10704
J. Phys. Chem. B 2008, 112, 10704–10713
Modeling and Molecular Dynamics Simulation of the Human Gonadotropin-Releasing Hormone Receptor in a Lipid Bilayer ´ ngel Pin˜eiro*,‡,§ Eduardo Jardo´n-Valadez,†,‡ Alfredo Ulloa-Aguirre,*,† and A Research Unit in ReproductiVe Medicine, Hospital de Ginecobstetricia “Luis Castelazo Ayala”, Instituto Mexicano del Seguro Social, Me´xico D.F. 01090, Mexico, and Facultad de Quı´mica, Depto de Fisicoquı´mica, UniVersidad Nacional Auto´noma de Me´xico, C.U., Me´xico D.F. 04510, Mexico ReceiVed: January 19, 2008; ReVised Manuscript ReceiVed: May 10, 2008
In the present study, a model for the human gonadotropin-releasing hormone receptor embedded in an explicit lipid bilayer was developed. The final conformation was obtained by extensive molecular dynamics simulations of a homology model based on the bovine rhodopsin crystal structure. The analysis of the receptor structure allowed us to detect a number of specific contacts between different amino acid residues, as well as waterand lipid-mediated interactions. These interactions were stable in six additional independent 35 ns long simulations at 310 and 323 K, which used the refined model as the starting structure. All loops, particularly the extracellular loop 2 and the intracellular loop 3, exhibited high fluctuations, whereas the transmembrane helices were more static. Although other models of this receptor have been previously developed, none of them have been subjected to extensive molecular dynamics simulations, and no other three-dimensional structure is publicly available. Our results suggest that the presence of ions as well as explicit solvent and lipid molecules are critical for the structure of membrane protein models, and that molecular dynamics simulations are certainly useful for their refinement. 1. Introduction G protein-coupled receptors (GPCRs) form a large and functionally diverse superfamily of plasma membrane receptors whose primary function is to transduce extracellular stimuli into the intracellular environment through the activation of one or more signal transduction pathways.1,2 Many signaling cascades use this class of receptors to convert a variety of external and internal stimuli, from photons of light, odorants, and ions to small molecule neurotransmitter, peptides, glycoproteins, and phospholipids, into intracellular responses.3,4 Signaling through GPCRs is mediated by guanine nucleotide-binding signaltransducing proteins (G proteins), which confer the name to these membrane receptors.5 Although G protein-coupled receptors may vary considerably in molecular size, all share a common molecular topology which consists of a single polypeptide chain of variable length that traverses the lipid bilayer seven times, forming characteristic transmembrane (TM) hydrophobic R-helices connected by alternating extracellular and intracellular loops.2 An extracellular NH2-terminus and an intracellular COOH-terminal tail are also typically present,1,2 and a disulfide bridge between the extracellular loop (EL) 1 and the EL2 has been found to be essential to maintain the stability of the heptahelical structure of the receptor.2,3 Agonist binding to GPCRs is followed by conformational rearrangements on the receptor structure, involving particularly changes in the relative orientation of the seven TM R-helices; these movements of the helices are propagated to the intracellular domains leading to activation of one or more G proteins and effector enzymes, * Towhomcorrespondenceshouldbeaddressed.E-mail:
[email protected] ´ .P.) and
[email protected] (A.U.-A.). (A † Instituto Mexicano del Seguro Social. ‡ Universidad Nacional Auto ´ noma de Me´xico. § Current address: Depto de Fı´sica Aplicada, Facultad de Fı´sica, Universidad de Santiago de Compostela, E-15782, Santiago de Compostela, Spain.
which in turn influence levels of second messengers that regulate a variety of biological functions including cell growth and differentiation, immune responses, and cellular metabolism.1–3,6 The relatively easy accessibility to these cell surface receptors from the exterior of cells, combined with their proven relationship with many diseases and physiological processes, make these membrane proteins an attractive potential target for drug design.7 Nevertheless, a major inconvenience that limits their study is the difficulty to solve GPCR structures in crystals, which might not preserve their native protein structure. As a result, among the hundreds of GPCRs identified so far,1 only bovine rhodopsin8 and, more recently, the human β2 adrenergic receptor9–11 structures have been experimentally determined. By using the rhodopsin structure as a template, a number of GPCR models have been recently developed.12 The resulting models have usually been validated by means of a limited number of specific interactions previously identified by mutagenesis and biochemical studies. To assess their stability, short length molecular dynamics (MD) simulations in vacuum or without explicit lipids have generally been performed. Only in a few particular cases, relatively long molecular dynamics trajectories in explicit solvent have been generated for GPCR models. Recently, a 20 ns long trajectory was obtained for a model of the human δ opioid receptor inserted in an explicit lipid membrane.13 Under similar conditions, a 31 ns long simulation was also performed for the human cholescystokinin receptor.14 These simulations proved to be useful for the understanding of these particular GPCRs. To our knowledge, no additional MD simulation studies have been performed for GPCR models in an explicit membrane environment during relatively longer time scales. The mammalian type I gonadotropin-releasing hormone receptor (GnRHR) (hereafter referred to only as “GnRHR”) belongs to the rhodopsin/β-adrenergic-like family of GPCRs (family A).2 The intracellular domains of this receptor are mainly coupled to the trimeric Gq/11 protein localized in the cytoplasm
10.1021/jp800544x CCC: $40.75 2008 American Chemical Society Published on Web 08/05/2008
Human Gonadotropin-Releasing Hormone Receptor
J. Phys. Chem. B, Vol. 112, No. 34, 2008 10705
Figure 1. Schematic representation of the hGnRHR structure. The amino acid residues located in the extra- and intracellular loops and those of the transmembrane domains are identified according to their particular location [extracellular (top), transmembrane (middle) and intracellular (bottom)]. Basic (red), acidic (blue), polar (green), nonpolar (white), and cysteine (yellow) residues are also highlighted.
of target cells.15 In addition, there is growing evidence that the GnRHR is also capable of coupling to other G proteins, including Gs and Gi proteins, thereby mediating different physiological functions.16,17 The human (h) GnRHR preferentially binds gonadotropin-releasing hormone-I (GnRH), a decapeptide produced by the hypothalamus and released in synchronized pulses to the anterior pituitary to regulate reproductive function.17–19 The GnRHR exhibits several particular features including the complementary role of Asp87 and Asn319 in receptor activation (observed in site-directed mutagenesis experiments by reciprocal mutations) at the transmembrane domain -2 and -7,20,21 the replacement of Tyr with Ser in the highly conserved Asp-Arg-Tyr motif located in the junction of the TM helix-3 and the intracellular loop (IL) 2, and the lack of the COOH-terminal extension into the cytosol.22 GnRH interacts with several amino acid residues of the receptor located mainly in the core of the TM helices and the carboxyl-terminal end of the EL3, and that includes Asp98, Asn102, Asp302, Trp101, Lys121, Asn212, and Tyr290, which define the binding pocket of the receptor for its natural ligand.19,22 Experimental evidence also supports that Asn53 and 87, Asp138 and 319, Arg139, and Tyr323 are involved in receptor activation and that intramolecular interactions exist between Glu90, Lys121, and Asp98, and between Asn53, Asn87, and Asp319 in the inactive conformation of the receptor.22 Several three-dimensional (3D) structures of the GnRHR have been proposed;23–26 these structures are based on its homology with bovine rhodopsin, on the secondary structure and/or
microdomains predictions of the GnRHR, and by imposing several experimentally detected direct interactions between different amino acid residues. However, none of those models has been subjected to extensive molecular dynamics simulations and no 3D structure has been made publicly available. In their recent study on the hGnRHR, So¨derha¨ll et al.27 proposed a complete receptor model employing one nanosecond-long molecular dynamics (MD) simulations without explicit lipids to assess its stability. Although the reported structure was carefully developed and considered previous models of the GnRHR as well as available information derived from experimental data, the time scale employed and the absence of explicit lipids in the simulations performed are clearly not suitable for an accurate assessment of the model. As in previous studies, the atomic coordinates of the resulting model have not been made available to other researchers. In the present study, a structure for the hGnRHR embedded in a lipid bilayer with explicit water molecules is proposed. Such structure was developed by a combination of homology modeling and extensive MD refinement. Taking the final conformation of the model as the starting structure, six 35 ns long independent trajectories (three at 310 K and three additional at 323 K) were generated to analyze the stability and dynamic behavior of the receptor. The full coordinates’ file of the refined model and those corresponding to the final conformations attained from the six replicas, including lipids, water molecules, and the ions needed to compensate for the net charge of the receptor at biological pH, are additionally supplied.
10706 J. Phys. Chem. B, Vol. 112, No. 34, 2008
Jardo´n-Valadez et al.
TABLE 1: List of Conditions Considered in the MD Trajectoriesa
stage
refinement trajectory
extended simulations (starting from the refined structure)
time interval (ns)
conditions
additional restrictions on top of the force field
slab geometry + NVT ensemble (T ) 310 K)
all protein backbone atoms frozen TM backbone atoms frozen TM-backbone H bonds restricted
70-120 120-160
constant pressure (T ) 310 K)
TM H bonds restricted none
0-35
constant volume (T ) 310 K)
none
0-35
constant pressure (T ) 323 K)
none
i
0-5
ii
5-10
iii
10-70
iV V R1 R2 R3 extended simulations (starting from the refined structure)
R1 R2 R3
a All simulations were performed with 192 DPPC molecules and 15 chlorine ions. The number of water molecules employed in stages i-iii of the refinement trajectory was 10 206. In the stages iV and V, as well as in the extended R1-3 trajectories, the total number of water molecules employed was 15 449.
2. Methods 2.1. Generation of the Initial GnRH Receptor Model. A homology model of the hGnRHR transmembrane domains was obtained employing the bovine rhodopsin structure (PDB ID 1F88)8 as a template and using the GPCR-specific software available in the SWISS-PROT server.28 The amino acid sequence alignment of the hGnRHR with GnRHRs from other species and with rhodopsin (including the identification of putative TM domains and their corresponding amino acid residues) has been previously published,22 and it is not shown in the present study. A schematic representation of the hGnRHR in two dimensions is depicted in Figure 1. Since the sequence similarity of the selected transmembrane region of the GnRHR with rhodopsin is relatively high (>75%), the corresponding homology model was considered as reasonably reliable. However, the similarity of the whole receptor sequence with rhodopsin is very low (
