Binding interaction of a gamma-aminobutyric acid derivative with serum albumin: an insight by fluorescence and molecular modeling analysis

Absorbance and fluorescence of the GABA derivative

Molecular structure of compound 5 is shown in Scheme 1: Figure S1 shows the absorption spectrum of the GABA derivative. Due to the presence
of conjugate systems, it showed absorption in UV region (below 300 nm). However, the
absorbance was very weak. The compound is non-fluorescent in nature.

Interaction with albumins

The fluorescence intensity of BSA and HSA decreased gradually with the increasing
concentration of ligand (Fig. 1). Thus, the quenching of the intrinsic tryptophan fluorescence of serum albumins
by the GABA derivative indicates its binding to the proteins. Figure 2 shows the Langmuir isotherm (Eq. 1) fitted to the quenching data for the determination of binding constants. The binding
dissociation constants were found to be in the low micromolar concentration range
(Table 1). Ligand shows a negligible absorbance at 295 nm wavelength. However, the experiments
were carried out at very low concentrations of protein and ligand to avoid the inner
filter effect.

Fig. 1. Fluorescence quenching of BSA and HSA with the GABA derivative. Serum albumin concentration
was kept constant at 0.5 ?M (in 20 mM Tris–HCl buffer of pH 7.0) and the ligand concentration
was varied from 0 to 5.5 ?M. a Spectral changes showing the quenching of intrinsic fluorescence emission of BSA
as a function of ligand concentration. b Quenching of intrinsic fluorescence of HSA in presence of ligand. These two spectra
were recorded at room temperature

Fig. 2. Determination of binding constants of the compound with serum albumins. a The changes in fluorescence intensity at the maximum emission wavelength and the
fitted Langmuir isotherm for the determination of binding constant with BSA. b Langmuir isotherm for HSA binding

Table 1. The K
_d
and K
_a
values for the binding of the GABA derivative to serum albumins as determined by the
fluorescence quenching experiments at room temperature

Thermodynamics of serum albumin binding

Equilibrium constant of a reaction changes with the temperature (Fig. 3), which is explained by van’t Hoff’s equation. The standard enthalpy and standard
entropy changes for the reaction can also be obtained from van’t Hoff’s equation.
The temperature depended fluorescence quenching study showed that the association
of compound 5 with serum albumins is thermodynamically favorable, which is evident
from the decrease in Gibbs free energy (Table 2). However, the binding with HSA was found to be enthalpy driven (negative ?H°) whereas
the binding with BSA was entropy driven (positive ?S°). It suggests that, despite
the structural similarity between the two proteins, the interactions with HSA are
thermodynamically different from those with BSA. The similar trend was also observed
for the binding of a naphthalene based fluorescent compound we previously reported
(Pal et al. 2015]). The molecule under investigation has structural similarity with the previously
reported molecule, however, it contains a phenyl substituent instead of a naphthyl
group.

Fig. 3. Determination of thermodynamic parameters of binding from van’t Hoff’s plot. a Decrease in the binding equilibrium constant with the decreasing temperature (303,
298 and 293 K) for the interaction with BSA and the fitted van’t Hoff equation. b Increase in the binding equilibrium constant with the decreasing temperature (308,
298 and 288 K) for the interaction with HSA and the fitted van’t Hoff equation

Table 2. Thermodynamics of the GABA derivative binding to serum albumins

Drug like properties of the GABA derivative

The molecular properties of the compound such as clogP, clogS, and polar surface area
(Bickerton et al. 2012]) are listed in Table 3. The clogP value of a compound is the logarithm of its partition coefficient between
n-octanol and water. It is a well established measure of the compound’s lipophilicity,
which influences its behaviour in a range of biological processes such as solubility,
membrane permeability, lack of selectivity and non-specific toxicity (Alam et al.
2011]). It has been shown for compounds to have a reasonable probability of being well
absorbed, their logP value must not be greater than 5.0 (Lipinski et al. 1997]). Besides, the aqueous solubility of a compound is also defined by logS, which significantly
affects its absorption and distribution characteristics. Typically, a low solubility
goes along with a bad absorption. Most of the drugs on the market have an estimated
logS value of about ?4. Table 3 lists the polar surface area of the compound as well, which should be less than 140
?
²
for a drug molecule (Lipinski et al. 1997]). Apart from lipophilicity/solubility and the polar surface area, the molecular weight
and the number of hydrogen bond acceptor/donor in the compound also follow the Lipinski’s
rule of five (Lipinski et al. 1997]).

Table 3. Molecular properties of the compound

Molecular modeling provides insight into the interaction with serum albumins

It has been established that serum albumin proteins have at least seven hydrophobic
grooves on their surface that provide a unique microenvironment and act as universal
receptors for many drug molecules (Curry et al. 1998]; Simard et al. 2006]; Reichenwallner and Hinderberger 2013]). Binding to these hydrophobic sites increases the solubility of hydrophobic ligands
in plasma and modulates their delivery to cells. The precise architecture of the binding
pockets is known from several crystallographic and NMR spectroscopic studies (Curry
et al. 1998]; Simard et al. 2006]; Hamilton 2013]). To gain a better insight into the interactions of compound 5 with serum albumins
molecular docking and dynamics analysis were carried out. Four different algorithms
were used to find the binding site of compound 5 on serum albumins: AutoDock 4.2,
AutoDock Vina, PatchDock/FireDock and SwissDock. These programs use different approaches
to model the ligand protein interactions, such as, PatchDock uses shape complementarity
whereas AutoDock 4.2 uses genetic algorithm. Molecular docking analysis by all these
four programs shows that the interactions of the compound 5 with serum albumins were
thermodynamically favorable (Table 4). The binding free energies computed by AutoDock Vina and SwissDock are very similar
to that of the experimentally obtained values (Table 2). Molecular docking also provides the insight into the most favorable binding site
for these compounds on the serum albumins (Fig. 4). The binding sites for the compound lie in the groove between domain I and domain
III of BSA, whereas it is within the domain I in case of HSA (Fig. 4). Thermodynamics analysis from temperature dependent quenching studies also suggest
differential nature of interaction with BSA and HSA. Thus, the docking studies which
produced two different binding sites for HSA and BSA also support the experimentally
obtained results (Table 2). Figure 4 also compares the best binding poses obtained by four different docking algorithms.
PatchDock is a rigid docking algorithm and, therefore, AutoDock 4.2, Vina and SwissDock
predicted ligand poses were used as input ligand structures for PatchDock. PatchDock
results suggests that vina outputs were the best solution among AutoDock 4.2, Vina
and SwissDock (Fig. 4). AutoDock Vina results, therefore, were used for further analysis and in molecular
dynamics simulation.

Table 4. Theoretical binding free energies as obtained by molecular docking experiments using
four different algorithms

Fig. 4. Interaction of the GABA derivative with serum albumins as obtained by molecular modeling.
a Binding site of compound 5 on BSA. b Comparison of the best docked conformations with BSA as obtained by AutoDock Vina
and AutoDock 4.2. c Comparison of Vina and SwissDock output for compound 5 binding with BSA. d PatchDock/FireDock shows Vina output has the best shape complementarity with BSA. e Overlap of best docked conformations of compound 5 with BSA. f Binding site of compound 5 on HSA. g Comparison of the best HSA binding poses as obtained by AutoDock Vina and AutoDock
4.2. h Comparison of Vina and SwissDock output for compound 5 binding with HSA. i PatchDock/FireDock showing Vina output has the best shape complementarity with HSA.
j Overlap of best docked conformations of compound 5 with HSA. Proteins are shown in
cartoon diagram and the ligands in stick model. The protein is colored in rainbow from N to C terminal. The three domains of serum albumin are marked with
I–III. Standard color representation is used to denote the elements, H, N and O, in ligand. Ligand C in
Vina, AutoDock 4.2, PatchDock/FireDock and SwissDock ouput are colored in white, green, cyan and magenta, respectively

Figure 5 shows the detailed interaction diagram for the interaction of compound 5 with BSA
and HSA. The figure shows that the interaction with BSA is mainly hydrophobic in nature,
however, a hydrogen bond formation was observed with Ser428. Residues from domain
I and domain III of BSA are involved in the interaction. On the other hand, interaction
with HSA was mediated by four hydrogen bonds. The side chain NH group of Arg117 of
HSA forms a hydrogen bond with the compound 5. Side chain NH of Lys137 was found to
form two hydrogen bonds and the Tyr161 was also involved to form a hydrogen bond with
the ligand through the phenolic OH group. Hydrophobic interactions also play a significant
role in the interaction of compound 5 and HSA.

Fig. 5. Detailed interaction of compound 5 with serum albumins. a 2D representation of the compound 5-BSA interaction diagram as obtained by LigPlot+.
Hydrophobic and hydrogen bonding interactions and the interacting protein side chain
residues are shown. b 2D representation of the compound 5-HSA interaction diagram as obtained by LigPlot+.
Hydrophobic and hydrogen bonding interactions and the interacting protein side chain
residues are shown

Dynamics of compound 5 binding with serum albumins

Molecular dynamics analysis was carried out to further investigate the stability of
the complex formation. It also allowed us to observe the effect of protein side chain
flexibility in the binding site as well as the effect of binding on the overall structure
of the protein. Figure 6, 7, 8, Additional files 2 and 3 summarize the changes observed during the 12 ns time scale of molecular dynamics
simulation. Figure 6a, b shows the root mean square deviation (RMSD) plots for the BSA and its complex
with the ligand. Changes in the RMSD values indicated the protein is undergoing a
conformational change. However, changes of the order of 1–3 Å are negligible for small,
globular proteins. RMSD changes also suggests that the simulation has converged very
rapidly and the protein/complex reached a stable conformation after around 1 ns. Figure 6c shows the changes in the radius of gyration of the protein in presence and in absence
of the compound. The radius of gyration is an indicator of the compactness of the
protein. Initially, in presence of the ligand BSA showed slightly relaxed conformation
(Fig. 6c), however, the structure converged after about 5 ns and attained a more compact
conformation. Figure 7 also shows the changes in RMSD and the radius of gyration of HSA in presence and
in absence of the compound.

Fig. 6. Molecular dynamics simulation of BSA-compound 5 complex. a Root mean square deviations (RMSD) of atomic positions of BSA backbone with respect
to the initial structure, in presence (blue) and absence (red) of compound 5. Ligand RMSD with respect to BSA is also shown (yellow). b RMSD of atomic positions of BSA side chains with respect to the initial atom positions,
in presence (blue) and absence (red) of compound 5. RMSD of compound 5 with respect to BSA is also shown. c Change in the radius of gyration of BSA with time in presence and absence of compound
5

Fig. 7. Molecular dynamics simulation of HSA-compound 5 complex. a RMSD of atomic positions of HSA backbone with respect to the initial structure, in
presence (blue) and absence (red) of compound 5. RMSD of compound 5 with respect to HSA is also shown (yellow). b RMSD of atomic positions of HSA side chains with respect to the initial atom positions,
in presence (blue) and absence (red) of compound 5. RMSD of compound 5 with respect to HSA is also shown. c Change in the radius of gyration of HSA with time in presence and absence of compound
5

Fig. 8. Fluctuations in the protein backbone and side chains when bound to compound 5. a Root mean square fluctuations (RMSF) in the backbone of BSA in presence (blue) and absence (red) of compound 5. Ligand contact sites are shown with vertical yellow lines. b RMSF in the side chains of BSA in presence and absence of compound 5. Ligand contact
sites are shown. c RMSF in the backbone of HSA in presence and absence of compound 5. Ligand contact
sites are shown. b RMSF in the side chains of HSA in presence and absence of compound 5. Ligand contact
sites are shown

Figure 8, on the other hand, highlights the residue-wise fluctuations in the backbone and
the side chains of BSA and HSA. The ligand contact sites are also highlighted. Figure 8 suggests that the backbone fluctuation slightly decreased near the ligand binding
site, both, in BSA and in HSA. Decrease in the backbone fluctuations near the ligand
indicated that the binding site attained a stable conformation. A video of the dynamics
of BSA-compound 5 complex is shown in Additional file 2: Video S1. Additional file 3: Video S2 shows the dynamics of HSA-compound 5 in water.

Over the course of simulation, the ligand makes stable as well as transient contacts
with the surrounding residues at the binding site. Such ligand contacts over the simulation
time scale are depicted in Figs. 9 and 10. Figure 9 summarizes the contacts with BSA. From this figure it appears that Arg458 of BSA
forms most contacts with the ligand. It forms hydrogen bonds, ionic interactions as
well as water bridges with the compound 5. Other residues that were fund to form hydrogen
bonds were Leu189 and Ser428. Tyr451 also maintained a consistent hydrophobic contact
with the compound. Figure 10 summarizes the contacts with HSA. The residues that maintained consistent contacts
over the simulation were Tyr161, Arg117 and Tyr138. The first two form predominantly
hydrogens bonds whereas the Tyr138 provides hydrophobic environment.

Fig. 9. Ligand contacts with BSA over the time frame of molecular dynamics simulation. Number
of hydrogen bonds, hydrophobic, ionic interactions and water bridges are measured.
The top panel shows the total number of specific contacts BSA makes with compound 5 over the course
of the trajectory. The bottom panel shows which residues of BSA interact with compound 5 in each trajectory frame. Some
residues make more than one specific contact with the ligand, which is represented
by a darker shade of orange, according to the scale to the right of the plot

Fig. 10. Ligand contacts with HSA over the time frame of molecular dynamics simulation. Number
of hydrogen bonds, hydrophobic, ionic interactions and water bridges are measured.
The top panel shows the total number of specific contacts HSA makes with compound 5 over the course
of the trajectory. The bottom panel shows which residues of HSA interact with compound 5 in each trajectory frame. Some
residues make more than one specific contact with the ligand, which is represented
by a darker shade of orange, according to the scale to the right of the plot