The temperature of maximum density for amino acid aqueous solutions. An experimental and molecular dynamics study

Highlights

• Temperature of maximum density (TMD) is determined for seven amino acids.

• Molecular dynamics simulation is performed for these systems.

• Good qualitative agreement is obtained between experimental and simulation results.

• All amino acids show TMD depression.

• Structural analysis from simulation reveals no enhancement of ice-like structures.

Abstract

The temperature of maximum density (TMD) for aqueous solutions of seven amino acids has been experimentally determined by means of density measurements versus temperature. The selected amino acids have been arginine, cysteine, glutamic acid, glutamine, lysine, methionine, and threonine. The TMD dependence against composition has been obtained from the experimental data and characterized through the Despretz constant. All amino acids induce a depression in the TMD as compared with that of pure water. If the mole fraction is selected as composition variable, a clear dependence against amino acid molar mass is observed, which disappears when TMDs are represented versus mass fraction; almost all data shrink onto a single straight line. It must be pointed out that the TMD depressions for all studied amino acids are quite larger than those previously observed for proteins. This suggests that TMDs for proteins cannot be explained as a simple, additive result of its constituents, and, therefore complex cooperative phenomena seem to take place. The partial molar volume at infinite dilution has been obtained from density data, and a consistency test between its temperature dependence and that of temperature of maximum density versus composition has been performed, obtaining satisfactory results. A molecular dynamics study for all the studied systems has been also carried out. Amino acids have been modeled through the OPLS-AA force field, whereas the TIP4P/2005 model was used for water. The temperature of maximum density and partial molar volume have been calculated from the simulated density, and the results are compared with experimental data. Although the agreement is only fair, similar qualitative trends were obtained. The simulated Despretz constants are smaller than the experimental ones, a result that was already previously observed for methanol aqueous solutions. A structural analysis of water molecules in solution along the MD trajectories showed no enhancement of ice-like structures in complete agreement with the TMD decrease with concentration.

Introduction

Ice Ih is the solid form of water below the melting point at atmospheric pressure [1]. It is a hexagonal structure where each water molecule takes part in four hydrogen bonds [2]. The microscopic form of water after melting resembles this ice-type structure, although with a significant degree of distortion. As temperature increases from the melting point, a new form of water structure emerges, structurally more disordered and denser. The occupancy of both structures by water molecules gradually changes as temperature increases, being the high density one the predominant form at higher temperatures. This feature is responsible for one of the most famous anomalies in the behavior of water, namely, the presence of a density maximum for a given temperature at atmospheric pressure [3]. The shape of this T-ρ curve changes slightly with pressure: the low density ice-type structure is partially distorted by increasing pressure, favoring the formation of the high density structure. This shifts the temperature of maximum density, TMD, towards lower temperatures [4].

Addition of solutes to water also affects the location of the TMD. For more than six decades, a large body of work has focused on the study of the variation of the TMD due to the addition of solutes to water, both electrolytes [[5], [6], [7], [8], [9]] and non-electrolytes [[10], [11], [12], [13], [14], [15], [16], [17], [18], [19], [20], [21], [22], [23], [24], [25]]. Wada and Umeda [11] have shown that the difference between the TMD of a mixture of solute mole fraction x from that of pure water, ΔT = TMD (mixture)- TMD (pure water), can be split into two contributions

$$\Delta T=\frac{-x{\alpha }_1{v}_1^{\%,273.15}}{2\left(1-x\right){\alpha }_2{v}_2^{\%,277.13}}-\frac{d{v}^E/dT}{2\left(1-x\right){\alpha }_2{v}_2^{\%,277.13}}$$

by considering the molar volume of the pure solute (component 1) and pure water (component 2) in the neighborhood of the TMD is given by

$$v_1^{\%}=v_1^{\%,273.15}\left(1+{\alpha }_1\left(T-273.15\right)\right),$$

$$v_2^{\%}=v_2^{\%,277.13}\left(1+{\alpha }_2{\left(T-277.13\right)}^2\right),$$

respectively, where $v^{\%}$ denotes molar volume of pure component, α is the thermal coefficient of the molar volume, and v^E is the excess molar volume. The first term of the second member of equation (1) is the ideal contribution ΔT^id, which defines the dependence on the solute mole fraction, x, (a monotonous decrease of ΔT as the solute mole fraction increases) when v^E is zero. The second term represents the structural contribution ΔT^st (non-ideal contribution) and it allows the classification of the solutes as “structure makers” (those that promote the low density ice-like water structure) if ΔT^st is positive, and “structure breakers” (those that promote the high density disordered structure) if ΔT^st is negative. This classification is, however, not unique [26] and in some cases ΔT is used instead ΔT^st. This last option is easier to apply since it does not require the previous knowledge of the parameters of Equation (2), which are not easily available for solid solutes such as electrolytes or biological macromolecules.

The characterization of the protein-water interaction is one of the great challenges of the biological sciences. The influence of small amounts of protein on the structure of solvation water will conversely affect the way in which hydrophobic and hydrophilic effects compete, and subsequently lead to protein folding in solution. As previously commented, the TMD variation is a good indicator of the changes induced by the solute in the structure of water. It is then of utmost importance to understand how single amino acids influence the structure of water and subsequently alter its TMD in order to deepen our understanding of water-protein interactions [27]. This work was initiated by Kuppers [28] and Kaulguld et al. [29],and recently continued by Romero et al. [30] and by Troncoso et al. [31], but to date the full 20 amino acid set present in natural proteins has yet to be completely studied experimentally. It is the aim of this work to contribute further to fill this gap, both experimentally and with the aid of molecular simulation.

Molecular simulations are nowadays a very powerful tool to facilitate a microscopic interpretation of macroscopic experimental results. The great increase of the computational power and improvements in modeling algorithms during the last decade has made feasible the simulation of complex systems, such as biological macromolecules, using fairly realistic molecular descriptions. Application of this methodology requires the use of accurate force fields (there is a wide variety of choices, such as OPLS-AA [32], AMBER [33], CHARMM [34], and MARTINI [35], among others) that adequately describe the inter and intramolecular interactions. The common practice is assessing first the ability of the force field to reproduce the required experimental macroscopic behavior for a given set of properties and thermodynamic states. This is crucial in mixtures, since most force fields are usually developed for pure components and their application to mixtures requires the use of somewhat arbitrary combining rules. In our case, we want to analyze the “structure maker/breaker” character of various amino acids from a microscopic stand point. This implies that the variation of the TMD with the solute concentration is the key quantity to be reproduced.

In this work we have thus focused on aqueous solutions of arginine, cysteine, glutamic acid, glutamine, lysine, methionine, and threonine in the dilute region. On the one hand, the variation of the TMD of water due to the addition of these seven amino acids was experimentally determined. This was done by carrying out density measurements in the temperature interval (273.65–283.15) K in 0.5 K steps using a vibrating tube densimeter DMA5000 from Anton Paar. Density data were also used to evaluate the partial molar volumes of the amino acids at infinite dilution. With this, we have performed a thermodynamic consistency check with the TMD values following the scheme devised by Armitage et al. [14]. On the other hand, extensive NpT Molecular Dynamics (MD) simulations of the studied systems, in the temperature range 238.15–308.15 K and at several solute concentrations (x = 0.0, 0.0025, 0.0050, 0.0075, 0.01), were performed using the OPLS-AA force field for amino acids together with the TIP4P/2005 model for water [36]. The purpose of the simulations was to evaluate the ability of the chosen model to account for the variation of the TMD with the solute concentration and also to analyze the promotion/destruction of ice-like water structures during mixing. This last point was carried out in a quantitative way by performing a specific structural analysis of water molecules in solution along the MD trajectories. We will see that although the simulation reproduces the correct qualitative trends, it provides a stronger depression of the TMD with the addition of a solute. Structural analysis showed no enhancement of the ice-like water structure confirming the ‘structure-breaker’ character of the solutes.