Subscriber access provided by Northern Illinois University
Article
Computational analysis of sterol ligand specificity of the Niemann Pick C2 protein Vasanthanathan Poongavanam, Jacob Kongsted, and Daniel Wüstner Biochemistry, Just Accepted Manuscript • DOI: 10.1021/acs.biochem.6b00217 • Publication Date (Web): 17 Aug 2016 Downloaded from http://pubs.acs.org on August 22, 2016
Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Biochemistry is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.
Page 1 of 47
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biochemistry
Computational analysis of sterol ligand specificity of the Niemann Pick C2 protein
Vasanthanathan Poongavanam§#, Jacob Kongsted§ and Daniel Wüstner†, * §
Department of Physics, Chemistry and Pharmacy and † Department of Biochemistry and Molecular Biology, University of Southern Denmark, DK-5230 Odense M, Denmark
#
Present address: Biomedicinskt Centrum (BMC), Department of Chemistry, Uppsala University, Husargatan 3, 75237 Uppsala, Sweden.
*To whom correspondence should be addressed: Department of Biochemistry and Molecular Biology, University of Southern Denmark, DK-5230 Odense M, Denmark Tel. +45-6550-2405, Fax +45-6550-2405,
[email protected] 1 ACS Paragon Plus Environment
Biochemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 2 of 47
ABSTRACT Transport of cholesterol derived from hydrolysis of lipoprotein associated cholesteryl esters out of late endosomes depends critically on the function of the Niemann Pick C1 (NPC1) and C2 (NPC2) proteins. Both proteins bind cholesterol but also various other sterols and both with strongly varying affinity. The molecular mechanisms underlying this multi-ligand specificity are not known. On the basis of the crystal structure of NPC2, we have here investigated structural details of NPC2-sterol interactions using molecular mechanics Poisson Boltzmann Surface Area (MM-PBSA) calculations. We found that an aliphatic side chain in the sterol ligand results in strong binding to NPC2, while sidechain oxidized sterols gave weaker binding. Estradiol and the hydrophobic amine U18666A had the lowest affinity of all tested ligands and at the same time showed the highest flexibility within the NPC2 binding pocket. The binding affinity of all ligands correlated highly with their calculated partitioning coefficient (logP) between octanol/water phases and with the potential of sterols to stabilize the protein backbone. From molecular dynamics (MD) simulations, we suggest a general mechanism for NPC2 mediated sterol transfer, in which Phe66, Val96 and Tyr100 act as reversible gate keepers. These residues stabilize the sterol in the binding pose via π-π stacking but move transiently apart during sterol release. A computational mutation analysis revealed that the binding of various ligands depends critically on the same specific amino acid residues within the binding pocket providing shape complementary to sterols, but also on residues in distal regions of the protein. Keywords: cholesterol, transport, lysosome, low density lipoprotein, MM-PBSA, binding affinity.
2 ACS Paragon Plus Environment
Page 3 of 47
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biochemistry
INTRODUCTION Cellular cholesterol homeostasis depends critically on cholesterol supply via the blood stream in form of low-density lipoprotein (LDL) in a process called forward cholesterol transport. LDL particles bind to the LDL receptor and become internalized by clathrinmediated endocytosis. Within the sorting endosome, LDL will dissociate from the receptor, which recycles to the cell surface for another round of LDL uptake. The released LDL follows the pathway of endosome maturation and becomes enriched in late endosomes and lysosomes (LE/LYSs) in which acid lipase hydrolyzes the LDL associated cholesteryl ester.1 How the liberated cholesterol becomes exported from LE/LYSs to meet cellular needs is not understood. Mutations in either the Niemann Pick C1 (NPC1) or Niemann Pick C2 (NPC2) proteins lead to almost identical phenotypes with strong accumulation of LDL-derived cholesterol in LE/LYSs.2 In these rare neurodegenerative diseases, i.e., NPC1 and NPC2 disease, respectively, transport of LDL and hydrolysis of its cholesterol ester appears to be normal but export of cholesterol from LE/LYSs is blocked making these disorders an important model for unraveling the molecular mechanisms of lysosomal cholesterol export. Fibroblasts lacking either functional NPC1 or NPC2 have a strongly reduced ability to elicit normal regulatory responses, as stimulation of esterification and suppression of synthesis of cholesterol.2, 3 The NPC1 protein is a large transmembrane protein located in various organelle membranes including LE/LYSs at steady state.4 The NPC2 protein is a small glycoprotein found in the lysosomal lumen, but NPC2 is also present in several body fluids including milk, bile and epididymal fluid. Both proteins bind cholesterol, though likely with opposite orientation: while in the NPC1 protein cholesterol binds with its hydroxyl group buried in the N-terminal binding region, in the NPC2 protein cholesterol is supposed to orient its 3’-hydroxyl group towards the aqueous interface.
3, 5-7
Based on
in vitro binding and sterol transfer experiments as well as the crystal structure of both, NPC2 and the N-terminal domain of NPC1, the following hand-over model has been proposed; 8, 9 NPC2 receives the liberated cholesterol after hydrolysis of cholesteryl ester 3 ACS Paragon Plus Environment
Biochemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 4 of 47
from acid lipase and donates the sterol ligand to NPC1 which is supposed to shuttle cholesterol over the glycocalyx-covered endo-lysosomal membrane and thereby out of the LE/LYSs. In support of this model, an alanine mutagenesis screen of both proteins found not only amino acids essential for sterol binding but also for transfer, and those residues are often found at the interface between the proposed molecular complex of NPC1 and NPC2.9 Recent molecular simulations of this proposed complex based on the crystal structure of the components have provided further insight into a possible path of cholesterol gliding from NPC2 to NPC1.
10, 11
These successes in unraveling the
molecular details of transport of LDL derived cholesterol hide the fact that many additional observations are not included in this model. For example, do both proteins bind other sterols, and in particular NPC1 has a stronger affinity to oxysterols than to cholesterol and was originally found in a screen for ligands to 25-hydroxy cholesterol.6 Also NPC2 binds various oxysterols and has the highest affinity not to cholesterol but to cholesterol-3-sulphate (CHSO4).12 NPC2 catalyzes the rapid exchange of cholesterol and intrinsically fluorescent sterols such as cholestatrienol (CTL) and dehydroergosterol (DHE) between model membranes in in vitro experiments, and this sterol transfer process is accelerated under conditions (anionic lipid composition, low pH and high ionic strength) as found inside LE/LYSs.
13-15
The structural details of the broad sterol ligand
specificity of NPC2 are not known, since only in the ligand-free form or in complex with CHSO4 its crystallization and structural analysis became possible.2, 12 The bovine NPC2 contains 7 strands, which are arranged into two sheets forming a hydrophobic pocket with a volume varying from 180 to 740 Å3 in the apo and sterol-bound form, respectively.5, 16 The X-ray structure of NPC2 with CHSO4 shows close proximity between the sterol ring system and lipophilic amino acids, such as Val59 and Val64, Phe66, Tyr100 and Pro101. To obtain insight into the broad sterol ligand specificity of NPC2, a chromatographic shift assay combined with inhibition of DHE binding to the NPC2 protein by cholesterol related molecules has been employed.12 DHE is an intrinsically fluorescent close analog of cholesterol, which has been shown to bind strongly to NPC2 and to be transferred by NPC2 between liposome membranes.5, 14, 16 In a competition binding assay delipidated human NPC2 was incubated with the respective sterol ligand for 30 min followed by incubation with DHE for another 30 min and analytical cation exchange chromatography 4 ACS Paragon Plus Environment
Page 5 of 47
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biochemistry
to remove any unbound sterols.
12
The amount of bound DHE was assessed from its
absorbance at 328 nm and taken as measure for the ability of a ligand to replace cholesterol in the NPC2 binding pocket. Although most of the tested compounds are structurally very similar, their potency to inhibit the DHE binding to NPC2 was found to differ greatly, ranging from no binding (e.g. cholesteryl oleate could not affect binding of DHE to NPC2) to high replacement activity (e.g. CHSO4 completely inhibited binding of DHE to NPC2). CHSO4 but not cholesterol could be co-crystallized with the human NPC2 protein.16 Here, we investigate the molecular basis for such a large difference in the binding affinity of cholesterol analogues in the NPC2 binding site as observed in the experimental studies.12,
17
For that purpose, we employ molecular dynamics (MD)
simulations based ligand binding free energy calculations and investigate the structural dynamics of the sterol-NPC2 complexes of the wild-type NPC2 protein and of various NPC2 mutations. Our study provides mechanistic insight into the sterol transfer activity of NPC2 and contributes to a better understanding of the structural aspects of the NPC2 disease phenotype.
COMPUTATIONAL METHODS PREPARATION OF NPC2 PROTEIN AND LIGANDS The molecular structure of bovine NPC2 was obtained from the Protein Data Bank (PDB ID: 2HKA, X-ray crystal structure resolution 1.81 Å).16 This protein structure was imported into the Maestro module available in the Schrödinger Suite18 and the protein was further optimized using the Protein Preparation Wizard.19 This optimization includes adding hydrogen atoms, assigning bond orders and building di-sulfide bonds. The protonation states of the ionizable residues (at pH=7.0 or pH=5.0) were predicted by the PROPKA tool20 provided in the Protein Preparation Wizard. An optimized structure model was finally found by energy minimization (i.e. only position of the hydrogen atoms) using the OPLS2005 force field.21 A set of 9 sterols (cholesterol: CHO, dehydroergosterol: DHE, cholesterol sulphate: CHSO4, 25-hydroxy cholesterol: 25OH, 24-hydroxy cholesterol: 24OH, estradiol: EST, cholesterol acetate: CHACE, U18666A: U186, cholestatrienol: CTL) that showed substantial differences in the experimental binding assays towards the NPC2 protein was 5 ACS Paragon Plus Environment
Biochemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
selected from the literature (Figure 1).
16
Page 6 of 47
These ligands were built using the Maestro
module.18 Subsequently, the ligands were pre-processed using the LigPrep module22 of the Schrödinger package (determination of protonation states using the Epik tool23 and energy minimization). The ligands were minimized using the MMFF95S force field. GENERATION OF LIGAND BINDING CONFORMATION The initial positioning of the ligands into the receptor was based on docking. The receptor grid generation module of Glide24 was used to define the active site for the docking. As the NPC2 crystal structure has a bound sterol ligand (i.e., CHSO4)16, the centroid of the grid box (of size 20 Å) was centered at this ligand. Water molecules at the active site beyond 3 Å from the bound ligand were deleted. The docking and scoring function parameters and settings used in this study are described in detail elsewhere.25 A proper starting configuration of a given ligand represents a crucial step for a final correct ranking of the ligands according to the experimental observation. Therefore, each ligand was docked and the best 10 ligand poses were saved for further binding pose analysis. In the crystal structure of NPC2 with tightly bound CHSO4, the tail of the sterol (Figure 1) is deeply inserted into the NPC2 binding pocket in order to establish hydrophobic interaction with the protein. After docking, the binding poses for each ligand were analyzed. It was found for many ligands that the sterol tails attached to the steroid D-ring deeply insert into the NPC2 pocket, exactly at the same site as it was found for CHSO4 in the X-ray structure of the NPC2-CHSO4 complex (Figure 2A and SF1). The docking process revealed that a few ligands (i.e. EST and U186) could bind to NPC2 either by orienting their D-ring attached ‘tail’ or their A-ring linked polar ‘head group’ towards the NPC2 binding pocket. In such cases we considered both orientations as relevant for subsequent MD simulations. Interestingly, the orientation of CHACE (which has low activity in the experimental DHE competition assay12) within the NPC2 binding pose was found to be very similar to that of CHSO4 (high activity; i.e., almost complete inhibition of DHE binding). The similar orientation of CHACE and CHSO4 in the NPC2 binding pocket despite their strongly differing binding affinity in experiments was unexpected. Therefore, in addition to the docking pose, a constrained ligand pose was also considered for MD simulations of CHACE i.e. CHACE was manually rotated 180o 6 ACS Paragon Plus Environment
Page 7 of 47
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biochemistry
so that its head group (the acetyl moiety, see Figure 1) is inserted into the pocket instead of the tail group (for more details see SF1 and below). The binding pose of the remaining compounds did not changed, as these compounds yielded only one type of binding pose (i.e. apolar side chain inserted into the binding pocket). MOLECULAR DYNAMIC SIMULATIONS All MD simulations were performed and analyzed using the Amber 14 software.
26
Before proceeding the MD simulations of the different NPC2-sterol complexes, all nine ligands shown in Figure 1 were geometry optimized at the level of HF/6-31G** using Gaussian 09
27
, and the atomic charges used for the sterols in the molecular dynamics
simulations were calculated from the electrostatic potential (ESP) using density functional theory at the level of B3LYP/cc-pVTZ including the IEF-PCM solvation model with settings for water as a solvent.28 These atomic charges were fitted using the RESP procedure as implemented in the Antechamber module of the Amber 14 software.26 After the geometry optimization and charge calculation for the ligands, the tleap tool in the Amber suite was used to build coordinate and parameter files using the Amber ff14SB force field. Subsequently, TIP3P water (solvent) molecules were added with a 10 Å buffering distance between the edges of the truncated octahedron box. Energy minimization was carried out in two steps; first, the system was minimized using a steepest descent minimization with all heavy atoms restrained. The maximum number of minimization cycles was set to 1000. In the second stage of the minimization, the entire system was energy minimized, and no positional restraints were applied at this stage. In order to avoid edge effects, periodic boundary conditions were applied during the MD simulations. In the process of thermalization, initial velocities were generated from a Maxwell-Boltzmann distribution at 100 K and the system temperature was gradually increased to 300 K at constant volume over a 200 ps MD simulation. After the thermalization process, the system was equilibrated at constant temperature (300 K) and pressure (1 bar) using the Berendsen coupling algorithm
29
for another 500 ps MD
simulation. After the equilibration step, the MD production run was started for 20 ns using a time step of 2fs. The SHAKE algorith30 was used to constrain the lengths of all 7 ACS Paragon Plus Environment
Biochemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 8 of 47
bonds involving hydrogen atoms. Coordinates were saved every 10 ps from the 20 ns simulation for analysis (in total 2000 snapshots). MM-PBSA CALCULATIONS Accurate prediction of binding affinities for ligands as well as their ranking with respect to each other represents a major challenge in computer-aided drug design in particular in lead identification/optimization processes.31, 32 Several methods have been developed to tackle this problem including the MM-PBSA (Molecular Mechanics Poisson Boltzmann Surface Area). The theory and methodology behind MM-PBSA have been described in detail in many publications.33-36 Briefly, MM-PBSA is a force-field based method which employs molecular mechanics (MM), the Poisson Boltzmann (PB) implicit solvation model, and solvent accessibility surface area to approximate the free energies of binding based on snapshots extracted from e.g. MD simulations. Under physiological conditions, the binding of a drug, i.e., a ligand (L), to the protein (P) forms a protein-ligand complex. In aqueous environment (indicated by the index, Aq), this process is reversible as shown below
PAq + LAq ⇔ PLAq (1) The binding free energy of a protein-ligand complex (PL) in an aqueous environment can be estimated from the Gibbs free energy difference between the bound and unbound states of the protein P and ligand L. This protocol is standard for measuring the binding affinity through computational methods:
∆GBind = GPL − GL − GP
(2)
However, significant computational time is required to calculate the solvent-solvent interactions. Therefore, the binding free energy is calculated based on the thermodynamic cycle (Eqn. 3).
∆GBind = ∆GBind,Vacuum + (∆GSolv,PL − (∆GSolv,L − ∆GSolv,P ) where ∆GBind and ∆G
Bind, Vacuum
(3)
correspond to the free energy difference between the
bound and unbound states of a complex in solvent and vacuum respectively. The terms 8 ACS Paragon Plus Environment
Page 9 of 47
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biochemistry
∆GSolv (∆GSolv,L, ∆GSolv,P and ∆GSolv,PL) represent the change in free energy between the solvated and vacuum states of the ligand, receptor or complex, respectively. These different contributions can be calculated as a sum of three terms:
G = EMM + GSolv( polar+nonpolar ) − T SMM (4)
EMM = EInt + EEl + EvdW
(5)
where EMM is the MM energy of the molecules, a measure of the enthalpic contribution to the Gibbs free energy. EMM is the sum of the internal energy (EInt) of the molecules (i.e. bonded terms including bond and torsion angles), EEl and EvdW represent the intermolecular electrostatic and van der Waals interactions energies between the protein and the ligand, respectively. In order to reduce the computational time, and to obtain stable energies, a single-trajectory is normally used for the ligand, protein and complex, i.e. only the PL complex is explicitly considered in the (NPT)-MD simulation. The energetic contributions of the ligand and protein are later subtracted from the simulation, thereby cancelling the EInt term from the calculation of ∆GBind. The Gibbs free energy of solvation, Gsolv, is composed of polar and non-polar solvation energies of the molecule, and those are estimated from the PB approximation combined with a solvent accessible surface area (SASA) calculation. In Eqn. 4, T is the temperature and SMM is the entropy (estimated for example from a normal-mode analysis calculated at the MM level).
37, 38
All the components in Eqn. 4 are averages, as obtained from snapshots based on the MD simulations (indicated by triangle brackets). The binding free energies (∆GBind) for all protein-ligand complexes were calculated using the MMPBSA.py script39 in Amber 14 based on the use of 2000 snapshots each extracted from the 20 ns MD simulation. All MM-PBSA calculations are based on the “singletrajectory MD simulation”, meaning that no separate MD simulations were run for free ligands or the receptor. As mentioned above, the entropic term is usually estimated from a normal mode analysis, but due to slow convergence, we excluded the entropy contribution to the free energy difference calculation, which is justified based on the fact that we only consider relative binding affinities. 9 ACS Paragon Plus Environment
Biochemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 10 of 47
COMPUTATIONAL MUTATION CALCULATION In order to understand the binding affinity differences of cholesterol-NPC2 complexes upon various mutations, additional MM-PBSA calculations were also run for NPC2 mutants as described above for the wildtype.39, 40. Before running such MD calculations, a specific residue is mutated using the “Mutate residue” option in the Maestro module in the Schrödinger Suite.18 Subsequently, the structure was energy optimized using the OPLS-2005 force field, and the resulting structure was used as starting pose for MD simulation followed by MM-PBSA calculation as described above.
RESULTS AND DISCUSSION PRELIMINARY ASSESSMENT OF THE BINDING POCKET All sterols used in this study were docked into the NPC2 binding pocket to obtain an initial binding orientation (pose) (Figure 1 and SF1). This was followed by MDsimulations based free energy binding calculations i.e. MM-PBSA. Initially, docking settings were assessed by the ability to reproduce the bound ligand (CHSO4) conformation in the bovine NPC2 X-ray crystal structure. 16 For this the observed RMSD (root-mean-square deviation) for the top 10 poses were less than 1.5 Å (the best pose gives an RMSD of only 0.8 Å). It has previously been demonstrated
16
that the ligand
binding pocket is quite shallow and primarily composed of hydrophobic amino acids such as Val (20, 38, 55, 59, 96, 105, 128), Leu (30, 94), Ile (103, 124, 126), Phe66, Pro (95, 101), Tyr (36, 100) and Trp (109, 122). Due to the lipophilic nature of the binding pocket, it is likely that sterols with apolar side chains (tail) are preferentially bound. In fact, CHSO4 is known to have the highest affinity and was co-crystallized with bovine NPC.
16
On the other extreme, steroid
hormones, such as EST having a hydroxyl- group instead of an alkyl side chain (Figure 1) showed almost no affinity to NPC2 in a fluorescence-based competition assay to DHE. 12
The comparison of binding pose of poor and high affinity compounds is shown in
Figure 2A. It is believed that cholesterol export from LE/LYSs takes place via translocation of the sterol from NPC2 to NPC1 via an intermediate structure (a NPC2Ligand-NPC1 complex), in which the cholesterol 3’hydroxy group has high affinity 10 ACS Paragon Plus Environment
Page 11 of 47
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biochemistry
towards the relatively polar binding site of NPC1 and the cholesterol alkyl chain experiences large conformational changes.10, 12, 16 It is not known, whether other sterols than cholesterol can be transferred via such an NPC2-ligand-NPC1 complex. Still, such simulations point to the importance of sterol ligand flexibility in NPC2-sterol interactions. It has been reported that the sterol-binding pocket of NPC2 exists in two different conformations; an open and a closed form. The cavity volume of the open and closed conformations varies widely from 180-720 Å3. This conformational transition is caused by the E and F loops of NPC2, particularly residues Tyr100 and Pro95, which move by ~3.5 Å and ~2.5 Å, respectively, away from their original position in the closed state.10, 16 In addition, when the sterol binds, residue Phe66 flips 80o away from the center of the ligand-binding pocket (Figure 2B). This suggests that when the tail of the sterol approaches the NPC2 active site, Phe66 and Tyr100 moves away allowing for easy entering of the ligand into the NPC2 binding pocket. Such a large conformational change of NPC2 might also favor the sterol translocation into the NPC1 binding pocket through simultaneous conformational change of the apolar tail and favorable polar interaction from head polar groups of the transferred sterols.10 In order to compare the flexibility of each residue in the apo and bound forms of NPC2, we investigated the B-factor, which is a measure of the mean square displacement of atomic fluctuations around the start structure (which is here calculated from the X-ray structure). We did this analysis for the backbone atoms (αC, N, C) from 2000 snapshots extracted from the respective 20 ns MD simulations (Figure 2C). It has been shown in the crystal structure, that the binding tunnel in NPC2 is formed primarily by residues with hydrophobic side chains, as V20, L30, Y36, V38, V55, G57, V59, V64, F66, L94, P95, V96, Y100, P101, I103, V105, V107, W109, W122, I124, I126 and V128.16 The underlined residues were found to change position in the crystal structure of NPC2 in the apo form compared to the CHSO4-bond NPC2. We observed that residues 6-29, 59-88 and 91-94 are highly flexible in the presence of ligand (Figure 2C). In other regions, as for residues 32-35 and 95-105, the backbone flexibility decreases after ligand binding. In particular, residues located on the entrance of the ligand binding pocket (as Val64, Phe66, Val96, Lys97, Tyr100, Pro101 or Ile103) are more flexible in the apo form than in the sterol bound form for most ligands, except for EST (compare black dashed and blue 11 ACS Paragon Plus Environment
Biochemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 12 of 47
straight lines in Figure 2C). Such residues are essential for the binding event and have very different positions in the crystal structure of the apo and sterol bound form of NPC2.16 A summary of the correlation of B-factors with binding affinity is provided in supplementary material, ST1. This is in contrast to residues deep inside the pocket or tunnel (residue number 20, 30, 35, 36, 54, 56, 59, 64, 66, 94, 100, 101, 103, 105, 107, 109, 124, 126). The backbone of such residues in the tunnel becomes more flexible when the sterols have bound to NPC2. The fluctuation analysis based on MD simulations also indicates that flexibility of certain regions (e.g., residues 22-29, 32-37, 59-70, 82-88 and 93-104) is anti-correlated with the binding affinity, i.e. high flexibility correlates with reduced binding affinity. Especially for EST, the flexibility goes up to peak B-values of 87-130Å2 in some regions followed by moderate flexibility for U186 and the two oxysterols, with B-values ~ 50-80Å2. In most regions, strong binders, as CHSO4 and cholesterol, give little protein flexibility as judged by low B-values. We conclude from that analysis that some regions of the NPC2 protein have to adapt their conformation to the sterol ligands in an “induced fit” manner and stay highly flexible in case of weak interactions, as observed in particular for EST. Other residues, in particular at the entrance of the binding pose become ‘locked’ in the complex, as residues Val64, Phe66, Lys97, Tyr100 or Ile103. This ‘freezing’ effect likely contributes to efficient shielding of the ligand from the surrounding polar medium thereby contributing to the stability of the complex. RELATIVE BINDING AFFINITY PREDICTION Initially, the top 10 docking poses were analyzed for each of the ligands and of these 10 top ranked poses, the pose, which occurs most frequently, was chosen as starting structure for the MD simulations (The initial docking poses used for MD simulations are shown in SF1). Since there is no experimental binding affinity (Ki) measurements available for these compounds, the experimentally determined percentage of inhibition (%) of DHE binding to NPC2 by the sterols16 was converted into logarithmic scale (log (affinity)) in order to get linear relationship between the observed and the calculated binding energies obtained from the MM-PBSA approach.
12 ACS Paragon Plus Environment
Page 13 of 47
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biochemistry
From the MD simulations, the binding free energy change for each complex was estimated based on snapshots extracted from each MD run using the MM-PBSA method. 33, 37
The summary of the free energies obtained for each complex is provided in Table 1,
which shows all the energy terms of eqn. 5 together with the experimentally measured activities as well as the predicted binding free energies (∆GBind). In general, the van der Waals (nonpolar) energy seems to favor the ligand binding more than the electrostatic interactions. CHO and CHSO4 have the highest affinity to NPC2 in our calculations, similar as found in experiments12. Lowering the pH to 5.0, as found in LE/LYSs did not alter the calculated binding affinity, as assessed for CHSO4 from additional simulations (not shown). Upon correlating the calculated binding affinity (MM-PBSA) with the experimentally measured inhibition of DHE binding to NPC212, the overall correlation found to be very poor correlation (R2 = 0.15) and this is mainly caused by deviations of the compounds EST, U183 and CHACE. EST is predicted to be an inactive compound (∆GBind > 0 kcal/mol). However, the correlation did not significantly improved (R2= 0.12) upon correction of the initial structure of EST by rotating EST in the binding pose manually by 180o (see Figure 2 and SF1, ST2). However, EST’s alternative pose significantly improved its binding affinity from 0.16 kcal/mol to -2.26 kcal/mol. This increased correlation between experimental data by Liou et al.12 and our simulations come from stronger binding of the rotated EST to NPC2, mainly due to the change in hydrophobic interactions of the ligand with the active site. The significant change in binding affinity of EST after rotating it in the binding pose is mainly due to the cyclo pentanol fragment which is slightly less hydrophobic (LogP = 0.86) compared to the phenolic fragment (logP = 1.64) (see SF2). The LogP for each fragment was calculated (theoretical) using the ChemBioDraw software.41 In case of CHACE, the correct ranking (i.e., the experimentally observed order of relative binding affinities in the DHE competition assay16), could not be achieved, and even if we used the constrained binding pose of CHACE as introduced for EST. According to the experimental finding (DHE competition assay), the compound CHACE poorly binds to NPC2 (low affinity), but according to the predicted binding energies, CHACE is found as a strong binder (high affinity, ∆GBind = -12.0 kcal/mol). Upon excluding CHACE from 13 ACS Paragon Plus Environment
Biochemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 14 of 47
the dataset for correlation of experimental and calculated binding affinities, the overall correlation improved remarkably from R2 = 0.12 to 0.83 (Figure 3A). To understand the reasons for the correlation of experimental and computational binding properties of various sterols to NPC2 (or lack thereof), we analyzed some key physicochemical properties of the sterol ligands and related those to the experimental and computational
relative
binding
affinities
(structure-activity
relationship;
see
Supplemental data, ST2). In particular, we calculated properties of the ligands such as molar refractivity (represents bulkiness of the ligand), atomic logP (AlogP, which is the atom-weighted logarithm of octanol/water partition coefficient, as calculated from a set of atom types), molecular weight, hydrogen bond donor/acceptor (polarity) and molecular surface area. As expected, descriptors that represent the hydrophilic-hydrophobic balance of the ligands correlated with the percentage of inhibition of DHE binding to NPC2 determined in experiments, except for CHACE.16 For instances, there is a strong correlation between experimental activities e.g. the potential of a sterol ligand to replace DHE in the competition assay12 and AlogP
42
(Figure 3B) AlogP is a measure of
hydrophobicity or lipophilicity of the sterol molecules, and this measure is strongly correlated with their binding affinity of the sterols to NPC2 (R2 = 0.93). Furthermore, the number of hydrogen bond acceptors (HBA) was inversely related to the experimental affinity in the DHE competition assay (R2 = 0.84, the lower the HBA, the higher the affinity). Absence of hydrogen bonds between NPC2 and sterols was also suggested based on the crystal structure of the protein with CHSO4.16 A moderate correlation was observed for the number of heavy atoms (HAC, R2 = 0.46), molar refractivity (MR, R2 = 0.53), and molecular weight (MW, R2 = 0.44) (see, ST3). From that analysis it can be concluded that the most relevant molecular property of sterols for being a suitable ligand to NPC2 is the hydrophobicity, and here especially the aliphatic character of the sterol side chain being buried inside the NPC2 pose. Still, we found that side-chain oxygenated sterols, as 24OH and 25OH have intermediate affinities to NPC2 (i.e., less than CHO, CHSO4, the fluorescent sterols DHE and CTL or CHACE but more than EST or U186). Those oxysterols were also shown to compete with DHE in the experimental study by Liou et al., though less compared to CHSO4 and CHO. We conclude that minor polar modifications in the side chain can be tolerated by NPC2, which could have important 14 ACS Paragon Plus Environment
Page 15 of 47
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biochemistry
physiological consequences (see Discussion section). A possible reason for the observed discrepancy between experimental data and our simulation results for CHACE can be attributed to the strongly differing water solubility of the sterol ligands. For example, we found for CHACE the highest an AlogP value of all ligands (7.76), while that of CHSO4 is with AlogP=6.65 much lower (see Figure 3B). In addition, we found a correlation between the calculated binding affinities and AlogP (R2=0.71) (Figure 3C). Thus, our results suggest that CHACE is in the same group as other strong binders, such as CHO, CTL or DHE, while CHACE performed poorly in the DHE competition experiments12. How can these discrepancies be reconciled? It has been shown that for neutral weak and strong detergents, the membrane/water partition coefficient is linearly related to the critical micelle concentration (CMC).42 For sterols, which have a low solubility in the aqueous phase, an intermediate micellar phase might form in the experimental completion assay, to an extent which is directly related to the hydrophobicity of the given sterol. This varying hydrophobicity is not accounted for in the estimation of binding affinities relative to DHE by Lobel and co-workers.
12, 44
In
other words, the experimental situation assesses the outcome of a two-step equilibrium process consisting of; q1
Sterol micelle ↔Sterol monomer
(6a)
q2
Sterol monomer ↔Sterol NPC2 complex
(6b)
Thus, in the DHE competition assay performed for various sterol ligands of NPC2, the total equilibrium constant, K=q1·q2, was estimated relative to DHE, which is K=[Sterol NPC2 complex]/[Sterol micelle], with square brackets indicating concentration.12 For very hydrophobic sterols, as CHACE, the CMC is very low, such that the concentration of micelles is higher at a given total sterol concentration than for more polar sterols, as CHSO4 but also as DHE (see AlogP values in Figure 3B). Accordingly, for such very hydrophobic sterol ligands, the experimental affinity is proportional to K and thereby underestimates the real affinity as given in Eq. 6b, above. In contrast, our simulations estimate the ligand affinity of NPC2 in isolation, i.e. the second reaction with equilibrium 15 ACS Paragon Plus Environment
Biochemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 16 of 47
constant q2=[Sterol NPC2 complex]/[Sterol monomer]. We therefore believe that our simulation results provide important complementary information to the experimental data, which does not take the different ligand solubility in the buffer solution into account. Our approach is not biased by eventual aggregation of the sterol ligand in the aqueous phase and directly reports about the relative affinity of various sterols to NPC2. DECOMPOSITION ANALYSIS OF THE BINDING FREE ENERGIES In order to determine the contribution of individual amino acid residues to the overall binding affinity of the compounds, a binding free energy decomposition analysis was performed. Here, the Gibbs free energies of interaction are decomposed from each protein residues and ligand i.e. sterol-residue pairs, which include ∆GvdW, ∆Gel and ∆Gsol(PB) energy contribution terms. Comparison of decomposition of binding energy for sterols (cholesterol and estradiol) used in the MM-PBSA calculations (final model) is shown in Figure 4. Energy contributions between each residue of NPC2 and the sterols are plotted in SF3. As seen from SF3 and as described in Supplementary information, respectively, all nine sterols showed quite similar interaction patterns with NPC2 residues, especially residues such as Leu30, Val59, Val64, Phe66, Leu94, Val96, Tyr100, Val105, Val107 and Ile126 contribute significantly (larger than -0.6 kcal/mol) to the overall binding affinity of all ligands. Except of Tyr100, all these residues participate in formation of the binding pose, though at different positions along the tunnel
16
(see above). Some of these residues are
conserved between human, murine, porcine, canine and bovine NPC2 and have been found to be essential in experimental binding studies of cholesterol to NPC2 and in cholesterol mobilization from LE/LYSs (i.e., Phe66, Val96, Tyr100).
5, 45
Others, as
Leu30, Val59, Val64, Val105 or Ile126, all residing in the binding tunnel, are not strictly conserved between mammalian NPC2 proteins, but contribute significantly to the binding event.16 The sequence identity among the mammalian NPC2 (human, bovin, rat and mouse) found to be 66.3% and there are 96 identical positions between these sequences (sequence alignment among these mammalian NPC2 is shown SF4). Interestingly, compounds such as CHACE, EST and U186 showed different interaction patterns with residues Tyr36, Ser37, Val38, Asn93, Lys94, Trp109 and Ile124 as 16 ACS Paragon Plus Environment
Page 17 of 47
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biochemistry
compared to the sterols with high affinity. For example, residue Ser37 shows an unfavorable (i.e. positive) energy contribution to seven compounds including EST and CHO (see Figure 4 and SF3). However, this residue’s energy contribution is favorable for U186, and a similar effect was also observed for residues Asn92 and Pro95. In contrast, Lys93 contributes (minor) favorably to all compounds except U186, which is unusually a large unfavorable contribution (~ 1.7 kcal/mol) to the overall binding affinity (see SF5). Together, this analysis indicates that mutating a certain amino acids, as Ser37 or Lys93 could cause reduced binding of the major sterol ligands but at the same time might give a gain-of-function in binding of some other sterol-like ligands to NPC2. Different protonation states of hydrophobic amines as U186 exist in the cell, and that could affect the binding affinity to NPC2 and its energetic composition. Since upon entering the acidic LE/LYSs, U186 will become protonated and thereby charged, it becomes trapped in LE/LYSs, where it could interact with NPC2. We therefore investigated the impact of different protonation states of U186 on NPC2 binding affinity and free energy decomposition. For this, MD simulations were initiated with or without a hydrogen atom at the tertiary nitrogen (see SF1 and SF5). Overall, when U186 is protonated, the protein backbone was found to be quite stable and showed a RMSD below 1 Å throughout the simulation as compared to U186 without a hydrogen atom at the tertiary nitrogen. In contrast, when deprotonated U186 binds in the active site of NPC2, the protein backbone moves from its original position, and this significantly increases the RMSD up to 1.4 Å, mainly due to the high flexibility in the ligand-binding region (residues 23 to 29 and 66 to 69 highlighted in Figure 5). Interestingly, protonation of U186 decreased the binding affinity to NPC2 from -14.34 kcal/mol to -6.01 kcal/mol. Thus, U186 is under physiological conditions a poor binder, similar as EST (see above). From the energy decomposition we can see that Asn92 yielded a favorable energy contribution (~ -1.5 kcal/mol) to U186 in the protonated but not in the deprotonated state (see SF3 and SF5). For the rest of the compounds, Asn92 did not showany significant energy contribution (< 0.01 kcal/mol). However, for Lys93, the energy contribution is unfavorable in the protonated state and slightly favorable in the deprotonated state (i.e., > +1.5 kcal/mol vs. ~ -0.05 kcal/mol, see SF5B). In addition, the relationship between sterol affinity and the protein residues forming the binding tunnel was also investigated 17 ACS Paragon Plus Environment
Biochemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 18 of 47
and found to be moderately correlated with the binding energy contribution. For example, the correlation (R2) between the binding affinity and energy contributions due to Tyr36, Val64, Leu94, and Ile126, all located in the binding pose and contributing to binding of deprotonated, neutral U186 (see SF5), was found to be more than 0.50 (see ST4). Overall, this analysis revealed that protonated and thereby charged U186, as it would exist inside LE/LYSs significantly reduces the flexibility of residues at the ligand-binding site in the protein. At the same time, protonated U186 binds weakly to NPC2 compared to cholesterol (CHO) or DHE, as found in the experimental assay.12 If, however, the extent of lysosomal trapping of U186 is significant, even the weak binder U186 could outcompete at least partly CHO and thereby induce the NPC2-disease like phenotype. Also, mutations of NPC2 in which certain residues, as Lys93 are replaced, could result in stronger binding of this hydrophobic amine to NPC2. STRUCTURAL DYNAMICS OF THE STEROL-NPC2 COMPLEXES Further insight can be obtained when studying the overall structural stability of NPC2 in its apo form or after binding the various sterol ligands. For that purpose, for all ligandprotein complexes and for the apo form of NPC2, the atom-positional RMSD of the backbone atoms (αC, N, C) relating to the initial structure was analyzed for the 2000 snapshots extracted from the 20 ns MD simulation (Figure 6). Except for EST, the RMSD for the backbone remains below 1.5 Å for all compounds, and no significant structural changes in the protein during the simulation were observed. However, the backbone RMSD of NPC2 in the presence of EST is significantly higher and this RMSD increased up to ~ 2.0 Å between 7 and 15 ns of the simulations. After 15 ns, the backbone of NPC2 reaches its normal position. This suggests that the protein undergoes major structural changes when EST binds in the active site, but after 15 ns, the protein returns to its original position. One can clearly see from analyzing the trajectories (snapshots representing the EST exit are shown in Figure 7; a movie is provided in the supplementary material) that residue Tyr100 and Phe66 are key players that induce the protein conformation changes in the active site and push the EST from its binding position outwards. Once the EST moved from the NPC2 hydrophobic pocket, the protein returns to its original position. This observation is likely not limited to EST but could resemble a general mechanism of NPC2 mediated sterol transport.12, 16 In fact, Tyr100 and 18 ACS Paragon Plus Environment
Page 19 of 47
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biochemistry
Phe66 are among the highly flexible residues for all ligands, as inferred from the B-factor plot (Figure 2C). A possible reason for why we did not observe such a structural dynamics for NPC2 in complex with the other sterols might be the fact that such large structural changes take place on a longer time-scale for more hydrophobic sterols with higher affinity to NPC2. Observing such events would accordingly require significantly longer simulations for NPC2 in complex with such sterols. However, given the large difference in aqueous solubility of sterol ligands (as assessed by the AlogP value in Figure 3), an alternative scenario could be that the more hydrophobic sterols actually need a membrane acceptor to leave the NPC2 binding pose. In fact, membrane interaction and even membrane fusion has been shown for NPC2 in various experimental studies14, 15, 46-48
and is further supported by the outcome of the mutational analysis discussed
below. COMPUTATIONAL MUTATIONAL STUDIES Mutagenesis studies have not only revealed amino acids in NPC2 of critical importance for sterol binding as discussed above, it has also previously been shown that mutations of certain amino acids in the vicinity of the ligand-binding region abolish or reduce the NPC2 cholesterol transfer function between membranes and at the same time impair the cholesterol mobilization from disease fibroblasts. 9,
45
Some other mutations significantly
influence the sterol transport between NPC2 and NPC1, and they were found either at the surface of the protein (residues that interact with membrane, e.g. Asp72, Lys75) or entrance/mouth of the protein (sterols recognition site, e.g. Phe66 or Tyr100) or tunnel (e.g. Val96) (Figure 4B). Some of the important mutations are alanine mutation of Pro24, Lys32, Val64, Phe66, Tyr100, Val105.9 In the wild-type, both Phe66 and Tyr100 interacts through π-π stacking with sterols, and mutation of any of these residues creates steric hindrance and neighboring residue-residue clash which blocks the sterol binding in the cavity.16 This is in strong support of the dynamic rearrangement, we found for these residues in the MD simulation of EST (see above and Figure 7). The decomposition analysis of MM-PBSA calculation also suggests that these residues contribute significantly to the sterol binding affinity (see Figure 4 and SF4). To further study the importance of such residues, the binding affinity of cholesterol was 19 ACS Paragon Plus Environment
Biochemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 20 of 47
estimated in the wild-type and mutated NPC2 protein. Initially, important protein residues, i.e. those that were experimentally shown to affect cholesterol binding, interbilayer transport or the ability of mutated NPC2 to rescue the sterol storage phenotype in cells (Pro24, Lys32, Val64, Phe66, Asp72, Lys75, Leu94, Val96, Tyr100, Val105) were selected and mutated either with alanine or phenylalanine. In addition, a triple mutant was generated, which contained three changed residues, i.e. F66A, V96F and Y100A (abbreviated ‘FVY’, see Figure 8). The mutated structures were energy optimized using the OPLS-2005 force field using the protein preparation wizard tool as implemented in the Schrödinger Suite. Subsequently cholesterol was docked into each mutated protein structures and the pose with highest docking score for each structure was further used for MM-PBSA calculation (for detail see the section on methods). A summary of relative binding affinities of cholesterol in various mutated forms of NPC2 is shown in Figure 8 (comparison of B-values and backbone RMSD for mutations and wild-type is provided in the SF6). Here, in order to compare cholesterol binding affinity differences due to mutational effects, the binding affinity of cholesterol to the wild-type (-16.94 kcal/mol) is subtracted from the binding affinities of the various mutations. Thus, a positive contribution in Figure 8 denotes that cholesterol binds poorly to the mutated NPC2 compared to the wild-type. In general, the relative affinity decreases when any of the active site residues are mutated either into alanine or phenylalanine. When highly hydrophobic residues such as Phe66, Leu94, Val96 and Val105 are mutated to alanine, the binding affinity decreased significantly, in close agreement with experimental findings (Figure 8). Phenylalanine is slightly more hydrophobic than valine, so one would expect that mutating Val64 into phenylalanine will either slightly improve the binding affinity of cholesterol, or leaves it unchanged, as observed in experimental studies.
45, 46
Indeed, we found that cholesterol
has a very similar binding affinity to NPC2 with the mutation V64F as to wildtype NPC2. Mutating residues in loop regions of NPC2 next to the binding pocket, as Pro24 → Ala24 decreases the binding affinity as well, while the wildtype residue Pro24 contributes little to the total binding free energy (compare Figure 4 and Figure 8). To explain this discrepancy, we took a closer look at the flexibility of this and adjacent residues in the loop region (SF7). Ala24 is less flexible than Pro24, while adjacent residues show 20 ACS Paragon Plus Environment
Page 21 of 47
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biochemistry
reduced flexibility (i.e., Thr25, Gln26 and Pro27). With the exception of Gln26, none of them contribute significantly to the total binding affinity (compare SF7 and Figure 4). Due to its cyclic structure, Pro24 locks the first backbone dihedral angle to about -65 degree, which changes upon mutation of this residue to alanine. This affects the flexibility of adjacent residue and impacts the overall binding affinity. Thus, the overall energy contribution of a given residue and the effect of its mutation on sterol binding are related properties but are not identical. Agreement between previous experimental findings and our computational analysis is also found for other residues; for example Xu et al. reported16, that mutation of Val96 with phenylalanine dramatically decreased the cholesterol affinity. We confirm that and can add that the reduced affinity in this mutant is not only due to change in the hydrophobicity but mostly a consequence of steric clashes. In other words, phenylalanine is blocking the proper orientation of cholesterol binding in the active site (Figure 9), which reduces the affinity of cholesterol in the mutant (Val96Phe) compared to the wildtype NPC2 from -16.94 to -15.92 kcal/mol. Such a difference in the binding affinity is mainly due to the van der Waals (vdW) energy contribution in the mutated, and this suggest that non-bonded interactions and shape complementarity play an important role in sterol binding to NPC2. Phe66Ala and Tyr100Ala mutations also showed less cholesterol binding affinity as compared to wild-type, but, interestingly, the difference was less pronounced compared to other mutations, though these residues play a central role as gate keepers in sterol release (see above and Figure 7). Since our calculations ignore entropic contributions to the binding free energy, large structural changes, as observed for Phe66 and Tyr100 in the apo form of NPC2 compared to the sterol bound form will not be accounted for in our calculations. This could explain the relatively low change in binding free energy upon mutating these residues to alanine. Very recently McCauliff et al.
46
have screened sterol transport properties of various
NPC2 mutants including residues at the surface of the protein (e.g. Lys32Ala, Asp72Ala, Lys75Ala) using fluorescence-based experiments in which DHE or CTL transfer between liposomal membranes was assessed by Förster resonance energy transfer from a Dansyltagged phosphatidylethanolamine in the donor membrane.
15, 46, 49
Their study indicates
that mutation of Aps72 and Lys75 to Ala in NPC2 dramatically changes the surface 21 ACS Paragon Plus Environment
Biochemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 22 of 47
charge distribution of the protein, thereby reducing its membrane interaction, because Ala75 favors negative charges from neighboring residues Glu70 and Asp72 on the surface. This and earlier studies also found that the mutations K32A, D72A and K75A have a minor effect on cholesterol binding to NPC2.
45, 46
Our MM-PBSA calculation
confirms that Lys32, Asp72 and Lys75 have a small contribution to NPC2’s overall binding affinity (see Figure 4) We also found that mutation of either of these residues to Ala results in a minor change in NPC2’s binding affinity to cholesterol (see Figure 8). Thus, we provide a coherent analysis of an amino acids energetic contribution and the effect of its mutation on sterol binding in NPC2 (see Table 2). Residues which have been shown to be essential in the binding step (as Tyr100, Phe66 and Val96) contributed significantly to the binding energy and resulted in high energy penalties upon mutation in our calculations. The flexibility of the protein backbone is higher in the alanine mutations in position 66 and position 100, which could contribute to the lowered binding affinity (see Table 2). Both positions are located at the entrance of the binding pose, and their conformational restriction upon sterol binding was found to be characteristic for strongly binding sterols (see Figure 4 and accompanying discussion). In contrast, residues whose mutation affect cholesterol binding only to a minor extent in experiments (as Lys32, Asp72 and Lys75) gave also only minor energy contributions in our calculations. Such residues are probably mostly important for the membrane interaction of NPC2 and thereby control the sterol transfer process between protein and membranes.46
CONCLUSIONS AND FUTURE PROSPECTS NPC2 is a sterol transfer protein in LE/LYSs of mammalian cells with broad sterol ligand specificity. Absence or dysfunction of this protein causes NPC2 diseases, a neurodegenerative disorder with endosomal accumulation of cholesterol and other lipids. In the present study, we used molecular simulations based on the crystal structure of NPC2 to explain the multi sterol ligand activity of NPC2 in direct comparison with experimental binding assays. Using MD simulations and MM-PBSA calculations we overall ranked the various sterol ligands correctly compared to the experimentally determined binding affinity relative to the fluorescent sterol DHE. We show that a structural requirement of high affinity sterol ligands to NPC2 is an aliphatic side chain 22 ACS Paragon Plus Environment
Page 23 of 47
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biochemistry
buried inside the NPC2 binding pocket, a high octanol/water partition coefficient and a low number of hydrogen acceptors, both properties being characteristic for hydrophobic ligands. Sterol binding by NPC2 strongly reduced the flexibility of the protein backbone at the entrance of the binding pose (its ‘mouth’), and this ‘freezing’ effect was correlated with the affinity of the sterols. Hydrophobic residues inside the binding pose increased in mobility upon binding, and this property was also necessary for efficient binding. This, together with the large displacement of such residues when the binding pose opens up to accommodate a sterol molecule, suggest a highly flexible ‘induced-fit’ mechanism as the underlying molecular process explaining NPC2’s high sterol binding potential. In addition, our study underlines the importance of considering a two-step transfer process in experimental binding assays, in which the studied sterol transfers first from eventual sterol micelles or aggregates to the aqueous phase (with equilibrium constant q1, Eq. 6a) followed by the actual binding step with binding constant q2, while in the simulations, the binding step is studied in isolation. Thus, differences in the aggregation state or CMC of various sterols will affect the experimental analysis as recently reviewed50, but not the simulation results presented here. Therefore, for hydrophobic ligands, as sterols, MMPBSA calculations can complement experimental studies very well, as suggested by the comparison of CHSO4 and CHACE, for which we find similar binding free energies, while both sterols behave very differently in a DHE competition-binding assay to human NPC2. Differences between human and bovine NPC2 or particularities of the DHE assay can be ruled out, since Okamura et al., found similar results for CHACE for bovine NPC2 in a different binding assay. 17 Side-chain oxysterols, as 24-OH or 25-OH bind less strongly to NPC2 compared to CHOL or CHSO4, while binding of the fluorescent sterols CTL and DHE was comparable supporting their close resemblance of cholesterol. Oxysterols as 25-OH are important metabolically active cholesterol derivatives, as 25-OH inhibits sterol response element binding protein (SREBP) maturation and thereby the transcription of genes involved in cholesterol homeostasis and uptake.51 The N-terminal sterol binding pocket of NPC1 binds 25-OH with higher affinity than CHO, and NPC1 was first detected as an oxysterol binding protein in a biochemical assay6,7. Furthermore, 25-OH binds the liver X receptors (LXRs) and thereby controls expression of transporters for cholesterol efflux 23 ACS Paragon Plus Environment
Biochemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 24 of 47
from cells. Up-regulating this pathway can bypass the cholesterol storage phenotype in NPC1 diseased cells but has no or only minor effects in cells from NPC2 disease patients.51-53 CTL hydroxylated at carbon 25 is a fluorescent derivative of 25-OH, which has recently been shown to be exported from LE/LYSs in a NPC1 dependent manner.54 Based on our simulation results, we propose that NPC2 can shuttle 25-OH and other sidechain oxidized cholesterol derivatives out of LE/LYSs, either directly or in concert with NPC1. Another study has recently shown that side-chain oxysterols can order monounsaturated phosphatidylcholine membranes to a similar degree as cholesterol but have additional tilted orientations in the bilayer.55 In contrast to cholesterol, they can even have the reversed orientation or span both leaflets via their hydroxyl groups sitting on opposite sides of the molecules.55 Even though, we found that NPC2 has a lower affinity to side-chain oxysterols, as 25-OH, it is possible that it can pick them up more easily from membranes given their more flexible orientation. This could play a role in oxysterol transfer from intraluminal vesicles either directly to the limiting endosomal membrane or to the luminal sterol binding domain of NPC1. Future studies could be envisioned to investigate these possibilities further. Replacing the alkyl-chain by a hydroxyl group (as in EST) or by a carbonyl group (as in U186) strongly reduced the affinity of the sterol ligands to NPC2. Nevertheless, both sterols could bind to NPC2 to some extent confirming experimental findings.12 Since, incubating cells with hydrophobic amines, as U186 but also with certain steroid hormones can mimic the NPC phenotype56, 57 our results cannot rule out that this effect is due to a direct blocking of the NPC2 binding pocket by these ligands. On the other hand, U186 has been shown to inactivate NPC1 by binding to its sterol sensing domain, distinct from the N-terminal sterol binding domain of NPC1.58 This, however, is not a contradiction, as sterols bind in the opposite orientation to NPC1 compared to NC2 (i.e., with the 3’hydroxy ‘head group inside the binding pose). The extended branched diethylaminoethyl ‘head group’ of U186 would likely prevent binding to the NPC1 sterol binding pose, while this group will not interfere with binding to NPC2. Interestingly, a free energy decomposition showed that the majority of amino acids contributing significantly to the overall binding energy are the same for different ligands and irrespective of their relative affinity (see Figure 4 and SF3). Thus, the overall binding 24 ACS Paragon Plus Environment
Page 25 of 47
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biochemistry
pattern of NPC2 to sterols seems to be ubiquitous with hydrophobic stabilization of the ligand and conformational restriction or ‘freezing’ of the protein backbone at the mouth as the central schemes. For one of the weakest ligands, EST, we observed a major structural transition in NPC2 during the MD simulation, in which residue Tyr100 and Phe66 contribute to push the ligand away from its binding position. Once the EST moves away the NPC2 hydrophobic pocket, the protein returns to its original position. Since both amino acids have been implicated in cholesterol binding and mobilization of endosomal cholesterol, our results point to a central mechanism role of such ‘gatekeeper’ residues in sterol transfer by NPC2. NPC2 has been shown to accelerate inter-bilayer transfer of cholesterol, CTL or DHE in a concentration-dependent manner.
3,15
Exchange of such sterols between NPC2 and
membranes was proposed to require transient membrane interaction of the protein, and key residues have been identified for NPC2’s interaction with lipid bilayers.
3, 15, 49
The
simulation approach presented here can be applied in future studies to characterize the molecular and energetic details of NPC2-membrane interaction. This will set the stage towards a thorough understanding of the mechanisms underlying NPC2 mediated sterol export from LE/LYSs and thereby give insight into the pathobiology of NPC2 disease.
ASSOCIATE CONTENT * Supporting Information Figure SF1: Initial conformation of protein-ligand complex used for MD simulations, number indicates the normal (1) and constrained (2) binding pose, Figure SF2: calculated physicochemical properties of fragments of estradiol and U186, Figure SF3. The energy contributions (obtained from binding energy decomposition) between each residue of NPC2 and the sterols are plotted, Figure SF4: comparison of various mammalian NPC2 sequences is provided, Figure SF5: Comparison of decomposition energies for important residues from different simulations of CHACE and U186 is plotted. Figure SF6: Comparison of residues flexibility upon various mutations and wild type (A) and comparison of backbone RMSD (in Å, B). Figure SF7: Flexibility analysis of a loop region in wildtype and mutant (Pro24Ala) NPC2. In the Table ST1: summary of Bfactors (Å2) and binding affinities is provided for the dataset, summary of energies 25 ACS Paragon Plus Environment
Biochemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 26 of 47
obtained from MM-PBSA methods using various MD runs is provided in ST2. The correlation between experimental activity (relative % of inhibition) with physicochemical properties of sterols, and residues decomposition of binding energies from the MMPBSA calculation (kcal/mol) plots are provided in tables ST3 and ST4, respectively.
ACKNOWLEDGEMENT The authors are thankful to the DeIC National HPC Center, SDU, for providing the computational/simulation facilities.
FUNDING Financial support from the Novo Nordisk and Villum foundations to DW, Denmark is gratefully acknowledged. In addition, this work was supported by a generous grant from the VILLUM Foundation to the VILLUM Center for Bioanalytical Sciences at the University of Southern Denmark.
REFERENCES 1.
Goldstein, J. L., and Brown, M. S. (2009) The LDL receptor, Arterioscler. Thromb. Vasc. Biol. 29, 431-438.
2.
Mukherjee, S., and Maxfield, F. R. (2004) Lipid and cholesterol trafficking in NPC, Biochim. Biophys. Acta 1685, 28-37.
3.
Storch, J., and Xu, Z. (2009) Niemann-Pick C2 (NPC2) and intracellular cholesterol trafficking, Biochim. Biophys. Acta 1791, 671-678.
4.
Ko, D. C., Gordon, M. D., Jin, J. Y., and Scott, M. P. (2001) Dynamic movements of organelles containing Niemann-Pick C1 protein: NPC1 involvement in late endocytic events, Mol. Biol. Cell, 12, 601-614.
26 ACS Paragon Plus Environment
Page 27 of 47
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biochemistry
5.
Friedland, N., Liou, H. L., Lobel, P., and Stock, A. M. (2003) Structure of a cholesterol-binding protein deficient in Niemann-Pick type C2 disease, Proc. Natl. Acad. Sci. USA, 100, 2512-2517.
6.
Infante, R. E., Abi-Mosleh, L., Radhakrishnan, A., Dale, J. D., Brown, M. S., and Goldstein, J. L. (2008) Purified NPC1 protein. I. Binding of cholesterol and oxysterols to a 1278-amino acid membrane protein, J. Biol. Chem. 283, 10521063.
7.
Infante, R. E., Radhakrishnan, A., Abi-Mosleh, L., Kinch, L. N., Wang, M. L., Grishin, N. V., Goldstein, J. L., and Brown, M. S. (2008) Purified NPC1 protein: II. Localization of sterol binding to a 240-amino acid soluble luminal loop, J Biol. Chem. 283, 1064-1075.
8.
Kwon, H. J., Abi-Mosleh, L., Wang, M. L., Deisenhofer, J., Goldstein, J. L., Brown, M. S., and Infante, R. E. (2009) Structure of N-terminal domain of NPC1 reveals distinct subdomains for binding and transfer of cholesterol, Cell 137, 1213-1224.
9.
Wang, M. L., Motamed, M., Infante, R. E., Abi-Mosleh, L., Kwon, H. J., Brown, M. S., and Goldstein, J. L. (2010) Identification of surface residues on NiemannPick C2 essential for hydrophobic handoff of cholesterol to NPC1 in lysosomes, Cell Metab. 12, 166-173.
10.
Elghobashi-Meinhardt, N. (2014) Niemann-Pick type C disease: a QM/MM study of conformational changes in cholesterol in the NPC1(NTD) and NPC2 binding pockets, Biochemistry 53, 6603-6614.
11.
Estiu, G., Khatri, N., and Wiest, O. (2013) Computational studies of the cholesterol transport between NPC2 and the N-terminal domain of NPC1 (NPC1(NTD)), Biochemistry 52, 6879-6891.
12.
Liou, H. L., Dixit, S. S., Xu, S., Tint, G. S., Stock, A. M., and Lobel, P. (2006) NPC2, the protein deficient in Niemann-Pick C2 disease, consists of multiple glycoforms that bind a variety of sterols, J Biol. Chem. 281, 36710-36723.
27 ACS Paragon Plus Environment
Biochemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
13.
Page 28 of 47
Babalola, J. O., Wendeler, M., Breiden, B., Arenz, C., Schwarzmann, G., Locatelli-Hoops, S., and Sandhoff, K. (2007) Development of an assay for the intermembrane transfer of cholesterol by Niemann-Pick C2 protein, Biol Chem. 388, 617-626.
14.
Cheruku, S. R., Xu, Z., Dutia, R., Lobel, P., and Storch, J. (2006) Mechanism of cholesterol transfer from the Niemann-Pick type C2 protein to model membranes supports a role in lysosomal cholesterol transport, J. Biol. Chem. 281, 3159431604.
15.
Xu, Z., Farver, W., Kodukula, S., and Storch, J. (2008) Regulation of sterol transport between membranes and NPC2, Biochemistry 47, 11134-11143.
16.
Xu, S., Benoff, B., Liou, H. L., Lobel, P., and Stock, A. M. (2007) Structural basis of sterol binding by NPC2, a lysosomal protein deficient in Niemann-Pick type C2 disease, J Biol Chem 282, 23525-23531.
17.
Okamura, N., Kiuchi, S., Tamba, M., Kashima, T., Hiramoto, S., Baba, T., Dacheux, F., Dacheux, J. L., Sugita, Y., and Jin, Y. Z. (1999) A porcine homolog of the major secretory protein of human epididymis, HE1, specifically binds cholesterol, Biochim. Biophys. Acta. 1438, 377-387.
18.
Maestro, Schrödinger, LLC, New York, NY, 2014.
19.
Sastry, G. M., Adzhigirey, M., Day, T., Annabhimoju, R., and Sherman, W. (2013) Protein and ligand preparation: parameters, protocols, and influence on virtual screening enrichments, J. Comput.-Aided Mol. Des. 27, 221-234.
20.
Li, H., Robertson, A. D., and Jensen, J. H. (2005) Very fast empirical prediction and rationalization of protein pKa values, Proteins 61, 704-721.
21.
Jorgensen. W. L, Maxwell. D.S, and Tirado-Rives, J. (1996) Development and Testing of the OPLS All-Atom Force Field on Conformational Energetics and Properties of Organic Liquids, J. Am. Chem. Soc. 118, 11225-11236.
22.
LigPrep, Schrödinger, LLC, New York, NY, 2014.
23.
Epik, Schrödinger, LLC, New York, NY, 2015. 28 ACS Paragon Plus Environment
Page 29 of 47
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biochemistry
24.
Friesner, R. A., Banks, J. L., Murphy, R. B., Halgren, T. A., Klicic, J. J., Mainz, D. T., Repasky, M. P., Knoll, E. H., Shelley, M., Perry, J. K., Shaw, D. E., Francis, P., and Shenkin, P. S. (2004) Glide: a new approach for rapid, accurate docking and scoring. 1. Method and assessment of docking accuracy, J. Med. Chem. 47, 1739-1749.
25.
Poongavanam, V., Olsen, J. M., and Kongsted, J. (2014) Binding free energy based structural dynamics analysis of HIV-1 RT RNase H-inhibitor complexes, Integr. Biol. 6, 1010-1022.
26.
Case, D. A., Darden, T. A., Cheatham, T. E., Simmerling, C. L., Wang, J., Duke, R. E., Luo, R., Walker, R. C., Zhang, W., Merz, K. M., Roberts, B., Hayik, S., Roitberg, A., Seabra, G., Swails, J., Goetz, A. W., Kolossváry, I., Wong, K. F., Paesani, F., Vanicek, J., Wolf, R. M., Liu, J., Wu, X., Brozell, S. R., Steinbrecher, T., Gohlke, H., Cai, Q., Ye, X., Wang, J., Hsieh, M. J., Cui, G., Roe, D. R., Mathews, D. H., Seetin, M. G., Salomon-Ferrer, R., Sagui, C., Babin, V., Luchko, T., Gusarov, S., Kovalenko, A., and Kollman, P. A. (2012) AMBER 12, University of California, San Francisco.
27.
Frisch, M. J., Trucks, G. W., Schlegel, H. B., Scuseria, G. E., Robb, M. A., Cheeseman, J. R., Scalmani, G., Barone, V., Mennucci, B., Petersson, G. A., Nakatsuji, H., Caricato, M., Li, X., Hratchian, H. P., Izmaylov, A. F., Bloino, J., Zheng, G., Sonnenberg, J. L., Hada, M., Ehara, M., Toyota, K., Fukuda, R., Hasegawa, J., Ishida, M., Nakajima, T., Honda, Y., Kitao, O., Nakai, H., Vreven, T., Montgomery Jr, J. A., Peralta, J. E., Ogliaro, F., Bearpark, M. J., Heyd, J., Brothers, E. N., Kudin, K. N., Staroverov, V. N., Kobayashi, R., Normand, J., Raghavachari, K., Rendell, A. P., Burant, J. C., Iyengar, S. S., Tomasi, J., Cossi, M., Rega, N., Millam, N. J., Klene, M., Knox, J. E., Cross, J. B., Bakken, V., Adamo, C., Jaramillo, J., Gomperts, R., Stratmann, R. E., Yazyev, O., Austin, A. J., Cammi, R., Pomelli, C., Ochterski, J. W., Martin, R. L., Morokuma, K., Zakrzewski, V. G., Voth, G. A., Salvador, P., Dannenberg, J. J., Dapprich, S., Daniels, A. D., Farkas, Ö., Foresman, J. B., Ortiz, J. V., Cioslowski, J., and Fox, D. J. (2009) Gaussian 09, Gaussian, Inc., Wallingford, CT, USA. 29 ACS Paragon Plus Environment
Biochemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
27.
Page 30 of 47
Tomasi, J., Mennucci, B., Cancès, E (1999) The IEF version of the PCM solvation method: an overview of a new method addressed to study molecular solutes at the QM ab initio level. J Mol Struct. (Theochem), 464, 211-226.
29.
Berendsen, H. J. C., Postma, J. P. M., Vangunsteren, W. F., Dinola, A., and Haak, J. R. (1984) Molecular-Dynamics with Coupling to an External Bath, J. Chem. Phys 81, 3684-3690.
30.
Ryckaert, J.-P., Ciccotti, G., and Berendsen, H. J. C. (1977) Numerical integration of the cartesian equations of motion of a system with constraints: molecular dynamics of n-alkanes, J. Comput. Phys. 23, 327-341.
31.
Ferrara, P., Gohlke, H., Price, D. J., Klebe, G., and Brooks, C. L., 3rd. (2004) Assessing scoring functions for protein-ligand interactions, J. Med. Chem 47, 3032-3047.
32.
Stjernschantz, E., and Oostenbrink, C. (2010) Improved ligand-protein binding affinity predictions using multiple binding modes, Biophys. J. 98, 2682-2691.
33.
Genheden, S., and Ryde, U. (2015) The MM/PBSA and MM/GBSA methods to estimate ligand-binding affinities, Exp. Op. Drug Discov. 10, 449-461.
34.
Hayes, J. M., and Archontis, G. (2012) MM-GB(PB)SA Calculations of ProteinLigand Binding Free Energies. Molecular Dynamics - Studies of Synthetic and Biological Macromolecules, Ed., Lichang Wang, INTECH Pub. Croatia
35.
Swanson, J. M., Henchman, R. H., and McCammon, J. A. (2004) Revisiting free energy calculations: a theoretical connection to MM/PBSA and direct calculation of the association free energy, Biophys. J. 86, 67-74.
36.
Weis, A., Katebzadeh, K., Soderhjelm, P., Nilsson, I., and Ryde, U. (2006) Ligand affinities predicted with the MM/PBSA method: dependence on the simulation method and the force field, J. Med. Chem. 49, 6596-6606.
37.
Kongsted, J., Ryde, U., An improved method to predict the entropy term with the MM/PBSA approach, J Comput.-Aided Mol. Des, 2008, 23, 1573-4951.
30 ACS Paragon Plus Environment
Page 31 of 47
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biochemistry
38.
Genheden, S., Kuhn, O., Mikulskis, P.,
Hoffmann, D., and Ryde, U., The
Normal-Mode Entropy in the MM/GBSA Method: Effect of System Truncation, Buffer Region, and Dielectric Constant, J. Chem. Inf. Model. 2012, 52 ,2079– 2088 39
Miller, B. R., 3rd, McGee, T. D., Jr., Swails, J. M., Homeyer, N., Gohlke, H., and Roitberg, A. E. (2012) MMPBSA.py: An Efficient Program for End-State Free Energy Calculations, J. Chem. Theory Comput., 8, 3314-3321.
40.
Irene, M., Contini, A., Explicit Ligand Hydration Shells Improve the Correlation between MM-PB/GBSA Binding Energies and Experimental Activities, J. Chem. Theory Comput., 2013, 9, 2706-2717
41.
ChemBioDraw.
(2014)
PerkinElmer
(CambridgeSoft
Corporation),
100
CambridgePark Drive, Cambridge, MA 02140. 42.
Ghose, A. K., Viswanadhan, V. N., and Wendoloski, J. J. (1998) Prediction of hydrophobic (lipophilic) properties of small organic molecules using fragmental methods: An analysis of ALOGP and CLOGP methods, J. Phys. Chem. A 102, 3762-3772.
43.
Heerklotz, H., and Seelig, J. (2000) Correlation of membrane/water partition coefficients of detergents with the critical micelle concentration, Biophys. J. 78, 2435-2440.
44.
Dixit, Sayali S., Sleat, David E., Stock, Ann M., and Lobel, P. (2007) Do mammalian NPC1 and NPC2 play a role in intestinal cholesterol absorption?, Biochem. J 408, 1-5.
45.
Ko, D. C., Binkley, J., Sidow, A., and Scott, M. P. (2003) The integrity of a cholesterol-binding pocket in Niemann-Pick C2 protein is necessary to control lysosome cholesterol levels, Proc. Natl. Acad. Sci. USA 100, 2518-2525.
46.
McCauliff, L. A., Xu, Z., Li, R., Kodukula, S., Ko, D. C., Scott, M. P., Kahn, P. C., and Storch, J. (2015) Multiple Surface Regions on the Niemann-Pick C2 Protein Facilitate Intracellular Cholesterol Transport, J. Biol. Chem. 290, 2732127331. 31 ACS Paragon Plus Environment
Biochemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
47
Page 32 of 47
Oninla, V.O., Breiden, B., Babalola, J.O., Sandhoff, K., (2014) Acid sphingomyelinase activity is regulated by membrane lipids and facilitates cholesterol transfer by NPC2, J Lipid Res, 55, 2606-2619.
48
Abdul-Hammed, M., Breiden, B., Adebayo, M.A., Babalola, J.O., Schwarzmann, G., Sandhoff, K. (2010) Role of endosomal membrane lipids and NPC2 in cholesterol transfer and membrane fusion, J Lipid Res, 51 1747-1760.
49.
John, K., Kubelt, J., Muller, P., Wustner, D., and Herrmann, A. (2002) Rapid transbilayer movement of the fluorescent sterol dehydroergosterol in lipid membranes, Biophys. J 83, 1525-1534.
50.
Wüstner, D., and Solanko, K. (2015) How cholesterol interacts with proteins and lipids during its intracellular transport, BBA - Biomembranes 1848, 1908-1926.
51.
Brown, M. S., and Goldstein, J. L. (2009) Cholesterol feedback: from Schoenheimer's bottle to Scap's MELADL, J Lipid Res 50 Suppl, S15-27.
52.
E. Boadu, R.C. Nelson, G.A. Francis, (2012) ABCA1-dependent mobilization of lysosomal cholesterol requires functional Niemann-Pick C2 but not Niemann-Pick C1 protein, Biochim Biophys Acta, 1821, 396-404.
53.
A. Frolov, S.E. Zielinski, J.R. Crowley, N. Dudley-Rucker, J.E. Schaffer, D.S. Ory, (2003) NPC1 and NPC2 regulate cellular cholesterol homeostasis through generation of low density lipoprotein cholesterol-derived oxysterols., J Biol Chem 278, 25517-25525.
54.
Iaea, D. B., Gale, S. E., Bielska, A. A., Krishnan, K., Fujiwara, H., Jiang, H., Maxfield, F. R., Schlesinger, P. H., Covey, D. F., Schaffer, J. E., and Ory, D. S. (2015) A novel intrinsically fluorescent probe for study of uptake and trafficking of 25-hydroxycholesterol, J. Lipid. Res. 56, 2408-2419.
55.
Kulig, W., Olżyńska, A., Jurkiewicz, P., Kantola, A. M., Komulainen, S., Manna, M., Pourmousa, M., Vazdar, M., Cwiklik, L., Rog, T., Khelashvili, G., Harries, D., Telkki, V.-V., Hof, M., Vattulainen, I., and Jungwirth, P. (2015) Cholesterol under oxidative stress—How lipid membranes sense oxidation as cholesterol is being replaced by oxysterols, Free Radic. Biol. Med 84, 30-41. 32 ACS Paragon Plus Environment
Page 33 of 47
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biochemistry
56.
Underwood, K. W., Andemariam, B., McWilliams, G. L., and Liscum, L. (1996) Quantitative analysis of hydrophobic amine inhibition of intracellular cholesterol transport, J. Lipid Res. 37, 1556-1568.
57.
Roff, C. F., Goldin, E., Comly, M. E., Cooney, A., Brown, A., Vanier, M. T., Miller, S. P., Brady, R. O., and Pentchev, P. G. (1991) Type C Niemann-Pick disease: use of hydrophobic amines to study defective cholesterol transport, Dev. Neurosci. 13, 315-319.
58.
Lu, F., Liang, Q., Abi-Mosleh, L., Das, A., De Brabander, J. K., Goldstein, J. L., and Brown, M. S. (2015) Identification of NPC1 as the target of U18666A, an inhibitor of lysosomal cholesterol export and Ebola infection, eLife 4, e12177.
33 ACS Paragon Plus Environment
Biochemistry
TABLE 1. Summary of (binding) energies obtained from final MM-PBSA model using various MD run
Energy (kcal/mol)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 34 of 47
Sterols
CHO
CHSO4
DHE
CTL
24OH
25OH
EST
U186
CHACE
Activity
90
100
90
90
50
50
10
20
0
EvdW
-49.67
-49.13
-53.62
-48.27
-51.21
-50.32
-32.10
-54.26
-50.46
EEle
-0.09
-1.18
-2.14
-0.51
-1.10
-1.68
-4.49
-14.92
0.55
∆Gsolv
32.83
34.43
43.16
35.74
44.66
44.38
34.33
63.16
-49.90
∆Ggas
-49.77
-50.32
-55.77
-48.78
-52.31
-52.00
-36.60
-69.18
37.81
-16.94
-15.88
-12.61
-13.04
-7.64
-7.61
-2.26
-6.01
-12.0*
±3.1
±3.0
±3.8
±3.4
±3.1
±3.4
± 3.0
±4.6
±3.1
∆G
Note: *constrained docking pose, All energy components are extracted from the differences (average) of ∆GComplex-∆GReceptor-∆GLigand. The results refer to averages over 2000 frames and all units are reported in kcal/mole. Abbreviation: EvdW= van der Waals energy, Eele=Electrostatic energy, ∆Ggas= Sum of Van der Waals energy + Electrostatic energy+ internal energy, ∆Gsolv= Solvation energy (polar and non-polar).
34 ACS Paragon Plus Environment
Page 35 of 47
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48
Biochemistry
TABLE 2: Comparison of experimental effects and the computational simulation results Effect of mutation in exp.$
Lys32
Effect of mutation in exp. binding* 0.76
Val64
Residues
21% of wt
Effect of mutation in our simulation (∆∆G) kcal/mol 0.28 -0.56
Energy contribution in our simulation (kcal/mol) -0.097 -0.167
Flexibility upon ligand binding in our simulation / wild type (Å) 37.43/ 48.43 22.51/ 16.13
Top surface Bottom surface
0.84**
93% of wt
0.20
-0.77
10.26/ 10.84
Mouth
Phe66 Asp72
Not detectable 0.61
5% of wt 12% of wt
1.11 0.21
-2.707 0.03
25.69/ 12.48 8.010 / 6.26
Mouth Top surface
Lys75
0.72
13% of wt + charge changed
0.59
-0.012
8.3883/ 9.03
Top surface
Val96 Tyr100
Not detectable
2% of wt
1.66
-0.583
06.53/ 9.030
Mouth
Not detectable
7% of wt
3.00
-1.903
23.59/ 14.10
Mouth
Pro24
Location
* Fraction of KD of WT (to Ala, except for V96, which is to F). Note that the standard deviation in such measurements was up to 40% of the mean.44, 45 ** Value taken from McCauliff et al. 45 $ Effect of mutation in exp. on cellular cholesterol efflux or on interbilayer sterol transfer (Ko, et al. 44)
35
ACS Paragon Plus Environment
Biochemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 36 of 47
FIGURE LEGENDS Figure 1.
2D representation of list of sterols used for NPC2-binding affinity predictions
Figure 2.
A) Comparison of binding pose (obtained from the docking) of estradiol (pink stick) and cholesterol sulphate (green stick); B) Open (pink) and closed (yellow) conformations of NPC2 are shown and important active site residues are highlighted; C) B-factor of protein backbone atoms as a function of residue number for all complexes, including ligand free (Apo) simulation. Inside cartoons represent Rfactor according to the X-ray crystal structure (blue: low flexibility, red: high flexibility).
Figure 3.
Relationship between the observed relative affinity (% of competition of DHE binding by a given sterol ligand is converted into logarithmic unit) and the MM-PBSA relative affinity (A) and AlogP property (B). Note, that CHACE is shown but not included in the correlation analysis and in calculation of the R2 value. The correlation analysis of calculated MM-PBSA binding affinity versus calculated AlogP is shown in panel (C), here, CHACE was included in the plot and in the calculation.
Figure 4.
Comparison of decomposition of binding energy for sterols (cholesterol and estradiol) from the MM-PBSA calculations, provided for the final model. Residues that are highlighted with “Red” and “Blue” denote the residues, which are located at the “surface” and “ligand binding-tunnel” of the protein, respectively. Some residues were removed from the plot A, due to insignificant energy (0.0 kcal/mol) contribution. B. Surface representation of binding mode of cholesterol (green stick) in the NPC2 is shown.
Figure 5.
RMSD of backbone heavy atoms relative to their initial structure (A) and B-factor of protein backbone atoms as a function of residue number (B) are compared for U186 in its protonated (red line) or deprotonated state (green line). Conformations from two different 36 ACS Paragon Plus Environment
Page 37 of 47
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biochemistry
simulations of U186 are shown in stick model (green: protonated, pink: not protonated) (C). The residues undergoing significant movements during the simulation are highlighted in blue in the molecular model and by arrows in panel (B). Figure 6.
The root-mean-square deviation (RMSD) of backbone heavy atoms relative to their initial structures. RMSDs were calculated for various NPC2-sterol complexes as indicated in the inset.
Figure 7.
Comparison of backbone heavy atoms RMSD of (relative to their initial structure) of EST and NPC2’s apo form is shown. Some of the representative snapshots that reveal the ligand’s outward movement are highlighted.
Figure 8.
Relative binding affinity of cholesterol in the various mutated NPC2 structures in relative to wild-type NPC2. The difference between the relative Gibb free energy change (∆∆G in kcal/mol) of wild-type and the mutated forms from separate simulations of the mutations Phe24, Lys32, Val64, Phe66, Asp72, Lys75, Leu94, Val96, Tyr100, Val105 to either alanine or phenylalanine are given. In addition, a triple mutant was generated, which contained three changed residues, i.e. F66A, V96F and Y100A (abbreviated ‘FVY’).
Figure 9.
Comparison of cholesterol (shown in green and violet space-filling model) binding poses in the wild type and mutated NPC2 (V96F) and important residues are highlighted
37 ACS Paragon Plus Environment
Biochemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Figure 1. 2D representation of list of sterols used for NPC2-binding affinity predictions 278x109mm (300 x 300 DPI)
ACS Paragon Plus Environment
Page 38 of 47
Page 39 of 47
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biochemistry
Figure 2. A) Comparison of binding pose (obtained from the docking) of estradiol (pink stick) and cholesterol sulphate (green stick); B) Open (pink) and closed (yellow) conformations of NPC2 are shown and important active site residues are highlighted; C) B-factor of protein backbone atoms as a function of residue number for all complexes, including ligand free (Apo) simulation. Inside cartoons represent R-factor according to the X-ray crystal structure (blue: low flexibility, red: high flexibility). 251x292mm (300 x 300 DPI)
ACS Paragon Plus Environment
Biochemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Figure 3. Relationship between the observed relative affinity (% of competition of DHE binding by a given sterol ligand is converted into logarithmic unit) and the MM-PBSA relative affinity (A) and AlogP property (B). Note, that CHACE is shown but not included in the correlation analysis and in calculation of the R2 value. The correlation analysis of calculated MM-PBSA binding affinity versus calculated AlogP is shown in panel (C), here, CHACE was included in the plot and in the calculation. 274x75mm (300 x 300 DPI)
ACS Paragon Plus Environment
Page 40 of 47
Page 41 of 47
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biochemistry
Figure 4. Comparison of decomposition of binding energy for sterols (cholesterol and estradiol) from the MM-PBSA calculations, provided for the final model. Residues that are highlighted with “Red” and “Blue” denote the residues, which are located at the “surface” and “ligand binding-tunnel” of the protein, respectively. Some residues were removed from the plot A, due to insignificant energy (0.0 kcal/mol) contribution. B. Surface representation of binding mode of cholesterol (green stick) in the NPC2 is shown. 260x181mm (300 x 300 DPI)
ACS Paragon Plus Environment
Biochemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Figure 5. RMSD of backbone heavy atoms relative to their initial structure (A) and B-factor of protein backbone atoms as a function of residue number (B) are compared for U186 in its protonated (red line) or deprotonated state (green line). Conformations from two different simulations of U186 are shown in stick model (green: protonated, pink: not protonated) (C). The residues undergoing significant movements during the simulation are highlighted in blue in the molecular model and by arrows in panel (B). 180x123mm (300 x 300 DPI)
ACS Paragon Plus Environment
Page 42 of 47
Page 43 of 47
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biochemistry
Figure 6. The root-mean-square deviation (RMSD) of backbone heavy atoms relative to their initial structures. RMSDs were calculated for various NPC2-sterol complexes as indicated in the inset. 280x147mm (300 x 300 DPI)
ACS Paragon Plus Environment
Biochemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Figure 7. Comparison of backbone heavy atoms RMSD of (relative to their initial structure) of EST and NPC2’s apo form is shown. Some of the representative snapshots that reveal the ligand’s outward movement are highlighted. 209x102mm (300 x 300 DPI)
ACS Paragon Plus Environment
Page 44 of 47
Page 45 of 47
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biochemistry
Figure 8. Relative binding affinity of cholesterol in the various mutated NPC2 structures in relative to wildtype NPC2. The difference between the relative Gibb free energy change (∆∆G in kcal/mol) of wild-type and the mutated forms from separate simulations of the mutations Phe24, Lys32, Val64, Phe66, Asp72, Lys75, Leu94, Val96, Tyr100, Val105 to either alanine or phenylalanine are given. In addition, a triple mutant was generated, which contained three changed residues, i.e. F66A, V96F and Y100A (abbreviated ‘FVY’). 89x72mm (300 x 300 DPI)
ACS Paragon Plus Environment
Biochemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Figure 9. Comparison of cholesterol (shown in green and violet space-filling model) binding poses in the wild type and mutated NPC2 (V96F) and important residues are highlighted 323x175mm (300 x 300 DPI)
ACS Paragon Plus Environment
Page 46 of 47
Page 47 of 47
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biochemistry
For Table of Contents Use Only 395x210mm (96 x 96 DPI)
ACS Paragon Plus Environment
