In silico study of levodopa in hydrated lipid bilayers at the atomistic level

Highlights

• Levodopa is simulated in the zwitterionic as well as in the neutral state.

• Levodopa zwitterions form a hydrogen bond network with water and phospholipids.

• Levodopa prefers to reside at the water/lipid interface and in the water phase.

• PMFs show that the hydrophobic region of the bilayers is impermeable by levodopa.

Abstract

This article presents atomistic molecular dynamics and umbrella sampling simulations of levodopa at various concentrations in hydrated cholesterol-free 1,2-dipalmitoyl-sn-glycero-3-phosphocholine (DPPC) and cholesterol-containing 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) bilayers. Levodopa is the standard medication for Parkinson's disease and is marketed under various trade names; in the context of this article, the levodopa molecule is mostly studied in its zwitterionic form but some results concerning the neutral levodopa are presented as well for comparison purposes. The motivation is to study in detail how levodopa behaves in different hydrated lipid membranes, primarily from the thermodynamic point of view, and reveal aspects of mechanism of its permeation through them. Dependencies of properties on the levodopa concentration are also investigated. Special attention is paid to the calculation of mass density profiles, order parameters and self-diffusion coefficients. Levodopa zwitterions, which form a hydrogen bond network with water and phospholipid molecules, are found to be preferentially located at the water/lipid interface, as well as in the aqueous phase surrounding the cholesterol-free and cholesterol-containing bilayers. This is concluded from the potentials of mean force calculated by umbrella sampling simulations as levodopa is transferred from the lipid to the aqueous phase along an axis perpendicular to the two leaflets of the membranes.

Introduction

The water-phospholipid interface constitutes a system of great significance in Biophysics and Biochemistry, since it is encountered in the plasma membranes of eukaryotic and prokaryoric cells and, among its many functions, it regulates the flow of substances in and out of the cell [1]. As a consequence, systematic studies of this interface are of paramount importance from both a scientific and a practical point of view. If one wishes to keep explicitly all or most of the degrees of freedom of the molecules (full-atom or united-atom atomistic models), unbiased molecular dynamics simulations may prove insufficient for sampling rare events due to rather high free energy barriers that cannot be overcome during the rather short simulation times. To this end, special computational techniques have been developed that allow the system to overcome free energy barriers and efficiently compute free energy differences. Such an accelerating sampling technique that is widely used in molecular modeling of biomolecular systems is umbrella sampling [[2], [3], [4], [5]]. The Weighted Histogram Analysis Method (WHAM) [5,6] can be utilized for the calculation of the potential of mean force (PMF) from a set of umbrella sampling simulations along the selected reaction coordinate.

Levodopa (3,4-dihydroxy-l-phenylalanine) is the standard medication for Parkinson's disease and is marketed under various trade names [[7], [8], [9], [10]]; References 7–10 contain useful information about the pharmacokinetics, pharmacodynamics, synthesis and therapeutic action of levodopa. Parkinson's disease is an incurable neurodegenerative disease of the central nervous system (CNS). It is believed that it is caused by the progressive loss of dopamine in the brain due to degeneration of the dopaminergic neurons [11]. Τhe role of the blood brain barrier (BBB) is determinant, inasmuch as the drugs aiming at the CNS must permeate the BBB [1,12,13]. Many drugs acting on the CNS pass the BBB passively through diffusion due to a concentration gradient [13]. Another mechanism of permeation is through transport proteins that are expressed in the BBB [13,14]. The role of levodopa is to permeate through the BBB and to be decarboxylated to dopamine with the aid of the enzyme aromatic l-amino acid decarboxylase (also known as DOPA decarboxylase) [10]. Carbidopa, which is coadministered with levodopa and is a known inhibitor of DOPA decarboxylase, prevents peripheral metabolism of levodopa to dopamine so that as much as possible levodopa reaches the brain [8,10]. Levodopa crosses the BBB with the help of the membrane protein LAT1 (L-type amino acid transporter 1) [8,15,16], whereas its absorption in the small intestine is effected by active long-chain neutral amino acid transporters [8].

The motivation of this article is to study in detail how levodopa behaves in different hydrated lipid membranes and reveal aspects of the permeation mechanism through them. The role of the lipid environment is of great significance, since the abovementioned transport proteins are embedded in lipid bilayers. DPPC, the constituent of the first membrane we study, is a benchmark for phospholipids, encountered in cell membranes, monolayers, liposomes, etc [17]. POPC, on the other hand, has been employed, either alone or along with cholesterol, as a constituent of very simple BBB mimetics in the context of some in silico studies [[18], [19], [20], [21]]. In the study of Thai and coworkers [21], a computational method is presented for the prediction of BBB permeability of various substances from steered molecular dynamics simulations. The article of Orlowski and coworkers [22] reports results from unbiased molecular dynamics simulations of levodopa and dopamine in a 1,2-dilauroyl-sn-glycero-3-phosphocholine (DLPC) bilayer; DLPC has the same headgroup as DPPC, and some findings from the aforementioned article are briefly discussed and compared with the results presented herein. Another recent paper focusing on dopamine in its neutral state is that of Megariotis et al. [23] in which the same methodology as described herein was applied. The active role of lipids in neurotransmitter dynamics is reviewed in a recent article published by Postilla et al. [24].

In this article we conduct biased and unbiased atomistic molecular dynamics simulations of cholesterol-free DPPC and cholesterol-containing POPC hydrated bilayers at various concentrations of levodopa to study in detail the interactions of levodopa with its environment from a nanoscopic point of view. Special attention is paid to the thermodynamic description of the abovementioned systems by calculating the PMFs of levodopa as it is taken, via biased simulations, from the lipid to the water phase along the selected reaction coordinate, namely the axis perpendicular to the two leaflets of the lipid membrane. The PMFs constitute free energy profiles which provide insight into how the drug behaves in the heterogeneous biological systems considered herein. Unbiased simulations, on the other hand, allow us to compute properties such as mass density profiles, self-diffusion coefficients, and specific order parameters. Our simulation strategy constitutes a powerful tool for the in silico study of the levodopa in different lipid environments, since it provides information at the nanoscopic level that is not easily extracted from in vivo, ex vivo or in vitro experiments.

The article is organized as follows. First, the models and methods along with all necessary details concerning biased and unbiased molecular dynamics simulations are discussed in Section 2. Next, the results are presented and discussed extensively in Section 3; the first part of this section focuses on the thermodynamic properties, whereas the rest of the section concerns properties calculated by the unbiased simulations. Section 4 summarizes the conclusions that are extracted from the presented results.