Computational Simulation On The Interactions Of Ionizable Halogenated Compounds With Thyroid Hormone Transport Proteins

A worldwide concern has grown regarding the potential endocrine disrupting effects of anthropogenic molecules. To date, more than140,000chemicals have been used commercially. For the majority of these chemicals, their endocrine disrupting effects information is either limited or unavailable. Due to time and cost limitations, testing of all potential endocrine disrupting chemicals (EDCs) through whole-animal assay is unrealistic. Thus, a faster and more cost-effective method is required to test and screen potential EDCs. Computational toxicology-based virtual screening method shows its benefits on this challenge.According to the identified toxicity pathways, whether the endocrine hormones exert their biological effects in human and wildlife or EDCs evoke endocrine-related diseases and endocrine dysfunction, the mechanisms of action (MOA) for both endocrine hormones and EDCs may be generalized as the interactions between a small molecule and one or more macromolecular targets. Revealing the binding mechanisms between the hormone receptors, transport proteins, enzymes and EDCs (even endocrine hormones) will pave the way for developing virtual screening methods of EDCs.Previous studies indicated that the ionizable group e.g.-OH,-COOH,-SO3H, is a key factor for tight binding to macromolecule in endocrine system such as thyroid hormone transport proteins. However, all of the previous modeling studies lose sight of the fact that the ionizable groups can ionize under physiological or experimental pH conditions. It is unclear whether the neutral and ionic forms of EDCs have distinct interaction mechanisms with the macromolecular.targets. In addition, many EDCs (especially thyroid disrupting chemicals) contain aromatic ring and halogen groups in their molecular structure. The aromatic ring and halogen moieties in EDCs may play an important role in determining the interactions between EDCs and macromolecular targets. Nevertheless, the exact role of aromatic ring and halogen groups is also unclear. Thus, the objectives of this thesis are:(1) to reveal the different binding mechanisms for the neutral and anionic forms of EDCs with macromolecular targets;(2) to probe the contributions of aromatic ring and halogen group in the interactions between EDCs and macromolecular targets; and (3) to develop mechanism-based quantitative structure-activity relationship (QSAR) models for virtual screening potential EDCs. In this thesis, halogenated ionizable compounds (phenolic compounds and aliphatic acids) and human transthyretin (hTTR) were selected as a model system and molecular modeling methods were employed to reveal the underlying mechanisms. Main contents and results are as follows: (1) Hybrid quantum mechanics/molecular mechanics optimizations and the relative competing potency of a chemical with T4binding to hTTR (logRP) were employed to probe the mechanisms between halogenated phenolic compounds and hTTR. The binding patterns of ionizable ligands in73hTTR crystal structures were also analyzed. Correlation analysis results indicated that there were statistically significant (p<0.005) negative correlations between the experimental logRP and pKa values of the halogenated phenolic compounds. Molecular simulation results indicated that the interaction energy (Eint) values between the anionic forms of halogenated phenolic compounds and hTTR are much lower than those of the corresponding neutral forms. The results indicated that the anionic forms bind stronger with hTTR than the neutral forms. Binding pattern analysis results from both hTTR crystal structures and simulated EDCs-hTTR complex structures indicated that the ionized groups (-O-) in halogenated phenolic compounds could form ionic-pair (electrostatic) interactions with the-NH+3groups of Lys15residues in hTTR and form hydrogen bonds with the residues of hTTR. Due to the dominant and orientational interactions, the-O-groups point toward the entry port of the binding site. The aromatic rings of the compounds can form cation-π interactions with the-NH3+group of Lys15residues in hTTR.A QSAR model for logRP of halogenated phenolic compounds was developed by partial least squares (PLS) regression. The model had high goodness-of-fit, robustness and predictive ability. The logRP of halogenated phenolic compounds related to pKa, adjusted values for the most negative net atomic charge on an oxygen atom (qO-adj) and logarithm of the n-octanol/water distribution coefficient (logD). pKa has a negative coefficient in the model, which encodes the information that the anionic forms of the compounds bind stronger than the neutral forms. In the model, a more negative value of qO-adj leads to a higher logRP value, as a compound with a more negative value of qO-adj has stronger electrostatic interactions between the-O-/-OH group and hTTR. logD may characterize hydrophobic interactions between the halogenated phenolic compounds and hTTR in the model.(2) To further understand the MOA for halogenated aliphatic acids and hTTR, the poly-and perfluorinated chemicals (PFCs) were selected as model compounds and the molecular dynamic simulation was employed to investigate the underlying binding mechanisms. It was found that the interactions between PFCs and hTTR were also related to the chemical forms. The ionized groups (-COO-,-SO3-, SO2-, NR-) in PFCs could form ionic-pair (electrostatic) with the-NH+3groups of Lys15residues in hTTR and form hydrogen bonds with the residues of hTTR. By analyzing the molecular orbital energies of PFCs, it was also found that the anionic groups (nucleophile) in PFCs could form electron donor-acceptor interactions with the-NH+3groups (electrophile) in Lys15. Forming the aforementioned orientational interactions resulted in the ionized groups of PFCs pointing toward the entry port of the binding site.Two QSAR models for logRP of PFCs were developed by PLS regression. The models had high goodness-of-fit, robustness and predictive ability. In the two QSAR models, logRP of PFCs related to leverage-weighted autocorrelation of lag6/weighted by mass (HATS6m), the fraction of ionized species at given pH (δA-) and qO-adj and adjusted values for the energies of highest occupied molecular orbital (EHOMO-adj).HATS6m describe the information of the molecular carbon chain length.δA-has a positive coefficient in the model, which encodes the information that the anionic forms of PFCs bind stronger than the neutral forms. A more negative value of qO-adj leads to a higher logRP value, as a compound with a more negative value of qO-adj has stronger electrostatic interactions between the ionizable groups and hTTR. While a more positive value of EHOMO-adj leads to a higher logRP value, as a compound with a more positive value of EHOMO-adj has stronger electron donor-acceptor interactions between the ionizable group and hTTR.(3) The effects of halogenation on the interactions between halogenated EDCs and TTR were investigated by analyzing the TTR crystal structures, the EDCs-TTR complex from molecular simulation, and logRP values. It was found that the halogen moieties in EDCs can affect the interactions by forming halogen bonds and halogen-hydrogen bonds with TTR, and through inductive effects and hydrophobic effects. The halogen bonds (mainly halogen-oxygen bonds) and halogen-hydrogen bonds enhance the binding between organic halogenated compounds and TTR. For halogenated phenolic compounds, the inductive effect of halogen moieties is a main factor determining the logRP values. The hydrophobic effect of halogen moieties is a critical factor governing the interactions between non-ionizable compounds (e.g. polybrominated diphenyl ethers and TTR.