Fatty acid binding into the highest affinity site of human serum albumin observed in molecular dynamics simulation

https://doi.org/10.1016/j.abb.2015.05.018Get rights and content

Highlights

  • Simulations were used to study the association of stearic acid in site FA5 of albumin.

  • Both kinetics and energetics of the ligand insertion were accurately determined.

  • A metastable state along the route towards the binding site was identified.

  • A comprehensive pathway is suggested for the binding of stearic acid in site FA5.

Abstract

Multiple molecular dynamics simulations were performed to investigate the association of stearic acid into the highest affinity binding site of human serum albumin.

All binding events ended with a rapid (<10 ps) lock-in of the fatty acid due to formation of a hydrogen bond with Tyr401. The kinetics and energetics of the penetration process both depended linearly on the positional shift of the fatty acid, with an average insertion time and free energy reduction of, respectively, 32 ± 20 ps and 0.70 ± 0.15 kcal/mol per methylene group absorbed. Binding events of longer duration (tbind > 1 ns) were characterized by a slow exploration of the pocket entry and, frequently, of a nearby protein crevice corresponding to a metastable state along the route to the binding site.

Taken all together, these findings reconstruct the following pathway for the binding process of stearic acid: (i) contact with the protein surface, possibly facilitated by the presence of an intermediate location, (ii) probing of the site entry, (iii) insertion into the protein, and (iv) lock-in at the final position. This general description may also apply to other long-chain fatty acids binding into any of the high-affinity sites of albumin, or to specific sites of other lipid-binding proteins.

Introduction

Fatty acids are structural components of cells and constitute a fundamental source for energy production when blood sugar is low. Only a tiny percentage (<0.01%) of fatty acids is free in the plasma, whereas a small fraction is bound to lipoproteins and the great majority to serum albumin [1]. This protein is the most abundant in the human blood and provides also the transport of other metabolites, drugs and ions [2], [3].

Long-chain non-esterified fatty acids attach non-covalently to albumin in seven binding sites [4]. Up to 6–7 fatty acids are bound to each albumin molecule under extreme conditions, such as fasting, intense stress or exercise, or in some disease conditions (see [5], [6], [7] and references therein). In contrast, an average of 0.1–2 molecules are present under normal physiological conditions [5]. The protein binding sites are not all equivalent: three have high affinity for fatty acids and four have lower affinity [6], [8]. High-affinity sites are named FA2, FA4 and FA5, and consist of long and narrow hydrophobic pockets in which the methylene chain of each fatty acid lies in an almost extended conformation [9], while the carboxylate head-group is hydrogen-bonded and additionally coordinated by a charged protein sidechain. In particular, the binding pocket with the highest affinity is site FA5 [8], [10], in which the fatty acid head-group forms a hydrogen bond (HB) with the Oη atom of Tyr401 and a salt bridge with the protonated amino group of Lys525.

Crystallographic structures [4], [9], [11] are available for human serum albumin complexed with saturated fatty acids ranging from capric (C10:0) to stearic acid (C18:0). Interaction with the latter is important not only for its nutritional relevance [12], but also from a molecular point of view. In fact, the relatively long chain of stearic acid is particularly suited to investigate the insertion of a fatty acid, because it can probe albumin binding sites in their full depth. Unfortunately, microscopic details concerning the dynamics of the lipid binding are currently not easily accessible to the experiment. Molecular dynamics (MD)1 represents an interesting tool to tackle this issue [13], [14] and has already been successfully employed in a few cases [15], [16], [17], [18], [19], [20].

Complete association events to fatty acid-binding proteins have been described in simulation in a small number of instances. Bello and García-Hernández performed a total of 20 unbiased MD runs to explore the binding of lauric and palmitic acid to β-lactoglobulin [18], starting with the ligand outside the protein binding site. The ligand was observed to have at least a partial penetration in approximately half of the simulations, and ended up forming a stable complex in three instances. In one case the penetration stage took approximately 1 ns, whereas in the other two cases the process was more gradual and slow. Tsfadia and coworkers reported an event of full insertion for palmitic acid [15] in a set of simulations on toad liver basic fatty acid binding protein (FABP). The estimated time for the completion of the process was on the order of 1 ns, and no uncertainty could be associated to this single occurrence.

In other studies, indirect information on the fatty acid association process was obtained ‘in reverse’ by simulating the dissociation process under non-equilibrium conditions. An extremely high temperature (1500 K) was used in an early report on intestinal FABP [21] to promote and speed up the dissociation process. More recently, a non-physical steering force was employed, either with constant direction [22] or randomly oriented [23], to favor dissociation of fatty acids from intestinal and liver FABP, respectively. Several details on structural intermediates and physical interactions involved along the association/dissociation pathway were obtained from these MD simulations, but no reliable kinetic features could be deduced.

Most of the previous results on protein-fatty acid association and dissociation involve proteins of the FABP family [24], which are relatively small (∼15 kDa) and with a simple structure (a single β-barrel), and generally bind only one fatty acid (two in the case of liver FABP). In contrast, in the present work MD simulations have been used to investigate the process of association of stearic acid into the highest-affinity binding site of albumin (66 kDa and multidomain), its major physiological carrier. By systematically varying the starting position of the ligand within the binding pocket, multiple association events were observed under equilibrium conditions. The results reveal key details of the kinetics and energetics of the association process, giving a comprehensive description of the binding pathway for the fatty acid in site FA5.

Section snippets

Molecular modeling and simulation

MD was performed by using the GROMACS simulation package [25] with the force field GROMOS 53a6 [26]. VMD was used for molecular modeling and visualization [27]. Human serum albumin was modeled on the basis of the fatty-acid loaded crystallographic structure [9], 1E7I entry of the Protein Data Bank. Ionizable amino acid residues were adapted to mimic protonation at neutral pH. The topology for stearic acid was built by combining the parameters used in GROMOS 53a6 for the carboxylate, methylene

Equilibrium dynamics of fully inserted stearic acid

An MD simulation was performed to characterize the coordination of a stearic acid molecule completely inserted into site FA5 in human serum albumin. The simulation indicates that albumin preserves its distinctive all-alpha secondary structure and heart-shaped tertiary structure, whereas the fatty acid molecule fluctuates around the starting configuration. Fig. 1 shows an ensemble of representative conformations of the fatty acid obtained during the simulation (Fig. 1a), compared with the

Discussion

The knowledge of the association process of fatty acids is fundamental to understand how transport proteins recognize and bind such ligands. In this work, the kinetic and energetic landscape of the binding reaction of stearic acid to site FA5 of human serum albumin was followed ‘backwards’. We first characterized the dynamics of the fatty acid in its unique binding position (i.e., fully inserted), and later investigated regions of the free energy funnel where the ligand is either only partly

References (46)

  • M. Dockal et al.

    J. Biol. Chem.

    (1999)
  • J.R. Simard et al.

    J. Mol. Biol.

    (2006)
  • G. Fanali et al.

    Mol. Aspects Med.

    (2012)
  • A.A. Bhattacharya et al.

    J. Mol. Biol.

    (2000)
  • S.I. Fujiwara et al.

    Biophys. J.

    (2008)
  • S. Curry et al.

    Lipids

    (1999)
  • S. Fujiwara et al.

    Biochim. Biophys. Acta Gen. Subj.

    (2013)
  • B. Rizzuti et al.

    Arch. Biochem. Biophys.

    (2013)
  • Y. Tsfadia et al.

    FEBS Lett.

    (2007)
  • R. Friedman et al.

    Biophys. J.

    (2006)
  • J.D. Ashbrook et al.

    J. Biol. Chem.

    (1975)
  • A.O. Pedersen et al.

    J. Biol. Chem.

    (1988)
  • C.N. Pace

    Methods Enzymol.

    (1995)
  • H.C. Berg et al.

    Biophys. J.

    (1977)
  • A. Saifer et al.

    J. Lipid Res.

    (1971)
  • M. Fasano et al.

    IUBMB Life

    (2005)
  • S. Curry et al.

    Nat. Struct. Biol.

    (1998)
  • T. Peters

    All About Albumin. Biochemistry, Genetics, and Medical Applications

    (1995)
  • J.R. Simard et al.

    Proc. Natl. Acad. Sci. U.S.A.

    (2005)
  • H. Sampath et al.

    Lipids

    (2005)
  • L.B. Levin et al.

    J. Mol. Model.

    (2010)
  • M. Bello et al.

    Biopolymers

    (2014)
  • S. Evoli et al.

    Proteins Struct. Funct. Bioinf.

    (2014)
  • Cited by (31)

    • Molecular simulations of proteins: From simplified physical interactions to complex biological phenomena

      2022, Biochimica et Biophysica Acta - Proteins and Proteomics
      Citation Excerpt :

      Studies of protein-drug complexes with ligands selectively bound in specific sites is difficult in ‘wet lab’ experiments, and much easier in MD simulations [89,90]. Classical MD simulation techniques are among the simplest methodologies to observe the dynamics of association of a ligand in a protein binding site [91–93], although a high computational cost is necessary to obtain unbiased binding events starting from the ligand in the bulk solvent [94]. An alternative and very common choice is to use simple end-point free energy calculations, which require an evaluation of the protein-ligand interactions only at the starting and ending state along the binding pathway.

    • Virtual screening in drug discovery: a precious tool for a still-demanding challenge

      2020, Protein Homeostasis Diseases: Mechanisms and Novel Therapies
    • Warfarin increases thermal resistance of albumin through stabilization of the protein lobe that includes its binding site

      2019, Archives of Biochemistry and Biophysics
      Citation Excerpt :

      Finally, both the pre- and post-transition baselines were fitted by a cubic function to obtain the excess heat capacity, Cpexc. All-atom molecular dynamics (MD) of the protein in unliganded form and in complex with warfarin was performed following a protocol we previously adopted for other simulations of albumin-ligand complexes [18,19]. The structure of albumin was obtained from the complex crystallized with R(+)-warfarin, deposited as 2BXD entry in the Protein Data Bank [1].

    • Atomistic Insights into the Functional Instability of the Second Helix of Fatty Acid Binding Protein

      2019, Biophysical Journal
      Citation Excerpt :

      Importantly, the rate of the transient unfolding (195 s−1) in this closed-cap variant nicely correlates with the maximal ligand association rate (180 s−1) at 13°C, providing strong support for the local unfolding of αII as a critical conformational reorganization required for ligand entry. All-atom computer simulations represent a powerful method that can provide a wealth of detailed information about protein dynamics and has been previously applied to study the structural dynamics of different FABP isoforms (12–17). However, transient unfolding of αII as part of the helical cap has not been observed in previous simulations.

    View all citing articles on Scopus
    View full text