NMR approaches to study proteins integrating globular and disordered domains: the case of c-Src

Src family of kinases

The c-Src protein (Src stands for “sarcoma”) is the product of the first discovered oncogene (a gene that is present in healthy individuals but that could eventually contribute to cancer processes). Indeed, overexpression or deregulation of c-Src is associated to poor prognosis in many cancers [10, 11]. c-Src is the leading member of the Src family of non-receptor tyrosine kinases formed by nine members: Src, Yes, Fyn, Fgr, Lck, Lyn, Hck, Blk and Frk [10]. In contrast to receptor kinases, that are transmembrane proteins sensing signals from the extracellular space, non-receptor tyrosine kinases transmit signals inside the cell by introducing a phosphate group in specific tyrosine residues of target proteins. The kinase domain, referred to as “Src homology 1” or SH1 is highly conserved in all the proteins of the Src family, and even in kinases of other families. In addition to the SH1 domain, all members of the Src family of kinases (SFK) have two additional globular domains, called SH2 and SH3, which are similarly highly conserved across the entire family. The N-terminal region of all SFKs is intrinsically disordered and contains the SH4 and Unique domains (Fig. 1). The SH4 domain is acylated with myristic (C14) in c-Src and with myristic and palmitic acid (C16) in most of the SFKs, and acts as a lipid anchor. The Unique domain receives this name because of the complete lack of homology of this region among the SFKs.

Fig. 1:

Domain structure of c-Src and comparison of the amino acid sequences of c-Src (top) and Yes (bottom). The background color in the sequence identifies the individual domains.

The SH2, SH3 and Unique domains are regulatory domains, however the degree of knowledge about their function is very different: the globular SH3, SH2 and SH1 participate in intramolecular contacts that can switch the protein between an inactive and an active form. The SH2 domain can interact with a phosphorylated tyrosine in the C-terminal tail of SFKs, while the SH3 domain interacts with the SH1 domain via a proline rich region located in the linker between the SH2 and kinase domains. This double interaction stabilizes a closed form, that is inactive. Dephosphorylation of the C-terminal tyrosine or competing interactions with the SH3 and SH2 domains may destabilize the closed form and render the open active kinase. Historically, the discovery of Src came from the analysis of the protein responsible for a viral induced sarcoma that was called viral-Src (v-Src). The viral protein was later found to be a modified version of the c-Src protein lacking the C-terminal tyrosine, and therefore being constitutively active. The X-ray structures of the closed (PDB 1FMK), and of an active form of Src (PDB 1Y57) confirmed this mechanism [12, 13]. The regulatory role of the Unique domain was not recognized for a long time.

De novo identification of functional regions by NMR

A first example of the power of NMR in the study of c-Src is the discovery of its Unique Lipid Binding Region (ULBR). We compared experimental and predicted residual dipolar couplings for the isolated disordered domain of c-Src. Residual dipolar couplings are observed in partially oriented samples that prevent the complete cancellation of the direct, through space, interaction between close spins (in our case we measured the ¹⁵N–¹H coupling that reports on the average orientation of each of the NH bonds along the sequence with respect to the external magnetic field). The prediction was based on a random coil model, in which individual residues were allowed to freely sample all the sterically allowed regions of the Ramachandran map. We detected significant deviations for a stretch of residues indicating that this region was sampling a high energy region of the conformational space in a naturally occurring sequence [14]. We reasoned that if evolution had selected a high energy situation, there should be a biological important function associated to this region. We checked the phenotypes associated to this region and the first observed effect was that this region supported lipid binding, from which the ULBR denomination was derived [15]. Later, in collaboration with cell biologists, we found that individual or multiple mutations of residues in this region resulted in a 50 % decrease in the invasive capacity of c-Src dependent cancer cells and a similar decrease in their division and the growth of implanted tumors [16].

Transferring information between disordered and ordered domains

To study the mechanism through which the Unique domain can regulate the kinase activity of Src we also used NMR. In this case, we combined chemical shift perturbations, that are sensitive to short-range interactions modifying the environment of individual residues, and paramagnetic relaxation enhancement (PRE) measurements that are sensitive to long range interactions. PRE are measured by chemically introducing a nitroxide radical containing species at the side chain of a cysteine residue. The proximity of an unpaired electron strongly enhances the relaxation of spins; therefore, it is possible to identify regions that are, even transiently, at short distances from the position where the paramagnetic center has been introduced [17]. This study led to the discovery of an intramolecular fuzzy complex involving the disordered regions of c-Src and the globular SH3 domain, that acts as a scaffold around which multiple weak interactions restrain the disordered region around the SH3 domain while retaining its disordered character [18] (Fig. 2).

Fig. 2:

Comparison of the volumes sampled by the disordered initial 85 residues of c-Src and those occupied by the globular SH3, SH2 and SH1 domains. The disordered region is represented by the projection of the coordinates of the alpha carbons in the ensemble that reproduces the experimental small angle X-ray scattering data of the construct formed by the initial 150 residues of c-Src, that include the disordered region and the SH3 domain.

These NMR studies also identified a set of aromatic residues in the Unique domain that are conserved in all the members of the Src family of kinases, providing another example that chemical singularities, such as the presence of hydrophobic groups in the middle of a highly polar and charged region, in a natural system subject to evolutionary pressure are usually indicative of a biological role. Indeed, the long-range interactions detected by PRE matched those derived using co-evolution analysis [19].

The overall charge of the Unique domain is naturally modulated by phosphorylation of serine and threonine residues by other kinases. Phosphorylation of serine 75 (S75) in Src causes changes in cell growth, cytoskeletal reorganization and mediates ubiquitination and degradation [20, 21] Phosphorylation of threonine 37 (T37) activates Src by disrupting the interaction between the SH2 domain and regulatory phosphotyrosine 530 [22]. Phosphorylation of serine 43 (S43) and serine 51 (S51) by Wnt3A have opposite effects on Src activation [23].

Interestingly, phosphorylated T37, S43, and S75 are located at three residue distance from basic residues lysine 40 and arginine 78. Changes of pKa of the phosphorylated residues upon mutation of the basic residues were measured using ³¹P NMR. These measurements showed the formation of salt bridges between S43 and lysine 40 and between S75 and arginine 78, and the lack of interaction of T37 with a basic residue [24].

Global protein breathing motions detected by NMR

Protein dynamics is not restricted to the intrinsically disordered regions. Globular domains show motions in time scales that range from nanosecond picosecond to microsecond-millisecond or even slower. Fast local motions are in the regime that can be accessed by NMR relaxation. Slower, long-range motions are more difficult to investigate. We have recently demonstrated that the chemical shifts of the methyl groups of methionine residues provide a dynamical tool that is sensitive to motions that average the environment of the methyl groups of this flexible side chain. To have a reference value for the chemical shifts expected in a rigid protein we used density functional theory (DFT) calculations of truncated models representing the natural environment of the methionine methyl groups of the protein in crystals used to obtain X-ray diffraction structures [25].

We observed a very good linear correlation between calculated and experimental ¹³C NMR chemical shifts for the methyl groups of the various methionine residues in several proteins. Interestingly, the slope of the linear regression line was protein dependent and correlated with the overall flexibility of the protein. The good correlation observed between experimental and calculated chemical shifts provides confidence on the accuracy of the DFT calculations using the truncated models derived from X-ray structures. The experimental ¹³C NMR chemical shifts are scaled towards the “random coil” values due to conformational averaging around the rigid model captured in the X-ray model. The fact that a linear correlation is maintained, indicates that all methionine residues in the same protein experience the same degree of conformational averaging. However, this averaging is different for each protein and depends on its overall flexibility. For a totally rigid protein with the structure captured in the X-ray model, the correlation line between experimental and calculated chemical shifts would have a slope of one. For an extremely flexible protein, all methionine methyl groups would have identical random coil chemical shifts, and the slope of the correlation line between experimental and calculated chemical shifts would be zero. For a protein undergoing and intermediate level of global breathing motions, the slope of the correlation between the experimental chemical shifts determined in solution and those calculated from the X-ray structure would have a value between zero and one. Thus, the slope of the correlation line defines an order parameter characterizing the extent of global conformational averaging.

Recently, this method has been used to characterize the effect of allosteric modulators of the neurotensin receptor 1, a G protein-coupled receptor that is a target for neurological diseases and cancer [26, 27].

In the 536 residues of c-Src there are only ten methionine residues, all of them located in the kinase domain (Fig. 3). The active site is located at the hinge between two lobes. The N-terminal one is mainly formed by β-sheets while the C-terminal lobe is helical. A good correlation (R ² = 0.94) between calculated and experimental values was obtained for the six methionine methyl groups located in the C-terminal lobe of the kinase domain using the ¹³C methyl methionine chemical shifts calculated from the 4mxo pdb structure [28]. The methionine methyl order parameter is 0.28 that corresponds with a moderately flexible domain. This order parameter is comparable to the one previously observed for ligand binding domain of the glucocorticoid receptor or the neurotensin receptor bound to the neurotensin 8–13 peptide [25, 26]. In contrast, the four residues located in the N-terminal lobe show much weaker correlation suggesting that the dynamics of the two lobes of the kinase domain of Src are independent. The C-terminal lobe interacts with the SH2 domain while the N-terminal lobe interacts mainly with the SH3 domain, which is part of the fuzzy complex with the disordered domain. We are currently exploring the effect of post-translational modifications in the Unique or SH4 domains in the dynamics and kinase activity of full-length c-Src.

Fig. 3:

Surface representation of the globular domains of human c-Src. The C-lobe and the N-lobe of the SH1 (kinase) domain, in which the ten methionine residues are located, are represented in red and blue color respectively. The experimental and calculated chemical shifts of the methyl groups of methionine residues in the C-lobe show a very good linear correlation (red line) suggesting a global breathing motion with an order parameter of 0.28. In contrast, methionine residues in the N-lobe (blue line) do not seem to participate in a global motion, showing a dynamic distinction between the two lobes of the kinase domain. The SH3 domain, that is the scaffold of the fuzzy complex with the disordered domain is in direct contact with the N-lobe of the SH1 domain.