Subscriber access provided by AUBURN UNIV AUBURN
Article
Structural Flexibility and Conformation Features of Cyclic Dinucleotides in Aqueous Solutions Xing Che, Jun Zhang, Yanyu Zhu, Lijiang Yang, Hui Quan, and Yi Qin Gao J. Phys. Chem. B, Just Accepted Manuscript • DOI: 10.1021/acs.jpcb.5b11531 • Publication Date (Web): 15 Feb 2016 Downloaded from http://pubs.acs.org on February 23, 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.
The Journal of Physical Chemistry B 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 39
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
The Journal of Physical Chemistry
Structural Flexibility and Conformation Features of Cyclic Dinucleotides in Aqueous Solutions Xing Che1,2, Jun Zhang1,2, Yanyu Zhu1, Lijiang Yang1, Quan Hui1, Yi Qin Gao1* 1. Institute of Theoretical and Computational Chemistry, College of Chemistry and Molecular Engineering, and Biodynamic Optical Imaging Center, Peking University, Beijing 100871, China 2. These authors contribute equally to this work. Correspondence Author:
[email protected] ABSTRACT Cyclic dinucleotides are able to trigger the innate immune system by activating STING. It was found that the binding affinity of asymmetric 2’3’-cGAMP to symmetric dimer of STING is three orders of magnitude higher than the symmetric 3’3’-cyclic dinucleotides. Such a phenomenon has not been understood yet. Here we show that the subtle changes in phosphodiester linkage of CDNs lead to their distinct structural properties which correspond to the varied binding affinities. 2’-5’ and/or 3’-5’ linked CDNs adopt specific while different types of ribose puckers and backbone conformations. That ribose conformations and base types have different propensities for anti or syn glycosidic conformations further affects the overall flexibility of CDNs. The counterbalance between backbone ring tension and electrostatic repulsion, both affected by the ring size, also contributes to the different flexibility of CDNs. Our calculations reveal that the free energy cost for 2’3’-cGAMP to adopt the STING-bound structure is smaller than that for 3’3’-cGAMP and cyclic-di-GMP. These findings may serve as a reference for design of CDN-analogues as vaccine adjuvants. Moreover, the cyclization pattern of CDNs closely related to their physiological roles suggests the importance of understanding structural properties in the study of protein-ligand interactions. Keywords cyclic dinucleotides, selective integrated tempering sampling, ligand solution conformation
1 / 27
ACS Paragon Plus Environment
The Journal of Physical Chemistry
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 39
1 INTRODUCTION Cyclic dinucleotides (CDNs), containing two purine nucleosides joined by two phosphate groups in a macro ring, serve as second messengers in both bacteria and metazoa. Cyclic di-GMP, which was found almost three decades ago, is ubiquitously involved in biological processes including differentiation, motility, virulence and cell cycle1. In metazoa, an endogenous second messenger cyclic GMP-AMP was recently discovered in cytosolic DNA induced immune pathway2. Unlike the uniform 3’-5’ phosphodiester linkages in c-di-GMP, this endogenous cyclic GMP-AMP (cGAMP) contains mixed phosphodiester linkages: noncanonical 2’-5’ phosphodiester bond at the GpA step and canonical 3’-5’ linkage at the ApG step
3, 4, 5
.
This molecule, denoted as 2’3’-cGAMP, triggers the innate immune system by binding to and activating Stimulator of Interferon Genes (STING), a vital adaptor protein which links the upstream cytosolic DNA detection and the downstream cytokine production6. As for other cGAMPs isomers, 3’3’-cGAMP was found in bacteria7, 8, whereas 2’2’-cGAMP and 3’2’-cGAMP have not been found in bacteria or metazoa. The binding affinity of 2’3’-cGAMP and 2’2’-cGAMP to STING is in the nM range, whereas that for 3’3’-cGAMP and c-di-GMP in the µM range
9, 10
. Such an observation raises a question: how do the subtle
differences in phosphodiester linkages among CDNs lead to binding affinities different in orders of magnitude? The crystallographic studies of the STING/CDN complexes provided some but limited insight on this issue9, 10. A recent study revealed that the difference in the free ligand structure of CDN isoforms is critical for understanding their varied binding affinities to STING and proposed that the analyses of free-ligand conformations should not be neglected in understanding protein-ligand interactions11. As has been realized, a thorough study of the ligand conformation ensemble is vital to investigate and predict ligand-substrate interactions. A ligand can adopt quite different conformations when it binds different substrates 12. For example, an analysis of endogenous cofactors ATP, NAD and FAD interacting with domains from respective 45, 33 and 38 homologous superfamilies demonstrated that their binding modes can be quite different13. Drugs such as methotrexate and Gleevec also interact with proteins through multiple binding modes, adopting either extended or compact conformations14. The stable structures alone, such as those obtained through structure optimization, may provide incomplete information on understanding ligand-substrate interaction. In fact, a study on 150 pharmaceutically related protein-ligand complexes found that over 60% of the ligands do not bind through a local minimum energy conformation15. Therefore, detailed investigation on the thermodynamics information of ligands in 2 / 27
ACS Paragon Plus Environment
Page 3 of 39
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
The Journal of Physical Chemistry
solution is also crucial to understand molecular interactions and binding free energy. Upon binding, a ligand also experiences entropy changes due to the confined pocket space and reduction in solvation degrees of freedom. Including only minimum energy structures in the binding free energy calculation would underestimate entropic forces12. A thorough configuration sampling could thus be vital to study the thermodynamic cycle of the entire binding process. In investigating detailed molecular features and thermodynamics of cyclic dinucleotides, molecular dynamics (MD) simulations can serve as a cost-effective tool. A number of MD simulations have been performed to study structures of various ribo- and deoxyribo-nucleosides as well as DNA, RNA and DNA/RNA hybrid duplexes, and reproduced a variety of experimental observations16. However, many observations remain elusive to be understood. One of the challenges is to achieve a complete and efficient conformation sampling of molecules with a large number of degrees of freedom, which is typically limited by the efficiency of unbiased MD simulations17. Ultrasonic relaxation measurements suggested that syn-anti glycosidic isomerization in purine nucleosides and nucleotides lies on a nanosecond time scale 18, 19
. Moreover, MD simulations implied that ribose conformation conversion between C3’-endo and
C2’-endo usually demands nano- to microseconds
20
. These slow structural transformations make a
thorough sampling of molecular configuration space expensive or even impractical for MD simulations. Therefore, enhanced sampling protocols are often used to facilitate conformation sampling. For example, Li and Szostak used umbrella sampling and metadynamics simulations to investigate the free energy landscape of pseudorotation in nucleic acids, using the phase angle (P) and pucker amplitude (τm) as collective variables20. Banavali and Roux studied A-DNA to B-DNA conversion using the root-mean-square-displacement as the collective variable21. In the current study, given that CDN structure changes involve multiple slow motions, and in case that the conformations might be inadequately sampled using enhanced sampling schemes depending on preselected low dimensional collective coordinates, we used a global enhanced method-Selective Integrated Tempering Sampling (SITS)22,
23, 24
to obtain a
thorough sampling of CDNs solution conformations. The SITS simulation results elucidate diverse structural preferences among CDNs in the presence of different phosphodiester linkage forms as well as different combinations of the linkages in cyclization. The resulted varying flexibilities among different CDNs account for their different free energies when changing from non-bound (inactive) states to bound-like (active) states. The detailed structural and thermodynamic information is then used to rationalize the different protein binding affinities of the CDN isoforms. 3 / 27
ACS Paragon Plus Environment
The Journal of Physical Chemistry
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 39
2 METHODS AND COMPUTATIONAL DETAILS 2.1 Ligand Preparation Ligand structures were extracted from PDB files (4LOI: 2’2’-cGAMP, 4LOH: 2’3’-cGAMP, 4F5D c-di-GMP). Since 3’3’-cGAMP in 4LOK complex features a α-glycosidic bond, but not the naturally occurring β-N9-glycosidic bond, we used a structure from a 3’3’-cGAMP-hSTINGS162A complex (unpublished). Geometry optimization was performed using Q-Chem 4.025 with 6-311G* basis set and the B3LYP functional. The force field of cyclic dinucleotide molecules were built based on the general AMBER force field (GAFF) with RESP charges using ANTECHAMBER26, 27. The force filed parameter files are listed in SI section I. We benchmarked the force filed parameters through detailed comparison between calculations and experiment data (X-ray and NMR) (section IV, SI). The second messengers were all solvated in cubic TIP3P water boxes. We also simulated another system containing 3’3’-cGAMP in the 0.5 mol/L MgCl2 solution (see below). 2.2 Simulation Details Each system was first subjected to an energy minimization (1000 steps of steepest descent minimization and 1500 steps of conjugate gradient minimization), a 500ps NVT heating-equilibration (from 300K to 330K), and then a 500ps NPT cooling from 330K to 300K. The SHAKE algorithm was used to constrain all bonds containing hydrogen. A cutoff of 10.0 Å was used for nonbonding interactions. The particle mesh Ewald method was employed for long-range electrostatic interaction calculations. The temperature was maintained by Langevin dynamics. The simulation step was 2 fs. A SITS implemented SANDER program in AMBER9 package was used for enhanced sampling MD simulations. Two regions, the solute region (CDN molecules) and the bath, were defined in SITS MD simulations separately. The bath includes sodium ions and water molecules. The simulation temperature was maintained at 300 K. Boltzmann distributions under 60 temperature values, Tk (k=1 to 60) evenly distributed between 220 K and 720 K, were summated to generate the effective potential ( , eq.1) for the enhanced sampling of CDNs: !
1 1 = + −
2
(1)
"
where Es, Eb, and Eint are the potential energies of the solute, the bath and the interactions between the solute and the bath, respectively. βk equals 1/(kBTk), where kB is the Boltzmann constant. T0 is the target 4 / 27
ACS Paragon Plus Environment
Page 5 of 39
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
The Journal of Physical Chemistry
temperature (300 K). The initial set of nk values was randomly selected. Then nk was iterated to achieve an efficient sampling over the desired large energy range28. With the determined values of nk, the production run lasted 300ns for each system under the NPT ensemble (1 atm, 300 K) and all the data were collected every 1.0 ps. The desired distribution function at the temperature T0 was calculated according to eq.2: #$%& = #'(( $%& )$*$+&* $+&& ,
(2)
with #'(( $%& calculated from the enhanced sampling simulation and $%& defined as in eq.1. Other
thermodynamics properties can be readily obtained once #$%& is known. For example, the probability for an event occurring at reaction coordinate r, P(r), was calculated by eq.3 : .$%& = / #'(( $%′& )$*$+
, &′ $+ , &&
δ$% − % &d%
(3)
The Gibbs free energy change ∆4$% → % &, along the reaction coordinate r (from r0 to r’), was then obtained though eq.4 : ∆4$% → % & = −67 8ln
.$% & .$% &
(4)
2.3 Structural Analysis Atoms and dihedral angles of CDNs are named following convention29. The conformation of the backbone of one nucleotide in cyclic dinucleotides, as illustrated in Figure 1, is described by torsion angles α, β, γ, δ (and δ’), ε, and ζ. Conformations with angle values around -60°, -120°, 60°, 120°, 180° are denoted as gauche- (g-, -30°~-90°), -anticlinal (-ac, -90°~-150°), gauche+ (g+, 30°~90°), +anticlinal (+ac, 90°~150°), and trans (t, 150°~180° and -150°~180°) respectively. The lower energy BI form29 is characterized with -160°
