![]()
PLASTICITY OF BH3 DOMAIN-BINDING HYDROPHOBIC
GROOVES IN THE ANTI-APOPTOTIC MCL-1 AND Al
PROTEINS
VIVEK MODI, DILRAJ LAMA1 AND RAMASUBBU SANKARARAMAKRISHNAN2
Department of Biological Sciences & Bioengineering
Indian Institute of Technology Kanpur, Kanpur - 208016, India
1Current Address: Bioinformatics Institute
30 Biopolis Street, #07-01, Matrix, Singapore 138671
2E-mail: [email protected] Mcl-1 and Al constitute a subclass within the group of anti-apoptotic Bcl-2 proteins and are shown to be overexpressed in several human cancers. Diverse BH3 domains of different pro-apoptotic Bcl-2 proteins bind to the hydrophobic groove of both these helical bundle proteins with different affinities. Development of any inhibitors for Mcl-1 and Al requires the knowledge of the behavior of the binding region. In this study, we have carried out molecular dynamics simulations of apo-Mcl-1 and holo-Al systems for a period of 100 ns and compared the results with our earlier studies on Bcl-XL, another anti-apoptotic protein from a different subclass. During the course of the simulation, the ???-containing helix H2 is destabilized in Mcl-1, a behavior observed in Bc1-Xl also. The unwinding is attributed to the presence of glycine residues in the segment containing H2. Additionally, this region also contains eight basic residues including three doublets. We have hypothesized that the dibasic motifs could be the cleavage sites for the enzymes that will generate smaller isoforms of Mcl-1 and the unwinding of H2 can aid in the exposure of these cleavage sites. As far as Al is concerned, the ???-containing helix H2 is mostly stable except the last two helical turns. However, it adopts a completely different orientation within the first 10 ns. This change in orientation realigns the loop region that links H2 and the next helix so that most of the exposed hydrophobic residues can be shielded. The charged residues in both the proteins that are projected towards the hydrophobic groove in the experimentally determined structures are exposed to the solvent during the simulations. This is due to the destabilization of helical regions in which they are present. The unwinding of helix H2 in Mcl-1 and the change in orientation of the same helix in Al point to the flexible nature of hydrophobic groove and this could be important for the binding of diverse BH3 peptides.
1. Introduction
Mitochondrial outer member permeabilization is a key event in the intrinsic pathway of apoptosis 1,2. Interactions between anti- and pro-apoptotic class of Bcl-2 family of proteins is an important step in regulating this event which eventually determines the fate of the cell 3,4. The pro-survival proteins, Bcl-2, Bcl-XL, Bcl-w, Mcl-1 and AI, have four conserved sequence motifs (BH1 to BH4 domains) and the pro-apoptotic proteins contain either three BH domains (BH1 to BH3) as observed in Bak and Bax or just a single BH3 domain as found in Bcl-2 proteins like Bim, Puma, Noxa, Bmf and Bid 5. The diverse anti-apoptotic and multi-domain pro-apoptotic Bcl-2 proteins adopt a similar helical fold with a pronounced hydrophobic groove 6,7. BH3 motifs of pro-apoptotic proteins and the BH3 peptides derived from these motifs bind to the hydrophobic grooves of anti-apoptotic proteins 8–10 and exert biological function. Anti-apoptotic proteins such as Bcl-XL and Mcl-1 are recognized by the BH3 domains of several pro-apoptotic proteins. Similarly, the BH3 domain of pro-apoptotic proteins like Bim and Puma can bind to a range of anti-apoptotic proteins 11,13. Other pro-apoptotic proteins like Noxa and Bad bind selectively to specific anti-apoptotic partners 14. Even when a specific anti/pro-apoptotic protein has multiple partners, they exhibit varying affinities. Over-expression of pro-survival Bcl-2 members in several human cancers has been reported and hence, developing therapeutic agents that target specific anti-apoptotic Bcl-2 members is an active area of research in cancer biology 15–18. In this context, understanding the molecular determinants of pro- and anti-apoptotic protein-protein interactions is an important step and both experimental and computational approaches are being used for this purpose.
Structures of several anti-apoptotic proteins in complex with pro-apoptotic BH3 peptides have been determined 8–10, 19,20. Examination of structures does not clearly reveal the reasons for differential affinities of BH3 peptides. Molecular dynamics (MD) simulations on isolated BH3 peptides and Bcl-XL complex structures have been carried out to investigate the factors important for high affinity of pro-survival proteins for selected BH3 peptides 21–23. MD technique was also employed to look into the features of hydrophobic groove of Bcl-XL which is the binding site for several BH3 peptides 24.
Mcl-1 and Al form a subclass within the anti-apoptotic Bcl-2 group. The binding pattern of these proteins is similar and shows distinct differences with other anti-apoptotic proteins like Bcl-2, Bcl-XL and Bcl-w 25. For example, the pro-apoptotic Bmf and Bad peptides exhibit weak binding affinities to Mcl-1 and Al whereas Noxa BH3 peptide binds strongly to the same proteins. For Bcl-2, Bcl-XL and Bcl-w, the binding profiles for these peptides are exactly the opposite. The other main difference is the presence of charged residues in the hydrophobic grooves of Mcl-1 and Al. While Al has Glu side-chains pointing towards the groove 826, Mel-1's hydrophobic groove is influenced by basic residues 927. In comparison, Bcl-XL's binding groove is uncharged andessentially dominated by the hydrophobic residues 20. The nature of the hydrophobic groove of an anti-apoptotic protein should be such that it should be able to bind certain BH3 peptides with strong affinities and at the same time other BH3 peptides should be selectively excluded. The selective binding and differential affinities of BH3 peptides can be understood at molecular level only if we have a full understanding of the nature of the hydrophobic groove to which the BH3 peptides bind. Earlier simulation studies from our laboratory on apo-and holo-forms of Bcl-XL helped to understand the hydrophobic groove of this protein 24. They revealed that the BH3-containing helix in Bcl-XL which is part of the groove is destabilized and indicated plasticity of the binding groove that can help to bind diverse BH3 peptides. The unwinding of this helix also explained its possible importance in homo- or hetero-dimerization of Bcl-2 proteins. Interactions among the hydrophobic residues in the groove and ability of the loop connecting the BH3-containing helix and the subsequent helix to shield the bulky hydrophobic residues are the major factors that explain how a large chunk of hydrophobic patch can exist in the solvent-exposed state. In this paper, we have performed MD simulations of free forms of two anti-apoptotic proteins Mcl-l and Al for a period of 100 ns. Two important questions are being addressed in this study. Does the BH3-containing helix in Mcl-l and Al undergo destabilization similar to that observed in Bcl-XL? How does the presence of charged residues in the hydrophobic grooves of these two proteins affect the behavior of the BH3-binding region?
2. Molecular Dynamics Simulations of Mcl-l and Al
The starting structures of Mcl-l and Al are those from mouse and the experimentally determined structures were downloaded from the Protein Data Bank (PDB) 28. The corresponding PDB IDs are 1WSX 27 and 2VOH 8 respectively for Mcl-l and Al. While Mcl-l is an apo structure determined using NMR, the Al is an X-ray structure in complex with Bak BH3 peptide (Resolution: 1.9 Å). Hence, we removed the Bak peptide from the structure and the holo-Al was used in the simulation. The extra five N-terminal residues as a result of cloning were removed from both Mcl-l and Al. The side-chain of Glu 141 in Al is not resolved in the X-ray structure. It was built using the Biopolymer module available in Insight ? software (Acclerys, San Diego, CA). The apo-Mcl-1 and holo-Al structures were simulated using the protocol developed in our laboratory to simulate Bcl-XL protein 21, 24. Briefly, MD simulations were carried out using GROMACS (ver. 4.0) suite of software 29,30 with the force field GROMOS96 43al 3132. The structures were initially placed in the center of a cubic box and the box size was determined with the criteria that the minimum distance between the protein and the box edge should be at least 15 Á. SPC water model 33 was used and all ionizable residues were assumed to have default protonation states. The solvated structures were first minimized followed by 2 ns of equilibration. The first 1 ns equilibration was carried out using NVT (constant number of atoms, volume and temperature) ensemble. After that, NPT (constant number of atoms, pressure and temperature) ensemble was used. A twin-range cut-off of 10 and 18 Å was used to evaluate the non-bonded interactions. The systems were simulated at 300 ? with 1 bar pressure. Berendsen's algorithm 34 was used to maintain the reference temperature and pressure of the systems. All other simulation details are same as those reported in the earlier simulation studies of Bcl-XL 21, 24 . Both the systems were simulated for a period of 100 ns after equilibration. Details of the simulations are summari...
