320
X. Yang et al.
16.2 Preparation of 3D Biomacromolecule Structures
There are two major sources for researchers to obtain 3D biomacromolecule structures in the endocrine system, i.e., selecting the crystal structure from various
databases or predicting the 3D structure from corresponding amino acid sequence.
Generally, various databases, e.g., Protein Data Bank (PDB, https://www.rcsb.org/)
[49] are the preferred sources for researchers to obtain their desired target structure.
However, there are many challenges and disadvantages in selecting the crystal structure of a biomacromolecule from public databases. If an inappropriate 3D target structure was selected during the molecular modeling, the accuracy and trustworthiness of
the modeling results will inevitably be affected. Recently, Yang et al. [50] proposed
six principles to guide appropriate crystal structure selection. The principles included
(1) species differences, (2) the MOA, (3) mutant amino acid residues, (4) protein chain
number, (5) the degree of structural similarity between the ligand in crystal structure
and the model compounds, and (6) others factors, e.g., the experimental pH conditions of the crystal structure determined and resolution [50]. After considering the
aforementioned factors, appropriate 3D structure can be selected more successfully.
As it is difficult and time-consuming to determinate the crystal structures from
experimental methods such as X-ray crystallography and nuclear magnetic resonance
for every target of interest, the available crystal structures of biomacromolecules are
unfortunately still limited. On the contrary, determination of the macromolecule
sequences is easier than structure determination. As of July 2018, the UniParc
database contained more than 221 million protein sequences (http://www.uniprot.
org/uniparc/). In contrast, as of July 2018 there were only 142,379 structures solved
experimentally in PDB. A huge gap between known annotated sequences and available 3D structures existed [51]. To bridge the gap between the demand for 3D
structures of biomacromolecules and limited experimental structures, computational
methods, e.g., homology modeling, protein threading, and ab initio methods can
be employed to predict the 3D structure of targets from corresponding amino acid
sequence [52–55]. Many tools could be used to perform this task, such as modeler
[56], Swiss-PdbViewer [57], and so on.
Before proceeding, the obtained 3D target structure should undergo treatment as
follows. Generally, the treatment steps include but are not limited to:
(1) Insert missing atoms in incomplete residues;
(2) Remove the ions, water molecules, and other substructures;
(3) Standardize the atom and/or residue names;
(4) Protonate or deprotonate the ionizable residues under given pH conditions;
(5) Find and define the binding sites. It deserves mentioning that some important
water molecules in the structure of targets should be kept. For example, it was
well documented that there were water molecules taking part in forming the
conserved hydrogen bond with Glu 353 and Arg 394 in the ligand-binding
domain (LBD) of estrogen receptors alpha (ERα) [58]. In this case, those water
molecules were recommended to be retained.
X. Yang et al.
16.2 Preparation of 3D Biomacromolecule Structures
There are two major sources for researchers to obtain 3D biomacromolecule structures in the endocrine system, i.e., selecting the crystal structure from various
databases or predicting the 3D structure from corresponding amino acid sequence.
Generally, various databases, e.g., Protein Data Bank (PDB, https://www.rcsb.org/)
[49] are the preferred sources for researchers to obtain their desired target structure.
However, there are many challenges and disadvantages in selecting the crystal structure of a biomacromolecule from public databases. If an inappropriate 3D target structure was selected during the molecular modeling, the accuracy and trustworthiness of
the modeling results will inevitably be affected. Recently, Yang et al. [50] proposed
six principles to guide appropriate crystal structure selection. The principles included
(1) species differences, (2) the MOA, (3) mutant amino acid residues, (4) protein chain
number, (5) the degree of structural similarity between the ligand in crystal structure
and the model compounds, and (6) others factors, e.g., the experimental pH conditions of the crystal structure determined and resolution [50]. After considering the
aforementioned factors, appropriate 3D structure can be selected more successfully.
As it is difficult and time-consuming to determinate the crystal structures from
experimental methods such as X-ray crystallography and nuclear magnetic resonance
for every target of interest, the available crystal structures of biomacromolecules are
unfortunately still limited. On the contrary, determination of the macromolecule
sequences is easier than structure determination. As of July 2018, the UniParc
database contained more than 221 million protein sequences (http://www.uniprot.
org/uniparc/). In contrast, as of July 2018 there were only 142,379 structures solved
experimentally in PDB. A huge gap between known annotated sequences and available 3D structures existed [51]. To bridge the gap between the demand for 3D
structures of biomacromolecules and limited experimental structures, computational
methods, e.g., homology modeling, protein threading, and ab initio methods can
be employed to predict the 3D structure of targets from corresponding amino acid
sequence [52–55]. Many tools could be used to perform this task, such as modeler
[56], Swiss-PdbViewer [57], and so on.
Before proceeding, the obtained 3D target structure should undergo treatment as
follows. Generally, the treatment steps include but are not limited to:
(1) Insert missing atoms in incomplete residues;
(2) Remove the ions, water molecules, and other substructures;
(3) Standardize the atom and/or residue names;
(4) Protonate or deprotonate the ionizable residues under given pH conditions;
(5) Find and define the binding sites. It deserves mentioning that some important
water molecules in the structure of targets should be kept. For example, it was
well documented that there were water molecules taking part in forming the
conserved hydrogen bond with Glu 353 and Arg 394 in the ligand-binding
domain (LBD) of estrogen receptors alpha (ERα) [58]. In this case, those water
molecules were recommended to be retained.