Conversion of MIG6 peptide from the nonbinder to binder of lung cancer-related EGFR by phosphorylation and cyclization

Abstract

Development of chemical drugs and biologic agents to target human epidermal growth factor receptor (EGFR) has long been established as a promising therapeutic strategy for lung cancer. Previously, it was found that the tumor-suppressor protein MIG6 is a negative regulator of the EGFR kinase, which can bind at the activation interface of asymmetric dimer of EGFR kinase domains to disrupt EGFR dimerization and then inactivate the kinase. The protein adopts two separated segments, that is, MIG6^s1 and MIG6^s2, to directly interact with EGFR. Here, by computational modeling and analysis of the intermolecular interaction between EGFR kinase domain and MIG6^s2 peptide, we demonstrated that the dephosphorylated MIG6^s2 peptide can be converted from nonbinder to weak binder and then to moderate binder of EGFR by phosphorylation and cyclization, respectively; the former introduces strong electrostatic potential to EGFR–peptide complex by forming two geometrically satisfactory salt bridges across the complex interface, while the latter minimizes entropy penalty upon the binding of highly flexible peptide to EGFR. Subsequently, the linear phosphorylated and dephosphorylated peptides and phosphorylated cyclic peptide were synthesized and purified, and their binding affinities to the recombinant protein of human EGFR kinase domain were determined by fluorescence polarization titration. As expected theoretically, the linear dephosphorylated peptide has no observable binding to the kinase, and further phosphorylation and cyclization can confer, respectively, low and moderate affinities to the peptide, suggesting a good consistence between the computational analysis and experimental assay.

Keywords:

• Cyclization
• human epidermal growth factor receptor
• lung cancer
• MIG6 protein
• peptide
• phosphorylation

Introduction

Receptor tyrosine kinase (RTK) superfamily represents a subgroup of transmembrane proteins with an intrinsic tyrosine kinase activity that determines various cellular functions as diverse as growth, differentiation, and cell motility or survival (Ai et al. 2014). The human epidermal growth factor receptor (EGFR) as an important member of RTK has received particular attention in recent years owing to its strong association with malignant proliferation, which has been shown to play a central role in the development and progression of nonsmall-cell lung cancer (NSCLC) (Bethune et al. 2010). EGFR activation results from the formation of an asymmetric dimer in which the C-terminal lobe of one kinase domain plays a role similar to that of cyclin in activated CDK/cyclin complexes. The CDK/cyclin-like complex formed by two kinase domains, thus explains the activation of EGFR-family receptors by homo- or hetero-dimerization (Zhang et al. 2006). EGFR dimerization stimulates its intrinsic intracellular protein–tyrosine kinase activity. As a result, autophosphorylation of several tyrosine residues in the C-terminal domain of EGFR occurs to elicit downstream activation and signaling by several other proteins that associate with the phosphorylated tyrosines through their own phosphotyrosine-binding SH2 domains.

The receptor-associated late transducer MIG6 has been reported as an important negative regulator of the EGFR signaling by binding at the activation interface of EGFR kinase domain to disrupt EGFR dimerization (Hackel et al. 2001). Accumulated evidences suggest that the MIG6 plays an essential role in suppressing the development, proliferation, and metastasis of NSCLC by inactivating EGFR (Li et al. 2014; Maity et al. 2015; Izumchenko and Sidransky 2015; Zhang et al. 2007). Knockout of the MIG6 gene has also been found to strongly associate with a variety of tumors such as breast cancer (Wendt et al. 2015), endometrial cancer (Kim et al. 2014) and skin tumor (Ferby et al. 2006). The crystal structure of EGFR–MIG6 complex revealed that two separated segments in MIG6 protein, called MIG6^s1 (residues 336–361) and MIG6^s2 (residues 376–398), interact with the C-lobe of EGFR kinase domain (Figure 1) (Park et al. 2015). The binding of MIG6^s1 to EGFR activation interface prevents an asymmetric dimer formation, leading to the inhibition of EGFR activation at micromolar-binding affinity level (Moonrin et al. 2015). However, the MIG6^s1 is only able to inhibit the kinase domain in the context of asymmetric dimer formation, but not directly relevant for shutting down kinase activity. The presence of MIG6^s2 C-terminal to MIG6^s1 exhibits ability to bind tightly to activated EGFR, which induces the EGFR conformation change from active to inactive states; MIG6^s1 alone cannot confer this property, because the kinase residues that interact with it do not change conformation upon activation (Zhang et al. 2007). Crystallographic analysis revealed that the MIG6^s2 can interact directly with activation loop and the regions close to kinase active site (Park et al. 2015). Recently, Yu et al. have successfully performed truncation, modification, and optimization for improving MIG6 affinity to EGFR protein (Yu et al. 2016). Here, we attempted to systematically investigate the intermolecular interaction between the EGFR kinase domain and MIG6^s2 peptide by integrating in silico modeling and in vitro assay. The isolated MIG6^s2 peptide cannot bind effectively to EGFR kinase domain, and we herein proposed a molecular design scheme to rationally convert it from the nonbinder to binder of EGFR by phosphorylation and cyclization. We also demonstrated that the scheme can work fairly well by using fluorescence polarization assay and elucidated the molecular mechanism and biological implication underlying the affinity improvement upon the structural modification and optimization of MIG6^s2 peptide.

Figure 1. The complex crystal structure of EGFR kinase domain with MIG6^s1 and MIG6^s2 (PDB: 4ZJV).

Materials and methods

Molecular simulation

Molecular dynamics (MD) simulations of isolated MIG^s2 peptide and EGFR–MIG^s2 complex (where the MIG^s2 peptide can be phosphorylated, dephosphorylated, linear, and/or cyclized) were carried out using AMBER03 force field (Duan et al. 2003) implemented in the AMBER11 package (Case et al. 2005). A truncated octahedral box of TIP3P waters (Jorgensen et al. 1983) was added with a 10 Å buffer around the complex. Counterions were added to make the simulated system electroneutral. The phosphotyrosine parameters for pTyr394 and pTyr395 of MIG^s2 were derived from the AMBER parameter database (Steinbrecher et al. 2012). The simulated system was heated from 0 to 300 K over 500 ps, followed by a MD production simulation performed in an isothermal isobaric ensemble with periodic boundary conditions. The time step was set to 2 fs. The particle mesh Ewald (PME) method (Darden et al. 1993) was employed to calculate long-range electrostatic interactions. The SHAKE method (Ryckaert et al. 1977) was employed to constrain all covalent bonds involving hydrogen atoms.

Binding free energy calculation

The EGFR–MIG^s2-binding free energy was calculated using molecular mechanics/Poisson–Boltzmann surface area (MM/PBSA) method based on 100 snapshots of EGFR–MIG^s2 peptide complex structure extracted evenly from MD equilibrium trajectories of the complex (Yang et al. 2015, 2016): Δ G = Δ E_int + Δ D_dslv - TΔ S, where ΔE_int is the intermolecular interaction energy between the EGFR and MIG^s2 peptide, which was calculated using force field approach, and ΔD_dslv is the desolvation free energy due to peptide binding, which can be computed by numerical solution of the nonlinear Poisson–Boltzmann equation (for polar contribution) plus surface area model (for nonpolar contribution). To consider entropy penalty upon peptide binding, the normal mode analysis (NMA) was employed to estimate the vibrational component of the entropy. For each EGFR–MIG^s2 complex system, the NMA was performed for twice; one on the MIG^s2 peptide in bound state and another on unbound state. Here, the entropy effect associated with EGFR kinase domain was ignored due to its high rigidity.

Peptide-binding assay

Peptides were synthesized using standard 9-fluorenyl methoxycarbonyl (Fmoc) solid-phase chemistry. Phosphotyrosine was directly incorporated as its N_α-fluorenylmethoxycarbonyl-O-phosphate-l-tyrosine derivative. For cyclic peptide, the PAM-Gly-Boc resin was used with an S-tritylmercaptopropionic acid linker to facilitate the cyclization, as described previously (Chan et al. 2011). The peptides were then purified by RP-HPLC and confirmed by mass spectrometry and amino acid analysis. The fluorescence polarization (FP) assays were performed at 298 K following a protocol modified from previous reports (Park et al. 2013; Zhang et al. 2007). Synthetic peptides were labeled with N-terminally conjugated fluorescein rhodamine, which were incubated with purified protein of human EGFR kinase domain in the binding affinity assay buffer (10 mM Tris-HCl, pH 7.5, 25 mM NaCl, and 2 mM DTT) in a 96-well assay plate. The plate was mixed on a shaker for 5 min and incubated at room temperature for an additional 30 min. Polarization, defined as millipolarization units, was measured at room temperature with a fluorescence microplate reader. A negative control (5 μM peptide and assay buffer) and a positive control (5 μM peptide, 60 μM kinase domain and assay buffer) were used to determine the range of the assay. The binding was monitored as titration curves and used to derive the equilibrium dissociation constants K_d by nonlinear curve fitting: F = F₀ + F_∝ (p/K_d)/1 + (p/K_d), where p is the protein concentration at each measurement point, F is the observed FP value at the given protein concentration, F₀ is the FP value of free peptide, and F_∞ is the maximal FP value saturated with protein. Each assay was performed in duplicate.

Results and discussion

The MIG6^s2 peptide is composed of 22 amino acids (³⁷⁶VPCILPIIENGKVCSTHYYLLP³⁹⁸), which is folded into two β-strands in cocrystallized complex with EGFR kinase domain (Figure 1). A recent study found that phosphorylation of intact MIG6 protein at Tyr394 and Tyr395 residues is essential for binding and inhibition of EGFR (Park et al. 2015). The two residues are just within the MIG6^s2 region, and phosphorylation renders them negatively charged that promotes the two residues to separately form two salt bridges with the positively charged residues Lys875 and Lys879 of EGFR kinase domain. According to the double-headed mechanism for EGFR inhibition by MIG6 protein proposed by Zhang et al. (2007), the MIG6^s1 first binds tightly to EGFR C-lobe to anchor MIG6 on EGFR, and then, the MIG6^s2 is dynamic balance between association and dissociation with EGFR activation loop by phosphorylation and dephosphorylation at its Tyr394 and Tyr395 residues. Thus, it is suggested that phosphorylation is the necessary but not sufficient condition for an isolated MIG6^s2 peptide binding to EGFR without MIG6^s1 assistance.

The phosphorylated MIG6^s2 was striped from its cocrystallized complex with EGFR kinase domain (Figure 2(A)), and then the phosphate moieties were manually removed from the phosphorylated MIG6^s2, resulting in a dephosphorylated counterpart of MIG6^s2 peptide (Figure 2(B)). Subsequently, the two MIG6^s2 peptides (phosphorylated and dephosphorylated) were separately subjected to a long-term MD simulation to reach at equilibrium after ∼120 ns and ∼140 ns simulations, respectively. In addition, the complex structures of EGFR kinase domain with the two peptides were also equilibrated by 50-ns MD simulations. The equilibrium state of phosphorylated and dephosphorylated MIG6^s2 peptides in solution was shown in Figure 2(A,B), respectively. As can be seen, there is a considerable difference between the bound and unbound conformations of the two peptides; the bound conformation is well structured that composed of two β-strands, just like that in crystal structure, while the unbound conformation exhibits typical disorder feature with two irregular helices present in N-terminal region. In fact, the intrinsic thermal motion of unbound peptides in equilibrium state is predominant and fast conformation changes can be observed during the entire MD simulations, suggesting that both the phosphorylated and dephosphorylated MIG6^s2 are intrinsically disordered that cannot form specific structure in solution. According to the indirect readout theory for protein–peptide recognition proposed by Yu et al. (2014), entropy penalty would be very significant upon binding of the highly flexible MIG6^s2 peptide to EGFR protein. This is because peptide is highly flexible molecule that would incur considerable loss in degrees of freedom (i.e. entropy decrease ΔS < 0) upon binding to protein receptor. According to the Gibbs’ equation ΔG = ΔH – TΔS, the ΔS < 0 means –TΔS > 0, which causes ΔG increase that is unfavorable to peptide binding (ΔG < 0 and >0 represent that the binding can be spontaneous and cannot be spontaneous, respectively). A large entropy penalty associated with peptide binding to protein has been observed in crystal structure analysis of protein–peptide complexes (London et al. 2010) and MD simulations of Tsg101 protein interaction with an HIV-derived PTAP peptide (Killian et al. 2009).

Figure 2. (A) The phosphorylated MIG6^s2 peptide was stripped from its cocrystallizated complex structure with EGFR kinase domain (PDB: 4ZJV), resulting in unbound phosphorylated peptide with the conformation as it in cocrystallizated complex. (B) The stripped peptide was dephosphorylated at residues pTyr394 and pTyr395 without conformation change, resulting in a dephosphorylated version of the peptide. (C) The Leu397 residue of the stripped peptide was virtually mutated to Cys397, which was then covalently connected to wild-type Cys378 residue by computationally modeling a disulfide bond between them, resulting in a phosphorylated cyclic peptide. Subsequently, the three artificially modeled peptides were subjected to MD simulations to reach at equilibrium state.

In order to clarify the effect of Tyr394 and Tyr395 phosphorylation on the intermolecular interaction between EGFR and MIG6^s2 peptides, the binding free energies ΔG of both phosphorylated and dephosphorylated MIG6^s2 peptides to EGFR kinase domain were calculated using MM/PBSA + NMA analysis. Based on the analysis, we can computationally derive the total binding free energy ΔG and decomposed energetic components of the two linear peptides to EGFR. The calculated energetics parameters are tabulated in Table 1. It is unraveled that: (i) phosphorylation can considerably improve the intermolecular interaction potency between EGFR kinase domain and MIG6^s2 peptide (ΔE_int changes from −176.4 to −240.7 kcal/mol upon the phosphorylation); this is expected because the introduced phosphates in MIG6^s2 can form two electrostatically attractive salt bridges with EGFR as revealed by crystal structure. (ii) The desolvation effect is unfavorable (ΔD_dslv > 0) since polar water molecules are stripped from the hydrophilic complex interface. The phosphorylation can substantially amplify the unfavorable desolvation effect (ΔD_dslv increases from 115.8 to 160.9 kcal/mol upon the phosphorylation) due to the presence of charged phosphates at the interface. (iii) As might be expected, entropy penalty is as much as ∼70 kcal/mol upon the binding of highly flexible MIG6^s2 peptide to EGFR, which is not different significantly between the phosphorylated and dephosphorylated peptides (–TΔS = 76.4 and 67.8 kcal/mol, respectively). Consequently, the total binding free energies ΔG of phosphorylated and dephosphorylated peptides are –3.4 and 7.2 kcal/mol, indicating that the former and the latter are weak binder and nonbinder of EGFR, respectively.

Table 1. The calculated energetics parameters and experimental affinities for the different forms of MIG6^s2 peptides to interact with EGFR kinase domain.

MIG6^s2		Calculated energetics (kcal/mol)				Experimental affinity Kd (μM)
phosphorylation^a	cyclization^b	ΔEint	ΔDdslv	−TΔS	ΔG
no	no	−176.4	115.8	67.8	7.2	n.d.^c
yes	no	−240.7	160.9	76.4	−3.4	95 ± 7
yes	yes	−226.2	175.7	34.2	−16.3	11.2 ± 1.4

^a yes, both the Tyr394 and Tyr395 residues of MIG6^s2 peptide are phosphorylated; no: both the Tyr394 and Tyr395 residues of MIG6^s2 peptide are dephosphorylated.

^b yes, the Leu397 residue of MIG6^s2 peptide is mutated to Cys397 and then covalently connected to wild-type Cys378 residue via a disulfide bond; no: no mutation and connection are addressed on the MIG6^s2 peptide.

^c n.d., not determined.

In order to improve the binding capability of phosphorylated MIG6^s2 peptide to EGFR kinase domain, we herein attempted to further optimize the peptide structure to obtain increased affinity. According to above calculations, the entropy penalty is an important energetic factor unfavorable to peptide binding, which cannot be influenced significantly by phosphorylation. To minimize the entropy penalty upon peptide binding, we tried to constrain the conformational flexibility of MIG6^s2 peptide in unbound state. Considering that the peptide is folded into two antiparallel β-stands in cocrystallizated complex where its C- and N-terminuses are spatially close to each other, it is assumed that the two ends of the peptide can be covalently connected by a disulfide bond to generate a cyclic peptide. A computational procedure for the cyclization of linear phosphorylated MIG6^s2 peptide is shown in Figure 3. First, the phosphorylated peptide with bound conformation was stripped from complex crystal structure (Figure 3(A)). By examining the conformation, it was found that a Cys378 residue is present at the peptide N-terminus, which can be considered as an “anchor” to create disulfide bond. Correspondingly, the C-terminal residue Leu397 can serve as a good candidate of another “anchor,” which is spatially vicinal to the N-terminal Cys378 residue. Here, the Leu397 residue was virtually mutated to Cys397 (Figure 3(B)) using the WHAT IF server (Vriend 1990), and then, a disulfide bond was modeled between the mutated Cys397 and wild-type Cys378 to cyclize the peptide (Figure 3(C)). Subsequently, the phosphorylated cyclic peptide was subjected to MD simulations to relax the artificial system. As might be expected, the cyclic peptide reached at its dynamics equilibrium only taking ∼10 ns simulations, which was much fast as compared to those linear peptides (∼120 ns and ∼140 ns simulations for linear phosphorylated and dephosphorylated peptides, respectively). In addition, the MD-equilibrated conformation of the cyclic peptide is roughly consistent with that of MIG6^s2 in cocrystallizated complex (Figure 2(C)), suggesting that the cyclic peptide is prestructured and its solution conformation is constrained to that in complex with EGFR. In this way, the entropy penalty upon the peptide binding to EGFR is expected to be largely minimized. This can be solidified by NMA analysis of entropy effect associated with the binding. As can be seen in Table 1, the cyclization of phosphorylated MIG6^s2 peptide does not cause significant influence on the intermolecular interaction (ΔE_int) and desolvation effect (ΔD_dslv) of EGFR–peptide binding but can largely reduce entropy penalty upon the binding (−TΔS changes from 76.4 to 34.2 kcal/mol due to the cyclization). Consequently, the cyclic peptide exhibits a high affinity toward EGFR kinase domain with calculated binding free energy ΔG of −16.3 kcal/mol, which is much higher than that of linear peptide (ΔG = −3.4 and 7.2 kcal/mol for linear phosphorylated and dephosphorylated peptides, respectively). It is therefore expected that the cyclic peptide can bind more tightly to EGFR kinase domain than its uncyclized counterpart, the linear phosphorylated MIG6^s2.

Figure 3. Cyclization of phosphorylated MIG6^s2 peptide. (A) The phosphorylated peptide was stripped from its cocrystallizated complex with EGFR kinase domain (PDB: 4ZJV). (B) The Leu397 residue of the peptide was virtually mutated to Cys397 (L397C). (C) The peptide was cyclized by introducing a disulfide bond between the mutated Cys397 and wild-type Cys378.

Next, we employed fluorescence polarization assay to measure the binding affinity of linear phosphorylated and dephosphorylated MIG6^s2 peptides and phosphorylated cyclic peptide to the recombinant protein of human EGFR kinase domain. The titration curves are shown in Figure 4. It is evident that the phosphorylated cyclic peptide impacts significantly on fluorescence emission, indicating that the peptide can bind potently to EGFR protein. The dissociation constant K_d of the peptide was determined as 11.2 ± 1.4 μM by curve fitting through the fluorescence titrations. In contract, the linear dephosphorylated peptide has no observable binding for EGFR, given a very modest fluorescence perturbation with titration of the peptide (K_d=n.d.). In addition, it was also found that the phosphorylation can effectively improve the binding potency for the linear peptide to obtain a moderate affinity (K_d = 95 ± 7 μM). In this respect, the experimental evidences are well consistent with that of computational findings, that is, the MIG6^s2 peptide can be changed from nonbinder (K_d=n.d.) to weak binder (K_d = 95 ± 7 μM) and then to moderate binder (K_d = 11.2 ± 1.4 μM) of EGFR by phosphorylation and cyclization, respectively. We therefore expected that strong binder (K_d at nanomolar level) could be obtained by a further sequence optimization and structure modification of the currently designed phosphorylated cyclic peptide.

Figure 4. Titration of recombinant EGFR kinase domain to different chemical forms of fluorescein-labeled MIG6^s2 peptides.

Disclosure statement

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript.