ARTICLE pubs.acs.org/JPCB
Characterization of Interactions between PilA from Pseudomonas aeruginosa Strain K and a Model Membrane Justin A. Lemkul and David R. Bevan* Department of Biochemistry, Virginia Tech, Blacksburg, Virginia 24061, United States ABSTRACT: Type IV pili are important adhesion and motility factors in both Gramnegative and Gram-positive bacterial species, making pilus assembly from pilin subunits an important biophysical mechanism to understand at an atomic level. Knowledge of the pilus assembly mechanism has applications in antibiotic development, microbial physiology, and systems biology. We applied molecular dynamics simulations to investigate the position and orientation of Pseudomonas aeruginosa strain K PilA in a model membrane as well as the interactions of PilA with the surrounding lipids, identifying several key residues that stabilize the position of PilA within the membrane. Furthermore, we performed umbrella sampling to determine the free energy of extracting this protein from the membrane. Our results provide insight into the molecular events and energetics associated with pilus assembly, thereby initiating a detailed examination of the dynamics of this process.
’ INTRODUCTION For many species of bacteria, attachment to the host environment and subsequent virulence depend in part on the formation of type IV pili (TFP) from pilin monomers,1 which in turn depends upon the translocation of the emerging pilus through the periplasm and outer membrane such that additional pilin subunits can be added at the base of the assembly in the inner membrane. Addition of pilin subunits is driven by ATP hydrolysis, a process that is carried out by a cytoplasmic ATPase that is a member of the GspE family.2 Mechanical force is transmitted through a GspF family integral membrane protein (IMP), which is involved in extraction of the pilin monomers from the inner bacterial membrane.3 The energy requirements for subunit addition, translocation, and any resulting conformational change of pilin subunits are currently unknown. Since the assembly rate of a pilus is approximately 350 nm s 1,4 it is important to use theoretical methods to gain atomistic insight into this process at time scales that are often inaccessible to most experimental techniques. While several studies have indicated that pilus assembly may generate forces in excess of 100 pN,4,5 making it one of the strongest known molecular motors, the underlying reasons for this force generation are thus far uncharacterized on an atomistic level. The position and orientation of pilin subunits within the membrane will influence the manner in which they are added to the growing pilus. Electrostatic interactions between the positively charged N-terminus of one pilin monomer and the highly conserved Glu5 of an adjacent monomer have been proposed as a driving force for this assembly.6 The present study analyzes the interactions of the Pseudomonas aeruginosa strain K (PAK) PilA subunit7 with a model palmitoyloleoylphosphatidylethanolamine (POPE) membrane and the energy barrier for its extraction from this environment. r 2011 American Chemical Society
’ METHODS MD Simulations. The starting structure for PAK PilA was taken from PDB entry 1OQW.7 One chain of the crystallized dimer was isolated, and crystal waters were removed from the coordinate file. The coordinates and topology for a POPE membrane containing 340 lipids were obtained from the Web site of D. Peter Tieleman (http://moose.bio.ucalgary.ca). PEcontaining lipids are the predominant lipid type in several pilusproducing bacteria8 and other bacterial species,9 and thus, POPE was chosen as the model for the present work. The N-terminal 19 residues of PilA were aligned with the z axis of the simulation cell, and Pro22 was placed coincident with the membrane water interface of the POPE membrane to produce its final orientation for membrane insertion. PilA was inserted into the POPE membrane using the InflateGRO method.10 Since water molecules in the pre-equilibrated POPE membrane were removed in this process, the simulation cell was resolvated with simple point charge (SPC) water11 such that no water molecules were placed in the membrane interior. The system was electroneutral, and thus, no ions were added. All simulations were subsequently performed with the GROMACS package, version 4.0.7.12 The force field applied for the simulations described here was GROMOS96 53A6,13 extended to include the lipid parameters developed by Berger et al.14 In all simulations, periodic boundary conditions were imposed in all three dimensions. Short-range van der Waals interactions were cut off at 1.2 nm, with dispersion correction applied to the energy and pressure terms to account for this truncation. Long-range electrostatic interactions were calculated Received: March 8, 2011 Revised: April 24, 2011 Published: June 01, 2011 8004
dx.doi.org/10.1021/jp202217f | J. Phys. Chem. B 2011, 115, 8004–8008
The Journal of Physical Chemistry B using the smooth particle mesh Ewald method,15,16 with the realspace contribution to this potential cut off at 1.2 nm. The neighbor list was cut off at 1.2 nm, as well, and updated every five simulation steps. Constraints were applied to all bonds using the P-LINCS algorithm,17 allowing an integration time step of 2 fs during molecular dynamics (MD). The PilA POPE system was energy-minimized using the steepest descent method, after which the system was equilibrated in two phases, during which position restraints were placed on all heavy atoms of PilA. The first phase of equilibration applied an NVT ensemble for 100 ps to maintain the system temperature at 100 K, allowing water molecules to soak into the POPE headgroups slowly. Temperature was controlled using a weak coupling method18 with a coupling constant of 0.1 ps. The protein, lipids, and water were all coupled to separate thermostats. Having stabilized at 100 K, an NPT ensemble was conducted for 1 ns at 310 K and 1 bar of pressure, using the Nose Hoover thermostat19,20 and Parrinello Rahman barostat.21,22 Pressure coupling was applied semi-isotropically, such that the x y and z dimensions were allowed to deform independently. Following equilibration, restraints were removed from the PilA heavy atoms and production simulations were allowed to proceed for 50 ns. Coordinates and energies were saved every 10 ps, and all subsequent analysis was performed using programs provided in the GROMACS package. Three independent simulations were conducted by initiating each NVT equilibration phase with different random velocities. Steered MD and Umbrella Sampling. The final configuration of a representative 50 ns trajectory was used as input for the steered MD (SMD) simulations. The unit cell was extended in the z dimension to accommodate the extraction of PilA such that all atoms were beyond the short-range cutoffs described above. Additional water molecules were added to this extended space. This new system was equilibrated in the same manner as described above. PilA was extracted from the POPE membrane by applying a constraint force to the Pro22 residue in a direction coincident with the z axis. A CR distance restraint matrix was used to maintain the R-helical configuration of residues 1 35 during this procedure. Lipid molecules were held in the plane of the membrane by a position restraint (kpr = 1000 kJ mol 1 nm 2) applied to the phosphorus atoms in the z dimension only; the lipids were allowed to diffuse freely in the x y plane. PilA was extracted at a rate of 0.01 nm ps 1. Center-of-mass (COM) pulling was carried out until the COM of Pro22 and POPE was in excess of 11 nm. At 6.8 nm of COM separation between Pro22 and POPE, none of the atoms of PilA were within the short-range nonbonded cutoff. Snapshots along this pulling direction were extracted at a COM spacing of 0.2 nm to generate starting configurations for umbrella sampling windows. Umbrella sampling windows were created over the range of 1.2 7.2 nm, spaced at 0.2 nm intervals, yielding 31 sampling windows. For reasons described below, an additional window corresponding to a configuration at 2.1 nm COM spacing was added, giving 32 total sampling windows in which data were collected. In each window, simulations were carried out for 10 ns. Potential of mean force (PMF) curves were obtained using the WHAM algorithm,23 as implemented in GROMACS.24
’ RESULTS AND DISCUSSION The most important characteristics of PilA in the POPE membrane are the position and orientation of the protein over
ARTICLE
Figure 1. Snapshots from a representative unrestrained PilA POPE simulation at (A) 0 ns and (B) 50 ns (inset, interfacial Tyr24 and Tyr27 and the position of Pro22). The protein is colored as a gradient (blue to red, N-terminus to C-terminus), and lipids are shown as gray lines with phosphorus atoms shown as gold spheres. For clarity, water molecules are not shown.
time and its interactions with POPE lipids, both hydrophobic and electrostatic. All averages and standard deviations reported in this study (representing all three simulations) reflect data collected over at least the last 25 ns of each simulation, unless otherwise indicated. Pilin monomers are believed to be positioned in the membrane with the N-terminal helical domain within the bilayer and the C-terminal domain in the periplasm.7 Thus, we oriented PilA with respect to the membrane such that Pro22 was coincident with the membrane interface (defined as the average z coordinate of phosphorus atoms of the outer, or periplasmic, leaflet of the membrane), with the N-terminal 22 residues (an R-helix) aligned with the membrane normal (Figure 1A). Such an alignment buried this predominantly hydrophobic segment of the protein structure in the nonpolar membrane without presupposing any tilt angle of the helix. Over the course of the MD simulations, PilA deviated from this position to move further into the membrane, with the N-terminal helix tilting with respect to the membrane normal (Figure 1B). The initial position of Pro22 was approximately 2 nm from the COM of the POPE membrane. Within 20 ns, PilA positioned itself such that Pro22 was below the membrane water interface at a COM distance of 1.6 ( 0.3 nm. This relocation caused approximately one additional turn of the N-terminal R-helix to be buried in the membrane, positioning Tyr24 and Tyr27 at the membrane water interface (Figure 1B, inset). The protein 8005
dx.doi.org/10.1021/jp202217f |J. Phys. Chem. B 2011, 115, 8004–8008
The Journal of Physical Chemistry B
ARTICLE
Figure 2. Phe1 Glu5 interactions at (A) 0 ns and (B) 50 ns. PilA is shown as a blue ribbon, with the atoms of Phe1 and Glu5 shown as sticks and colored by element (C, blue; H, white; O, red; N, dark blue). Phosphorus atoms (gold spheres) are shown to illustrate the location of the membrane water interface.
sequence in this region (Q23YQNYVARSEGAS35) contains a number of charged and polar residues, which interacted with the membrane via hydrogen bonding with the POPE lipids. The combination of these interactions and the hydrophobicity of the N-terminal sequence likely serve to anchor PilA in the membrane. The loop region connecting β-strands β3 and β4 in PilA also came in close contact with the POPE headgroups, specifically through hydrogen bonding and electrostatic interactions involving Lys110. The C-terminal receptor-binding loop25 27 also bound to the POPE membrane water interface via hydrogen bonding involving Asp134, Glu135, and Gln136. These observations indicate that protein lipid hydrogen bonding is an important force in establishing the position of PilA in the POPE membrane. At the outset of the simulations, only 7.0 ( 1.7 hydrogen bonds were present between PilA and POPE lipids, but 27.0 ( 2.5 hydrogen bonds developed over time during the three simulations performed here. As the C-terminal region has been implicated in binding to epithelial cells in the host organism,25 27 it is reasonable that this sequence may have some affinity for the P. aeruginosa lipid membrane as well. The bacterium must thus expend some energy to disrupt these interactions to allow PilA to bind to the host cell membrane or receptors therein. Our simulations indicate that these C-terminal residues, as well as the interfacial sequence described above, are particularly important in stabilizing the position of PilA within the P. aeruginosa inner membrane, an observation that likely extends to other bacteria that produce Type IVa pili. Tyr24, Tyr27, Arg30, Lys110, and Asp134 are all highly conserved in many similar bacterial strains, including Neisseria gonorrheae strain MS11 and P. aeruginosa strain K122-4.7 The polar sequence S31EGAS35 present in PAK PilA is not highly conserved across these organisms, but the presence of many polar residues of the polypeptide sequences indicates that polar and charged residues in this region may be a common feature of pilin proteins that allow for membrane anchoring, as observed here. As the vertical position of PilA stabilized over the course of the simulations, the N-terminal 22 residues adopted a tilt of 9.2 ( 1.8° with respect to the membrane normal, such that both Glu5 and the positively charged N-terminus became more proximal to the polar membrane water interface of the cytofacial leaflet of the membrane (Figure 2). Initially, the N-terminal amino group of Phe1 participated in a salt bridge with the side chain of Glu5 at a distance of 0.36 ( 0.03 nm (Figure 2A),7 shielding these free
Figure 3. PMF curve for extracting PilA from the POPE membrane. Vertical red lines indicate error bars generated by the Bayesian bootstrap method implemented in the g_wham program.24
charges from the hydrophobic core of the membrane. The change in position and orientation of the PilA N-terminal helix caused the region encompassing Phe1 and Glu5 to enter a membrane environment with a higher dielectric, breaking this interaction, and the amino and carboxylate groups stabilized at a distance of 0.46 ( 0.03 nm (Figure 2B). Separating these charges may have functional implications for the recruitment of a new PilA subunit during pilus formation, as these charged moieties have been implicated as having a role in stabilizing the assembly and thus must be accessible to nearby PilA subunits.6 Tilting of transmembrane R-helices is an entropically favorable phenomenon,28 providing an increase in configurational volume accessible to the protein due to helix precession around the membrane normal. The tilting of PilA observed here is driven by polar contacts between tyrosine and other polar amino acids in PilA and the periplasmic face of the membrane. Association of tyrosine with the membrane water interface is energetically favorable and is frequently observed in membrane proteins, with a ΔG of 14.0 kJ mol 1 for transferring tyrosine from water to the interface.29 The gain in configurational entropy by the protein as a whole may be an important factor in the accessibility of PilA to the pilus assembly. By adopting this tilted orientation in the membrane, it may be easier for subunits to be recruited to the pilus, since tilting promotes the proximity of Phe1 and Glu5 to the interface, breaking their intramolecular interaction. The potential of mean force (PMF) curve for extracting PilA from the POPE membrane (Figure 3) indicates that complete extraction of the protein would require 264.9 ( 2.8 kJ mol 1, approximately equivalent to the energy released by the hydrolysis of 7 8 ATP molecules, assuming a ΔG of 35.7 kJ mol 1 for ATP hydrolysis under physiological conditions. The free energy minimum of the PMF curve is located at approximately 1.3 nm, the distance between the COM of the POPE membrane and Pro22 that allowed for optimal interfacial alignment of Tyr24 and Tyr27 and hydrogen bonding during the unrestrained MD simulations discussed above. The inclusion of an additional sampling window at a COM distance of 2.1 nm (between two existing windows at 2.0 and 2.2 nm) was based on poor sampling that was produced in this region of the reaction coordinate as polar residues crossed the membrane water interface (Figure 4). Thus, sampling at 2.1 nm 8006
dx.doi.org/10.1021/jp202217f |J. Phys. Chem. B 2011, 115, 8004–8008
The Journal of Physical Chemistry B
ARTICLE
view three ATP molecules as an upper limit for this displacement, with much of the energy being utilized to extract Tyr24, Tyr27, and other polar, interfacial amino acids from the periplasmic membrane water interface. Vertical displacement of 1.0 nm along the z axis (as required by the model of pilus extension discussed above) positions this region just above the membrane water interface (Figure 4).
Figure 4. Positions of Tyr24 and Tyr27 along the z axis at the energy minimum of the PMF (blue) and at 1.0 nm vertical displacement (red). Tyrosine residues are shown as sticks and colored by element (C, tint color, blue or red; H, white; O, dark red; N, dark blue). The interface is indicated by the position of phosphorus atoms (transparent gold spheres).
’ CONCLUSIONS The mechanism by which PilA monomers are added to the growing pilus produced by P. aeruginosa remains to be established. The results presented here are the first atomic-resolution biophysical characterization of the interactions of PilA with a surrounding membrane and the first theoretical examination of the energetics of PilA extraction, both of which are crucial insights into a better understanding of the dynamics of pilus assembly. From this information, future studies can be designed to determine the means by which energy is transmitted from the cytoplasmic ATPase, through IMP, and applied to PilA to affect pilus extension. ’ AUTHOR INFORMATION Corresponding Author
*Phone: (540) 231-5040. Fax: (540) 231-9070. E-mail: drbevan@ vt.edu.
’ ACKNOWLEDGMENT We thank Dr. Lisa Craig for a critical review of this manuscript, Advanced Research Computing at Virginia Tech for computing time on the SystemX supercomputer, and the Institute for Critical Technology and Applied Science (ICTAS) at Virginia Tech for financial support. ’ REFERENCES
Figure 5. Umbrella sampling histograms from the 32 sampling windows.
of COM spacing was inefficient, so an additional window was added, corresponding to this COM distance. Good sampling was achieved upon doing so (Figure 5), allowing for the assembly of a reliable PMF curve with a low error estimate. To align with the turn of a pilus structure, a single PilA protein would have to be displaced only approximately 1 nm along the membrane normal.6 From the PMF curve, such a translocation would require 113.0 kJ mol 1, or approximately three ATP molecules. Maier et al. proposed a model in which up to six ATPases work in concert to affect stepwise elongation of the growing pilus.4 The presence of other proteins, such as the IMP and the growing pilus, may affect the free energy of displacing a PilA subunit from the inner membrane, and thus, the actual energy barrier for translocation may be different under physiological conditions. Presumably, the other components of the pilus complex would interact with PilA units favorably, thus reducing the energy barrier for PilA extraction. In this respect, we
(1) Pelicic, V. Mol. Microbiol. 2008, 68, 827. (2) Planet, P. J.; Kachlany, S. C.; DeSalle, R.; Figurski, D. H. Proc. Natl. Acad. Sci. U.S.A. 2001, 98, 2503. (3) Collins, R. F.; Saleem, M.; Derrick, J. P. J. Bacteriol. 2007, 189, 6389. (4) Maier, B.; Koomey, M.; Sheetz, M. P. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 10961. (5) Maier, B.; Potter, L.; So, M.; Long, C. D.; Seifert, H. S.; Sheetz, M. P. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 16012. (6) Craig, L.; Li, J. Curr. Opin. Struct. Biol. 2008, 18, 267. (7) Craig, L.; Taylor, R. K.; Pique, M. E.; Adair, B. D.; Arvai, A. S.; Singh, M.; Lloyd, S. J.; Shin, D. S.; Getzoff, E. D.; Yeager, M.; Forest, K. T.; Tainer, J. A. Mol. Cell 2003, 11, 1139. (8) Rahman, M. M.; Kolli, V. S. K.; Kahler, C. M.; Shih, G.; Stephens, D. S.; Carlson, R. W. Microbiology 2000, 146, 1901. (9) Cronan, J. E. Annu. Rev. Microbiol. 2003, 57, 203. (10) Kandt, C.; Ash, W. L.; Tieleman, D. P. Methods 2007, 41, 475. (11) Berendsen, H. J. C.; Postma, J. P. M.; van Gunsteren, W. F.; Hermans, J. Interaction models for water in relation to protein hydration. In Intermolecular Forces; Pullman, B., Ed.; Reidel: Dordrecht, The Netherlands, 1981; p 331. (12) Hess, B.; Kutzner, C.; van der Spoel, D.; Lindahl, E. J. Chem. Theory Comput. 2008, 4, 435. (13) Oostenbrink, C.; Villa, A.; Mark, A. E.; van Gunsteren, W. F. J. Comput. Chem. 2004, 25, 1656. (14) Berger, O.; Edholm, O.; J€ahnig, F. Biophys. J. 1997, 72, 2002. (15) Darden, T.; York, D.; Pedersen, L. J. Chem. Phys. 1993, 98, 10089. 8007
dx.doi.org/10.1021/jp202217f |J. Phys. Chem. B 2011, 115, 8004–8008
The Journal of Physical Chemistry B
ARTICLE
(16) Essmann, U.; Perera, L.; Berkowitz, M. L.; Darden, T.; Lee, H.; Pedersen, L. G. J. Chem. Phys. 1995, 103, 8577. (17) Hess, B. J. Chem. Theory Comput. 2008, 4, 116. (18) Berendsen, H. J. C.; Postma, J. P. M.; van Gunsteren, W. F.; DiNola, A.; Haak, J. R. J. Chem. Phys. 1984, 81, 3684. (19) Nose, S. J. Chem. Phys. 1984, 81, 511. (20) Hoover, W. G. Phys. Rev. A 1985, 31, 1695. (21) Parrinello, M.; Rahman, A. J. Appl. Phys. 1981, 52, 7182. (22) Nose, S.; Klein, M. L. Mol. Phys. 1983, 50, 1055. (23) Kumar, S.; Rosenberg, J. M.; Bouzida, D.; Swendsen, R. H.; Kollman, P. A. J. Comput. Chem. 1992, 13, 1011. (24) Hub, J. S.; de Groot, B. L.; van der Spoel, D. J. Chem. Theory Comput. 2010, 6, 3713. (25) Lee, K. K.; Doig, P.; Irvin, R. T.; Paranchych, W.; Hodges, R. S. Mol. Microbiol. 1989, 3, 1493. (26) Irvin, R. T.; Doig, P.; Lee, K. K.; Sastry, P. A.; Paranchych, W.; Todd, T.; Hodges, R. S. Infect. Immun. 1989, 57, 3720. (27) Lee, K. K.; Sheth, H. B.; Wong, W. Y.; Serburne, R.; Paranchych, W.; Hodges, R. S.; Lingwood, C. A.; Krivan, H.; Irvin, R. T. Mol. Microbiol. 1994, 11, 705. (28) Lee, J.; Im, W. Phys. Rev. Lett. 2008, 100, 018103. (29) MacCallum, J. L.; Bennett, W. F. D.; Tieleman, D. P. J. Gen. Physiol. 2007, 129, 371.
8008
dx.doi.org/10.1021/jp202217f |J. Phys. Chem. B 2011, 115, 8004–8008
