1. Introduction
There are two important biological targets related to breast cancer: Human Epidermal Growth Factor Receptor 2 (HER-2) and Epidermal Growth Factor Receptor (EGFR) [
1]. Overexpression of ErbB (this abbreviation is derived from the name of a viral oncogene to which these receptors are homologous: erythroblastic leukemia viral oncogene) family members has implicated in many human cancers, and HER-2 expression is predictive of recurrence of human disease and prognosis. Inhibitors of the kinase domain of EGFR and HER-2 have been approved for the treatment of cancer, for example, erlotinib, lapatinib and trastuzumab [
2]. Receptors of the HER (ErbB) family are critical for the development of various organs and systems. When activated, these receptors bind to dimers, transphosphorylate and become capable of transducing intracellular signals that can affect cell growth, the inhibition of apoptosis, the migration and invasiveness, and angiogenesis, among other processes that lead to progression of malignant tumors [
3]. The simple overexpression of HER-1 (EGFR) does not transform cells, since the HER1:HER1 dimer is only capable of being transphosphorylated when one of its extracellular ligands is coupled in its active site. HER-2, for which an extracellular ligand has not yet been described, may spontaneously form dimers, a characteristic conferred by the peculiar structure of its extracellular portion.
Usually, two copies of the HER-2 gene are found in each cell, which must produce an adequate amount of protein on the cell surface. In breast cancer, one can find 25–50 copies of the HER-2 gene and an increase of the protein amount by 40–100 times, resulting in 2 million receptors expressed in the tumor cell; the amplification is what defines a subtype of cancer, with a gene signature, and is maintained during the cancer progression [
4].
The protein, after binding to a ligand, is activated by means of homo- or heterodimerization, leading to a cascade of events that activate its tyrosine kinase domain and promoting the rapid cell growth, differentiation, survival and migration associated with HER-2 positive breast cancer [
5]. Thus, the HER2:HER2 dimer can be transphosphorylated independently of the absence of ligand, stimulating morphological transformation and cell growth, either in its mutated form or not [
6]. There is evidence of a preferred binding partner between HER-2 and EGFR, and the HER-2/EGFR heterodimer shows an increase in the relative signaling potency for the EGFR homodimers. In contrast to most tyrosine kinase receptors, the loop phosphorylation is not required for the kinase activation, whereas kinase is intrinsically self-inhibitory in the cell [
3].
It has been suggested that HER-2 can play an important role in the oncogenic activity of EGFR. Preclinical studies have shown that both EFGR and HER-2 act in a synergetic way in the cellular transformation [
7]. HER-2 is the main and most common partner on heterodimerization of EGFR [
8]. Thus, HER-2 contributes to an extension of the EGFR activity by improving the affinity by ligands [
8], reducing its degradation [
9] and increasing its predisposition recycling [
10]. Moulder et al. showed that specific EGFR inhibitors can reduce the HER-2 signalization and the growth of breast cancer cells that overexpress HER-2 [
11]. Lapatinib is a good dual inhibitor of EGFR and HER-2 and it is approved by FDA in combination with capecitabine for the treatment of patients in advanced stage or metastatic breast cancer who have not responded to other drugs. However, not all cells that overexpress HER-2 also respond to the treatment with lapatinib and some patients have presented resistance to this drug [
12]. Thus, the proposal of new inhibitors of both EGFR and HER-2 can be more effective than simply targeting one of them alone.
Several studies attempt to inhibit the biological targets under study and one way to study the interaction processes between HER-2/EGFR and inhibitor molecules is employing molecular modeling methods, which are often employed in medicinal chemistry [
6,
13,
14,
15,
16]. Using these techniques, it is possible to identify the interactions that occur between bioactive molecules and biological receptors. To quantify the structure and activity relationships of diverse compounds, two important techniques have been widely employed elsewhere: Comparative Molecular Fields Analysis (CoMFA) and Comparative Molecular Similarity Index Analysis (CoMSIA) [
17,
18,
19,
20,
21]. The main objective of this study was to assess the interactions that occur between HER-2/EGFR and dual inhibitors (acting on both HER-2 and EGFR) and, consequently, understand their inhibition mechanisms and propose new models of drugs to treat related diseases, such as breast cancer.
3. Discussion
Based on the receptor-based molecular alignment (
Figure 6), we performed analyses of the results for the most (compound 24) and least active (compound 15) inhibitors of HER-2 and EGFR. The poses generated in the molecular docking can provide valuable information on the key ligand–receptor interactions related to the inhibition of EGFR and HER-2 receptors.
Figure 3 shows that all compounds studied performed a polar interaction with Met793 at EGFR and Met801 at HER-2, which are important residues involved in the receptor–ligand crystallographic interactions. The most active compound at both targets has a large substituent group attached to the general structure, which provides a hydrophilic interaction between the aromatic system and both hydrophobic and hydrophilic pockets (according to
Figure 2B). In addition, the most and the least active compounds (24 and 15, respectively) perform extra polar contact with Met793 (EGFR) and Met801 (HER-2). It is important to note that, for this active compound (24), the binding mode observed in the docking simulations is the same as that proposed by the authors who synthesized and tested the compounds [
1,
23,
24]. Alternatively, the main interactions at the active site occur between this ligand and the following residues: Met793/Thr854 and Met801/Asp863 for EGFR and HER-2, respectively. Compound 15 (the least active) exhibits only the interaction with Thr854 and it does not interact with any of the major residues. In addition, compound 15 does not perform interactions in the hinge region of EGFR.
The most robust and predictive CoMFA model can also be used to rationalize the major ligand–receptor interactions from stereochemical and electrostatic contour maps. The stereochemical maps obtained from the CoMFA analyses are shown in
Figure 8 and present favorable steric contributions in green and unfavorable steric contributions in yellow. Analyzing the electrostatic maps in
Figure 8, we can see that, in the benzoisothiazole region, close to the nitrogen and sulfur atoms, substitutions by negative electrostatic groups are suggested, as well as
N-hydroxypyrrolidine substituents. However, for compound 24 (the most active) at the HER-2 target, the region close to benzoisothiazole suggests a substitution by positive and negative groups. However, as in EGFR, the
N-hydroxypyrrolidine region indicates favorable electrostatic interactions and, in fact, this region interacts with Asp863, which is a negatively charged polar residue. Compound 15 (the least active), considering its activity against EGFR, showed a favorable region in the ring containing nitrogen of the indolone group, whereas for HER-2, the electrostatic map suggests several modifications of negative character. In addition to the indolone region, close to the 7-(2-hydroxyethyl)-1-(phenylamine), the contour map also suggests substitutions by electronegative groups.
In relation to the stereochemical map, in the benzoisothiazole region, large green polyhedral are shown for both targets. These outlines indicate that, in these regions, substitutions by bulky groups can be carried out potentiating the biological activity of the compound. On the other hand, close to the 6-(2-chlorophenoxy) region, substitutions by less bulky groups are suggested. In the stereochemical analyses for both targets, it is possible to note that the maps suggest a substitution in the hydroxyl region by larger groups. Already, the position of the oxygen atom of 1-chloro-2-(cyclohexyloxy) indicates that less bulky groups would be more favorable.
5. Conclusions
Protein kinases (in this case, HER-2 and EGFR) are involved in cancer-related diseases. Due to the deregulation of genes that control cell growth, substances that inhibit protein kinases can be employed in the treatment of breast cancer, for example. In this study, a set containing 63 dual inhibitors of HER-2 and EGFR was analyzed with the following computational approaches: molecular docking (to obtain the structural alignment at each biological target), CoMFA and CoMSIA analyses. The models obtained using these techniques presented good predictive capacity, since the internal and external validations carried out showed that these models have a good correlation between the computationally predicted and the experimental biological values.
From the CoMFA analyses, it was possible to verify that the model showed a good statistical quality for HER-2 (q2LOO = 0.827) and EGFR (q2LOO = 0.728). To evaluate the predictive power of the best models generated, external validation was performed using the test compounds. The significant agreement between the predicted and predicted pIC50 values for HER-2 (r2pred = 0.999) and EGFR (r2pred = 0.998) indicate the predictive capacity of the CoMFA models. For the CoMSIA models, it was verified that the model has statistical quality for HER-2 (q2LOO = 0.744) and EGFR (q2LOO = 0.718), indicating the predictive capacity of the CoMSIA models. After the statistical analyses and its validations, the CoMFA and CoMSIA contour maps of the most and the least active compounds at each target were analyzed, indicating the regions with better contributions to the biological activity. The analysis of these maps suggested substitutions by groups that can potentiate the biological activity of new compounds, where four new compounds were proposed, prioritizing the contribution map of each biological target, for the most and least active compounds. From this, the new compounds were aligned to the CoMFA and CoMSIA models and the values of biological activity were predicted, showing significant improvements, mainly for the least active compound. Therefore, the results obtained in this study indicate the main interactions that occur between the inhibitors studied and the biological targets and may help the proposition of new potential dual inhibitors of HER-2 and EGFR, candidates to treat breast cancer.