Y527 dephosphorylation event is an essential step prior to autophosphorylation at Y416. Subsequently, c-Src kinase activity is largely dependent on the phosphorylation status of the negative and positive regulators that bind at Y527 and Y416 residues, respectively (Boerner et al., 1996;Cowan-Jacob et al., 2005;Faraldo-G omez & Roux, 2007;Lydon et al., 1992;Roskoski, 2005;Summy & Gallick, 2003). Generally, an intra-molecular interaction supports the inactive form of c-Src, while external interactions like binding other proteins to SH2 or SH3 domains may promote its active state.
The current study is based on the elucidation of the constitutive kinase activity of c-Src through monitoring imperative conformational changes in the active versus inactive states in parallel to inducing mutagenesis (W118A, R175L and W118A þ R175L) in the c-Src SH2/SH3 domain using the modeling and molecular dynamics approaches. Our results reveal convincing clues regarding the close-to-open state of c-Src upon the gain of function mutations. The knowledge will primarily add helpful information about the key residues involved in the kinase activity regulation at an intramolecular level that may prove to be a milestone for devising novel therapeutic options against cancer progression.

Mutational study
Function blocking point mutations in SH3 (W118A) and SH2 (R175L) domain of c-Src were selected through an extensive literature study (Griffiths et al., 2004;Mart ın-P erez & Garc ıa-Mart ınez, 2015;Mayoral-Varo et al., 2021). These mutations were further used for detailed structural studies using c-Src open and close conformations. Multiple sequence alignment (MSA) of SH3 and SH2 domains of human SFKs family members were performed by ClustalW (https://www.genome.jp/ tools-bin/clustalw) to estimate their degree of conservation.
Three-dimensional (3D) structure prediction 3D structures of the c-Src mutants were predicated using Modeler 9.14 (Eswar et al., 2008) structures of c-Src W118A-close and c-Src R175L-close were predicted through homology modeling by using a crystal structure of c-Src WT-close (PDB ID: 2SRC) as a template. Upon superposition with template structure, c-Src W118A-close and c-Src R175L-close exhibited calculated RMSD values of 0.233 Å and 0.170 Å, respectively. Modeled structures were evaluated by Molecular Operating Environment (MOE) (Vilar et al., 2008) by computing Phi-Psi angles, rotamers, and contact energy values. Furthermore, tools like MolProbity (Chen et al., 2010), Verify3D (Bhattacharya et al., 2007) ProQ (Wallner & Elofsson, 2003) and ProSAWeb (Wiederstein & Sippl, 2007) were used for detailed analysis. UCSF Chimera 1.11 (E. C. Meng et al., 2006) was used for phosphorylation. A-loops of c-Src WT-open and its mutant forms (c-Src W118A-open , c-Src R175L-open , and c-SrcW118AþR175L-open ) were phosphorylated at Y416 residue, whereas regulatory C-terminal tails of c-Src W118A-close and c-Src R175L-close structures were phosphorylated at Y527, respectively. The crystal structure of c-Src WT-close was already phosphorylated at Y527.  (Abraham et al., 2015) using GROMOS9643a1 extended phosphorylated force field (Abraham et al., 2015). All structures were solvated by the SPC216 water model in the periodic box dimension ranging from 8.72 Â 7.28 Â 6.88 Å and 6.74 Â 7.03 Â 6.29 Å, respectively. Na þ and Clcounter ions were added to neutralize the system. Energy minimization was performed via the steepest descent algorithm for 500 steps to eliminate the initial steric clashes. Subsequently, systems were equilibrated for 1000 ps at 300 K and 1 bar pressure in NVT and NPT ensembles, respectively. Finally, MD simulations were run for a 200 ns time scale. Particle Mesh Ewald algorithm was employed to analyze long-range electrostatic interactions. Consequently, snapshots were collected for all systems throughout MD simulation, and PDB files were generated for every 50 ns time interval to investigate the time-dependent behavior and stability. The root means square deviation (RMSD), root mean square fluctuation (RMSF), and secondary structure calculation analyses were performed using g_rms, g_rmsf, and do dssp utilities of GROMACS.

Principal component analysis
The calculations of eigenvalues and eigenvectors were carried out through the Essential dynamic (ED) method as described elsewhere using the GROMACS built-in utilities, such as gmx covar and gmx anaeig (GROMACS Documentation Release 2019 GROMACS Development Team, 2019). Principle component analysis is the foremost approach that alleviates data complication and pulls out the collective motion in simulation that is substantially associated and probably meaningful for biological functions (Amadei et al., 1993). Briefly, a variance/covariance matrix is generated through trajectories to eliminate the rotational and translational movements using fundamental analysis. The eigenvalues and eigenvectors were used to decompose the matrix through diagonalization. The eigenvalue constitutes the eigenvector amplitude and multidimensional spaces; the displacement of atoms with every eigenvector represents the coordinated protein motion within each direction (Jolliffe & Cadima, 2016). The covariance matrix was constructed for wild-type and mutated c-Src backbone Ca atoms of individual trajectories by ED analysis. The eigenvectors and eigenvalues of the covariance matrices and the projections of the first two principal components were computed.

Dynamic correlation analysis
Dynamic correlation matrices were generated through dynamic correlation functions to determine the comparative correlations among open and close conformations of c-Src backbones (normal and mutant forms). This correlation is either a direct correlation, i.e., residue A moves in the same direction as residue B, or it is inversely correlated, i.e., movement of residue A decreases the movement of residue B.
Dynamic correlation functions are given as follows (Kasahara et al., 2014).
Where Cij denotes the given correlation between the i th residue against the jth residue. This is then incorporated into the correlation matrix named Cij (Abid et al., 2020). The formula for each r is given as: The x and y are the two atoms whose trajectories are tracked concerning time throughout a simulation timeframe.

Sequence and geometry analysis
All SFKs members exhibit a similar domain arrangement. Human c-Src comprises SH3(84-145 aa) and SH2 (151-248 aa) domains, a linker region (249-269 aa), a kinase domain (270-523 aa), and a C-terminal tail (524-533 aa), respectively (residue numbers were referred corresponding to chicken c-Src) ( Figure 1) (Roskoski, 2004). The MSA exploration of SH3 and SH2 domains of human SFK family members indicted that both W118 and R175 residues are relatively conserved throughout the family ( Figure S1). Ramachandran scores for c-Src WT-open , c-Src W118A-open , c-Src R175L , c-Src W118A-openþR175L , c -Src WT-close , c-Src W118A-close, and c-Src R175L-close structures suggested that most of the residues accommodated in the sterically allowed regions (Table S1 and Figure S2). Verify3D plots revealed that c-Src structures were good quality, indicating an average 3D-1Dscore> ¼0.2 (Table S1 and Figure  S3). Atomic contact energy (ACE) values were calculated for c-Src WT-open , c-Src WT-close, and its mutant forms were in a contact range of 6 Å, assigning energy terms (in kcal/mol). These energies were calculated for all residues in the system. In general, the enormous negative terms designate that the residues are mainly in contact with hydrophobic atoms and therefore probably localize in the buried protein environment. However, residues with positive energy values indicate their association with predominantly hydrophilic atoms and are expected to be present in more solvent-exposed regions ( Figure S4). The rotamer plots illustrate the strain energy (kcal/mol), whereas X-axis demonstrates the residues in the sequential order. Residues having strain energies >5.0 kcal/ mol are rare in the structures and warrant a closer inspection ( Figure S5). Other structural evaluation results (ProSA web) were illustrated in (Table S1).

Dynamic behavior study
Another exciting aspect of this study is to investigate the comparative structural details through simultaneous phosphorylations at the A-loop (pY416) and C-terminal tail (pY527), leading to activation and inactivation, respectively

Structural stability analysis
RMSD values of backbone atoms in c-Src WT and mutant structures were calculated to assess the conformational stability of   (Figure 2A). In contrast, RMSD profile analysis for c-Src WT-close and its mutant forms demarcated more stable behaviors in a range of 2.5-6.0 Å, respectively ( Figure 2B). c-Src WT-close , c-Src W118-close, and c-Src R175L-close structures achieved stability at an 80 ns time scale and later remained stable throughout MD simulation ( Figure 2B). These findings demonstrated higher average RMSD values for c-Src W118A-open and c-Src R175L-open than c-Src W118-close and c-Src R175L-close structures. Since a stable structure of c-Src open conformation is necessary for a proper function, its mutant forms may exhibit less stability and activity.

Principal component analysis
ED or PCA is commonly used to predict a protein dynamic behavior. To identify large-scale collective motions (in both open and close conformations) of c-Src WT and its mutant forms, PCA was performed using the last 10 ns MD trajectories. We selected the first two principal components (PC1 and PC2) to analyze the trajectory projections in the phase space ( Figure 4A Figure 5A and D). In contrast, due to W118 and R175L mutations, multiple structural changes at the secondary level were observed, including bend to the coil and turn to bend ( Figure 5B and 5C). In the close conformation of c-Src and its mutant forms, non-significant changes were observed ( Figure 5E-G). The SH2 domain region was lying between 144-248AA, followed by a recurring coil, b-sheet, bend, and a turn. The c-Src WT-open region ranging from 224-244AA was transformed from A-helix to 5-helix, while this region remained stable in c-Src   (Figure 5A-D).
As described earlier, W118A and R175L mutations in c-Src open conformation influenced the stability and flexibility. Furthermore, these mutations resulted in an apparent conformational drift in the c-Src open conformation in the secondary structure analysis. In order to investigate how these substantial conformational switches in c-Src WT-open and c-Src WT-close occur due to W118A, R175L and W118A þ R175L mutations, PDB files were generated at regular time intervals   Figure S7).
Based on the finding that electrostatic interaction between pY527 and R175/K203 residues becomes stabilized during the transition of active-to-inactive c-Src (Yoon et al., 2018), we intended to measure distances among these residues. We observed a substantial difference in distances of open versus close c-Src variants ( Figure 6A-G). The observed  Figure 6A-D). In contrast, due to phosphorylation at Y527, significantly reduced distances (4.38 Å, 11.08 Å, and 11.38 Å) were observed in c-Src WT-close , c-Src W118A-close, and c-Src R175L-close between pY527 and R175 residues, respectively

Dynamical cross-correlation analysis
The dynamical cross-correlation map (DCCM) (H€ unenberger et al., 1995) was generated to analyze the correlations among the positions of residues between the active and inactive conformations of c-Src and its mutant forms. In this map, Cij ranges between À1 and þ1. Positive values mean that two Ca atoms are correlated: and tend to move during the same period and phase. Conversely, negative values mean that two Ca atoms are anticorrelated: the two Ca atoms move in the same period but towards the opposite phase. Motions can be positively correlated (in the same direction), anticorrelated (opposite), or uncorrelated. The DCCMs corresponding from 190-200 ns of the MD simulation runs is shown in Figure (7A-D). In c-Src WT-open , regions including the SH3 domain, a linker region, KD and C-terminal tail, were highly correlated. SH2-SH2 domain correlations were positive, while the SH2 domain was uncorrelated, corresponding to the SH3 domain, a linker region, KD and C-terminal tail ( Figure 7A). In c-Src W118A-open , both SH2 and SH3 domains exhibited more correlations with respect to each other. Additionally, SH2 and SH3 domains were also correlated with the linker region (249-260AA), while they were anticorrelated with the KD and C-terminal tail ( Figure 7B). We observed a strong correlation in c-Src R175L-open ( Figure  7C). In contrast, the correlation map demonstrated a highly

Discussion
Src family kinases (SFKs) play essential roles in cellular signal transduction by regulating cell growth, differentiation, proliferation, migration, and survival (Tse & Verkhivker, 2015). Generally, SFK activity is regulated through the allosteric conformational readjustment of multiple domains due to phosphorylation of two distant tyrosine residues (Y416 and Y527). The active form of Src kinase is maintained by autophosphorylation of Y416 residue that is located at the central "activation or A-loop" of the catalytic domain (Yadav & Miller, 2008), while Y527 phosphorylation induces c-Src deactivation (Alba et al., 2021). Recently, it has been demonstrated that a double mutant of c-Src having both R175L and W118A mutations fails to adopt close conformation and remains constitutively active (Griffiths et al. 2004;Hirai & Varmus, 1990;Shvartsman et al., 2007;Yadav & Miller, 2008). Though kinase activity remains constitutive, R175L results in the functional inactivation of only the SH2 domain and keeps the SH3 domain functional, while W118A mutation in the SH3 domain confers the functional inactivity of the SH3 domain (Hirai & Varmus, 1990;Mayoral-Varo et al., 2021). To understand the molecular mechanism of Src constitutive kinase activity at atomistic scale, active states of c-Src WT , c-Src R175L , c-Src W118A and c-Src R175LþW118A (phosphorylated at Y416) were compared for their conformational space against the inactive c-Src (phosphorylated at Y527) through MD simulation and dynamic cross-correlation assays.
Evidently, in the c-Src open forms, more conformational readjustments were observed in the SH3-SH2 domain segment than that of its close variants. KDs of c-Src mutants (c-  (Figure 8). In contrast, L533 side chains in the mutant forms were shifted to the lateral gate of the myristoyl pocket, and cavity sizes were not uniform ( Figure 8). Additionally, the Y527 residue side-chain position was significantly altered in c-Src W118AþR175L-open as compared to c-Src WT-open . We speculate that due to the altered conformational space of the myristoyl pocket, c-Src W118AþR175L-open Y527 residue may not be recognized by C-terminal Src Kinase (Figure 8) (Levinson et al., 2008) for phosphorylation resulting in a failure to adopt autoinhibitory state. Indeed L533 binding site gets free only by the myristoylation that promotes the phosphorylation of Y527, and any hinderence in the displacement of L533 may perturb Y527 phosphorylation status in c-Src W118AþR175L-open ( Figure 8). As nonmyristoylated c-Src is cytosolic and exhibits reduced tyrosine kinase activity (Bagrodia et al., 1993;Patwardhan & Resh, 2010), it is tempting to speculate on the c-Src W118A-open and c-Src W118AþR175L-open constitutive activation that may presumably be achieved by the prolonged membrane binding due to unusual conformations of C-ter and myrisoyl switch residues, resulting in the reduction of cell proliferation (Eling & Kremsner, 1994). Membrane attachment of c-Src via its C-ter tail may also restrict the accessibility of Src to the ubiquitination machinery and enhance its activity through docking with phosphatase for Y527 dephosphorylation. A higher dephosphorylation rate at pY527 in the myristylated c-Src induces mitotic activation (Bagrodia et al., 1993).
Given that the intramolecular associations of SH2 domain to pY527 and SH3 domain to polyproline PPII helix pack SH3 and SH2 domains against the back of KD to maintain c-Src close conformation, we examined conformational dynamics of these domains. Both c-Src WT-open and c-Src R175L-open demonstrated strong anticorrelated motions between SH2 and SH3 domains, while KD and C-terminal tail motions were correlated. Such behavior of the SH2 and SH3 domain could assist in balancing the kinase active conformation space. In contrast, c-Src W118A-open revealed an utterly opposite behavior, where SH2 was correlated with SH3, while KD and C-terminal tail had anticorrelated motions, indicating a favoring role of W118 residue in the kinase regulation. Thus, the W118A mutation restricted the conformational flexibility of the KD and C-terminal tail, while the SH3 domain would attain more flexibility. These findings are in good agreement with the RMSF analysis. In c-Src W118AþR175L-open , no correlations were observed in the motions of KD with SH2, SH3 and the linker region, suggesting that higher anticorrelated dynamics of KD may shift SH3, SH2 and the linker region away from the KD.
Collectively, our findings describe the intramolecular connections (Table S2)  largely uncover conformational signatures leading to its constitutive kinase activity. This study will largely help in devising novel therapeutic strategies for the treatment of cancer metastasis.