This article was downloaded by: [134.117.10.200] On: 29 November 2014, At: 02:23 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Journal of Biomolecular Structure and Dynamics Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/tbsd20
Structural insights into the interactions of xpt riboswitch with novel guanine analogues: a molecular dynamics simulation study a
b
b
Swapan S. Jain , Uddhavesh B. Sonavane , Mallikarjunachari V.N. Uppuladinne , Emily C. a
a
a
b
McLaughlin , Weiqing Wang , Sheneil Black & Rajendra R. Joshi a
Department of Chemistry, Bard College, 30 Campus Rd, Annandale-on-Hudson, NY 12504, USA b
Bioinformatics Group, Centre for Development of Advanced Computing, Pune University Campus, Pune, Maharashtra 411007, India Published online: 03 Jan 2014.
To cite this article: Swapan S. Jain, Uddhavesh B. Sonavane, Mallikarjunachari V.N. Uppuladinne, Emily C. McLaughlin, Weiqing Wang, Sheneil Black & Rajendra R. Joshi (2015) Structural insights into the interactions of xpt riboswitch with novel guanine analogues: a molecular dynamics simulation study, Journal of Biomolecular Structure and Dynamics, 33:2, 234-243, DOI: 10.1080/07391102.2013.870930 To link to this article: http://dx.doi.org/10.1080/07391102.2013.870930
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions
Journal of Biomolecular Structure and Dynamics, 2015 Vol. 33, No. 2, 234–243, http://dx.doi.org/10.1080/07391102.2013.870930
Structural insights into the interactions of xpt riboswitch with novel guanine analogues: a molecular dynamics simulation study Swapan S. Jaina*, Uddhavesh B. Sonavaneb*, Mallikarjunachari V.N. Uppuladinneb, Emily C. McLaughlina, Weiqing Wanga, Sheneil Blacka and Rajendra R. Joshib a
Department of Chemistry, Bard College, 30 Campus Rd, Annandale-on-Hudson, NY 12504, USA; bBioinformatics Group, Centre for Development of Advanced Computing, Pune University Campus, Pune, Maharashtra 411007, India Communicated by Ramaswamy H. Sarma
Downloaded by [134.117.10.200] at 02:23 29 November 2014
(Received 17 July 2013; accepted 27 November 2013) Ligand recognition in purine riboswitches is a complex process requiring different levels of conformational changes. Recent efforts in the area of purine riboswitch research have focused on ligand analogue binding studies. In the case of the guanine xanthine phosphoribosyl transferase (xpt) riboswitch, synthetic analogues that resemble guanine have the potential to tightly bind and subsequently influence the genetic expression of xpt mRNA in prokaryotes. We have carried out 25 ns Molecular Dynamics (MD) simulation studies of the aptamer domain of the xpt G-riboswitch in four different states: guanine riboswitch in free form, riboswitch bound with its cognate ligand guanine, and with two guanine analogues SJ1 and SJ2. Our work reveals novel interactions of SJ1 and SJ2 ligands with the binding core residues of the riboswitch. The ligands proposed in this work bind to the riboswitch with greater overall stability and lower root mean square deviations and fluctuations compared to guanine ligand. Reporter gene assay data demonstrate that the ligand analogues, upon binding to the RNA, lower the genetic expression of the guanine riboswitch. Our work has important implications for future ligand design and binding studies in the exciting field of riboswitches. Keywords: riboswitch; guanine; ligand analogues; MD simulations
Introduction A vast majority of drugs target proteins in bacteria (Blount & Breaker, 2006). The widespread use of protein-targeting antibiotics has resulted in the rise of bacteria with antibiotic resistance, a major problem for public health that requires a fresh approach to combating disease. RNA, due to its rich structural diversity, presents a target much like proteins (Zaman, Michiels, & van Boeckel, 2003). Riboswitches are highly structured segments within the untranslated regions of specific mRNA molecules (Serganov & Patel, 2007). They have been shown to act as powerful gene regulatory elements in prokaryotes and certain eukaryotes as well (Mandal, Boese, Barrick, Winkler, & Breaker, 2003; Montange & Batey, 2008; Sudarsan, Barrick, & Breaker, 2003; Winkler & Breaker, 2005). Small molecules binding to the aptamer (sensor) domain of riboswitches cause conformational changes in the adjacent expression (actuator) domain of mRNA, which subsequently leads to either premature termination of transcription or inhibition of translation (Fuchs, Grundy, & Henkin, 2006; Mandal & Breaker, 2004). Numerous riboswitches that have been discovered in the last decade are capable of selectively binding coenzymes (TPP: thiamine pyrophosphate, FMN: flavin mononucleotide, SAM: S-adenosylmethionine, and cobalamine), amino acids
(glycine, lysine, and glutamine), sugars (glucosamine-6phosphate), nucleotide derivatives (guanine, adenine, cyclic di-GMP, and ATP) and even small ions (magnesium and fluoride) (Breaker, 2011; Ren, Rajashankar, & Patel, 2012; Watson & Fedor, 2012). Purine-sensing mRNA molecules were one of the first riboswitches to be discovered in bacteria (Winkler, 2005). Xanthine phosphoribosyl transferase (xpt) is a key enzyme in the purine metabolic pathway of bacteria catalyzing the synthesis of xanthine monophosphate from xanthine. It has been shown that binding of guanine to the xpt mRNA leads to the formation of a distinctly folded RNA segment, which leads to transcription termination (Batey, Gilbert, & Montange, 2004; Serganov et al., 2004). Guanine has numerous H-bond donor and acceptor groups, which facilitate site-specific binding to nucleotide bases in the aptamer domain of the riboswitch (Figure 1(a)). Guanine forms eight non-covalent interactions at a distance of < 3 Å to the aptamer domain of the riboswitch resulting in a tightly bound complex with a dissociation constant of ~5 nM (Mandal & Breaker, 2004). Figure 1(b) shows that guanine base forms a Watson Crick pair with C74 nucleotide of the riboswitch aptamer domain. Three more hydrogen bonds are observed with N3, N9, and the exocyclic –NH2 group of
*Corresponding authors. Emails: [email protected] (S.S. Jain); [email protected] (U.B. Sonavane) © 2013 Taylor & Francis
Downloaded by [134.117.10.200] at 02:23 29 November 2014
Interactions of xpt Riboswitch with Novel Guanine Analogues
Figure 1. (a) Secondary structure of the guanine riboswitch aptamer domain (b) Space filling model of guanine bound to the residues of the riboswitch. Dashed lines indicate hydrogen bonding and other non-covalent interactions. White ovals show empty space that can be exploited by rational ligand design.
U51 nucleotide. An additional hydrogen-bonding interaction occurs with U47 nucleotide. U22 of the riboswitch is involved in an interaction where the 2′–OH group of its sugar makes a hydrogen bonding contact with the N7 atom of the guanine ligand. Crystallography data by Batey and coworkers have shown that hypoxanthine, which is a precursor of guanine, is completely surrounded by the xpt riboswitch. Furthermore, ~97% of the surface area of hypoxanthine is inaccessible to the bulk solvent (Batey et al., 2004). Longrange tertiary interactions between aptamer domain loops L2 and L3 (Figure 1(a)) are essential for guanine binding to the riboswitch. These interactions are formed even in the ligand-free state of the riboswitch and are further stabilized by the stacking of two base quadruples (U34A65— G37-C61 and A33A66–G38-C60). Even though the residues in the loop region do not contact the ligand directly, the interactions are nonetheless essential for the overall stability and architecture of the riboswitch. Ligand recognition and binding occurs by a hierarchically induced fit process where binding of guanine leads to structural rearrangements in other parts of the RNA molecule.
235
Recent efforts in the area of purine riboswitch research have focused on ligand analogue studies. Synthetic molecules that mimic purines may have the potential of tightly binding and regulating riboswitch mRNA in prokaryotes (Mulhbacher & Lafontaine, 2007). Work by Lafontaine group shows that a purine ring system is not necessary for effective binding of synthetic compounds to the xpt riboswitch (Mulhbacher et al., 2010). Pyrimidine-based analogs that present H-bonding groups at similar position as guanine also bind with high selectivity to the aptamer domain of the RNA. In 2009, Breaker and coworkers conducted a ligand analogue study where guanine was modified either at the C2 or the C6 position using a variety of functional groups (Kim et al., 2009). White ovals in Figure 1(b) show empty spaces that can be exploited by modifying guanine at the C2 or C6 position. In-line probing assays showed that some of the compounds bound to the riboswitch even more tightly than guanine. In particular, modification of guanine at the C2 position with an acetyl group resulted in N-acetyl guanine that bound to the riboswitch RNA with a 10-fold higher binding affinity (Kd = 0.5 nM) compared to guanine. Out of all the C6-modified analogues tested by Breaker and coworkers, only the 6-hydrazine guanine demonstrated a similar binding affinity to guanine. We have designed our compounds, SJ1 and SJ2, by taking cues from previous work done by the Breaker group. Guanine analogs simultaneously modified at both the C2 and C6 position of the purine ring (right hand column in Figure 2) may serve as effective analogs in regulating the xpt riboswitch in bacteria. Synthesis and biochemical studies illustrating the interactions of these compounds to riboswitch RNA will be presented in a forthcoming paper. In the current work, 25 ns molecular dynamics simulations in explicit solvent were carried out after guanine, SJ1 and SJ2 were docked to the aptamer domain of the guanine riboswitch. Brenk and coworkers have shown that an RNA–ligand docking approach, much like a protein–ligand docking approach, can be effectively used to discover novel compounds displaying high affinity to the guanine-sensing riboswitch (Daldrop et al., 2011). Stock and coworkers conducted molecular dynamics simulations of the guanine riboswitch with guanine, adenine and the ligand-free state (Villa, Wohnert, & Stock, 2009). They demonstrated that there is a stepwise process where ligand binding to a preformed pocket is followed by structural rearrangement and conformational changes leading to a globally stable RNA ligand complex. A recent molecular dynamics study of the adenine riboswitch (structurally identical to the guanine riboswitch) reveals the importance of the L2/L3 loop–loop kissing interactions in stabilizing the overall tertiary structure of the aptamer domain (Allner, Nilsson, & Villa, 2013).
236
S.S. Jain et al. H2N NH
NH N N H
O N
6
N N
N
NH2
6-hydrazine guanine Kd ~ 8 nM
N H
N
O N H
N
SJ1
NH 2
N H
N
H2N
NH2
NH
O
Guanine Kd ~ 5 nM
N
Downloaded by [134.117.10.200] at 02:23 29 November 2014
N H
N H N
N H
N-acetyl guanine Kd ~ 0.5 nM
Figure 2.
O
N N H
N N
N
NH2
N H
SJ2
Chemical structure of guanine, 6-hydrazine guanine, N-acetyl guanine and C2/C6 modified SJ1 and SJ2.
We have used molecular dynamics simulations to investigate the interactions of novel ligand analogues with the guanine riboswitch mRNA. In this work, we have shown that in addition to contacting core nucleotides bound by guanine ligand, analogues SJ1 and SJ2 bind other nucleotide bases and do so with a greater number of overall hydrogen bonds. These ligands, upon binding to the aptamer domain of the riboswitch, yield a lower overall free energy and root mean square deviation (RMSD) of the complex than that of the guanine-bound riboswitch. Root mean square fluctuation (RMSF) analysis shows that these ligands restrict the conformational flexibility of the aptamer domain, which potentially has important implications on overall riboswitch function. To date, several experimental and theoretical studies have investigated the effect of ligand analogue binding to different types of riboswitches and nucleic acid aptamers (Banas, Walter, Sponer, & Otyepka, 2010; Kumar, Endoh, Murakami, & Sugimoto, 2012; Schill & Koslowski, 2013; Xin & Hamelberg, 2010). To the best of our knowledge, this work represents the first report where ligand analogues bound to a guanine riboswitch molecule have been studied by molecular dynamics simulations. Materials and methods Initial structures The crystal structure (PDB ID 1Y27) of guanine-bound G-riboswitch (guanine riboswitch) was considered as the starting structure for control system (Batey et al., 2004).
We removed the guanine molecule from the control structure and designated it as the ligand-free G-riboswitch system. The G-riboswitch with SJ1 and SJ2 as the bound ligands in place of guanine were modelled by fitting SJ1 and SJ2 to the ligand position in the crystallographic structure. All these four complex systems were considered as the starting structures for the simulations. Force-field parameters Parameters for all the three ligands guanine, SJ1, and SJ2 were derived from AMBER force field (ff03) (Duan et al., 2003). Partial atomic charges were derived by performing geometry optimization at Hartree–Fock level using 6–31G* and single-point energy calculations at DFT level using 6–311G** basis sets (Ditchfield, Hehre, & Pople, 1971; Hehre, Ditchfield, & Pople, 1972). The electronic structure calculations were carried out with Gaussian 03 package (Case et al., 2008; Frisch et al., 2004). Molecular dynamics simulations The MD simulations were carried out using AMBER10 program with ff03 force field (Case et al., 2008; Duan et al., 2003). All the four systems were neutralized by adding counter ions and solvated with TIP3P water box (Jorgensen, Chandrasekhar, Madura, Impey, & Klein, 1983). Energy minimization was carried out for all the systems by using steepest descent method for 5000 steps. Then simulated annealing was carried out at temperature
Downloaded by [134.117.10.200] at 02:23 29 November 2014
Interactions of xpt Riboswitch with Novel Guanine Analogues range of 50 K to 300 K for 100 ps, by using constraints of 100 kcal/mol. Equilibration was achieved by gradually reducing the constraints for 500 ps. The equilibration protocol of Cheatham was followed similar to previous simulations of nucleic acids (Cheatham & Kollman, 1997). Simulations were performed under periodic boundary conditions by employing the particle-mesh Ewald technique to account for long-range electrostatics (York, Darden, & Pedersen, 1993). MD integration was carried out using a 2.0 fs time step, employing the SHAKE algorithm on all the hydrogen atoms and non-bonded cut-off of 10.0 Å (Ryckaert, Ciccotti, & Berendsen, 1977). The pair list was updated every 100 steps. Constant pressure (1 atm) and temperature (300 K) were maintained using the Berendsen coupling algorithm throughout the production simulation run (Berendsen, Postma, van Gunsteren, Dinola, & Haak, 1984). The systems were allowed to equilibrate under production run conditions for 1 ns before collection of data over 25 ns simulation time. Four simulations were carried out for 25 ns each and the trajectories were used for the analysis. All-atom classical MD simulations of four systems, namely Free-Ribo, Ribo-Gua, Ribo-SJ1 and Ribo-SJ2, were carried out on the Bioinformatics Resources and Application Facility (BRAF) of C-DAC, Pune.
237
expression was measured using the standard Miller Unit assay (Miller, 1992). Samples were run in quadruplicates. Results and discussion The present simulation study of the four complexes reveals a good overall representation of the important interactions between ligand and G-riboswitch along with the effects of the ligand in stabilizing the overall complex. Four simulations were carried out in order to understand the effect of different ligands and compared with control system (Ribo-Gua) along with ligand-free G-riboswitch system (Free-Ribo). The initial simulation structure of the riboswitch–guanine complex (Ribo-Gua) is shown in Figure 3. The starting structures for the Free-Ribo, Ribo-SJ1, and Ribo-SJ2 are presented as supplementary information (Figure S1). During the simulation run, the plots for temperature, pressure and total energy maintain stable average values with allowed fluctuations indicating that all the four systems were in
Data analysis All statistics were collected on 1-picosecond snapshots over 25 ns of simulation data. Trajectories and structures were visualized using VMD (Humphrey, Dalke, & Schulten, 1996). PTRAJ module of AmberTools was used for RMSD, RMSF, radius of gyration, and hydrogen bond analyses. The criterion for hydrogen bonding was set at ≤3.5 Å distance between electron donor atom and hydrogen of electron acceptor atom with 120-degree angle cut-off. Strong hydrogen bonds were categorized as those having a distance of ≤ 2.5 Å whereas weak bonds were categorized as those with distance of 2.5–3.5 Å. Molecular Mechanics Generalized Born Surface Area (MMGBSA) module of AMBER10 was used for calculating free energy. XMGRACE software was used for generating all the plots. Analysis of riboswitch function Bacillus subtilis cells with guanine riboswitch gene fused upstream of promotorless lacZ reporter gene were used to quantify riboswitch mRNA expression in vivo. B. subtilis cells were provided by Dr. Ronald Breaker (Yale University). Ligand stock solutions were prepared in DMSO solvent. Cells were grown overnight in minimal medium with constant shaking at 37 °C. Cells at an OD600 of ~0.1 were incubated with 200 µM of the designated ligands for 8 h at 37 °C. Beta galactosidase
Figure 3. Starting structure of Ribo–Gua complex. Riboswitch RNA molecule is shown as coloured sticks and the guanine ligand is shown as a sphere rendering.
S.S. Jain et al.
equilibrium (Supplementary Figure S2). These plots suggest that the thermodynamic ensemble is maintained and energies are not showing any abnormality in the system. All the four trajectories are analysed for RMSD, RMSF, radius of gyration, hydrogen bonding, and free energy. The radius of gyration data is presented in the supplementary information section (Figure S3). The backbone RMSD of the free riboswitch was maintained at 3 Å with fluctuations for all four complexes (Figure 4(a)). Though all four complexes demonstrated similar RMSD values, there is a slight difference in the average RMSDs. The average RMSD with respect to the X-ray structure was only 2.92 ± 0.44 Å for the control system (Ribo–Gua). The Free-Ribo system yielded an average RMSD value of 2.97 ± 0.57 Å, which was the highest among all four complexes. The average RMSD value of Ribo-SJ1 and Ribo-SJ2 was 2.73 ± 0.41 Å and 2.65 ± 0.35 Å, respectively. Since all four complexes displayed average RMSD values that were similar to each other, the residue-specific RMSF calculations provide more detailed information regarding local deviations that occur upon ligand binding to the riboswitch. The RMSF parameter measures the average atomic mobility of backbone atoms during the molecular dynamic simulations of a system. The residue-wise RMSF of the four complexes was calculated and plotted in Figure 4(b). Our results show that RMS fluctuations were lower for Ribo-SJ1 and Ribo-SJ2 complex across the entire aptamer domain when compared to the
RMSD (Å)
(a)
5 4 3 2 1
Free-Ribo
Ribo-Gua
Ribo-SJ1
Ribo-SJ2
0 0
5
10
15
20
25
time (ns)
(b)
25 Ribo-Gua
Free-Ribo
Ribo-SJ2
Ribo-SJ1
20
RMSF (Å)
Downloaded by [134.117.10.200] at 02:23 29 November 2014
238
15 10 5 0 10
20 _____
30 ____
P1
P2 J1-2
____ 40 ____ L2
50
P2
___60___ P3
J2-3
L
70 ___
80 _____
P3
P1 J3-1
Residue #
Figure 4. (a) Backbone RMSD of the riboswitch RNA complexes. (b) Residue-wise RMSF of the four complexes.
Ribo-Gua complex. In particular, U47 residue that is present in the J2/3 junction and contacts the ligand shows significantly lower fluctuations than the Ribo-Gua complex. Fluctuations at residue C74, which is necessary for the specific recognition of the guanine ligand, are very similar for all of the simulated complexes. High-resolution studies have indicated that loops L2 and L3 form interconnecting hydrogen bonds (kissing-loop interaction) and these interactions are important for the overall folded architecture of the riboswitch (Batey et al., 2004; Serganov et al., 2004). Residues 32–38 present in L2 region directly interact with residues 60–66 present in L3 region. RMSF peaks in the loop regions show lower fluctuations for Ribo-SJ1 and Ribo-SJ2 compared to Ribo-Gua indicating that these novel ligands bind to the riboswitch without major structural perturbations whilst retaining the overall folded architecture of the complex. We have also looked at region-wise RMSD values for the ligand complexes and observe that there are regions such as the L3 loop where the Ribo-Gua system is showing more than 1Å high RMSD compared to the other three systems. In an effort to probe the intramolecular (within the riboswitch) and intermolecular (between the riboswitch and the ligand) interactions, hydrogen bond calculations were carried out after the simulations were completed. The stability of a complex was evaluated by calculating hydrogen bonding percentage (HB%) which was defined as the time over which a hydrogen bond satisfied the two standard criteria (d(H…O)
Journal of Biomolecular Structure and Dynamics Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/tbsd20
Structural insights into the interactions of xpt riboswitch with novel guanine analogues: a molecular dynamics simulation study a
b
b
Swapan S. Jain , Uddhavesh B. Sonavane , Mallikarjunachari V.N. Uppuladinne , Emily C. a
a
a
b
McLaughlin , Weiqing Wang , Sheneil Black & Rajendra R. Joshi a
Department of Chemistry, Bard College, 30 Campus Rd, Annandale-on-Hudson, NY 12504, USA b
Bioinformatics Group, Centre for Development of Advanced Computing, Pune University Campus, Pune, Maharashtra 411007, India Published online: 03 Jan 2014.
To cite this article: Swapan S. Jain, Uddhavesh B. Sonavane, Mallikarjunachari V.N. Uppuladinne, Emily C. McLaughlin, Weiqing Wang, Sheneil Black & Rajendra R. Joshi (2015) Structural insights into the interactions of xpt riboswitch with novel guanine analogues: a molecular dynamics simulation study, Journal of Biomolecular Structure and Dynamics, 33:2, 234-243, DOI: 10.1080/07391102.2013.870930 To link to this article: http://dx.doi.org/10.1080/07391102.2013.870930
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions
Journal of Biomolecular Structure and Dynamics, 2015 Vol. 33, No. 2, 234–243, http://dx.doi.org/10.1080/07391102.2013.870930
Structural insights into the interactions of xpt riboswitch with novel guanine analogues: a molecular dynamics simulation study Swapan S. Jaina*, Uddhavesh B. Sonavaneb*, Mallikarjunachari V.N. Uppuladinneb, Emily C. McLaughlina, Weiqing Wanga, Sheneil Blacka and Rajendra R. Joshib a
Department of Chemistry, Bard College, 30 Campus Rd, Annandale-on-Hudson, NY 12504, USA; bBioinformatics Group, Centre for Development of Advanced Computing, Pune University Campus, Pune, Maharashtra 411007, India Communicated by Ramaswamy H. Sarma
Downloaded by [134.117.10.200] at 02:23 29 November 2014
(Received 17 July 2013; accepted 27 November 2013) Ligand recognition in purine riboswitches is a complex process requiring different levels of conformational changes. Recent efforts in the area of purine riboswitch research have focused on ligand analogue binding studies. In the case of the guanine xanthine phosphoribosyl transferase (xpt) riboswitch, synthetic analogues that resemble guanine have the potential to tightly bind and subsequently influence the genetic expression of xpt mRNA in prokaryotes. We have carried out 25 ns Molecular Dynamics (MD) simulation studies of the aptamer domain of the xpt G-riboswitch in four different states: guanine riboswitch in free form, riboswitch bound with its cognate ligand guanine, and with two guanine analogues SJ1 and SJ2. Our work reveals novel interactions of SJ1 and SJ2 ligands with the binding core residues of the riboswitch. The ligands proposed in this work bind to the riboswitch with greater overall stability and lower root mean square deviations and fluctuations compared to guanine ligand. Reporter gene assay data demonstrate that the ligand analogues, upon binding to the RNA, lower the genetic expression of the guanine riboswitch. Our work has important implications for future ligand design and binding studies in the exciting field of riboswitches. Keywords: riboswitch; guanine; ligand analogues; MD simulations
Introduction A vast majority of drugs target proteins in bacteria (Blount & Breaker, 2006). The widespread use of protein-targeting antibiotics has resulted in the rise of bacteria with antibiotic resistance, a major problem for public health that requires a fresh approach to combating disease. RNA, due to its rich structural diversity, presents a target much like proteins (Zaman, Michiels, & van Boeckel, 2003). Riboswitches are highly structured segments within the untranslated regions of specific mRNA molecules (Serganov & Patel, 2007). They have been shown to act as powerful gene regulatory elements in prokaryotes and certain eukaryotes as well (Mandal, Boese, Barrick, Winkler, & Breaker, 2003; Montange & Batey, 2008; Sudarsan, Barrick, & Breaker, 2003; Winkler & Breaker, 2005). Small molecules binding to the aptamer (sensor) domain of riboswitches cause conformational changes in the adjacent expression (actuator) domain of mRNA, which subsequently leads to either premature termination of transcription or inhibition of translation (Fuchs, Grundy, & Henkin, 2006; Mandal & Breaker, 2004). Numerous riboswitches that have been discovered in the last decade are capable of selectively binding coenzymes (TPP: thiamine pyrophosphate, FMN: flavin mononucleotide, SAM: S-adenosylmethionine, and cobalamine), amino acids
(glycine, lysine, and glutamine), sugars (glucosamine-6phosphate), nucleotide derivatives (guanine, adenine, cyclic di-GMP, and ATP) and even small ions (magnesium and fluoride) (Breaker, 2011; Ren, Rajashankar, & Patel, 2012; Watson & Fedor, 2012). Purine-sensing mRNA molecules were one of the first riboswitches to be discovered in bacteria (Winkler, 2005). Xanthine phosphoribosyl transferase (xpt) is a key enzyme in the purine metabolic pathway of bacteria catalyzing the synthesis of xanthine monophosphate from xanthine. It has been shown that binding of guanine to the xpt mRNA leads to the formation of a distinctly folded RNA segment, which leads to transcription termination (Batey, Gilbert, & Montange, 2004; Serganov et al., 2004). Guanine has numerous H-bond donor and acceptor groups, which facilitate site-specific binding to nucleotide bases in the aptamer domain of the riboswitch (Figure 1(a)). Guanine forms eight non-covalent interactions at a distance of < 3 Å to the aptamer domain of the riboswitch resulting in a tightly bound complex with a dissociation constant of ~5 nM (Mandal & Breaker, 2004). Figure 1(b) shows that guanine base forms a Watson Crick pair with C74 nucleotide of the riboswitch aptamer domain. Three more hydrogen bonds are observed with N3, N9, and the exocyclic –NH2 group of
*Corresponding authors. Emails: [email protected] (S.S. Jain); [email protected] (U.B. Sonavane) © 2013 Taylor & Francis
Downloaded by [134.117.10.200] at 02:23 29 November 2014
Interactions of xpt Riboswitch with Novel Guanine Analogues
Figure 1. (a) Secondary structure of the guanine riboswitch aptamer domain (b) Space filling model of guanine bound to the residues of the riboswitch. Dashed lines indicate hydrogen bonding and other non-covalent interactions. White ovals show empty space that can be exploited by rational ligand design.
U51 nucleotide. An additional hydrogen-bonding interaction occurs with U47 nucleotide. U22 of the riboswitch is involved in an interaction where the 2′–OH group of its sugar makes a hydrogen bonding contact with the N7 atom of the guanine ligand. Crystallography data by Batey and coworkers have shown that hypoxanthine, which is a precursor of guanine, is completely surrounded by the xpt riboswitch. Furthermore, ~97% of the surface area of hypoxanthine is inaccessible to the bulk solvent (Batey et al., 2004). Longrange tertiary interactions between aptamer domain loops L2 and L3 (Figure 1(a)) are essential for guanine binding to the riboswitch. These interactions are formed even in the ligand-free state of the riboswitch and are further stabilized by the stacking of two base quadruples (U34A65— G37-C61 and A33A66–G38-C60). Even though the residues in the loop region do not contact the ligand directly, the interactions are nonetheless essential for the overall stability and architecture of the riboswitch. Ligand recognition and binding occurs by a hierarchically induced fit process where binding of guanine leads to structural rearrangements in other parts of the RNA molecule.
235
Recent efforts in the area of purine riboswitch research have focused on ligand analogue studies. Synthetic molecules that mimic purines may have the potential of tightly binding and regulating riboswitch mRNA in prokaryotes (Mulhbacher & Lafontaine, 2007). Work by Lafontaine group shows that a purine ring system is not necessary for effective binding of synthetic compounds to the xpt riboswitch (Mulhbacher et al., 2010). Pyrimidine-based analogs that present H-bonding groups at similar position as guanine also bind with high selectivity to the aptamer domain of the RNA. In 2009, Breaker and coworkers conducted a ligand analogue study where guanine was modified either at the C2 or the C6 position using a variety of functional groups (Kim et al., 2009). White ovals in Figure 1(b) show empty spaces that can be exploited by modifying guanine at the C2 or C6 position. In-line probing assays showed that some of the compounds bound to the riboswitch even more tightly than guanine. In particular, modification of guanine at the C2 position with an acetyl group resulted in N-acetyl guanine that bound to the riboswitch RNA with a 10-fold higher binding affinity (Kd = 0.5 nM) compared to guanine. Out of all the C6-modified analogues tested by Breaker and coworkers, only the 6-hydrazine guanine demonstrated a similar binding affinity to guanine. We have designed our compounds, SJ1 and SJ2, by taking cues from previous work done by the Breaker group. Guanine analogs simultaneously modified at both the C2 and C6 position of the purine ring (right hand column in Figure 2) may serve as effective analogs in regulating the xpt riboswitch in bacteria. Synthesis and biochemical studies illustrating the interactions of these compounds to riboswitch RNA will be presented in a forthcoming paper. In the current work, 25 ns molecular dynamics simulations in explicit solvent were carried out after guanine, SJ1 and SJ2 were docked to the aptamer domain of the guanine riboswitch. Brenk and coworkers have shown that an RNA–ligand docking approach, much like a protein–ligand docking approach, can be effectively used to discover novel compounds displaying high affinity to the guanine-sensing riboswitch (Daldrop et al., 2011). Stock and coworkers conducted molecular dynamics simulations of the guanine riboswitch with guanine, adenine and the ligand-free state (Villa, Wohnert, & Stock, 2009). They demonstrated that there is a stepwise process where ligand binding to a preformed pocket is followed by structural rearrangement and conformational changes leading to a globally stable RNA ligand complex. A recent molecular dynamics study of the adenine riboswitch (structurally identical to the guanine riboswitch) reveals the importance of the L2/L3 loop–loop kissing interactions in stabilizing the overall tertiary structure of the aptamer domain (Allner, Nilsson, & Villa, 2013).
236
S.S. Jain et al. H2N NH
NH N N H
O N
6
N N
N
NH2
6-hydrazine guanine Kd ~ 8 nM
N H
N
O N H
N
SJ1
NH 2
N H
N
H2N
NH2
NH
O
Guanine Kd ~ 5 nM
N
Downloaded by [134.117.10.200] at 02:23 29 November 2014
N H
N H N
N H
N-acetyl guanine Kd ~ 0.5 nM
Figure 2.
O
N N H
N N
N
NH2
N H
SJ2
Chemical structure of guanine, 6-hydrazine guanine, N-acetyl guanine and C2/C6 modified SJ1 and SJ2.
We have used molecular dynamics simulations to investigate the interactions of novel ligand analogues with the guanine riboswitch mRNA. In this work, we have shown that in addition to contacting core nucleotides bound by guanine ligand, analogues SJ1 and SJ2 bind other nucleotide bases and do so with a greater number of overall hydrogen bonds. These ligands, upon binding to the aptamer domain of the riboswitch, yield a lower overall free energy and root mean square deviation (RMSD) of the complex than that of the guanine-bound riboswitch. Root mean square fluctuation (RMSF) analysis shows that these ligands restrict the conformational flexibility of the aptamer domain, which potentially has important implications on overall riboswitch function. To date, several experimental and theoretical studies have investigated the effect of ligand analogue binding to different types of riboswitches and nucleic acid aptamers (Banas, Walter, Sponer, & Otyepka, 2010; Kumar, Endoh, Murakami, & Sugimoto, 2012; Schill & Koslowski, 2013; Xin & Hamelberg, 2010). To the best of our knowledge, this work represents the first report where ligand analogues bound to a guanine riboswitch molecule have been studied by molecular dynamics simulations. Materials and methods Initial structures The crystal structure (PDB ID 1Y27) of guanine-bound G-riboswitch (guanine riboswitch) was considered as the starting structure for control system (Batey et al., 2004).
We removed the guanine molecule from the control structure and designated it as the ligand-free G-riboswitch system. The G-riboswitch with SJ1 and SJ2 as the bound ligands in place of guanine were modelled by fitting SJ1 and SJ2 to the ligand position in the crystallographic structure. All these four complex systems were considered as the starting structures for the simulations. Force-field parameters Parameters for all the three ligands guanine, SJ1, and SJ2 were derived from AMBER force field (ff03) (Duan et al., 2003). Partial atomic charges were derived by performing geometry optimization at Hartree–Fock level using 6–31G* and single-point energy calculations at DFT level using 6–311G** basis sets (Ditchfield, Hehre, & Pople, 1971; Hehre, Ditchfield, & Pople, 1972). The electronic structure calculations were carried out with Gaussian 03 package (Case et al., 2008; Frisch et al., 2004). Molecular dynamics simulations The MD simulations were carried out using AMBER10 program with ff03 force field (Case et al., 2008; Duan et al., 2003). All the four systems were neutralized by adding counter ions and solvated with TIP3P water box (Jorgensen, Chandrasekhar, Madura, Impey, & Klein, 1983). Energy minimization was carried out for all the systems by using steepest descent method for 5000 steps. Then simulated annealing was carried out at temperature
Downloaded by [134.117.10.200] at 02:23 29 November 2014
Interactions of xpt Riboswitch with Novel Guanine Analogues range of 50 K to 300 K for 100 ps, by using constraints of 100 kcal/mol. Equilibration was achieved by gradually reducing the constraints for 500 ps. The equilibration protocol of Cheatham was followed similar to previous simulations of nucleic acids (Cheatham & Kollman, 1997). Simulations were performed under periodic boundary conditions by employing the particle-mesh Ewald technique to account for long-range electrostatics (York, Darden, & Pedersen, 1993). MD integration was carried out using a 2.0 fs time step, employing the SHAKE algorithm on all the hydrogen atoms and non-bonded cut-off of 10.0 Å (Ryckaert, Ciccotti, & Berendsen, 1977). The pair list was updated every 100 steps. Constant pressure (1 atm) and temperature (300 K) were maintained using the Berendsen coupling algorithm throughout the production simulation run (Berendsen, Postma, van Gunsteren, Dinola, & Haak, 1984). The systems were allowed to equilibrate under production run conditions for 1 ns before collection of data over 25 ns simulation time. Four simulations were carried out for 25 ns each and the trajectories were used for the analysis. All-atom classical MD simulations of four systems, namely Free-Ribo, Ribo-Gua, Ribo-SJ1 and Ribo-SJ2, were carried out on the Bioinformatics Resources and Application Facility (BRAF) of C-DAC, Pune.
237
expression was measured using the standard Miller Unit assay (Miller, 1992). Samples were run in quadruplicates. Results and discussion The present simulation study of the four complexes reveals a good overall representation of the important interactions between ligand and G-riboswitch along with the effects of the ligand in stabilizing the overall complex. Four simulations were carried out in order to understand the effect of different ligands and compared with control system (Ribo-Gua) along with ligand-free G-riboswitch system (Free-Ribo). The initial simulation structure of the riboswitch–guanine complex (Ribo-Gua) is shown in Figure 3. The starting structures for the Free-Ribo, Ribo-SJ1, and Ribo-SJ2 are presented as supplementary information (Figure S1). During the simulation run, the plots for temperature, pressure and total energy maintain stable average values with allowed fluctuations indicating that all the four systems were in
Data analysis All statistics were collected on 1-picosecond snapshots over 25 ns of simulation data. Trajectories and structures were visualized using VMD (Humphrey, Dalke, & Schulten, 1996). PTRAJ module of AmberTools was used for RMSD, RMSF, radius of gyration, and hydrogen bond analyses. The criterion for hydrogen bonding was set at ≤3.5 Å distance between electron donor atom and hydrogen of electron acceptor atom with 120-degree angle cut-off. Strong hydrogen bonds were categorized as those having a distance of ≤ 2.5 Å whereas weak bonds were categorized as those with distance of 2.5–3.5 Å. Molecular Mechanics Generalized Born Surface Area (MMGBSA) module of AMBER10 was used for calculating free energy. XMGRACE software was used for generating all the plots. Analysis of riboswitch function Bacillus subtilis cells with guanine riboswitch gene fused upstream of promotorless lacZ reporter gene were used to quantify riboswitch mRNA expression in vivo. B. subtilis cells were provided by Dr. Ronald Breaker (Yale University). Ligand stock solutions were prepared in DMSO solvent. Cells were grown overnight in minimal medium with constant shaking at 37 °C. Cells at an OD600 of ~0.1 were incubated with 200 µM of the designated ligands for 8 h at 37 °C. Beta galactosidase
Figure 3. Starting structure of Ribo–Gua complex. Riboswitch RNA molecule is shown as coloured sticks and the guanine ligand is shown as a sphere rendering.
S.S. Jain et al.
equilibrium (Supplementary Figure S2). These plots suggest that the thermodynamic ensemble is maintained and energies are not showing any abnormality in the system. All the four trajectories are analysed for RMSD, RMSF, radius of gyration, hydrogen bonding, and free energy. The radius of gyration data is presented in the supplementary information section (Figure S3). The backbone RMSD of the free riboswitch was maintained at 3 Å with fluctuations for all four complexes (Figure 4(a)). Though all four complexes demonstrated similar RMSD values, there is a slight difference in the average RMSDs. The average RMSD with respect to the X-ray structure was only 2.92 ± 0.44 Å for the control system (Ribo–Gua). The Free-Ribo system yielded an average RMSD value of 2.97 ± 0.57 Å, which was the highest among all four complexes. The average RMSD value of Ribo-SJ1 and Ribo-SJ2 was 2.73 ± 0.41 Å and 2.65 ± 0.35 Å, respectively. Since all four complexes displayed average RMSD values that were similar to each other, the residue-specific RMSF calculations provide more detailed information regarding local deviations that occur upon ligand binding to the riboswitch. The RMSF parameter measures the average atomic mobility of backbone atoms during the molecular dynamic simulations of a system. The residue-wise RMSF of the four complexes was calculated and plotted in Figure 4(b). Our results show that RMS fluctuations were lower for Ribo-SJ1 and Ribo-SJ2 complex across the entire aptamer domain when compared to the
RMSD (Å)
(a)
5 4 3 2 1
Free-Ribo
Ribo-Gua
Ribo-SJ1
Ribo-SJ2
0 0
5
10
15
20
25
time (ns)
(b)
25 Ribo-Gua
Free-Ribo
Ribo-SJ2
Ribo-SJ1
20
RMSF (Å)
Downloaded by [134.117.10.200] at 02:23 29 November 2014
238
15 10 5 0 10
20 _____
30 ____
P1
P2 J1-2
____ 40 ____ L2
50
P2
___60___ P3
J2-3
L
70 ___
80 _____
P3
P1 J3-1
Residue #
Figure 4. (a) Backbone RMSD of the riboswitch RNA complexes. (b) Residue-wise RMSF of the four complexes.
Ribo-Gua complex. In particular, U47 residue that is present in the J2/3 junction and contacts the ligand shows significantly lower fluctuations than the Ribo-Gua complex. Fluctuations at residue C74, which is necessary for the specific recognition of the guanine ligand, are very similar for all of the simulated complexes. High-resolution studies have indicated that loops L2 and L3 form interconnecting hydrogen bonds (kissing-loop interaction) and these interactions are important for the overall folded architecture of the riboswitch (Batey et al., 2004; Serganov et al., 2004). Residues 32–38 present in L2 region directly interact with residues 60–66 present in L3 region. RMSF peaks in the loop regions show lower fluctuations for Ribo-SJ1 and Ribo-SJ2 compared to Ribo-Gua indicating that these novel ligands bind to the riboswitch without major structural perturbations whilst retaining the overall folded architecture of the complex. We have also looked at region-wise RMSD values for the ligand complexes and observe that there are regions such as the L3 loop where the Ribo-Gua system is showing more than 1Å high RMSD compared to the other three systems. In an effort to probe the intramolecular (within the riboswitch) and intermolecular (between the riboswitch and the ligand) interactions, hydrogen bond calculations were carried out after the simulations were completed. The stability of a complex was evaluated by calculating hydrogen bonding percentage (HB%) which was defined as the time over which a hydrogen bond satisfied the two standard criteria (d(H…O)
