16 Molecular Modeling Method Applications …
325
confirmed that the anionic groups of the ligands formed ionic interactions with the
−NH 3
+ group of Lys15 in hTTR [83, 84]. On the basis of the obtained conformations (200 frames) from MD simulation for each compound, we also found that the
d for >91% conformations was ≤5 Å, indicating that the formed ionic interaction
was stable [84].
(5) π interactions
The π interaction included π–π, cation–π, anion–π, and sigma–π interactions. For
the endocrine system targets, π–π and cation–π interaction have been observed up
to now. For example, Li et al. [103] observed π–π interactions between the phenyl
group of hydroxylated polybrominated diphenyl ethers (HO-PBDEs) and Phe272,
Phe442, and Phe455 of TRβ. Yang et al. [83] observed cation–π interaction between
the phenyl group of phenolic compounds and –NH 3
+ group of Lys15 in hTTR.
As a best practice, it is recommended to analyze the binding pattern and noncovalent interaction of a crystal ligand with the corresponding target before performing molecular modeling in order to validate the reliability of simulation results.
For example, the noncovalent interaction analysis indicated that the bisphenols only
formed hydrogen bonds and hydrophobic interactions with human androgen receptor
(hAR). There were four amino acid residues (Asn 705, Gln 711, Arg 752 and Thr
877) involved in forming hydrogen bonds (Fig. 16.4a). Among them, the Gln 711
and Asn 705 were identified as the most important amino acid residues by calculating
the hydrogen bond formation rates. After analyzing the 76 hAR crystal structures,
a similar result was observed (Fig. 16.4b), which confirmed the reliability of the
simulation results [104].
16.5.3 Calculating Binding Energy
Based on the simulated conformation, the binding energy (E) of EDCs with a target
or other scores could be calculated. Theoretically, there was a significant correlation between E or other scores and the biological targets activities, e.g., estrogenic
activity and thyroid hormone activity. A linear correlation of E or other scores with
biological target activities could be derived, which could be further used to screen
potential EDCs or fill the data gap. For example, Ng et al. [78] developed a good linear
relationship between the median relative binding affinity values (logRBA) of estrogenic activity for bisphenol A replacement compounds and E (logRBA = −7.719 −
0.0860 E (p = 0.007)). Then, the missing estrogenic activity of other bisphenol A
replacement compounds was filled by employing the developed equation.
In addition, lots of software and tools could decompose the total binding energy
(E total ) into different components, such as electrostatic energy (E ele ), van der Waals
interaction energy (E vdw ), and so on. After analyzing the energy components, we
could identify the dominant driving force. For example, Lu et al. [106] decomposed
the E total into three components using Amber software, and the results indicated that
the E vdw was the major component of the total binding energy. This implied that
325
confirmed that the anionic groups of the ligands formed ionic interactions with the
−NH 3
+ group of Lys15 in hTTR [83, 84]. On the basis of the obtained conformations (200 frames) from MD simulation for each compound, we also found that the
d for >91% conformations was ≤5 Å, indicating that the formed ionic interaction
was stable [84].
(5) π interactions
The π interaction included π–π, cation–π, anion–π, and sigma–π interactions. For
the endocrine system targets, π–π and cation–π interaction have been observed up
to now. For example, Li et al. [103] observed π–π interactions between the phenyl
group of hydroxylated polybrominated diphenyl ethers (HO-PBDEs) and Phe272,
Phe442, and Phe455 of TRβ. Yang et al. [83] observed cation–π interaction between
the phenyl group of phenolic compounds and –NH 3
+ group of Lys15 in hTTR.
As a best practice, it is recommended to analyze the binding pattern and noncovalent interaction of a crystal ligand with the corresponding target before performing molecular modeling in order to validate the reliability of simulation results.
For example, the noncovalent interaction analysis indicated that the bisphenols only
formed hydrogen bonds and hydrophobic interactions with human androgen receptor
(hAR). There were four amino acid residues (Asn 705, Gln 711, Arg 752 and Thr
877) involved in forming hydrogen bonds (Fig. 16.4a). Among them, the Gln 711
and Asn 705 were identified as the most important amino acid residues by calculating
the hydrogen bond formation rates. After analyzing the 76 hAR crystal structures,
a similar result was observed (Fig. 16.4b), which confirmed the reliability of the
simulation results [104].
16.5.3 Calculating Binding Energy
Based on the simulated conformation, the binding energy (E) of EDCs with a target
or other scores could be calculated. Theoretically, there was a significant correlation between E or other scores and the biological targets activities, e.g., estrogenic
activity and thyroid hormone activity. A linear correlation of E or other scores with
biological target activities could be derived, which could be further used to screen
potential EDCs or fill the data gap. For example, Ng et al. [78] developed a good linear
relationship between the median relative binding affinity values (logRBA) of estrogenic activity for bisphenol A replacement compounds and E (logRBA = −7.719 −
0.0860 E (p = 0.007)). Then, the missing estrogenic activity of other bisphenol A
replacement compounds was filled by employing the developed equation.
In addition, lots of software and tools could decompose the total binding energy
(E total ) into different components, such as electrostatic energy (E ele ), van der Waals
interaction energy (E vdw ), and so on. After analyzing the energy components, we
could identify the dominant driving force. For example, Lu et al. [106] decomposed
the E total into three components using Amber software, and the results indicated that
the E vdw was the major component of the total binding energy. This implied that