Molecular Dynamics Simulations of Membrane ... - ACS Publications

Article pubs.acs.org/jcim

Molecular Dynamics Simulations of Membrane-Bound STIM1 to Investigate Conformational Changes during STIM1 Activation upon Calcium Release Sreya Mukherjee,† Aleksandra Karolak,† Marjolaine Debant,‡,§,∥ Paul Buscaglia,§,∥ Yves Renaudineau,‡,§ Olivier Mignen,§,∥ Wayne C. Guida,† and Wesley H. Brooks*,† †

Department of Chemistry, University of South Florida, Tampa, Florida 33620, United States INSERM ESPRI, ERI29/EA2216 Laboratory of Immunotherapy and B Cell Pathologies, Laboratory of Immunology and Immunotherapy, CHRU Morvan, European University of Brittany, F29609 Brest, France § Network “Ion channels and cancer-Cancéropole Grand Ouest (IC-CGO)”, F29609 Brest, France ∥ INSERM U1078, Brest University Medical School, F29609 Brest, France ‡

ABSTRACT: Calcium is involved in important intracellular processes, such as intracellular signaling from cell membrane receptors to the nucleus. Typically, calcium levels are kept at less than 100 nM in the nucleus and cytosol, but some calcium is stored in the endoplasmic reticulum (ER) lumen for rapid release to activate intracellular calcium-dependent functions. Stromal interacting molecule 1 (STIM1) plays a critical role in early sensing of changes in the ER’s calcium level, especially when there is a sudden release of stored calcium from the ER. Inactive STIM1, which has a bound calcium ion, is activated upon ion release. Following activation of STIM1, there is STIM1-assisted initiation of extracellular calcium entry through channels in the cell membrane. This extracellular calcium entering the cell then amplifies intracellular calcium-dependent actions. At the end of the process, ER levels of stored calcium are reestablished. The main focus of this work was to study the conformational changes accompanying homo- or heterodimerization of STIM1. For this purpose, the ER luminal portion of STIM1 (residues 58−236), which includes the sterile alpha motif (SAM) domain plus the calcium-binding EF-hand domains 1 and 2 attached to the STIM1 transmembrane region (TM), was modeled and embedded in a virtual membrane. Next, molecular dynamics simulations were performed to study the conformational changes that take place during STIM1 activation and subsequent protein−protein interactions. Indeed, the simulations revealed exposure of residues in the EF-hand domains, which may be important for dimerization steps. Altogether, understanding conformational changes in STIM1 can help in drug discovery when targeting this key protein in intracellular calcium functions.

■

the Ca2+ concentration and, upon sensing a drop in the ER lumen Ca2+ level, activates the C-terminal located in the cytosol. Activation then enables STIM1 to bind to ORAI calcium channels located in the plasma membrane. ORAI1 channels are the pore-forming subunits of the Ca2+-releaseactivated Ca2+ (CRAC) channel. As a result, ORAI1 channels localize to the puncta (i.e., sites where the cell membrane and ER membrane are in close proximity) and allow Ca2+ influx into the cell. Ca2+ release from the ER can be triggered by cell membrane receptors that convert an external stimulus to an intracellular signal by, for example, receptor-associated phospholipase C activation, which generates inositol triphosphate (IP3). IP3 then induces opening of IP3-responsive Ca2+ channels in the ER membrane to release the ER-stored Ca2+ into the cytosol.8 STIM1 has been established as the main Ca2+ ion sensor in nonexcitable cells, making it a key component early in many

INTRODUCTION Calcium ions (Ca2+) act as secondary messengers in numerous intracellular signaling pathways and as cofactors in enzyme activation.1 Some of the events controlled by Ca2+ include cell replication and cell division via Ca2+/calmodulin-stimulated protein kinases I and II,2 cell death (such as apoptosis and NETosis),3 activation of lymphocytes,4 activation of mast cells,5 and protein folding.6 In order to prevent inappropriate enzyme activation and signaling among the many diverse and important Ca2+-dependent cellular processes, Ca2+ is kept at low concentrations (∼100−200 nM) in the cytosol and nucleus. Nevertheless, Ca2+ is stored in the endoplasmic reticulum (ER) at ∼800 nM7 and is available for rapid release into the cytosol to trigger Ca2+-dependent actions. Stromal interacting molecule 1 (STIM1) and its homologue STIM 2 are Ca2+-binding singlepass transmembrane proteins. The N- and C-termini of these proteins extend into the ER lumen and cytosol, respectively,8 connected via the transmembrane fragment. The mechanism of action of STIM1 is depicted in Figure 1. The N-terminal part of STIM1 present in the ER lumen acts as a sensor of changes in © 2017 American Chemical Society

Received: August 16, 2016 Published: February 2, 2017 335

DOI: 10.1021/acs.jcim.6b00475 J. Chem. Inf. Model. 2017, 57, 335−344

Article

Journal of Chemical Information and Modeling

Figure 1. (A) Schematic model of STIM1 domains. cEF and nEF (the EF-hand domains 1 and 2, respectively), sterile alpha motif domain (SAM), transmembrane region (TM), coiled-coil domains 1−3 (CC1−3), STIM1−ORAI1 activating region (SOAR), serine/proline-rich region (SP), and lysine-rich region (K) are labeled. The calcium ion is shown bound to EF-hand 1. The gaps between labeled regions signify coils. (B) Events that occur when STIM1 acts to initiate calcium flux: (I) An external signal triggers intracellular IP3 signaling, which opens IP3 receptors to release ERstored Ca2+. (II) Release of ER-stored Ca2+ into the cytosol leads to loss of bound Ca2+ from the STIM1 EF-hand domain and facilitates oligomerization of STIM1. (III) STIM1 interacts with ORAI1 to open Ca2+-release-activated calcium (CRAC) channels to increase intracellular Ca2+. (IV) storage of Ca2+ in the ER is reestablished by sarco/endoplasmic reticulum calcium ATPase (SERCA) channels. Red spheres represent calcium ions.

intracellular Ca2+ signaling pathways. The STIM1 domains and the events that occur during initiation of intracellular Ca2+ signaling are shown in Figure 1. The ER luminal portion of STIM1 (the N-terminal domain of STIM1) has two major domains: the EF-hand domain and a sterile alpha motif (SAM) domain. In its inactive state, STIM1 has a single Ca2+ ion bound in the EF-hand domain, the EF hand being a frequently used helix−loop−helix Ca2+-binding motif in proteins. However, as the stored Ca2+ ions are released from the ER lumen through IP3-responsive channels, as mentioned above, STIM1 loses its bound Ca2+ and becomes active. The current postulate for STIM1 activation suggests that upon Ca2+ release from the ER, hydrophobic regions in the EF−SAM domains are exposed and dimerize with a neighboring STIM1 N-terminus. The formation of STIM1−ORAI1 interactions in the puncta and CRAC channel activation critically depend on STIM1 oligomerization, a process involving the luminal N-terminus and mediated by coiled-coil (CC) domains in the C-terminus of STIM1. Among the three CC domains (CC1, CC2, and CC3) present in the STIM1 cytosolic portion, CC2 and CC3 are required for STIM1 oligomerization after depletion of stored Ca2+. This domain, alternatively called the CRAC activation domain (CAD), STIM1−ORAI activating region (SOAR), or coiled-coil domain b9 (CCb9), is sufficient to activate CRAC channels when expressed as a soluble protein. At the end of the SOAR is a lysine-rich sequence that, when the SOAR is extended, can interact with cytosolic and cell membrane proteins, such as the ORAI1 transmembrane protein. ORAI1 forms Ca2+ channels in the cell membrane that are responsible for store-operated calcium entry (SOCE), and the channels are activated when STIM1 and ORAI1 interact. After this Ca2+ influx, the cell uses ATP to pump the Ca2+ back into the ER through sarco/endoplasmic reticulum

calcium ATPase (SERCA) channels to reestablish the stored Ca2+ levels in preparation for the next round. STIM1 is reported to have an association with SERCA in modulating the rate of this replenishment of ER Ca2+ levels.9−11 Variations in the strength and frequency of the stimulation, cell surface receptors receiving the ligand stimulation, intracellular signal pathway (e.g., IP3), STIM1 partners, persistence of the signals, ATP levels, and other factors help determine the actual intracellular targets and the resulting effect of the Ca2+ flux. Aberrant Ca2+ activities have been associated with autoimmune diseases such as systemic lupus erythematosus (lupus) with overly sensitive T cell activation.12 Also, association of overexpression of STIM1 with abnormal Ca2+ flux in T and B cells in lupus has also been reported,13 while a deficiency characterizes lymphocytes from patients with Sjö gren’s syndrome. Such abnormal activity could potentially lead to loss of tolerance of endogenous material due to lymphocyte dysregulation. The facts that STIM1 has a key role in Ca2+ flux and that STIM1 abnormalities have been associated with numerous other diseases14 such as Stormorken syndrome, York syndrome, and tubular aggregate myopathy (TAM) suggest that STIM1 is a promising target for drug discovery to develop new therapeutics for many diseases.15 As a consequence, we have postulated that understanding the conformational changes occurring in the ER luminal portion of STIM1 would be pivotal in designing future drugs for the protein. To achieve this goal, we report here the development of a refined computational model comprising the luminal and transmembrane portions of STIM1. Next, using molecular dynamics (MD) simulations, we analyzed the hypothesis of Stathopulos et al.8 that the collective solvent exposure of EFhand domain 2 and also the α10 helix of the SAM domain lead to an unstable state of STIM1.8,16 MD is a powerful tool used 336


Article


Figure 2. Structure built prior to insertion into the membrane showing the N-terminus of STIM1 positioned in the ER luminal portion. It consists of the EF-hand domain harboring the Ca2+ ion and the SAM domain connected to the TM. The cytosolic part is not shown.

Figure 3. Luminal part of the protein (in purple) with the transmembrane part embedded in the POPC membrane.

in the study of molecular system motions at the atomistic level of resolution and provides computational descriptions of phenomena that are difficult to view experimentally. Interpretation and validation of experimental data with MD simulations support these results and can be further used to improve computational models in drug discovery efforts.17 The structures of biomolecules of interest required for MD simulations are often obtained from NMR or X-ray studies. However, since conditions for successful crystallization or NMR analysis can vary from the in vivo conditions and involve dense protein stacking or high salt concentrations, some of the resulting structural data may be incomplete, e.g., with missing or distorted amino acid side chains. To address this issue, additional computational tools are employed to refine the models toward more in vivo-like states by allowing the structure to relax toward lower energy states, thus reducing the impact of the harsh crystallization conditions. Given the diverse nature of the membrane−protein interactions along with force field parametrization of the membrane components and contrasting phases in and around the membrane, constructing a realistic in silico model for membrane proteins remains a challenging

process. To overcome these difficulties, single-species lipid bilayers or bilayers containing two or more lipid species, such as cholesterol, are typically used.18−20 One of the most common single-species lipid bilayers for in silico simulations is the 1palmitoyl-2-oleoylphosphatidylcholine (POPC) membrane. POPC resembles the naturally occurring phospholipids containing the saturated sn-1 and unsaturated sn-2 acyl chains.21−23 The major structural lipids in the ER membrane are phosphatidylcholines (PCs), glycerophospholipids that act as a permeability barrier. Our computational study of model system dynamics to develop a model closer to putative in vivo states assisted us in understanding mechanisms by which STIM1 unfolds upon Ca2+ loss in the presence of the POPC membrane, considering that it is the relevant lipid membrane most commonly used in computational studies.21−23 This information represents a key factor in determining potential sites on the protein suitable for targeting in drug discovery.

■

METHODS The ER luminal portion of human STIM1 (containing residues 58−201, which include the EF-hand calcium binding domain 337


Article


Figure 4. (A) Block representation and secondary structures of STIM1 domains. (B) SAM and EF-hand domains of the STIM1 N-terminal luminal portion inside the endoplasmic reticulum: EF-hand domain 1 (green), which harbors Ca2+; EF-hand domain 2 (orange); and the SAM domain helices (various colors, helices 5 to 10). The transmembrane helix is not shown.

directions for embedding from VMD (Figure 3). The membrane plugin 1.1 of VMD was used to build a rectangular matrix as a two-dimensional hexagonal lattice of lipids. The lipid tails were (almost) fully extended because they were run previously for 1 ps in a vacuum, allowing for easy insertion of proteins into the membrane and therefore reducing the required equilibration time. The distance between the layers was set to fit the actual membrane thickness, and the lattice period was set to fit the actual surface density of lipid molecules being used. The lipid head groups were also properly solvated, as they would be in vivo. The simulated box with protein embedded in the membrane was filled with TIP3 water molecules and Na+ and Cl− ions to give a concentration of 0.2 mol/L.28 The 120 Å × 120 Å × 120 Å simulation box contained over 170 000 atoms. Systems were minimized to remove steric clashes, gradually heated to 300 K, and equilibrated for 10 ns. All of the simulations used the NAMD 2.7 force field and CHARMM topology and parameter files29 in the presence or absence of Ca2+. Coordinates were written to the output trajectory file every 100 ps. The particlemesh Ewald method was used to treat long-range electrostatics with a time step 2 fs. The H-bond lengths were estimated using VMD, and graphs were made with XMGRACE.30 The clustering analyses of the protein structures were performed with Wordom31 with the criteria set to a root-mean-square deviation (RMSD) of 1.5 Å, and the method used was QT.32 The top three conformational clusters were selected for the protein secondary structure analysis with VMD. The bilayer thickness of the POPC systems was measured using PyMOL33 at the pre-equilibration and post-equilibration stages of the simulations. The thickness was approximated using the distance between the phosphorus atoms of the upper and lower membrane layers.

and the SAM domain) as determined from NMR data by Stathopulos et al.8 and available as Protein Data Bank (PDB) entry 2K60 was selected.24 All STIM homologues are not essentially proteins embedded in a membrane but contain a single helical transmembrane portion that connects the two termini of the protein, one on either side of the membrane (cytosolic or ER luminal), in which the major motions occur. The Schrödinger suite 2014-3 was used to prepare the initial in silico protein structures,25 and since STIM1 is a transmembrane protein that spans the ER membrane to the cytosol, residues 202−236 were added to provide the transmembrane residues, which were modeled as an α-helix using Schrödinger Prime26 (Figure 2). Accordingly, the model built herein included the ER luminal portion from PDB entry 2K60 (NMR structure) containing residues 58−201 and the helical transmembrane portion obtained using Prime and data from PDB entry 3SR7 with 68% similarity to the amino acid sequence of the transmembrane region. This allowed placement of the structure in a simulated membrane, which in the in vivo environment anchors the protein and restricts the direction and extent of protein structural fluctuations. The ER luminal residues of STIM1 were positioned as protruding from the membrane’s surface, with Gly 201 initiating the luminal portion of the protein−membrane interface and with no overlap between the EF-hand or SAM domain and lipid headgroups. This part of the protein immediately followed the modeled transmembrane region (residues 202 and above). Since the ER opening residues of STIM1 (199−201) lack organized secondary structure, this initial luminal portion was allowed to relax during simulation. ER luminal residues were adjoining from the membrane. An initial simulated membrane was created using the membrane tool in VMD,27 and the protein was inserted into it as a single-pass transmembrane protein following the 338


Article


Figure 5. Superimposition of the NMR structure (cyan) with the simulated top cluster pose in the presence of Ca2+ (orange).

Figure 6. Calcium-binding zone around the EF-hand domain of (A) the simulated top cluster pose in the presence of Ca2+ and (B) the NMR structure.

■

STIM1 activation process.16 In the study presented here, we added the transmembrane part, which not only links to the luminal portion of the structure but also more realistically represents the system. Still, in order to reduce the overall computation time and because of the lack of high-resolution data, the cytosolic portion was omitted. Figure 4 shows the different domains of the STIM1 N-terminal luminal portion as present in the NMR structure. Simulated Model in the Presence of Ca2+. The clustering analysis of MD simulations for the system with Ca2+ revealed that the simulated protein has a conformation similar to the NMR structures. The NMR conformers were aligned, and a ±1.103 Å difference in RMSD was observed between them. The binding site was conserved among the conformers. The alignment of the top pose superimposed with

RESULTS

Modeling the ER Luminal Portion of STIM1 Together with Its Transmembrane Portion. Since STIM1 is a transmembrane protein, modeling the transmembrane portion is vital for understanding what triggers the conformational changes taking place in the absence of Ca2+. Studies discussing the process accompanying the movement of domains in the absence of Ca2+, which initiates activation, have been reported.34 One of them incorporated a replica-exchange method based on the NMR structure (PDB entry 2K60), where the transmembrane and cytosolic regions were not modeled. The results revealed cooperativity between the Ca2+ concentration and unfolding of the protein. The authors suggested that such cooperativity could be critical for the 339


Article


Figure 7. Calcium-binding zone around the EF-hand domain of the simulated model in the absence of Ca2+ gathered from the top cluster protein pose (magenta ribbons). Residues involved in hydrogen bonding and a salt bridge between Glu 94 and Arg 93 are shown.

Table 1. Average Distances (Å) between Glu 94 and Phe 108, between Phe 108 and Val 117, between Asp 77 and Glu 87, and between Val 85 and Asp 95 simulations without Ca2+ distance

NMR structure

Glu 94−Phe 108 Phe 108− Val 117 Asp 77−Glu 87 Val 85−Asp 95

15.74 11.81 8.62 15.43

from the top three clusters 22.54 17.38 8.45 15.78

± ± ± ±

simulations with Ca2+

from the last 2 ns

0.45 0.61 0.06 0.75

22.5 16.86 10.67 14.08

± ± ± ±

0.04 0.34 2.91 0.16

from the top three clusters 15.97 10.95 8.24 17.27

± ± ± ±

1.48 1.24 0.61 1.72

from the last 2 ns 15.44 7.61 7.06 14.31

± ± ± ±

0.08 0.03 0.44 0.15

Upon removal of the Ca2+ ion, this negatively charged patch becomes available for hydrogen-bonding interactions with residues in the neighboring helices. The new hydrogen-bonding patterns observed among residues Asp 76, Asp 78, Asn 80, Val 83, and Asp 84 compensate for the absence of the strong positive charge of the Ca2+ ion (Figure 7). Also, it may assist the uncoiling of loops and formation of prominent helices, as discussed later. Arg 93 forms a salt bridge with Glu 94, which may stabilize the α2-helix of EF-hand domain 1. Overall, as a result of the absence of the strong positively charged cation and concurrent hydrogen bonding, the residues involved in interacting with Ca2+ appear to deviate from the β-strand (β1 and β2) to an organized complete α-helix (α2) and loop variation. Referring to the STIM1 structural geometry in the NMR structure paper by Stathopolous et al.,8 the α2-helix (spanning residues 89−97) elongates, and the ordered helix uses residues from the β-turn (residues 82 and 83), which is also discerned on the basis of distance calculations run throughout the trajectory. Secondary structure assessment shows the loss of the β-turn and coils to a helix via residues 83−97. Furthermore, we performed the distance analysis for the last 2 ns of the simulations and for the three top clusters. The comparison with the NMR structure is summarized in Table 1. The distance between Val 85 and Asp 95, which forms the helix

Maestro onto one of the original wild-type systems is presented in Figure 5. The EF-hand and SAM domains are held together in the same fashion as seen in the original NMR structure, and the only area of flexibility is around the tail portion, which harbors the amino acid residues missing in the original NMR structure. Next, the interactions around Ca2+ were analyzed and compared with those in the NMR structure. During the simulation, the Ca2+ ion held between residues Asp 77 to Glu 87 and Arg 93 in the NMR structure becomes only slightly more exposed (Figure 6). All of the other residues in the NMR structure and the simulation retained their major interactions, including hydrogen bonds and secondary structures as well. Also, no hydrogen bonding between the POPC membrane and the residues near the membrane is seen. This shows that the presence of the Ca2+ ion is vital in maintaining the wild-type structure. Simulated Model in the Absence of Ca2+. The simulations in the absence of Ca2+ show initiation of the unfolding process of STIM1 via a few important conformational changes, which help in understanding the protein dynamics. The canonical EF-hand domain 1, which has the form of a helix−β-loop−helix motif, surrounds the Ca2+ ion with six negatively charged amino acid residues: Asp 77, Asp 78, Asp 82, Asp 84, Glu 86, and Glu 87 located along the β-loop. 340


Article


Figure 8. (A) STIM1 with a calcium ion interacts with negatively charged residues from the NMR structure. Val 117, Phe 108, Glu 94, and hydrophobic amino acid residues on EF-hand domain 2 are seen to form the packed conformation. (B) STIM1 in the absence of calcium. The negatively charged amino acid residues Asp 77 and Glu 87 are further apart. (C) STIM1 in the presence of calcium. The negatively charged amino acid residues Asp 77 and Glu 87 interact similarly as in (A).

oligomerization process. We observed that α10 moves away from the EF-hand zone by ∼7 Å mediated by interactions with other SAM helices (Figure 8). Also, α9 uncoils, as revealed by the secondary structure analysis (Figure 9), which shows that helix 9 residues and part of helix 10 unfold from a helix to a βturn. The α to β transition of the α10 helix of the SAM domain could lead to its uncoiling and aid in oligomerization. Hydrogen-bonding interactions of Leu 204 (located in the early transmembrane portion) and Thr 107 and Gln 139 (located on helices α3 and α6) with phosphate groups present in the POPC membrane are observed.

as discussed above, was monitored for the simulations and in the NMR structure and was found to be ∼15 Å in the system with Ca2+ ion and ∼7 Å longer without Ca2+. The next change occurs when the α3- and α4-helices, which form part of EFhand domain 2, are seen to drift apart, exposing a few hydrophobic residues. Movement between α4 and α3 is quantified by the change in the distance between Val 117 and Phe 108, which is around 10 Å longer in the Ca2+-depleted system (increases from 7 to 17 Å). We also observed movement of EF-hand domain 2 away from EF hand 1 helices, exposing the hydrophobic residues; Phe 108 in α3 and Glu 94 in α2 move further away by almost 7 Å with no calcium ion present. Thus, the unfolding of STIM1 in the absence of Ca2+ is mediated through the EF-hand domain and then translates to the next portion of the protein, the SAM domain. The SAM domain in STIM1 consists of five helices that take part in the

■

DISCUSSION Understanding the mechanism by which a protein works is key to understanding how to target it for drug discovery.36,37 Generally proteins that have EF hands like calmodulin are 341


Article


Figure 9. Ca2+ ion held by a few negatively charged residues that rest in between α-helices and a β-turn in EF-hand domain 1 (depicted in orange).

found in two conformations, “closed” or calcium-free and “open” or calcium-bound. An EF-hand protein mechanism states that in the absence of calcium, the helices forming the EF hand would close, forming a very compact or closed structure. When calcium binds to the protein again, the two helices of each EF-hand motif would be pushed apart, inducing the exposure of hydrophobic residues otherwise involved in the original helix packing in the calcium-free form. These residues then become available for further interactions with other proteins.36,37 Many proteins are also able to form intrinsically disordered conformations to allow for cooperation of distant domains.39,40 STIM1 and other proteins such as adenovirus protein E1A and other EF-hand-containing proteins are seen to do this.38,40 Upon binding to a ligand (or Ca2+ ion in the case of EF-hand-containing proteins), domain interactions are established that lead to a state through which various functions are achieved. Another example is a neuronal calcium sensor, the NCS1 protein, which is activated by binding of three Ca2+ ions, leading to restructuring of the domains to form a pocket where binding of neuronal receptors occurs.41 In STIM1, disordering of the structure allows the system to oligomerize and initiate a cascade of rearrangements along the entire sequence of STIM1 for puncta localization, CRAC channel interactions, and eventual reestablishment of stored Ca2+ in the ER. MD simulations can provide molecular-level insights into the events, especially during the early activation steps. To our knowledge, this is the first time the luminal and transmembrane portions of STIM1 have been modeled together in the absence/presence of Ca2+ to study the process of its activation. This STIM1 structure represents the ER luminal portion of the protein well but lacks the N-terminal 57 residues for which structure could not be discerned from the NMR data because of either random positioning during NMR analysis or loss of the residues during preparation. These missing residues may in fact have importance in STIM1 functions, interactions, and structure, but that will remain to be determined from future analysis when sufficient experimental and structural data are available to define the positions of the missing residues. The results suggest that conformational changes taking place in the absence of Ca2+ lead to an unstable disordered state that may promote oligomerization. Major changes are observed in the EF-hand domains, followed by rearrangements of the SAM helices. Secondary structure assessment demonstrates the loss of a β-strand in EF-hand domain 1 due to the absence of the

Ca2+ via the negatively charged residues that form a hexagonal chelate around it, as shown in Figure 9. This change in secondary structure directs helices α3 (103− 108) and α4 (117−126), along with the β-turn loop β2, to adjust into a coil and start moving apart from EF-hand domain 1 as a result of the increased flexibility in the absence of the electrostatic charges from calcium. PROPKA42 analysis of the negatively charged residues showed that these negatively charged residues remain deprotonated during the simulation. The simulations imply that the EF-hand domain 2 rearrangements initiate the oligomerization process and continue down to the SAM domain, indicating cooperativity between domains. The POPC headgroups are expected to be a major factor that contributes to the differences in the simulations. The headgroup of POPC is a −N+(CH3)3 group, and the sterically hindered quaternized nature of the nitrogen in the group reduces the potential for favorable interactions (e.g., hydrogenbonding and electrostatic interactions) between the lipid and the protein to occur. The surface area of a biomolecule that is accessible to a solvent is known as the solvent-accessible surface area (SASA). It is usually calculated using the “rolling ball” algorithm developed by Shrake and Rupley,47 which uses a sphere (of solvent) of radius 1.4 Å to “probe” the surface of the molecule. The SASA for the entire protein calculated with Maestro35,43 increases to 6888.041 Å2, as opposed to 6381.440 Å2 in the Ca2+-bound system. We also looked at the SASA for the EFhand and SAM domains, considering that they are the major domains where the oligomerization happens, and it was seen that without the presence of calcium the EF-hand domain increases to 4548.714 Å2 from 3743.162 Å2, while for the SAM domain the increase is from 3478.529 Å2 to 3752.801 Å2. Overall we do see an increase in the surface area across the domains of STIM1, signifying unfolding. Although the transmembrane portion increases the rigidity in the system, it is seen to change during the simulation. In the presence of Ca2+, the transmembrane domain takes the form of a helix as suggested in the NMR structure. In the absence of Ca2+ however, this region interacts with the membrane, and a few residues are seen to twist from the helix, which signifies that signal transduction occurs in the absence of Ca2+ via the mechanism postulated. Since the transmembrane domain is the key to translating this change through the membrane to the cytosolic part, understanding changes in that will also help us understand the role of the various domains in signal 342


Article

Journal of Chemical Information and Modeling transduction. During the simulation, since the transmembrane portion is a single-pass polypeptide attached on one end to the cytosolic portion and on the other end to helix 10 of the SAM domain, the entire helix 10 is not turned away unless the residues also interact with the membrane itself. It is plausible that once Ca2+ is lost and uncoiling has occurred, helix α10, from which some residues change to a β-turn, could be pulled into the membrane. The transmembrane residues 200−210 change to a coil in the absence of the Ca2+, while residues 225− 228 change to a β-turn. The transmembrane portion does not recede into the membrane, and these structural changes could be sufficient to trigger the CC1, CC2, and CC3 regions observed experimentally for SOAR activation. The EF-hand and SAM domains are seen to participate in the initiation of the oligomerization process. STIM1’s EF-hand domain most closely resembles C-CAM44 and the SAM domain that of the EphB2 receptor.45 Opening of the EF hand usually exposes hydrophobic surfaces, which participate in binding of target ligands or sequences. Here the exposure of the residues helps unfold the protein. SAM domains are known to unfold for oligomerization in three different ways: (1) via exchanging N-terminal arms and utilization of additional interface contacts at the C-terminus; (2) via exchanging N-terminal arms but with rotation of the Cterminus away; and (3) as predicted earlier and seen from the simulations, the oligomerization could occur via the nonpolar surfaces of the helix close to the C-terminus and the mid loop.46,8 On the basis of our observations, the change in secondary structure leads to the opening of the hydrophobic residues of the STIM1 luminal portion, which potentially precedes oligomerization. This highlights the importance of the transmembrane portion of STIM1, since the activation must be conveyed to the cytosolic part of STIM1. Since our simulations show only the ER luminal portion of STIM1, it is possible that after the luminal portion starts to unpack, the cytosolic part begins to unravel as well. Druggable sites, such as the EF-hand domain 1 area harboring the Ca2+ and EF-hand domain 2 areas that contribute to the opening of the protein will hence be considered for further drug discovery methods. Our future work will focus on determining how the entire protein structure changes upon activation as well as exploring compounds that could be potential leads for rational drug design targeting STIM1.

comprehensive understanding of these changes could help to identify potential druggable sites for targeting STIM1.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Wesley H. Brooks: 0000-0002-5604-3769 Notes

The authors declare no competing financial interest.

■ ■

ACKNOWLEDGMENTS M.D. was funded by the Region Bretagne and the Ligue contre le cancer. REFERENCES

(1) Raraty, M.; Ward, J.; Erdemli, G.; Vaillant, C.; Neoptolemos, J. P.; Sutton, R.; Petersen, O. H. Calcium-dependent Enzyme Activation and Vacuole Formation in the Apical Granular Region of Pancreatic Acinar Cells. Proc. Natl. Acad. Sci. U. S. A. 2000, 97, 13126−13131. (2) Skelding, K. A.; Rostas, J. A. P.; Verrills, N. M. Controlling the Cell Cycle: The Role of Calcium/Calmodulin-Stimulated Protein Kinases I and II. Cell Cycle 2011, 10, 631−639. (3) Orrenius, S.; Zhivotovsky, B.; Nicotera, P. Regulation of Cell Death: The Calcium-Apoptosis Link. Nat. Rev. Mol. Cell Biol. 2003, 4, 552−565. (4) Oh-hora, M.; Yamashita, M.; Hogan, P. G.; Sharma, S.; Lamperti, E.; Chung, W.; Prakriya, M.; Feske, S.; Rao, A. Dual Functions for the Endoplasmic Reticulum Calcium Sensors STIM1 and STIM2 in TCell Activation and Tolerance. Nat. Immunol. 2008, 9, 432−443. (5) Baba, Y.; Nishida, K.; Fujii, Y.; Hirano, T.; Hikida, M.; Kurosaki, T. Essential Function for the Calcium Sensor STIM1 in Mast Cell Activation and Anaphylactic Responses. Nat. Immunol. 2008, 9, 81− 88. (6) Coe, H.; Michalak, M. Calcium Binding Chaperones of the Endoplasmic Reticulum. Gen. Physiol. Biophys. 2009, 28, F96−F103. (7) Samtleben, S.; Jaepel, J.; Fecher, C.; Andreska, T.; Rehberg, M.; Blum, R. Direct Imaging of ER Calcium with Targeted-Esterase Induced Dye Loading (TED). J. Visualized Exp. 2013, 75, 50317. (8) Stathopulos, P. B.; Zheng, L.; Li, G. Y.; Plevin, M. J.; Ikura, M. Structural and Mechanistic Insights into STIM1-Mediated Initiation of Store-Operated Calcium Entry. Cell 2008, 135, 110−122. (9) Park, C. Y.; Hoover, P. J.; Mullins, F. M.; Bachhawat, P.; Covington, E. D.; Raunser, S.; Walz, T.; Garcia, K. C.; Dolmetsch, R. E.; Lewis, R. S. STIM1 Clusters and Activates CRAC Channels via Direct Binding of a Cytosolic Domain to Orai1. Cell 2009, 136, 876− 890. (10) Wu, M. M.; Buchanan, J.; Luik, R. M.; Lewis, R. S. Ca2+ Store Depletion Causes STIM1 to Accumulate in ER Regions Closely Associated with the Plasma Membrane. J. Cell Biol. 2006, 174, 803− 813. (11) Liou, J.; Kim, M. L.; Heo, W. D.; Jones, J. T.; Myers, J. W.; Ferrell, J. E., Jr.; Meyer, T. STIM is a Ca2+ Sensor Essential for Ca2+Store-Depletion-Triggered Ca2+ Influx. Curr. Biol. 2005, 15, 1235− 1241. (12) Kyttaris, V. C.; Zhang, Z.; Kampagianni, O.; Tsokos, G. C. Calcium Signaling in Systemic Lupus Erythematosus T-Cells: A Treatment Target. Arthritis Rheum. 2011, 63, 2058−2066. (13) Renaudineau, Y.; Mignen, O.; Fali, T.; Burgos, M.; Cornec, D.; Jousse, S.; Saraux, A.; Pers, J. Abnormal Calcium Influx in T and B Lymphocytes from Systemic Lupus Erythematosus Patients is Related to STIM-1 Over-Expression. Ann. Rheum. Dis. 2013, 72, A30. (14) Mukherjee, S.; Brooks, W. H. Stromal Interaction Molecules as Important Therapeutic Targets in Diseases with Dysregulated Calcium Flux. Biochim. Biophys. Acta, Mol. Cell Res. 2014, 1843, 2307−2314. (15) Feske, S.; Picard, C.; Fischer, A. Immunodeficiency Due to Mutations in ORAI1 and STIM1. Clin. Immunol. 2010, 135, 169−182.

■

CONCLUSIONS We have performed an MD study of STIM1 in the active and nonactive forms. One difficulty was generating a representation of the transmembrane region of STIM1 in order to collect accurate dynamics. The inclusion of the TM region of STIM1 was essential in order to preserve the interactions occurring in the real system. We aimed to understand the early events upon conformational changes in the ER luminal portion of the protein, which could lead to potential rearrangements in the cytosolic part. The study revealed a series of structural fluctuations and the exposure of hydrophobic residues in the EF-hand domain and additional rearrangements of the distant SAM domain as well as changes in interaction patterns with the membrane. Potentially, in addition to the shifts in interaction patterns involving hydrophobic residues and structure destabilization, the observed reorganization is an important factor in the process of dimerization. Notably, the transition from a βturn to an α-helix may trigger conformational changes, and a 343


Article

Journal of Chemical Information and Modeling (16) Furukawa, Y.; Teraguchi, S.; Ikegami, T.; Dagliyan, O.; Jin, L.; Hall, D.; Dokholyan, N. V.; Namba, K.; Akira, S.; Kurosaki, T.; Baba, Y.; Standley, D. M. Intrinsic Disorder Mediates Cooperative Signal Transduction in STIM1. J. Mol. Biol. 2014, 426, 2082−2097. (17) van Gunsteren, W. F.; Berendsen, H. J. C. Computer Simulation of Molecular Dynamics: Methodology, Applications, and Perspectives in Chemistry. Angew. Chem., Int. Ed. Engl. 1990, 29, 992−1023. (18) Filizola, M.; Wang, S. X.; Weinstein, H. Dynamic Models of GProtein Coupled Receptor Dimers: Indications of Asymmetry in the Rhodopsin Dimer from Molecular Dynamics Simulations in a POPC Bilayer. J. Comput.-Aided Mol. Des. 2006, 20, 405−416. (19) Martinez-Archundia, M.; Cordomi, A.; Garriga, P.; Perez, J. J. Molecular Modeling of the M3 Acetylcholine Muscarinic Receptor and Its Binding Site. J. Biomed. Biotechnol. 2012, 2012, 789741. (20) Lai, P. C.; Crasto, C. J. Beyond Modeling: All-Atom Olfactory Receptor Model Simulations. Front. Genet. 2012, 3, 61. (21) Ceccarelli, M.; Marchi, M. Molecular Dynamics Simulation of POPC at Low Hydration Near the Liquid Crystal Phase Transition. Biochimie 1998, 80, 415−419. (22) Li, S.; Lin, H. N.; Wang, Z. Q.; Huang, C. Identification and Characterization of Kink Motifs in 1-Palmitoyl-2-Oleoyl-Phosphatidylcholines: A Molecular Mechanics Study. Biophys. J. 1994, 66, 2005−2018. (23) Barenholz, Y.; Thompson, T. E. Sphingomyelin: Biophysical Aspects. Chem. Phys. Lipids 1999, 102, 29−34. (24) Berman, H. M.; Westbrook, J.; Feng, Z.; Gilliland, G.; Bhat, T. N.; Weissig, H.; Shindyalov, I. N.; Bourne, P. E. The Protein Data Bank. Nucleic Acids Res. 2000, 28, 235−242. www.rcsb.org (accessed January−May 2016). (25) Schrödinger, release 2014-3: Maestro, version 9.9; Schrödinger, LLC: New York, 2014. (26) Andrec, M.; Harano, Y.; Jacobson, M. P.; Friesner, R. A.; Levy, R. M. Complete Protein Structure Determination Using Backbone Residual Dipolar Couplings and Sidechain Rotamer Prediction. J. Struct. Funct. Genomics 2002, 2, 103−111. (27) Humphrey, W.; Dalke, A.; Schulten, K. VMD: Visual Molecular Dynamics. J. Mol. Graphics 1996, 14, 33−38. http://www.ks.uiuc.edu/ Research/vmd/ (accessed January−December 2016) (28) Jorgensen, W.; Chandrasekhar, J.; Madura, J.; Impey, R.; Klein, M. Comparison of Simple Potential Functions for Simulating Liquid Water. J. Chem. Phys. 1983, 79, 926−935. (29) Phillips, J. C.; Braun, R.; Wang, W.; Gumbart, J.; Tajkhorshid, E.; Villa, E.; Chipot, C.; Skeel, R. D.; Kale, L.; Schulten, K. Scalable Molecular Dynamics with NAMD. J. Comput. Chem. 2005, 26, 1781− 1802. (30) Grace User’s Guide (for Grace-5.1.22). http://plasma-gate. weizmann.ac.il/Grace/doc/UsersGuide.html (accessed March−July 2016). (31) Seeber, M.; Cecchini, M.; Rao, F.; Settanni, G.; Caflisch, A. Wordom: A Program for Efficient Analysis of Molecular Dynamics Simulations. Bioinformatics 2007, 23, 2625−2627. (32) Seeber, M.; Felline, A.; Raimondi, F.; Muff, S.; Friedman, R.; Rao, F.; Caflisch, A.; Fanelli, F. Wordom: A User-Friendly Program for the Analysis of Molecular Structures, Trajectories, and Free Energy Surfaces. J. Comput. Chem. 2011, 32, 1183−1194. (33) The PyMOL Molecular Graphics System, version 1.8; Schrödinger, LLC: New York. (34) Feske, S.; Prakriya, M. Conformational Dynamics of STIM1 Activation. Nat. Struct. Mol. Biol. 2013, 20, 918−919. (35) Munteanu, C. R.; Pimenta, A. C.; Fernandez-Lozano, C.; Melo, A.; Cordeiro, M. N.; Moreira, I. S. SASA-Based Hot-Spot Detection 2 (SBHD) Method for Protein-Protein and Protein-Nucleic Acid Interfaces. J. Chem. Inf. Model. 2015, 55, 1077−86. (36) Araujo, R. P.; Liotta, L. A.; Petricoin, E. F. Proteins, Drug Targets and the Mechanisms they Control: The Simple Truth about Complex Networks. Nat. Rev. Drug Discovery 2007, 6, 871−880. (37) Herzberg, O.; Moult, J.; James, M. N. A Model for the Ca2+Induced Conformational Transition of Troponin C. A Trigger for Muscle Contraction. J. Biol. Chem. 1986, 261, 2638−2644.

(38) Ferreon, A. C.; Ferreon, J. C.; Wright, P. E.; Deniz, A. A. Modulation of Allostery by Protein Intrinsic Disorder. Nature 2013, 498, 390−394. (39) Reichheld, S. E.; Yu, Z.; Davidson, A. R. The Induction of Folding Cooperativity by Ligand Binding Drives the Allosteric Response of Tetracycline Repressor. Proc. Natl. Acad. Sci. U. S. A. 2009, 106, 22263−22268. (40) Hilser, V. J.; Thompson, E. B. Intrinsic Disorder as a Mechanism to Optimize Allosteric Coupling in Proteins. Proc. Natl. Acad. Sci. U. S. A. 2007, 104, 8311−8315. (41) Heidarsson, P. O.; Otazo, M. R.; Bellucci, L.; Mossa, A.; Imparato, A.; Paci, E.; Corni, S.; Di Felice, R.; Kragelund, B.; Cecconi, C. Single-Molecule Folding Mechanism of an EF-Hand Neuronal Calcium Sensor. Structure 2013, 21, 1812−1821. (42) Sondergaard, C. R.; Olsson, M. H. M.; Rostkowski, M.; Jensen, J. H. Improved Treatment of Ligands and Coupling Effects in Empirical Calculation and Rationalization of pKa Values. J. Chem. Theory Comput. 2011, 7, 2284−2295. (43) Durham, E.; Dorr, B.; Woetzel, D.; Staritzbichler, R.; Meiler, J. Solvent Accessible Surface Area Approximations for Rapid and Accurate Protein Structure Prediction. J. Mol. Model. 2009, 15, 1093−1108. (44) Chin, D.; Means, A. R. Calmodulin: A Prototypical Calcium Sensor. Trends Cell Biol. 2000, 10, 322−328. (45) Stapleton, D.; Balan, I.; Pawson, T.; Sicheri, F. The Crystal Structure of an Eph Receptor SAM Domain Reveals a Mechanism for Modular Dimerization. Nat. Struct. Biol. 1999, 6, 44−49. (46) Kim, C. A.; Bowie, J. U. SAM Domains: Uniform Structure, Diversity of Function. Trends Biochem. Sci. 2003, 28, 625−628. (47) Shrake, A.; Rupley, J. A. Environment and Exposure to Solvent of Protein Atoms. Lysozyme and Insulin. J. Mol. Biol. 1973, 79, 351− 371.

344


Molecular Dynamics Simulations of Membrane ... - ACS Publications

Recommend Documents