Is styrene competitive for dopamine receptor binding?

Introduction

Styrene is an organic molecule extensively used in industrial processes, for example, as a reactive diluent in epoxy resins, as an intermediate in the preparation of a variety of agricultural and biological chemicals, cosmetics, surface coatings, and treatment of textiles and fibers, and as a raw material for the production of phenyl stearyl alcohol in perfume industries. Styrene enters the human body through several routes, mainly through the respiratory system. More than 80% of the inhaled styrene undergoes bioactivation to styrene oxide by cytochrome P450 monooxygenases [1]. A considerable amount of evidence has been collected in the literature showing that styrene and styrene oxide are neurotoxic, although their precise mechanism of activity is unclear and quantitative data are lacking [2].

Styrene and styrene oxide are supposed to block the functioning of various neurotransmitters in the brain including dopamine and serotonine [3,4,5]. Several studies carried out on rodent postmortem brain tissue suggest that styrene may alter dopaminergic neurotransmission in rabbit or rat brain. Moreover, in vitro studies suggest that they inhibit the uptake of dopamine in purified synaptic vesicles prepared from rat brain striatum [5].

Significant scientific evidence has been collected in the last decades demonstrating that exposure to styrene oxide, either alone or in concert with noise exposure, has severe ototoxic effects also on humans. Occupational medicine studies, performed among workers exposed to this solvent, show a significant decrease of the distortion product oto-acustic emission (DPOAE) levels in styrene-exposed ears, a situation that normally means lowered cochlear response [6,7,8].

In the outer hair cells membrane of the normal ear, proteins able to bind dopamine, have been indeed detected. They exist in five variants, known as dopaminergic receptors (Dr) D1, D2, D3, D4, and D5. Interestingly, studies on rodents suggested that cochlear function was normal in DrD3, DrD4, and DrD5 knock-outs models. On the contrary, DrD1 and DrD2 knock-outs showed slight, but significant enhancement and suppression of cochlear responses, respectively, in both the neural output auditory brainstem response wave 1 and outer hair cell function DPOAE [9].

Pharmacological studies of cochlear dopaminergic signaling using classic agonist/antagonist have shown that DrD2 antagonists are most effective in altering the cochlear microphonics and reducing the DPOAE response in the guinea pig [9].

In this work, we have investigated the potential role of styrene oxide in altering the dopaminergic pathway in the ear. We have used a computational approach to estimate the binding affinity of styrene oxide to the dopaminergic receptor DrD2 and have compared with that of dopamine. We have used molecular docking to identify the poses on the DrD2 receptor where the two compounds have the highest binding affinity. We have then performed classical molecular dynamics (MD) simulations to provide a quantitative evaluation of their binding affinities. The ligand–receptor affinity has been evaluated by measuring the difference of the binding free-energy (Gibbs potential, ∆G) between the configuration where the ligand is bound to the receptor and the situation in which it is “infinitely” far away. Binding free-energy computations are performed by following the umbrella sampling approach [10,11] with the distance between the center of mass (COM) of ligand and receptor as the reaction coordinate.

We find that the binding affinity of dopamine is larger but comparable with that of styrene oxide meaning that at high concentrations the latter can compete with dopamine for DrD2 binding.

Materials and methods

Classical MD simulations represent a powerful tool to analyze dynamic and thermodynamic properties of biochemical systems. In particular, it allows computation of chemical affinities in terms of free-energy (Gibbs potential) differences (∆G).

The computational work is logically divided into three parts: (1) molecular docking analysis, (2) a set of straightforward classical MD simulations, and (3) a subsequent binding free-energy calculation. We now briefly illustrate these three items in turn.

Molecular docking

Molecular docking is a reliable strategy to identify the poses on the DrD2 receptor where either styrene oxide or dopamine has the highest binding affinity. Molecular docking calculations have been carried out using both the SwissDoc server [12] and AutoDockTools4.2.6 (ADT) suite of codes [13] as in ref. [14], obtaining identical results.

The DrD2 receptor structure was taken from the crystal data of ref. [15] (see Figure 1). Dopamine and styrene oxide structures (see Figure 2) were downloaded from the Automatic Topology Builder (ATB) repository [16,17], where dopamine is referred to as ATB_molid: 337402 and styrene oxide as ATB_molid: 38701.

Figure 1

The DrD2 receptor [15] embedded in the POPC membrane. Upper and lower membrane planes are depicted in red and blue, respectively.

Figure 2

Dopamine (top) and styrene oxide (bottom) structures. Hydrogen atoms are white, carbon atoms are green, oxygen atoms are red, and nitrogen atoms are blue.

Molecular dynamics

The highest affinity regions identified by the molecular docking analysis were used as starting configurations to perform classical all atoms MD simulations to identify the stable and well-equilibrated configurations of the two receptors plus ligand (DrD2 plus dopamine and DrD2 plus styrene oxide) systems at T = 300 K. The simulations were carried out with the GROMACS 2016.1 package [18,19] employing the gromos54a7 force field [20].

The simulation strategy we used, identical for the two systems we are interested in (DrD2 embedded in a bilayer plus either styrene oxide or dopamine), is the following. First, each system (receptor plus ligand) was relaxed in vacuum via a steepest descent minimization. Second, after having identified the correct orientation of the DrD2 receptor within the membrane by means of the Orientations of Proteins in Membranes (OPM) database [21], the receptor plus ligand system was embedded in an explicit POPC lipid bilayer [22]. We used a tetragonal simulation box with size (19 × 19 × 13) nm³ and periodic boundary conditions. The box was sufficiently large to prevent interaction with the system images. At this point, water and an appropriate number of counterions (necessary to get system neutrality) are added. The simple point-charge (SPC) description was used to model water molecules. Each of the two systems contains approximately 4 × 10⁵ atoms.

The simulation procedure was started by minimizing the potential energy of the whole system, solute, lipid membrane, solvent, and counterions, using the steepest descent algorithm. Then, the system was equilibrated for 12 ns in the NpT ensemble at 150 K. The successive production run was a 250 ns long NpT simulation at 300 K. The temperature was kept constant by using the V-rescale thermostat [23] with a 0.1 ps coupling time. The pressure was kept constant to 1.0 bar by using the Berendsen barostat [24] with a 1 ps coupling time and an isothermal compressibility of 4 × 10⁻⁵ bar⁻¹. MD equations of motion were integrated with a time step of 2 fs.

Bonds with hydrogen atoms were constrained according to the LINCS [25] algorithm. The radius of the sphere defining the list for pairwise interactions was set to 1.0 nm. Consistently, the van der Waals interaction was also cutoff at 1.0 nm. The Particle Mesh Ewald method [26] for the long-range electrostatic interactions was employed using a 1.0 nm real space cutoff. Initial velocities were extracted from a Maxwell distribution at T = 300 K.

The analysis of the collected MD trajectories was carried out by using standard GROMACS tools, VMD tools [27], as well as some ad-hoc written python scripts.

Umbrella sampling and potential of mean force

The binding affinity (∆G) was computed by making use of the consolidated umbrella sampling procedure [10,11] as implemented in the GROMACS suite.

As extensively explained in [28], an efficient way to implement the umbrella sampling in the case of receptor–ligand interaction is to take as a reaction coordinate the distance, ζ, between the ligand and the receptor COMs, conventionally taken at ζ ₀ = 0. Sets of configurations (“windows”) were successively generated, each set corresponding to simulations performed with the ligand bound by a harmonic potential centered at increasing values of the receptor–ligand distance from ζ ₀ to some final, large value, ζ _F , where the ligand–receptor interactions can be considered vanishingly small.

This dynamical constraint allows the ligand to sample the configuration space in a region along with the reaction coordinate at increasing receptor–ligand distances (pulling). Taking |ζ _F − ζ ₀| = 4.4 nm, a convenient choice is to use 22 “windows” with the coordinate of the center of the harmonic potential spaced by 0.2 nm.

One gets in this way the potential of mean force (PMF) profile in a set of 22 “overlapping windows” each centered around one of the 22 ζ _n values (ζ _n = ζ ₀ + 0.2n, n = 1, …, 22) spanning the [ζ ₀, ζ _F ] interval. In each one of the windows, the system was simulated for a total of 24 ns of which only the last 16 ns were taken into account for the Gibbs free energy calculations. The data collected along the 22 windows were finally combined with the help of the weighted histogram analysis method (WHAM) [10,29] algorithm. To carry out this step, we used the routine gmx.wham of the GROMACS package.

All numerical simulations were performed using the Marconi cluster of the CINECA Consortium (CPU Intel Xeon 8160 @2.1 GHz).

Results molecular docking

By using the Swiss Dock server and ADT, we were able to determine binding pose regions for dopamine and styrene oxide on the DrD2 receptor. Both ligands occupy the same pocket as risperidone in pdb-id:6cm4 [15] and haloperidol in pdb-id:6luq [30], hence corroborating the molecular docking identification of this pocket as a binding site for small molecules in DrD2. In Figure 3, the highest affinity poses are shown. Quite interestingly one finds that these regions are almost identical for the two ligands and, as shown in Figure 4 and in the first column of Table 1, the amino acids involved in the interaction are almost the same.

Figure 3

Binding poses for styrene oxide (a) and dopamine (b) on the DrD2 receptor.

Figure 4

Magnification of the structure of the initial DrD2 binding pocket of styrene oxide (a) and dopamine (b). The DrD2 amino acids in contact with the ligands are explicitly drawn and are listed in the leftmost column of Table 1.

Table 1

List of the receptor amino acids belonging to the interaction sphere of styrene oxide (top panel) and dopamine (bottom panel) at the beginning (left column) and at the end (right column) of the MD simulation. The amino acids in orange are the ones belonging to both the ligands interaction spheres at the beginning of the MD simulations. The amino acids in purple are the ones belonging to both the ligands interaction spheres at the end of the MD simulations. The amino acids that are underlined are the ones remaining in the ligands interaction spheres during the whole simulation. The receptor amino acids belonging to the ligand interaction sphere are those having atoms at a distance shorter than 3.5 Å from any of the ligand atom

Molecular dynamics

As mentioned in Material and Methods, the receptor plus ligand structures identified by the molecular docking algorithm were used as starting configurations for the MD simulations of the DrD2 plus styrene oxide and DrD2 plus dopamine systems.

A standard way to monitor equilibration is to look at the root mean square displacement (RMSD) evolution of the systems against the simulation time. In Figure 5, we plot the RMSD of the DrD2 Cα atoms for the two systems. We see that at T = 300 K both systems reach equilibrium after about 180 ns and the DrD2 structures, in both cases, differ by about 5.5 Å with respect to the starting one. In Figure 5 inset, we plot the RMSD of the DrD2 Cα atoms of the amino acids constituting the ligands interactions sphere (the pocket). As shown, the structure of the pocket behaves differently in the two systems. In the case of dopamine, the RMSD of the pocket is very small (less than 1 Å). On the contrary, in the case of styrene oxide, we notice a higher, although still very small (about 2 Å), value of the RMSD.

Figure 5

RMSD of the DrD2 Cα atoms for the two systems, computed in the NpT ensemble at T = 300 K. The black vertical line marks the time t = 180 ns from which the two systems can be considered in thermodynamic equilibrium. In the inset, the RMSD of the Cα atoms of the amino acids constituting the interaction sphere of dopamine and styrene oxide.

This behavior is in good agreement with the results shown in Table 1, where a detailed comparison of the DrD2 amino acids in contact with the two ligands is presented. At the beginning of the MD simulation (first column), the two ligand interaction spheres (as determined by the Molecular Docking algorithm) are almost identical, while at the end (second column), only two receptor amino acids, namely, Phe363 and Trp367, remain in contact with both ligands. We notice that the dopamine interaction sphere is more stable than the styrene oxide one. In fact, during the MD simulation styrene oxide moves to a binding pocket whose structure differs from the initial one, as it is detailed in the top panel of Table 1, while the dopamine pocket remains almost the same. One can compare the amino acids of the styrene oxide binding pocket at the beginning of the simulation displayed in the left panel with those at the end of the simulation displayed in the right panel. Only two (the underlined ones) are conserved.

∆G evaluation

Following the standard pulling strategy outlined in Materials and Methods, we have proceeded to the computation of the binding affinity of the DrD2 plus dopamine and DrD2 plus styrene oxide.

In Figure 6, the PMF profile for the two systems is displayed. The ∆G values associated with the two pulling processes are estimated by computing the average of the PMF data from the point in which the plateau is reached (ζ > 3.8 nm) to ζ = 4.4 nm, a position where both ligands are outside the membrane and completely in water. The errors on these numbers are evaluated by means of the bootstrapping technique [31], as implemented in gmx.wham routine of GROMACS. The results for the binding free energy of the two systems are as follows:

(1) Δ G DrD2 − styrene oxide = ( 111 ± 8 ) k J / m o l .

(2) Δ G D r D 2 − d o p a m i n e = ( 287 ± 12 ) k J / m o l .

Figure 6

Potential of mean force for the binding of the two studied ligands to the DrD2 receptor.

The key result of this investigation is that dopamine has a higher affinity than styrene oxide for DrD2. However, since the two affinities only differ by a factor 2.5, we argue that at high concentration styrene oxide is able to compete with dopamine for DrD2 binding, thus being able to play a neurotoxic role by occupying the DrD2 dopaminergic receptor.

Conclusions

Occupational medicine studies, performed on workers exposed to environmental toxins and chemicals, show evidence of ototoxic effects. Among the ototoxic compounds to which workers are exposed, styrene is one of the most largely detected and investigated.

A signicant amount of evidence [2,3,4,5] show that styrene oxide (the bio-activated product of styrene) has a significant impact on the dopaminergic neurotransmission with the DrD2 receptor playing a crucial role in the dopamine activity of the auditory system.

In the present study, we have computed the binding affinities of dopamine and styrene oxide to the DrD2 receptor with the standard methods of classical MD. The main purpose of this study is to investigate whether styrene oxide is able to compete with dopamine for DrD2 binding and hence whether it can be held responsible for the impairment of cochlear responses.

Although a direct comparison of ∆G _{DrD2−styrene oxide} and ∆G _{DrD2−dopamine} (see equations (1) and (2)) shows that dopamine has a higher binding affinity for DrD2, the difference is only a factor of about 2.5. This means that, if present at sufficiently high concentration, exogenous styrene oxide can compete with endogenous dopamine for dopamine receptor binding, altering dopaminergic neurotransmission with consequent neuro-ototoxic effects.