pubs.acs.org/Langmuir © 2010 American Chemical Society
Effect of Poly(ethylene glycol) (PEG) Spacers on the Conformational Properties of Small Peptides: A Molecular Dynamics Study Ying Xue,†,‡,^ Megan L. O’Mara,†,^ Peter P. T. Surawski,† Matt Trau,§ and Alan E. Mark*,†,‡ †
School of Chemistry and Molecular Biosciences (SCMB), and the Institute for Molecular Biosciences (IMB), University of Queensland, Brisbane QLD 4072, Australia, ‡Groningen Biomolecular Sciences and Biotechnology Institute (GBB), Department of Biophysical Chemistry, University of Groningen, The Netherlands, and § Australian Institute for Bioengineering and Nanotechnology, The University of Queensland, Brisbane QLD 4072, Australia. ^ Co-first authors Received September 22, 2010. Revised Manuscript Received November 18, 2010 Poly(ethylene glycol) (PEG) is used as an inert spacer in a wide range of biotechnological applications such as to display peptides and proteins on surfaces for diagnostic purposes. In such applications it is critical that the peptide is accessible to solvent and that the PEG does not affect the conformational properties of the peptide to which it is attached. Using molecular dynamics (MD) simulation techniques, we have investigated the influence of a commonly used PEG spacer on the conformation properties of a series of five peptides with differing physical-chemical properties (YGSLPQ, VFVVFV, GSGGSG, EEGEEG, and KKGKKG). The conformational properties of the peptides were compared (a) free in solution, (b) attached to a PEG-11 spacer in solution, and (c) constrained to a two-dimensional lattice via a (PEG-11)3 spacer, mimicking a peptide displayed on a surface as used in microarray techniques. The simulations suggest that the PEG spacer has little effect on the conformational properties of small neutral peptides but has a significant effect on the conformational properties of small highly charged peptides. When constrained to a twodimensional surface at peptide densities similar to those used experimentally, it was found that the peptides, in particular the polar and nonpolar peptides, aggregated strongly. The peptides also partitioned into the PEG layer. Potentially, this means that at high packing densities only a small fraction of the peptide attached to the surface would in fact be accessible to a potential interaction partner.
Introduction Poly(ethylene glycol) (PEG) is an inert polymer composed of repeating units of CH2CH2O. PEG is nontoxic, nonionic, and hydrophilic. In addition to being used as a vehicle or base in foods, cosmetics, and pharmaceuticals, PEG is widely used to stabilize, immobilize, or modify the physical properties of biological molecules. Specifically, peptides, proteins, and other biological molecules can be chemically linked to one or more PEG chains, a process commonly referred to as PEGylation. PEGylated peptides and proteins are increasingly important therapeutically. PEGylation can result in longer therapeutic half-lives, improved circulation, and lower immunogenicity and antigenicity compared to those of the native compounds.1-3 PEGylation can also reduce the propensity for proteins to aggregate, thus increasing solubility and stability.2-6 Some effects of PEGylation, such as the increased half-life in the circulation, stem simply from the increased molecular weight of the PEGylated protein. However, as the excluded volume of PEG is between 5 and 10 times that of a typical protein of similar molecular weight,7,8 PEGylation can also act to shield peptides and *Corresponding author. E-mail:
[email protected]. (1) Cua, H. Y.; Kwan, V.; Henriquez, M.; Kench, J.; George, J. Gut 2006, 55, 1521–1522. (2) Veronese, F. M.; Harris, J. M. Adv. Drug Delivery Rev. 2002, 54, 453–456. (3) Veronese, F. M.; Pasut, G. Drug Discovery Today 2005, 10, 1451–1458. (4) Morar, A. S.; Schrimsher, J. L.; Chavez, M. D. BioPharm Int. 2006, 19, 34– 49. (5) Fee, C. J.; Van Alstine, J. M. Chem. Eng. Sci. 2006, 61, 934–939. (6) Katre, N. V. Adv. Drug Delivery Rev. 1993, 10, 91–114. (7) Kozlowski, A.; Milton Harris, J. J. Controlled Release 2001, 72, 217–224. (8) Milton Harris, J.; Chess, R. B. Nat. Rev. Drug Discovery 2003, 2, 214–221. (9) Elbert, D. L.; Hubbell, J. A. Annu. Rev. Mater. Sci. 1996, 26, 365–394.
296 DOI: 10.1021/la103800h
proteins from electrostatic and/or hydrophobic interactions with other proteins and reduce protein adsorption onto surfaces.9-13 PEG is widely used as a linker to immobilize proteins or peptides on surfaces.14,15 This can be done to functionalize a specific surface by coupling proteins such as enzymes or to display peptide fragments for recognition by antibodies or even whole cells. For example, Trau and colleagues15 have used PEG to display antigenic peptide fragments on the surface of microspheres for diagnostic purposes. Such applications require not only that the peptide be accessible to a potential receptor such as an antibody but also that the peptide be able to adopt an appropriate conformation. The conformational properties of small peptides are strongly dependent on the local environment. In the case of PEGylated peptides the conformation will be affected by not only the nature of the covalent link but also the extent to which the peptide can directly interact with the PEG spacer. Although PEG is chemically inert and PEGylation widely used to present peptides on surfaces, little is known in regard to the extent to which PEGylation might affect the structural properties of small peptides. In this study, the effect of PEGylation on the conformational properties of a series of five peptides with differing physical-chemical properties is examined using molecular dynamics simulation (10) Amiji, M.; Park, K. J. Biomater. Sci., Polym. Ed. 1993, 4, 217–234. (11) Jeon, S. I.; Lee, J. H.; Andrade, J. D.; De Gennes, P. G. J. Colloid Interface Sci. 1991, 142, 149–158. (12) Arakawa, T.; Timasheff, S. N. Biochemistry 1985, 24, 6756–6762. (13) Hermans, J. J. Chem. Phys. 1982, 77, 2193–2203. (14) Nolan, J. P.; Lauer, S.; Prossnitz, E. R.; Sklar, L. A. Drug Discovery Today 1999, 4, 173–180. (15) Miller, C. R.; Vogel, R.; Surawski, P. P. T.; Corrie, S. R.; R€uhmann, A.; Trau, M. Chem. Commun. 2005, 4783–4785.
Published on Web 12/01/2010
Langmuir 2011, 27(1), 296–303
Xue et al.
Article
techniques. The conformational properties of the peptides free in solution, covalently linked to PEG-11 (O-(amino ethyl)-O0 -2carboxyethyl)undecaethylene glycol) or tethered to a 2-dimensional surface via a (PEG-11)3 spacer designed to mimic the experimental system of Trau and colleagues,15 have been examined. In particular, work focused on the analysis of differences in the relative populations of specific conformations adopted by the peptides in the different environments. The effect of the size of the PEG spacer on the accessibility of the peptide and the effect of the peptide termini on the conformational properties of the peptides were also examined.
Materials and Methods Simulations. All MD simulations were performed using the GROMACS (Groningen Machine for Chemical Simulation) package, version 3.2.1,16 using the GROMOS 53A6 force field17 for peptides which has been specifically parametrized to reproduce both the structural and thermodynamic properties of peptides in polar and weakly polar environments. The bond lengths, bond angles, and nonbonded Lennard-Jones and Coulomb parameters of the PEG spacer were taken from the GROMOS 53A6 force field.17 The PEG torsion parameters were based on the ab initio calculations of Anderson and Wilson.18 The simple point charge (SPC) water model19 was used to describe the solvent water. All simulations were performed under periodic boundary conditions. The dimensions of the boxes were chosen such that the minimum distance to the box wall was 1.0 nm in the case of the molecules free in solution and 0.9 nm for the systems constrained to a 2-dimensional lattice. The molecules free in solution were simulated in a dodecahedral box, which is the optimal box geometry for a periodic spherically symmetric system, while the systems constrained to a 2-dimensional lattice were simulated in a rectangular box. The nonbonded interactions were evaluated using a twin-range method. Interactions within the short-range cutoff of 0.9 nm were updated every step. Interactions within the long-range cutoff of 1.4 nm were updated every 5 steps together with the pair list. To minimize the effect of truncating the electrostatic interactions beyond the 1.4 nm long-range cutoff, a reaction field correction was applied using a relative dielectric constant of εr = 78.20 No counterions were included as is appropriate for a system at low salt concentration where the statistical probability of finding any counterions in the volume simulated becomes negligible. The LINCS algorithm21 was used to constrain the lengths of the covalent bonds in the peptide and PEG. The SETTLE algorithm22 was used to constrain the geometry of the water molecules. In order to extend the time scale that could be simulated, explicit hydrogen atoms in the peptide were replaced with dummy atoms. The forces acting on the hydrogens were calculated every step and projected onto the heavy atoms to which they were attached. The new positions of the hydrogens are then recalculated each step based on the positions of these heavy atoms. In essence, the hydrogen and heavy atom are propagated as a single unit. This eliminates high-frequency degrees of freedom associated with the bond angle vibrations involving hydrogens, allowing a time step of 4 fs to be used to integrate the equations of motion without affecting thermodynamic properties of the system significantly.23 The simulations were carried (16) Lindahl, E.; Hess, B.; van der Spoel, D. J. Mol. Model. 2001, 7, 306–317. (17) Oostenbrink, C.; Villa, A.; Mark, A. E.; van Gunsteren, W. F. J. Comput. Chem. 2004, 25, 1656–1676. (18) Anderson, P. M.; Wilson, M. R. Mol. Phys. 2005, 103, 89–97. (19) Berendsen, H. J. C.; Postma, J. P. M.; van Gunsteren, W. F.; Hermans, J. In Intermolecular Forces; Reidel: Dordrecht, 1981. (20) Tironi, I. G.; Sperb, R.; Smith, P. E.; van Gunsteren, W. F. J. Chem. Phys. 1995, 102, 5451–5459. (21) Hess, B.; Bekker, H.; Berendsen, H. J. C.; Fraaije, J. G. E. M. J. Comput. Chem. 1997, 18, 1463–1472. (22) Miyamoto, S.; Kollman, P. A. J. Comput. Chem. 1992, 13, 952–962. (23) Feenstra, K. A.; Hess, B.; Berendsen, H. J. C. J. Comput. Chem. 1999, 20, 786–798.
Langmuir 2011, 27(1), 296–303
out in the NPT-ensemble at T = 298 K and P = 1 bar. The temperature and pressure were maintained close to the reference values by weakly coupling the system to an external temperature and pressure bath using a relaxation time constant of 0.1 ps24 and 0.5 ps, respectively. The pressure coupling was isotropic. Data were collected every 200 ps for analysis. Images were produced using VMD.25 Systems. To determine the effect of the PEG spacer on the conformational properties of small peptides, a set of five peptides with differing chemical and physical properties were examined. The peptides were simulated under three different conditions: (1) free in solution, (2) attached to a PEG spacer, and (3) attached to PEG spacer constrained to a 2-dimensional lattice to mimic the display of a peptide on the surface of a microsphere. The five peptides were YGSLPQ (of mixed physical-chemical properties), VFVVFV (nonpolar), GSGGSG (polar), EEGEEG (negatively charged), and KKGKKG (positively charged). The peptide YGSLPQ is an encephalitogenic peptide corresponding to the conserved residues 68-74 of myelin basic protein26 and recognized by the antibody MAB384 (MBP: Chemicon International). The peptide GSGGSG is a commonly used flexible linker peptide.27,28 The sequence EEGEEG is derived from the glutamation of the R-tubulin, an important post-translational modification promoting the stability of microtubules.29,30 It has been used as a minimal system to study substrate specificity in tubulin ligases.31 The peptide KKGKKG is derived from the bipartite nuclear localization signal sequence of HMG proteins which are actively transported into the nucleus.32 The peptide YGSLPQ was used to examine the effects of capping the N- and C-termini of the peptide on its conformational properties free in solution prior to PEGylation. Simulations of YGSLPQ were performed with both termini charged (NH3þYGSLPQ-COO-; system I); with the N-terminus charged and the C-terminus capped by a neutral methylamino group, as occurs during PEGylation (NH3þ-YGSLPQ-CONH-CH3; system II); and finally with the C-terminus capped by a neutral methylamino group and the N-terminus acetylated (Ac-YGSLPQCONH-CH3; system III). The remaining four peptides were simulated in solution with the N-terminus charged and the C-terminus capped by a neutral methylamino group, i.e., NH3þVFVVFV-CONH-CH3 (system IV), NH3þ-GSGGSG-CONHCH3 (system V), NH3þ-EEGEEG-CONH-CH3 (system VI), and NH3þ-KKGKKG-CONH-CH3 (system VII). In each case, the initial structure was fully extended and the dimensions of the box were chosen so as to ensure that the peptide could not interact directly with its periodic image. Two alterative PEG spacers were examined: PEG-11 (O-(aminoethyl)-O0 -2-carboxyethyl)undecaethylene glycol) and (PEG-11)3, which consisted of three covalently linked repeats of PEG-11. The spacers chosen were the same as those used by Trau and co-workers.33 To determine the conformational preference of the spacer itself, PEG-11 was simulated free in solution (PEG-11; system VIII). To determine the effect of the peptide on the (24) Berendsen, H. J. C.; Postma, J. P. M.; van Gunsteren, W. F.; Dinola, A.; Haak, J. R. J. Chem. Phys. 1984, 81, 3684–3690. (25) Humphrey, W.; Dallke, A. J. Mol. Graphics Modell. 1996, 14, 33–38. (26) Mannie, M. D.; Paterson, P. Y.; U’Prichard, D. C.; Flouret, G. Proc. Natl. Acad. Sci. U.S.A. 1985, 82, 5515–5519. (27) Pryor, P. R.; Jackson, L.; Gray, S. R.; Edeling, M. A.; Thompson, A.; Sanderson, C. M.; Evans, P. R.; Owen, D. J.; Luzio, J. P. Cell 2008, 34, 817–827. (28) Wu, C.-H.; Balasubramanian, W. R.; Ko, Y.-P.; Hsu, G.; Chang, S.-E.; Prijovich, Z. M.; Chen, K.-C.; Roffler, S. R. Biotechnol. Appl. Biochem. 2004, 39, 1–7. (29) Ikegami, K.; Heier, R. L.; Taruishi, M.; Takagi, H.; Mukai, M.; Shimma, S.; Taira, S.; Hatanaka, K.; Morone, N.; Yao, I.; Campbell, P. K.; Yuasa, S.; Janke, C.; MacGregor, G. R.; Setou, M. Proc. Natl. Acad. Sci. U.S.A. 2007, 104, 3213–3218. (30) Kuriyama, R.; Levin, A.; Nelson, D.; Madl, J.; Frankfurter, A.; Kimble, M. Cell Motil. Cytoskeleton 1995, 30, 171–182. (31) Rudiger, M.; Wehland, J.; Weber, K. Eur. J. Biochem. 1994, 220, 309–320. (32) Hock, R.; Scheer, U.; Bustin, M. J. Cell Biol. 1998, 143, 1427–1436. (33) Surawski, P. P. T.; Battersby, B. J.; Vogel, R.; Lawrie, G.; Trau, M. Mol. Biosyst. 2009, 5, 826–831.
DOI: 10.1021/la103800h
297
Article
Xue et al.
Figure 1. Chemical structure of the peptide YGSLPQ covalently bound to a PEG-11 spacer. conformational properties of the spacer, the peptide YGSLPQ attached to a PEG-11 spacer was also simulated free in solution (NH3þ-YGSLPQ-CONH-C2H4-PEG-11; system IX). To generate the peptide-spacer complex, the C-terminus of the peptide was covalently bound at the N-terminus of the PEG spacer, as shown in Figure 1. The remainder of the systems simulated consisted of the peptides covalently linked to either a PEG-11 or (PEG-11)3 spacer constrained to a 2-dimensional array. These systems were designed to mimic peptides attached a solid support via a PEG linker as used in microarray or microsphere technologies.34,35 Note: the microspheres that these simulations aim to mimic have been shown experimentally to have very low surface roughness and low microporosity. They are thus appropriately represented on an atomic scale as a planar surface. The distance between adjacent peptides was set to 1.8 nm, which corresponded to a packing density of one strand per 3.24 nm2. This was equivalent to the average packing density determined by Trau and co-workers.34,35 Experimentally, two alternative PEG linkers were used by Trau and co-workers:34 a PEG-11 spacer and a (PEG-11)3 spacer, the longer spacer being more effective. To determine the effect of the length of the spacer on the behavior of the peptide, the peptide YGSLPQ linked to a PEG-11 spacer was simulated in a 3 3 array, NH3þ-YGSLPQ-CONH-C2H4-PEG-11 (system X), and attached to a (PEG-11)3 spacer in a 6 6 array, NH3þ-YGSLPQCONH-C2H4-(PEG-11)3 (system XI). The remaining four peptides were simulated as 6 6 arrays with the peptides covalently bound to a (PEG-11)3 spacer, namely, NH3þ-VFVVFV-CONH-C2H4(PEG-11)3 (system XII), NH3þ-GSGGSG-CONH-C2H4-(PEG11)3 (system XIII), NH3þ-EEGEEG-CONH-C2H4-(PEG-11)3 (system XIV), and NH3þ-KKGKKG-CONH-C2H4-(PEG-11)3 (system XV). In systems X-XV the last three carbon atoms of the ethylene glycol linker were harmonically restrained (K = 5000 kJ mol-1 nm-2) to a square-planar lattice, as illustrated in Figure 2. In addition, water molecules lying in the plane between of the terminal carbons were also restrained to mimic a solid surface. Note: in the chemical procedure used by Trau and coworkers only the terminal atom of the ethylene glycol is coupled to the organosilica microsphere surface.34,35 In the simulations the positions of several atoms were constrained so as to mimic better a solid surface below the point of attachment. We would also note that given to the length of the linker precisely which atoms were restrained would have little effect on the conformation of the peptide. An overview of the systems simulated is given in Table 1. Analysis. Backbone Conformation. The backbone conformations of the peptide were analyzed in terms of the distribution of j and ψ dihedrals36 of the central four residues. The allowed regions for the non-glycine residues were taken to be: region I, -180 < j < 30 and -80 < ψ < 180; region II, -30 < j < 90 and -10 < ψ < 120; region III: -180 < j < 30 and -180 < ψ < 150.37 (34) Kozak, D.; Surawskiaa, P. P. T.; Thoren, K. M.; Lu, C.-Y.; Marcon, L.; Trau, M. Biomacromolecules 2009, 10, 360–365. (35) Chen, A.; Kozak, D.; Battersby, B. J.; Forrest, R. M.; Scholler, N.; Urban, N.; Trau, M. Langmuir 2009, 25, 13510–13515. (36) van Gunsteren, W. F.; Daura, X.; Mark, A. E. Helv. Chim. Acta 2002, 85, 3113–3129. (37) Kabsch, W.; Sander, C. Biopolym.;Pept. Sci. Sect. 1983, 22, 2577–2637.
298 DOI: 10.1021/la103800h
Figure 2. Illustration of the model used to simulate the behavior of a 2-dimensional peptide display system. (a, b) Top and side views of the initial configuration of system XI consisting of a 6 6 array of the peptide YGSLPQ (purple) covalently attached to (PEG-11)3 (cyan) surrounded by periodic images (gray) of the central system. (c, d) Top and side views of system XI after 12.5 ns of simulation. Note the last three carbon atoms of the ethylene glycol were harmonically restrained in the x-y plane. The terminal carbon in each case is represented as an orange sphere.
Analysis of Side-Chain Conformations. The sampling of side-chain conformations was analyzed by the comparing the free energy distributions around the χ1 torsion angle Ni-CRi-Cβi-Cγi for the side chains of Tyr-1, Ser-3, Leu-4, Pro-5, and Gln-6. The χ1 angles were binned using 10 windows, and the relative free energy as a function of χ1 was estimated as ΔG(χ) = -kBT ln[NA(χ)/N(χ)], where NA(χ) is the number of samples within a given window and N is the total number of samples in all windows, kB is Boltzmann’s constant, and T is the temperature (298 K). Cluster Analysis. To determine the relative populations of specific conformations, the trajectories were clustered using the method of Daura et al.38,39 First, a matrix of the positional rootmean-square deviation (rmsd) between all conformations was constructed. The conformation with the most neighbors within a specified cutoff was then determined. This structure (the center or representative configuration of the first cluster), together with all of its neighbors, was then removed from the ensemble, and the procedure was repeated to obtain the second and higher clusters until the set of structures was empty. In this work, two conformations were considered neighbors if the backbone between the two conformations rmsd was
