5702
J. Phys. Chem. B 2008, 112, 5702-5709
Stabilizing Helical Polyalanine Peptides with Negative Polarity or Charge: Capping with Cysteine Silvya Oommachen, Jianhua Ren, and C. Michael McCallum* Department of Chemistry, UniVersity of the Pacific, 3601 Pacific AVenue, Stockton, California 95211 ReceiVed: April 30, 2007; In Final Form: January 20, 2008
Alanine-based peptides are widely known for their propensity to form helices, whether in the gas phase or in aqueous solution. Interactions of substituent groups or peptides with the helical macrodipole may either encourage or discourage the formation or stability of a helix, depending upon the placement of these groups. We report the first study of the inductive stabilization of a number of peptides through electronegative or anionic N-terminal residue capping. Using Charmm27/CMAP equilibrium and replica-exchange (REX) molecular dynamics (MD) simulations with Generalized Born implicit solvation methods, we find that the N-terminal cysteine capping of alanine peptides strongly enhances the helicity, even allowing the helical moiety to remain at temperatures beyond the denaturing temperature. Though the overall number of hydrogen bonds is enhanced, this stabilization seems to occur indirectly through interaction with the helical macrodipole rather than as a direct result of hydrogen bonding involving the cysteine, though the nature of the hydrogen bonding changes.
Introduction The ubiquitous helical secondary structures include R-, 310-, and rarely π-helices.1,2 The former and latter helical moieties may also be described as 413- and 516-helices, constituent with the 310 label, which describes how many residues (i, i + n) and atoms are involved.3,4 It is well known that R-helices are the main form found in natural proteins, usually spanning ten or more residues, whereas 310-helices tend to cap R-helices and are much shorter;5,6 the function of π-helices is not well understood.7,8 Other statistical information such as which residues show a propensity to populate certain helices or the location within the helix has been compiled.9-11 Exactly how helices matter in the process of protein folding itself has been intensely studied.12-17 Understanding the dynamics of helix formation should lead to greater insight in the protein-folding problem. Accepted theory18,19 describes R-helices as small polar fragments cooperating to form an overall macrodipole, running from the C-terminal end to the N-terminal end, with an overall strength of approximately 3.5 D per residue. These monopolemacrodipole interactions contribute to the overall stabilization of the helix motif. Other helices also benefit from this cooperative alignment of each residue dipole. The 310-helix does not have nearly the same number of geometrically favorable hydrogen bonds as the R-helix,20 which appears to be one of the factors diminishing its population in natural proteins. Although 310- and π-helices have not received the same level of theoretical attention as R-helices, one can say that the essential physical nature of any helix as expressed by the macrodipole should be similar. The dynamics, characterization, and stability of helices has received much attention in recent years, including studies on the structural features and stability of alanine-based peptides.5,21-35 Studies have shown that the helical macrodipole can be exploited to stabilize or destroy helical moiety. Early work included observing the effect on stabilization through varying capping * To whom correspondence should be addressed.
residues,36 while specific enhancement through the addition of an electropositive residue such as a protonated lysine positioned at the C-terminus came somewhat later.21 Other works include interconversion and stability studies,5 ion-mobility and molecular dynamics,22-24 free-energy simulations,25 effects of capping and spacing with different groups,26,27 and investigations of lefthanded zwitterionic λ-helices.28 Later, it was found that placing excess positive charge (as protons) either enhances or interferes with the helical structure, depending on their placement with respect to individual subhelical dipoles.29-31 One recent report indicates that protonation of the N-terminus destroys or severely destabilizes R-helical structure.32 In addition to these structural studies, computational and experimental work has been carried out33-35,37-39 on amino acids and cysteine-capped peptides that show unusually low pKa values, elucidating how the R-helical macrodipole enhances the acidity of the thiol proton. Thus, there exists a large body of work that supports the idea that helical secondary structures may be influenced or enhanced through interaction with charge or polarization. Placement of a negative charge or an electronegative species at the N-terminus should show an enhancement of the helicity through interaction with the macrodipole analogous to the effect of positive charge at the C-terminus. In examining this behavior, cysteine is a logical residue to provide either electronegative or anionic interaction. Cysteine has a low natural abundance in native proteins2 (only tryptophan is less prevalent), and it therefore might be expected to have a disproportionately important functionality when it does appear. Aside from disulfide bridges, which provide connective tissue to a protein, cysteine provides enzymatic activity,40,41 a selective-detection site for glutathione,42 an important role in the absorption of cytochrome P450,43 and a high relative concentration in important active sites.44,45 Cysteine does not show a high propensity to occur at either end of a helix,46,47 and only an average helical propensity itself.48,49 From an engineered standpoint, the placement of cysteine within a native R-helix, together with the question of the nature of the sulfur (either a covalent disulfide bond or an
10.1021/jp073315a CCC: $40.75 © 2008 American Chemical Society Published on Web 04/17/2008
Stabilizing Helical Polyalanine Peptides: Capping with Cysteine anionic salt bridging bond), could be important in determining if a helix is stable and could help determine how the helix forms in the first place. In this article, we discuss the stabilizing effect of cysteine residues on alanine peptides. We focus on polyalanine and polyglycine capped with single cysteine or serine residues (negatively charged or neutral) at the C-terminus or N-terminus or substituted centrally within these peptides. The effects of these substitutions are described and contrasted through overall secondary-structural effects and through individual hydrogen bond locations. The stabilizing effect is found to be robust in both increasing helical content and allowing the helicity to be maintained at elevated temperatures compared to the unsubstituted peptides. The nature of the hydrogen bond capping of each end of the peptides affects the overall structure and thus is investigated in some detail. Some predictions for more realistic models of native peptide chains are made, and future work in this area is laid out. Methods Model Compounds. The polyalanine-based structures included N-terminal cysteine-capped alanine Ac-Cys-Alan and AcCys--Alan, C-terminal cysteine-capped alanine Ac-Alan-Cys and Ac-Alan-Cys-, mid-helix cysteine Ac-Alan-Cys-Alan and AcAlan-Cys--Alan, cysteine-capped polyglycine polymers, Ac-CysGlyn and Ac-Glyn-Cys, and serine-capped peptides Ac-Ser-Alan and Ac-Alan-Ser. In all cases, the C-terminal end of the peptides were amidated, whereas N-termini were parametrized when necessary and constructed with a neutral amine N-terminus. Thiolate cysteine charges used were determined through modeling and were similar to those used by Foloppe via MacKerell.50 Our results were determined through molecular dynamics (MD) and replica-exchange MD (REX) simulations51 using the CHARMM27/CMAP52,53 force field, along with the Generalized Born (GB) implicit solvation method54,55 when it was necessary to include solvent effects. Polar hydrogens (those attached to nitrogen) were parametrized by setting their Born radii to 0.8 Å. The peptide structures were generated as R-helices for the standard MD simulations to observe the hydrogen bonding and equilibrium structural effect; fully extended conformations were used as a starting point for the REX simulations to obtain information on the helical folding (winding) events. The standard leapfrog Verlet algorithm was used and employed a nonbonded cutoff at 12 Å with an electrostatic shift at 12 Å and a van der Waals switching employed between 10 and 12 Å. The lengths of the bonds with hydrogen atoms were kept constant through use of the standard SHAKE algorithm.56 In the standard MD simulations, the minimized base structures were heated over 105 steps with a time step of 1 fs, heating from 50 K to either 210, 310, 400, or 600 K. This was followed by 105 equilibration steps with a time step of 1 fs, followed by statistics segments on the order of 107 steps, also at 1 fs time step. Trajectory coordinates were generally saved every 1000 timesteps with energy and other statistics saved every 100 steps. The REX simulations were performed with the aid of the MMTSB toolset.57 Each REX simulation consisted of 5000 exchange cycles; each cycle consisted of 500 timesteps of 2 fs each. Four temperatures ranging from 250 to 650 K (either 250450 K or 450-650 K) were used to classify the starting configurations (known as “exchange clients”). CHARMM parameters were matched to those in the conventional MD runs. Explicit hydrogen bonding was analyzed using CHARMM’s HBOND and CONTACT routines. The CONTACT routine allows specification of donors and acceptors in a general way,
J. Phys. Chem. B, Vol. 112, No. 18, 2008 5703 which is needed for sulfur-containing hydrogen bonds when they are not defined in the structure files. The hydrogen bonds were defined through the default definitions: maximum hydrogen bond distance of 4.5 Å, maximum out-of-line angle of 90°, the switching function (needed to smooth cutoff transitions) on between 3.5 and 4.0 Å and 50 to 70°. All hydrogen bonds were defined with a minimum 5 ps lifetime. In all simulations, the resulting structures were characterized with the aid of VMD58 and Simulaid.59 Quantum chemical calculations were carried out at a B3LYP/ 6-31G(d) model chemistry,60-63 which has been shown to give good results in these cases.53 Parameter Set Discussion. The CHARMM27/CMAP parameter set was introduced to counter a bias toward π-helices observed with the older CHARMM22 parameter set. Aside from achieving the correct relative stability of R- versus π-helices, there was a marked effect on the sulfur alignment in the cysteine in our MD simulations. In other simulations (not directly reported here) on the systems reported here but using the standard CHARMM22 parameters, there was significant competition between the R- and π-helical structures, often resulting in a strong preference for the π-helix. In addition, preferred structural orientation of the cysteine sulfur showed marked differences between the two parameter sets. These differences required some investigation into the correct representation of cysteine, including parametrization. To determine the relative accuracy of the CHARMM27/CMAP parameter sets with respect to the cysteine/thiolate cysteine orientation, unconstrained geometry optimizations from an R-helical starting structure were made on Cys-Ala2 tripeptides in an R-helical conformer (which have been shown to be the most stable conformer for dialanines53). The results for {-}Cys-Ala-Ala and Cys-Ala-Ala yield S-CB-CR-C(carbonyl) dihedral angles of 58 and 63°, respectively. Overall, simulations using the CHARMM27/CMAP parameters show much better agreement with the QC calculations. In the simulations, the thiolate cysteine sulfur dihedral measurements yield a relatively narrow distribution, as is expected from the higher charge density, and also exhibits three distinct peaks. These occur just below 57, at 59, and at nearly 61°, corresponding to the different possible hydrogen bond arrangements from the donors on the Cys. At the smaller of the three angles, the thiolate sulfur forms hydrogen bonds with the neighboring Ala HN (an (i, i + 1) hydrogen bond). At larger angles, the prevalent hydrogen bond from the cysteine cap originates from the acetyl oxygen, binding to the HN on the third alanine (an i, i + 3, or 310 hydrogen bond). When this moiety is established, the Cys sulfur may form the (i, i + 1) hydrogen bond, yielding the dihedral around 59°; otherwise, in the absence of this hydrogen bond a dihedral broadly centered at 62° is found. Results and Discussion Polyalanine Hydrogen-Bonding Patterns and Helical Structure. (Homo)polyalanine peptides are often described as having a high propensity for helices, even in vacuum.64,65 In helical peptides, the total number of possible hydrogen bonds approaches the number of residues, excluding only the end residues (as hydrogen bonds that support the helical structure must point in from the ends). Generally, the fraction of hydrogen bonding observed is much less than this maximum, and this smaller number is sufficient to maintain the helix. Whether these hydrogen bonds are spread out evenly over the whole helix, or whether certain pairs of residues tend to be strongly hydrogen bonded over relatively long time scales can depend on the nature
5704 J. Phys. Chem. B, Vol. 112, No. 18, 2008
Figure 1. Alanine fractional hydrogen bond time series from simulations of Ac-Ala15 from equilibrium MD simulation (107 MD steps at 310 K with GB implicit solvation). Top, total helical content; middle, R-helical content; bottom, 310-helical content. ) 1 (black line) and ) 78.5 (red line) for all plots. Each point is an average over 1000 MD time steps or 10 output steps.
of the solvent, the polarity of the side chains, and the presence of any charged groups. In the case of polyalanine, the nature of the helicity as defined through hydrogen bonds is described in Figure 1. In a vacuum situation (obtained here by setting the implicit GB solvent to ) 1.0), a polyalanine peptide has a relatively low degree of helicity, consisting largely of 310 and a roughly equal amount of R-helical moiety. The peptide appears to be helical overall, although the variation in hydrogen bond number is wide, and the overall hydrogen bond fraction remains close to 0.25. However, in an implicit GB solvent of ) 78.5 (corresponding water) the degree of helicity increases solely due to the increase in R-helical hydrogen bonds. As the R-helix has a larger diameter than the 310, it could be surmised that this change in helical preference arises from the hydrophobic interaction of the methyl groups with the implicit water solvent, thus driving these groups toward the interior of the helix. The R-helix, having a larger cross-sectional area, would be able to accommodate this slight change better than the 310-helix. However, this is not seen in our simulations; the position of the methyl sidechains is mostly unchanged as the implicit solvent is changed from hydrophobic ( ) 1.0) to hydrophilic ( ) 80.0). Effect of Cysteine Capping on Polyalanine Secondary Structure. In the case of helical enhancement of polyalanines, it is necessary to work with smaller lengths and investigate the hydrogen-bonding patterns along the barrel of the helix as well as the capping hydrogen bonds. Homopolyalanines beyond a length of 10 residues exhibit such a high degree of helicity that it is difficult to quantify any increases in this inherent stability. Each succeeding full turn (three to five residues) of a helix adds stability, both directly through (i, i + 4) hydrogen bonding and indirectly through the lengthening of the helical macrodipole. Possible solutions include the major structural change of removing the side-chain methyl groups from the outside of the helix, but this is not energetically favored unless the peptide is in a polar solvent environment. Thus, when measuring changes in the gas phase changes may be subtle when attempting to enhance helicity by capping. Destabilizing the helix through charge opposition (in this case, capping with a cysteine at the C-terminus) can be readily observed, even in the gas phase. To investigate any enhancement on polyalanine in the gas phase, it is useful to investigate short chains, as any enhancement either through the co-operational helical dipole or charge-dipole interaction will have a greater influence when only one- or two-
Oommachen et al.
Figure 2. Fractional hydrogen bond time series for n ) 8 alanine peptides from equilibrium MD simulation (107 MD steps at 310 K with ) 1.0). Top, total helical content; middle, R-helical content; bottom, 310-helical content. Cysteine-capped Ac-Cys-Ala7 (red line), serinecapped Ac-Ser-Ala7 (black dashed line), and uncapped Ac-Ala8 (black solid line). Each point is an average over 1000 MD time steps or 10 output steps.
Figure 3. Hydrogen bond time series for n ) 10 alanine peptides from equilibrium MD simulation (107 MD steps at 310 K with ) 1.0). Top, total helical content; middle, R-helical content; bottom, 310-helical content. Cysteine-capped Ac-Cys-Ala9 (red line) and uncapped AcAla10 (black solid line). Each point is an average over 1000 MD time steps or 10 output steps.
full helical turns are possible. Verification is possible (1) through counting actual hydrogen bonds (as defined above) from trajectories and (2) through structural determination programs such as DSSP.66 Both methods are used here. To this end, we have examined peptide chains with n ) 8 and n ) 10, including pure alanine (Ac-Alan), pure glycine (Ac-Glyn), cysteine-capped alanine (Ac-Cys-Alan) and glycine (Ac-Cys-Glyn), as well as serine-capped alanine chains (Ac-Ser-Alan) through gas-phase MD simulations. We observe dramatic differences between these model peptides: an enhancement when cysteine caps the chain on the N-terminus and no significant change when either serine caps polyalanine or when cysteine caps polyglycine. Figures 2 and 3 illustrate these differences. The cysteinecapped polyalanines have much more helical content compared with both the serine-capped and uncapped polyalanine peptides. This increase in helicity is due solely to an increase in (i, i + 4), R-helical, hydrogen bonds. The net result of N-terminal cysteine capping of polyalanine is to strongly increase the R-helicity in the gas phase, where polyalanine normally does not show such strong R-helicity. Polyglycine is a homopeptide that generally shows little propensity to form helices, though even a few alanine residues can change this by forming local helical structure.67 However, we found no inducement of helicity
Stabilizing Helical Polyalanine Peptides: Capping with Cysteine
J. Phys. Chem. B, Vol. 112, No. 18, 2008 5705
Figure 4. Evolution of secondary structure for the n ) 10 alanine peptides (as in Figure 3). Top, Ac-Cys-Ala9; bottom, Ac-Ala10. Pink, R-helix; blue, 310-helix; red, π-helix; cyan, turn. N-terminus of each peptide is at the top of each 2D time series.
in polyglycine when either cysteine or serine is used as N-terminal caps on homopolyglycine peptides. The uncapped peptides have disjointed groups of residues with significant 310-helical content and in the case of the n ) 10 polyalanine significant, R-helical content, but the overall moiety is not a stable helix. The periodic nature of the helical units in these uncapped polyalanines prevent them from forming a stabilized helix for their full length. To begin to understand how each specific residue is engaged in particular hydrogen bonds, Figure 4 presents a DSSP time series of n ) 10 peptides. From this figure, the location of the different helical moiety may be found; in the homopolyalanine, the 310-helical content is largely found at the C-terminus, capping or replacing the R-helical content, which begins only from the third alanine from the N-terminus. Conversely, the cysteine-capped peptide exhibits R-helical structure throughout its length, often originating with the cysteine itself. What cannot be seen in the DSSP representation is the larger fluctuations observed in the cysteine-capped polyalanine. Average Residue-to-Residue Hydrogen-Bonding Patterns. The DSSP analysis provides a good overview of how the secondary structure of the peptide changes through a time trajectory but remains somewhat qualitative. It is useful to examine the hydrogen-bonding patterns in more detail for these nearly homogeneous peptides. Hydrogen bonds in native helices (for example in proteins) show much variation along the length of the helix; typical occupations as measured through simulation of short miniproteins are near or below 50%,68-70 as only a fraction of all possible (i, i + 4) hydrogen bonds are necessary along the helix barrel to keep a R-helix structure stable over long times. This transient behavior is less evident in the gas phase, as the lack of solvent makes hydrogen bonds relatively stronger than in the condensed phase. Thus, although the helical hydrogen bond fraction remains about the same, the distribution of hydrogen bonds among possible pairs of residues may remain more fixed. The time-averaged per-residue helical contribution extracted from equilibrium MD trajectories for cysteine-capped and uncapped polyalanines supports this prediction. The timeaveraged per-residue hydrogen bonds in homopolyalanine, shown in Figures 5 and 6 are mostly “frozen” in certain positions due to the lack of a screening solvent. For example, in AcAla8, the helical hydrogen bonds are substantially fixed at
Figure 5. Per-residue average hydrogen bonding from equilibrium MD simulation, implicit gas phase for n ) 8 peptides, averaged over 1 × 107 1 fs timesteps, GB ) 2.0, 310 K. Solid lines, Ac-Cys-Ala7; broken lines, Ac-Ala8. Representative snapshot of each peptide shown for clarity.
Figure 6. Per-residue average hydrogen bonding from equilibrium MD simulation, implicit gas phase for n ) 10 peptides, averaged over 1 × 107 1 fs timesteps, GB ) 2.0, 310 K. Solid lines, Ac-Cys-Ala9; broken lines, Ac-Ala10.
residues 1, 4, 7, and 8. Though some size effects may be inferred from these distributions, it is clear that the cysteine-capped polyalanines have a smoother distribution of hydrogen bonds across the whole length of the helix with each residue contributing more or less equally to the hydrogen bond distribution.
5706 J. Phys. Chem. B, Vol. 112, No. 18, 2008
Figure 7. Per-residue average hydrogen bonding from equilibrium MD simulation, implicit GB solvent ( ) 78.5). Solid lines represent AcCys-Ala15 peptide, dotted lines represents Ac-Ala15 peptide.
The mean lifetime of hydrogen bonds compiled from the equilibrium MD simulations also show some differences. For hydrogen bonds with significant populations (above 0.05), the mean lifetime for the Ac-Cys-Ala7 and Ac-Cys-Ala9 peptides averaged 8.4 ps with a range of 6.9 to 10.5 ps for (i, i + 4)/R and 5.9 ps with a range of 5.0 to 7.4 ps for (i, i + 3)/310 hydrogen bonds. For the Ac-Ala8 and Ac-Ala10 peptides, the averages are skewed by the extremely long lifetimes for certain residue pairs: 16.1, 78.3, and even 330 ps for the (i, i + 4)/R hydrogen bond between the first and fourth residues in AcAla8. There are two main definitions of a hydrogen bond lifetime possible:71,72 continuous (where a hydrogen bond must exist for the entire checking interval ∆t) or interrupted (where a hydrogen bond is allowed to break and reform). We have used the default CHARMM definition, which is the continuous form. Although the interrupted form appears to be more satisfactory, our results are used only in a relative comparison. The intermittant hydrogen bond correlation method73 will be used in the future for more relevant comparisons. Other observations from the MD simulations include differences in hydrogen bond angle and length. The cysteine-capped polyalanines had average hydrogen bond geometries of 155° with a O‚‚‚H length of close to 2.0 Å for R-hydrogen bonds, whereas the 310-hydrogen bonds averaged 142° with a O‚‚‚H distance of 2.2 Å or more. However, in the homopolyalanine peptides, the hydrogen bond O‚‚‚H distances generally followed the same trend in which the angles did not show a clear difference. The comparison of these different peptide lengths allows the inference that the freezing of the hydrogen bonds in the gas phase is not due to the length of the peptide forcing certain hydrogen bond patterns. Interestingly, these differences are somewhat minimized when the peptides are placed in solvent. In implicit water solvent ( ) 78.5), the time-averaged hydrogen-bonding pattern shows a smoother distribution for both peptides. Even in peptides this short, the hydrogen bonds are spreading out along the length of the helix barrel, as the solvent provides some screening, presumably allowing more flexibility in the hydrogen bond distribution. In all these figures, it is notable that the cysteinecapped polyalanine adds a small amount of favorable hydrogen bonding at the N-terminus, both in the form of 310 (i, i + 3) and R (i, i + 4) connections. The amount of added (i, i + 4) hydrogen bonds is actually smaller than at the corresponding N-terminus of the homopolyalanine. Replica Exchange Simulations. The dynamic evolution of the helical structures can be efficiently studied through replicaexchange (REX) MD simulations.51 Whereas usually applied
Oommachen et al.
Figure 8. Time series of the measured RMSD from R-helix for the cysteine-capped alanine eightmer Ac-Cys-Ala7 and homopolyalanine eightmer Ac-Ala8, T ) 450-650 K. Lowest-scoring cysteine-capped replica in black (35.98% at 450 K); lowest-scoring homopolyalanine replica in red (38.48% at 450 K). Most common secondary structure as determined by DSS procedure listed also. Final 2000 cycles of 5000 shown.
to the folding of peptides and proteins, the REX method can also describe the stability of the polyalanines at high temperatures. Running REX simulations at relatively high starting temperatures, such as 450 K, and comparing calculated rootmean squared deviations (RMSDs) from a reference helix can give information on the overall stability and stable lifetimes of model structures. REX simulations are organized into separate client structures, which are set up to equilibrate and run at different starting temperatures, logarithmically separated from low to high. Each replica may be ranked according to how much time it spends at the lowest temperature, either through a percentage or a score weighted by temperature. Thus a score of 1.00 would be given to a replica that spends 100% of its time at the lowest temperature. Gas-phase REX simulations running from 450 to 650 K were performed for n ) 8 peptides: homopolyalanine, cysteinecapped polyalanine, and thiolate-cysteine-capped polyalanine; these higher temperatures were chosen to test the stability of the capped peptides relative to the polyalanine. The results of each of these simulations were quite different, though the lowestscoring (best) result of each simulation was generally the same R-helical secondary structure. A plot of the RMSD versus the corresponding R-helical structure of the lowest-scoring replicas of neutral eightmers (cysteine-capped and homopolyalanine) of the peptide is given in Figure 8. The two lowest-scoring cysteine-capped replicas account for over 70% of the simulation time spent at the lowest temperature of 450 K. Each of these two replicas spends considerable time in an R-helical conformation, as determined by the RMSD measurements, with an average mean lifetime of the helix of 1000 REX cycles (50 000 MD steps or 100 ps). In comparison, the two lowest-scoring homopolyalanine replicas spend 66% of the simulation time at 450 K, and the mean lifetime of the helical structure is much less on the order of 200-500 REX cycles (10 000 to 25 000 MD steps or 20-50 ps). For folding pathways that are not highly dimensional, transition-state theory may be applied to quantify the apparent differences in helical (“folded”) lifetimes. To that end, a helicaldenatured time correlation function has been calculated as68,74,75
C(t) ) C(f(t)) )
〈f(0)f(t)〉 - 〈f 〉2 〈f 2〉 - 〈f 〉2
(1)
Stabilizing Helical Polyalanine Peptides: Capping with Cysteine
Figure 9. Normalized helix correlation function C(t) ) C(f(t)) with f(t) ) 1 for a helix RMSD e 1.5 Å and f(t) ) 0 for a helix RMSD > 1.5 Å. Averages over all replicas of that type, as well as time. Dashed lines correspond to a best fit to an exponential decay. Relative folded lifetimes derived from exponential decay constants τH for each polyalanine are indicated (see text).
where the angled brackets denote both time-averaging as well as an average over all replicas of a given molecule. Here, we define the two-state function f(t) as
f(t) )
{
1, RMSD e 1.5Å 0, RMSD > 1.5Å
Figure 10. Time series of the measured RMSD from R-helix for the thiolate cysteine-capped alanine eightmer, Ac-{-}Cys-Ala7, T ) 450650 K. Lowest-scoring replica in black (44.11% at 450 K); next-lowestscoring replica in red (34.73% at 450 K). The third-ranked replica is in blue; this had a rank identical to the second-ranked replica, 21.43% at 450 K. For the two best replicas, the most common secondary structure as determined by DSS procedure listed alongside data. First 2000 cycles of 3000 shown.
(2)
This yields an approximately decaying-exponential behavior for C(t), with overall helix-to-denatured relaxation times τ of 289 fs (cysteine-capped) and 86.2 fs (homopolyalanine); see Figure 9. (The RMSD cutoff is arbitrary and is meant to provide a reasonably clear cutoff for a “folded” or helical peptide; we found that using cutoffs between 1.0 and 2.0 Å did not change the following calculations to any great extent.) Extracting the relative lifetime of the helical state requires the use of detailed balance.74 First, the overall relaxation time is related to the helical and denatured lifetimes through
τ-1 ) τH + τd
J. Phys. Chem. B, Vol. 112, No. 18, 2008 5707
(3)
where τH ) τHfd is the helix-to-denatured lifetime (and viceversa). Given the number fractions of helical and denatured replicas at any give time step (xH and xd, respectively), the lifetimes must obey the steady-state number fraction condition τH/τd ) xH/(1 - xd), and these fractions are found from the simulation itself. While both number fractions are found to be similar (0.757 for the cysteine-capped and 0.728 for the homopolyalanine), the differing relaxation times for each polyalanine simulation yield markedly different helix lifetimes τH of 1190 and 317 fs for the capped and uncapped helices, respectively. Thus, there is some numerical evidence for the qualitative behavior shown in Figure 8. It should be noted that these lifetimes are relative, and they should not be expected to correspond to experimentally observed winding/unwinding times. In contrast to the neutral peptides, the lowest-scoring thiolate cysteine-capped peptide (Figure 10) showed very little propensity to change its R-helical moiety over the course of the simulation. On the other hand, the second-best replica formed a hairpin moiety, stabilized by the charge on the sulfur. As might be expected, once formed in this way the hydrogen bonds are difficult to break in the gas phase. The thiolate cysteine-capped peptide does show a distinct curvature in the helix, which is not found in the neutral peptides at all. This makes the formation of a hairpin structure more likely.
Figure 11. Optimized geometries of (a) Ac-{-}Cys-Ala2 and (b) AcCys-Ala2. Bond lengths are given in angstroms and bond angles are given in degrees. Yellow, S; gray, C, blue, N, white, H.
Quantum Chemistry and Capping Hydrogen Bonds. Recently, Woo reported electronic structure calculations on cysteine thiolate hydrogen bonding.35 We have made similar observations using B3LYP/6-31G(d) on the tripeptides {-}Cys-Ala-Ala and Cys-Ala-Ala (Figure 11). Notably, we find that the thiolate sulfur engages in two hydrogen bonds with the nearest H-N hydrogens, forming five- and six-membered hydrogen-bonded rings. Both of these hydrogen bonds are somewhat long (2.30 and 2.37 Å) and can be considered a “local” effect (in contrast to any backbone hydrogen bonding that may occur). Together these capping hydrogen bonds seem to supply enough favorable energy to keep the thiolate sulfur bound. The remaining backbone hydrogen bonds in both tripeptides occur between the carbonyl oxygen and the HN proton with the second HN proton coming from the amide terminal group. These may be described as 310-hydrogen bonds; these are the only helical hydrogen bonds available to the tripeptides. These hydrogen bonds are shorter than the thiolate sulfur-containing hydrogen bond: 2.08 Å, 157.4° from the oxygen on the first (cysteine) residue and 2.02 Å, 162.8° from the oxygen on the second (alanine) residue. The neutral cysteine-capped alanine peptide shows no hydrogen bonding through the sulfur at all, and its backbone hydrogen bonds are weaker: 2.15 Å and 125.6° from the cysteine oxygen and 2.29 Å and 158.2° from the second (alanine) oxygen. The situation observed from MD simulations is somewhat different due mainly to the length of these peptides. When hydrogen bonding is found with the thiolate sulfur, the effect
5708 J. Phys. Chem. B, Vol. 112, No. 18, 2008
Figure 12. Comparison of the capping behavior of thiolate Cys. Left: generated and minimized R-helical Ac-{-}Cys-Ala14 peptide. Right: representative equilibrated snapshot from gas-phase MD simulation. Pertinent hydrogen bonds are marked with dotted lines.
Figure 13. Comparison of the capping behavior polyalanine (left), cysteine-capped polyalanine (center), and thiolate-cysteine-capped polyalanine (right). Colors are the same as in Figure 5.
is again to hydrogen bond with the i + 1 residue through the HN hydrogen. In the dynamic simulations, this markedly changes the capping of the R-helix, as it allows the cap to maintain the apparent (i, i + 4) nature rather than the usual 310-hydrogen bond that a homopolyalanine would form at this end of the helix. Because the thiolate sulfur strongly competes with oxygen for the formation of hydrogen bonds, the nature of the helix at the N-terminus is changed compared to homopolyalanine. Whereas there is local benefit by forming these two weak hydrogen bonds at the N-terminus, the overall R-helical pattern is disrupted as each succeeding (i, i + 4) connection is pushed back. This leads to an eventual loss of the R-helical moiety at the C-terminus (see Figure 12). Though it is clear that a neutral cysteine cap is responsible for stabilizing helicity of the homopolyalanines, in contrast with the thiolate cysteine case, the sulfur itself is not directly involved in any hydrogen bonding. Any hydrogen bonding by the neutral cysteine capping comes from the carbonyl and/or acetyl oxygens. These oxygens are oriented in such a way to encourage (i, i + 4) or R-hydrogen bonding. Interestingly, the equivalent N-terminal polyalanine oxygens are involved in more hydrogen bonds on average than in the neutral cysteine but in an (i, i + 3) or 310 fashion. This implies that these N-terminal hydrogen bonds involving oxygen are not fixing the helicity for the entire peptide. If the cysteine was to merely change the pattern of the hydrogen bonding by creating a differing starting point (similar to the effect observed in a thiolate cysteine-capped polyalanine), the effect would be seen as different lengths of peptide (alanine) were added. However, at least up to n ) 15, we see no evidence of such an effect. As seen in the average hydrogen bond fractions by residue (Figures 6-7), the hydrogen bond pattern becomes less granular and more consistent across the whole peptide length. We surmise that this is an effect induced from the properties of the neutral sulfur on cysteine. Serine is clearly similar to cysteine, though the pKa values are much different (8.3 for cysteine versus 13 for serine), which implies some difference in electronegativity. In addition, sulfur is more polarizable than oxygen, which can only be included in the CHARMM force field through the parametrization. The apparent result of this is the ability of cysteine sulfur to interact with the helical macrodipole precisely because it is not involved
Oommachen et al. in hydrogen bonds unlike the serine hydroxy oxygen. The serine oxygen is a much better hydrogen-bond acceptor than the thiol sulfur in cysteine. Again, this allows cysteine to contribute efficiently to (i, i + 4) hydrogen bonding from the N-terminus of the peptide, as both donor (carbonyl oxygen) and acceptors (HN hydrogen) on the cysteine are favorably oriented. Indeed, a serine-capped polyalanine exhibits secondary features much more similar to polyalanine than the corresponding cysteinecapped polyalanine. In our simulations, the hydrogen bond density and overall helicity of Ac-Ser-Alan-1 did not differ significantly from Ac-Alan. Native helices show much more diversity than these simple polyalanines, and more realistic model helical peptides must be considered. There is experimental data on more sophisticated models, such as glutamate-, glycine-, and lysine-substituted alanines,76 and we have begun work studying the cysteinecapping effect on these peptides. There are many examples of cysteine-containing native helices, and some of these are being considered as well. Conclusions We have demonstrated that cysteine is able to enhance the helicity of polyalanine in the gas phase when placed at the N-terminus. This R-helical enhancement occurs without the formation of a hydrogen bond involving the thiol sulfur. In addition, neutral N-terminal cysteine is able to maintain coherent helical moiety at a higher temperature compared with uncapped polyalanine. The mean lifetimes of full helicity as determined through REX simulations are much lower for pure alanine peptides when compared with N-terminal cysteine polyalanine peptides. Neutral cysteine does not have a strong disrupting effect when placed either at the C-terminus or within a larger polyalanine sequence. N-terminal anionic cysteine enhances the overall helicity of polyalanine by forming a strong (i, i + 1) hydrogen bond and creating an overall R-helical hydrogenbonding pattern directly from the N-terminus. N-terminal serine is unable to stabilize polyalanine in the same way. When anionic cysteine is placed at the C-terminus or within a polyalanine sequence, the helical structure is completely or partially disrupted. In the gas phase, the location and number of hydrogen bonds in short polyalanines are nearly fixed with only two or three residues participating substantially in hydrogen bonding. The amount of (i, i + 3) or 310-hydrogen bonds is substantial in polyalanine peptides with a length of ten residues or less. However, when an N-terminal neutral cysteine caps polyalanine peptides of the same length, more residues throughout the primary sequence participate in (i, i + 4) hydrogen bonds, corresponding to an R-helical pattern. Thus, the fluctuations in both hydrogen bond position and number increase, but the overall R-helical moiety is stabilized. Though most of our simulations were performed in vacuum, we are able to report some similar results in implicit (Generalized Born) solvent corresponding to water. In polyalanine peptides capped at the N-terminus by neutral cysteine, we again observe an overall mediation of the hydrogen bond distribution, compared with homopolyalanine. Although the hydrogen bond density at the N-terminus is lessened by the cysteine cap, the overall helicity is lengthened throughout the helix, mostly in the form of added (i, i + 4) or R-hydrogen bonding toward the C-terminus. Cysteine is an example of an amino acid that can have a strong effect on the secondary structure of protein segments. In the goal of fully understanding the dynamics of protein folding,
Stabilizing Helical Polyalanine Peptides: Capping with Cysteine cysteine’s role in stabilization and formation dynamics of helices could help toward this global understanding. Acknowledgment. The authors gratefully acknowledge helpful conversations with Mihaly Mezei (Mount Sinai School of Medicine) and additional help from Ed Lau and Felice Chu Lightstone (Lawrence Livermore National Laboratory). J.R. acknowledges support from the American Chemical Society Petroleum Research Fund. Supporting Information Available: Tables of CHARMM charge parameters, topology file for deprotonated cysteine compounds, and CHARMM patch file for deprotonated cysteine. This material is available free of charge via the Internet at http:// pubs.acs.org. References and Notes (1) Hunter, T.; Pines, J. Cell 1994, 79, 573-582. (2) Voet, D.; Voet, J. G. Biochemistry, 3rd ed.; Wiley: New York, 2005. (3) Stickle, D.; Presta, L.; Dill, K.; Rose, G. J. Mol. Biol. 1992, 226, 1143-1159. (4) Aurora, R.; Creamer, T.; Srinivasan, R.; Rose, G. J. Biol. Chem. 1997, 272, 1413-1416. (5) Topol, I. A.; Burt, S. K.; Deretey, E.; Tang, T.-H.; Perczel, A.; Rashin, A.; Csizmadia, I. G. J. Am. Chem. Soc. 2001, 123, 6054-6060. (6) Wu, Y.-D.; Zhao, Y.-L. J. Am. Chem. Soc. 2001, 123, 5313-5319. (7) Fodje, M.; Al-Karadaghi, S. Protein Eng. 2002, 15, 353-358. (8) Weaver, T. M. Protein Sci. 2000, 9, 201 - 206. (9) Serrano, L. J. Mol. Biol. 1995, 254, 322-33. (10) Bowie, J. J. Mol. Biol. 1997, 272, 780-789. (11) Ulmschneider, M.; Sansom, M. Biochim. Biophys. Acta 2001, 1512, 14. (12) Creighton, T. Biochem. J 1990, 270, 1-16. (13) Dill, K.; Fiebig, K.; Chan, H. Proc. Natl. Acad. Sci. U.S.A. 1993, 90, 1942-1946. (14) Munoz, V.; Eaton, W. A. Proc. Natl. Acad. Sci. U.S.A. 1999, 96, 11311-11316. (15) Onuchic, J.; Wolynes, P. Curr. Opin. Struct. Biol 2004, 14, 7075. (16) Vu, D.; Myers, J.; Oas, T.; Dyer, R. Biochemistry 2004, 43, 35823589. (17) Snow, C.; Sorin, E.; Rhee, Y.; Pande, V. Annu. ReV. Biophys. Biomol. Struct. 2005, 34, 43-69. (18) Hol, W.; van Duijnen, P.; Berendsen, H. Nature 1978, 273, 443446. (19) Park, C.; Goddard, W. A., III J. Phys. Chem. B 2000, 104, 77847789. (20) Tran, T.; Zeng, J.; Treutlein, H.; Burgess, A. J. Am. Chem. Soc. 2002, 124, 5222-5230. (21) Hudgins, R. R.; Ratner, M. A.; Jarrold, M. F. J. Am. Chem. Soc. 1998, 120, 12974-12975. (22) Counterman, A. E.; Clemmer, D. E. J. Phys. Chem. B 2004, 108, 4885-4898. (23) Counterman, A. E.; Clemmer, D. E. J. Phys. Chem. B 2003, 107, 2111-2117. (24) Counterman, A. E.; Clemmer, D. E. J. Phys. Chem. B 2002, 106, 12045-12051. (25) Hiltpold, A.; Ferrara, P.; Gsponer, J.; Caflisch, A. J. Phys. Chem. B 2000, 104, 10080-10086. (26) Maison, W.; Kennedy, R. J.; Miller, J. S.; Kemp, D. S. Tetrahedron Lett. 2001, 42, 4975-4977. (27) Miller, J. S.; Kennedy, R. J.; Kemp, D. S. Biochemistry 2001, 40, 305-309. (28) Son, H. S.; Hong, B. H.; Lee, C.-W.; Yun, S.; Kim, K. S. J. Am. Chem. Soc. 2001, 123, 514-515. (29) Kohtani, M.; Jones, T. C.; Schneider, J. E.; Jarrold, M. F. J. Am. Chem. Soc. 2004, 126, 7420-7421. (30) Kohtani, M.; Jarrold, M. F. J. Am. Chem. Soc. 2004, 126, 84548458. (31) Kohtani, M.; Jarrold, M. F.; Wee, S.; O’Hair, R. A. J. J. Phys. Chem. B 2004, 108, 6093-6097. (32) Wieczorek, R.; Dannenberg, J. J. J. Am. Chem. Soc. 2004, 126, 12278-12279.
J. Phys. Chem. B, Vol. 112, No. 18, 2008 5709 (33) Roos, G.; Loverix, S.; Geerlings, P. J. Phys. Chem. B 2006, 110, 557-562. (34) Tan, J. P.; Ren, J. J. Am. Soc. Mass Spectrom. 2007, 18, 188194. (35) Woo, H.-K.; Lau, K.-C.; Wang, X.-B.; Wang, L.-S. J. Phys. Chem. A 2006, 110, 12603-12606. (36) Forood, B.; Feliciano, E.; Nambiar, K. Proc. Natl. Acad. Sci. U.S.A. 1993, 90, 838-842. (37) Ren, J.; Patel, C. G. J. Am. Soc. Mass Spectrom. 2005, 16, 535541. (38) Ren, J. J. Phys. Chem. A 2006, 110, 13405-13411. (39) Tian, Z.; Pawlow, A.; Poutsma, J.; Kass, S. J. Am. Chem. Soc. 2007, 129, 5403-5407. (40) Unkles, S.; Rouch, D.; Wang, Y.; Siddiqi, M.; Okamoto, M.; Stephenson, R.; Kinghorn, J.; Glass, A. Biochemistry 2005, 44, 5471-5477. (41) Kinjo, T.; Szerencsei, R.; Winkfein, R.; Schnetkamp, P. Biochemistry 2004, 43, 7940-7947. (42) Sudeep, P.; Joseph, S.; Thomas, K. J. Am. Chem. Soc. 2005, 127, 6516-6517. (43) Uno, T.; Yukinari, A.; Tomisugi, Y.; Ishikawa, Y.; Makino, R.; Brannigan, J.; Wilkinson, A. J. Am. Chem. Soc. 2001, 123, 2458-2459. (44) Wood, M.; Andrade, E.; Storz, G. Biochemistry 2003, 42, 1198211991. (45) Wood, M. J.; Storz, G.; Tjandra, N. Nature 2004, 430, 917-921. (46) Richardson, J.; Richardson, D. Science 1988, 240, 1648-1652. (47) Doig, A. J.; Baldwin, R. L. Protein Sci. 1995, 4, 1325-1335. (48) Wang, J.; Feng, J.-A. Protein Eng. 2003, 16, 799-807. (49) Engel, D. E.; DeGrado, W. F. J. Mol. Biol. 2004, 337, 1195-1205. (50) Foloppe, N.; Sagemark, J.; Nordstrand, K.; Berndt, K. D.; Nilsson, L. J. Mol. Biol. 2001, 310, 449-470. (51) Sugita, Y.; Okamotoa, Y. Chem. Phys. Lett. 1999, 314, 141-151. (52) MacKerell, A. D.; Bashford, D.; Bellott, M.; Dunbrack, R. L.; Evanseck, J. D.; Field, M. J.; Fischer, S.; Gao, J.; Guo, H.; Ha, S.; JosephMcCarthy, D.; Kuchnir, L.; Kuczera, K.; Lau, F. T. K.; Mattos, C.; Michnick, S.; Ngo, T.; Nguyen, D. T.; Prodhom, B.; Reiher, W. E.; Roux, B.; Schlenkrich, M.; Smith, J. C.; Stote, R.; Straub, J.; Watanabe, M.; Wiorkiewicz-Kuczera, J.; Yin, D.; Karplus, M. J. Phys. Chem. B 1998, 102, 3586-3616. (53) Feig, M.; MacKerell, A. D., Jr.; Brooks, C. L., III J. Phys. Chem. B 2003, 107, 2831-2836. (54) Still, W. C.; Tempczyk, A.; Hawley, R. C.; Hendrickson, T. J. Am. Chem. Soc. 1990, 112, 6127-6129. (55) Bashford, D.; Case, D. A. Annu. ReV. Phys. Chem. 2000, 51, 129152. (56) Ryckaert, J.-P.; Ciccotti, G.; Berendsen, H. J. C. J. Comp. Phys. 1977, 23, 327-341. (57) Feig, M.; Karanicolas, J.; Brooks, C. L., III J. Mol. Graphics Modell. 2004, 22, 377-395. (58) Humphrey, W.; Dalke, A.; Schulten, K. J. Mol. Graphics 1996, 14, 33-38. (59) Mezei, M. Simulaid: Fortran-77 program with simulation setup utilites. URL:http://inka.mssm.edu/mezei/simulaid (accessed June, 2007). (60) Becke, A. D. J. Chem. Phys. 1988, 88, 2547-2553. (61) Kohn, W.; Becke, A.; Parr, R. J. Phys. Chem. 1996, 100, 1297412980. (62) Bauschlicher, C. W., Jr.; Partridge, H. J. Chem. Phys. 1995, 103, 1788-1791. (63) Francl, M. M.; Pietro, W. J.; Hehre, W. J.; Binkley, J. S.; Gordon, M. S.; DeFrees, D. J.; Pople, J. A. J. Chem. Phys. 1982, 77, 3654-3665. (64) Chou, P. Y.; Fasman, G. D. Biochemistry 1974, 13, 222-245. (65) Chakrabartty, A.; Baldwin, R. L. AdV. Protein Chem. 1995, 46, 141-176. (66) Kabsch, W.; Sander, C. Biopolymers 1983, 22, 2577-2637. (67) Breaux, G. A.; Jarrold, M. F. J. Am. Chem. Soc. 2003, 125, 1074010747. (68) Paschek, D.; Nymeyer, H.; Garcia, A. E. J. Struct. Biol. 2007, 157, 524-533. (69) Couch, V. A.; Cheng, N.; Nambiar, K.; Fink, W. J. Phys. Chem. B 2006, 110, 3410-3419. (70) Yang, W. Y.; Pitera, J. W.; Swope, W. C.; Gruebele, M. J. Mol. Biol. 2004, 336, 241-251. (71) van der Spoel, D.; van Maaren, P.; Larsson, P.; Timneanu, N. J. Phys. Chem. B 2006, 110, 4393-4398. (72) Luzar, A. J. Chem. Phys. 2000, 113, 10663-10675. (73) Luzar, A.; Chandler, D. Nature 1996, 379, 55-57. (74) Chandler, D. J. Chem. Phys. 1978, 68, 2959-2970. (75) Chandler, D. Modern Statistical Mechanics; Oxford University Press: New York, 1987. (76) Sudha, R.; Kohtani, M.; Breaux, G. A.; Jarrold, M. F. J. Am. Chem. Soc. 2004, 126, 2777-2784.