The Effects of Flanking Sequences in the Interaction of Polyglutamine

Dec 20, 2013 - We observe significant membrane thinning, disorderliness of lipid molecules, and compensation effects between the top and the bottom le...
0 downloads 10 Views 2MB Size
Article pubs.acs.org/JPCB

The Effects of Flanking Sequences in the Interaction of Polyglutamine Peptides with a Membrane Bilayer Anu Nagarajan, Sudi Jawahery, and Silvina Matysiak* Fischell Department of Bioengineering, University of Maryland, College Park, Maryland 20742, United States S Supporting Information *

ABSTRACT: Huntington’s disease (HD) is caused by the presence of an extended polyglutamine (polyQ) region at the N-terminus of the huntingtin (htt) protein. The presence of flanking sequences adjacent to the polyQ region has been reported to modulate the effects of potentially toxic protein− membrane interactions. In this study, we consider four peptide systems with various combinations of flanking sequences (KKQ35KK, KKQ35P11KK, N17Q35KK, N17Q35P11KK) and use atomistic molecular dynamics simulations to study the interactions with a DOPC lipid bilayer. We observe significant membrane thinning, disorderliness of lipid molecules, and compensation effects between the top and the bottom leaflets of the bilayer depending on the presence of particular flanking sequences. Overall, we find that the presence of the N-17 flanking sequence is crucial for membrane interactions. Polyproline decreases the interaction with the membrane in the absence of N-17, but enhances it when present along N-17.



phospholipids that are involved in cell signaling.18 Numerous kinetic and thermodynamic studies have shown that proteins undergo conformational changes when associating with surfaces.19 These changes in protein structure could play a role in the formation of aggregates and even stabilize them.20 Lipid interactions have been specifically associated with aggregate formation in the studies of amyloid-β(Aβ)21−25 and α-synuclein.26−28 Lipid molecules have also been shown to inhibit the effect of small molecule therapeutic agents that prevent amyloid formation. Epigallocatechin gallate (EGCG), which is known to inhibit amyloid formation in the absence of surfaces,29,30 had a lesser effect in inhibiting aggregate formation in the presence of a lipid interface.31 While lipid membranes may provide the environment for these proteins to undergo conformational changes leading to aggregate formation, it is also possible that the association of these peptides with a lipid bilayer may lead to changes in morphology and structure of cell membranes, ultimately leading to cell death. Recent experiments have shown that the N-terminal flanking sequence directly interacts with phospholipids and is responsible for the interaction of polyQ peptides with membrane bilayers.32,33 The flanking sequences, the N-17 and polyproline domains, and the length of polyQ domain all modulate the interactions of htt with lipids.34 It is

INTRODUCTION Many neurodegenerative diseases including Huntington’s, Alzheimer’s and Parkinson’s disease are caused by the accumulation of protein aggregates that have β-sheet-rich structures as a prominent feature.1−3 In Huntington’s disease (HD) and Spinocerebellar ataxis (SCAs), mutations that cause an expansion of CAG triplet repeats, which encode polyglutamine (polyQ), are responsible for the pathogenesis of these neurodegenerative disorders.4−7 It is hypothesized that after proteolytic cleavage, the first exon 1 of huntingtin (htt), which contains the polyQ region, misfolds and forms inclusions in neuronal cells. The onset of HD and its severity is correlated with polyQ length.8−10 PolyQ domains shorter than 35 residues are not associated with the disease, while lengths greater than 35 residues (especially 40−60), are associated with adult onset of HD.11 Although the genetics of HD are very well studied, the exact mechanism of pathogenic peptide aggregation still remains contentious. Various factors influencing the rate of aggregation have been widely studied.12−15 Flanking sequences that are adjacent to the polyQ region such as the first 17 amino acids of the htt exon 1 (N-17) have been shown to increase aggregation in these peptides.12,14 On the other hand, the polyproline segment on the C-terminus of the polyQ domain has the opposite effect by reducing the kinetics of aggregation of these peptides and the formation of β-sheets in the polyQ region.15,16 Interactions of the flanking sequences and polyQ domain with lipids have been shown to influence misfolding and alter toxicity in yeast.17 Also, the htt protein is known to associate itself with membranes through direct interaction with © XXXX American Chemical Society

Special Issue: William C. Swope Festschrift Received: August 7, 2013 Revised: December 18, 2013

A

dx.doi.org/10.1021/jp407900c | J. Phys. Chem. B XXXX, XXX, XXX−XXX

The Journal of Physical Chemistry B

Article

Figure 1. Colors on the graphs correspond to the following secondary structures: red - β-sheets; black - β-bridges; green - bends; yellow - turns; white - coils; blue; and violet - helix structures. KKQ35KK has 39 total residues; KKQ35P11KK has 50 total residues; N17Q35KK has 54 total residues; N17Q35P11KK has 65 total residues. Panels a−b, c−d, e−f, and g−h represent KKQ35KK, KKQ35P11KK, N17Q35KK, and N17Q35P11KK proteins while free in water and in contact with the membrane, respectively.

therefore crucial to address this area, as it could help identify novel therapeutic and neuroprotective targets for HD. While several experimental studies have established the existence of interactions between the N-17 domain, polyQ, and lipid bilayers,18,32,33,35 the exact nature of atomic interactions responsible for the binding and aggregation of the htt protein in the presence of membranes have not been fully studied. Recently, AFM studies have shown the role of these flanking sequences on the aggregation of polyQ peptides in the presence of lipid bilayers.20 It was shown that flanking sequences altered both the rate of accumulation and morphology of aggregates on membranes. Those studies also provided direct evidence that the N-17 flanking sequence is necessary for the aggregation of polyQ peptides on a lipid bilayer. Although several experimental studies have shown the effect of flanking sequences in htt peptide-membrane interactions, a mechanistic understanding of the observed behavior is still lacking. Soluble polyQ peptides are intrinsically disordered,36−38 whereas aggregates exhibit amyloid-like β-strands. Aggregation kinetics studies have shown that the aggregation nucleus is a monomer and that the conversion of random coil structures to β-strands occurs in a single monomer.39,40 As a consequence, the dynamics of a single monomer determine its aggregation mechanism and the structure of the aggregates. Therefore, we focus here on characterizing the conformational

dynamics of an individual polyQ chain in the presence of a DOPC lipid bilayer with different flanking sequences. To study in detail the nature of interactions of these peptides with lipids and their effects, we employ a computational approach using molecular dynamics simulations. The aim of these simulations is to examine early events in the surface binding of polyQ peptides, as well as subsequent disordering effects on the membrane and changes in peptide structure. The peptide systems are modeled using the sequences from the work done by Legleiter et al20 - KKQ35KK, KKQ35P11KK, N17Q35KK, and N17Q35P11KK.



RESULTS AND DISCUSSION Each peptide sequenceKKQ35KK, KKQ35P11KK, N17Q35KK, and N17Q35P11KKis simulated in close proximity to a DOPC lipid bilayer. Since polyQ peptides are intrinsically disordered,36−38 to select an initial peptide structure for the peptide-membrane simulations, we have performed implicit solvent simulations. We have simulated the folding of these peptides without the presence of a membrane starting from a completely stretched conformation. As seen in Table T1, only a small percentage of low energy structures (less than 30%) are populated in the course of the implicit peptides simulations. A representative structure of the low energy ensemble is relaxed by a subsequent 20 ns explicit B

dx.doi.org/10.1021/jp407900c | J. Phys. Chem. B XXXX, XXX, XXX−XXX

The Journal of Physical Chemistry B

Article

Figure 2. Membrane thickness is presented as a function of two-dimensional position along the lipid bilayer. Color bar represents the thickness in nm from 3.2 nm (purple) to 4.2 nm (yellow). (a) Control (no protein) membrane thickness data. (b) Membrane thickness with KKQ35KK peptide. (c) Membrane thickness with KKQ35P11KK peptide. (d) Membrane thickness with N17Q35KK peptide. (e) Membrane thickness with N17Q35P11KK peptide.

of this peptide with the membrane, the β-sheet formation in the C-terminal region increases with time, with new β-sheet formation around residue 35 after 150 ns. Residues 6 and 7 of the N-17 region exhibit β-sheet formation in water, and the rest of the N-17 region fluctuates between β-bridges, bends, coils, and turns. Interestingly, after 40 ns when the N17Q35KK peptide associates with the membrane (Figure 1e), the N-17 region exhibits a dramatic reduction in the content of β-bridges, bends, and turns, having only one residue in a β-sheet conformation. The structural content of N17Q35P11KK in water (Figure 1g) was found to be similar to that of KKQ35KK except in two residues in the N-17 region, whose β-bridge structure occasionally fluctuated to β-sheets. When this peptide associated with the membrane (Figure 1h), the β-sheet content minimally increased after 40 ns, transitioning from β-bridge to β-sheet for residues 15−16 and 32−34. It is worth mentioning that DSSP does not capture PPII helix conformations.42 That is why the polyproline regions in Figure 1c−d and Figure 1g−h mainly show a coil structure. Figures S1a and S1b show the Ramachandran plot of the proline region for KKQ35P11KK and N17Q35P11KK, respectively, when these peptides are in contact with the membrane. The most populated phi and psi angles of the polyproline segment correspond to the PPII helix region with a phi around −75° and psi around 150°. Similar results are obtained for the peptides in water. Since the polyproline regions exhibit a PPII-helix in all cases, it is likely that these helices are formed independently of other

water simulation. The last conformation of this run is chosen and placed close to a DOPC membrane to study the interactions of these peptides with lipids. Contact with Membrane Bilayer Increases β-Sheet Content in the Peptides. A detailed description of the conformational dynamics of the polyQ peptides with and without the presence of the membrane is provided by analysis of the time-dependent secondary structure. Figure 1a−h show the secondary structural content as defined by the dictionary of secondary structures of proteins (DSSP).41 For the KKQ35KK peptide in water (Figure 1a), the N-terminal segment together with regions around residues 20 and 35 switch dynamically between bends and turns. The structure of the rest of the peptide dynamically fluctuates between coil, β-bridges, and bends. When the KKQ35KK peptide comes in contact with the lipid bilayer, the structure of the N-terminal segment together with a region around residue 20 vacillate between β-bridges and β-sheets after 25 ns (Figure 1b). For the KKQ35P11KK peptide in water (Figure 1c), there is β-sheet structure formation in the N-terminal region and around residue 20. These two regions of β-sheets remain the same when this peptide associates with the membrane (Figure 1d) exhibiting no increase in β-sheet content. This result indicates that polyproline does not induce the formation of new β-sheet structure. A different scenario is observed for the N17Q35KK peptide. The C-terminal segment and the region around residues 25 to 30 show β-sheet content in water (Figure 1e). Upon association C

dx.doi.org/10.1021/jp407900c | J. Phys. Chem. B XXXX, XXX, XXX−XXX

The Journal of Physical Chemistry B

Article

3.42 nm (Figure 2e). Minimum thickness followed the same trends- the average minimum thicknesses of the KKQ35KK, KKQ35P11KK, N17Q35KK and N17Q35P11KK systems were 3.46 nm, 3.53 nm, 3.52 nm, and 3.23 nm. The average membrane thickness of the control membranes was 3.88 nm, and the average minimum thickness was 3.63 nm. The statistical significance of the difference between the average thickness of the control membrane and the average thin spot thicknesses of the peptide−membrane systems was verified using the student’s t test. Table 1 clearly shows that

interactions. Previous studies have suggested that polyproline segments may disrupt the β-sheet structure in the polyQ region due to the formation of PPII-like helix.43 As seen in Figure 1d, when the polyproline domain is added to the polyQ peptide, the β-sheet content does not increase when this peptide associates with the membrane. However, a reduction of β-sheet content compared to polyQ without the polyproline domain is not observed as in other studies.43 Interestingly, when the N-17 domain is present, the addition of the polyproline domain leads to a dramatic reduction of β-sheet content, which is evident by comparison of Figure 1e with Figure 1g and Figure 1f with Figure 1h. There are more important observations to make about the N-17 flanking sequence. Studies have reported both helical44−46 and compact random coil12,46 conformations for this region in isolation. We also observe a helical conformation for the N-17 region when simulated with and without the presence of the DOPC membrane, as shown in Figures S2a and S2b. On the other hand, the N-17 segment exhibits several residues in a random coil conformation when simulated with polyQ, with or without the presence of the polyproline region. This is consistent with computational studies that show a loss of helical structure in the N-17 segment with polyQ length; for polyQ lengths greater than 25, the helical content has been shown to drop below 5%.46 Further, we find that when a nontoxic length of polyQ is flanked only by N-17, the polyQ region has more helical region than other peptide systems. This helicity is retained throughout the course of the simualtion with the lipid membrane (Figure S8). Overall, we find that the contact of polyQ peptides with the membrane promotes the growth of existing β-sheet structures and the creation of new β-sheets, with the exception of the KKQ35P11KK peptide. It is known that surfaces tend to increase the β-sheet propensity of peptides when in close contact with a surface.47 In addition to the presence of a flat surface, it is possible that the interactions with the head groups of the membrane promote the enhancement of β-structures. Also, in the most β-rich proteins, most of the β-structure is in glutamine residues. Flanking Sequences Alter Bilayer Thickness. To characterize the changes in membrane structure induced by the presence of the different polyQ peptides, we have computed the average bilayer thickness. Figure 2a shows the thickness of the DOPC bilayer without the presence of a peptide (control simulation). While uniform thickness is observed for the control simulation, bilayer thinning is observed in areas surrounding the region where the peptide is in contact with the membrane in all the peptide−membrane systems. The extent to which thinning occurs in the membrane depends on the flanking sequences present in the peptide. The KKQ35KK peptide causes thinning of the bilayer over about 4.8 nm2 (Figure 2b), exhibiting an average thin spot thickness of 3.54 nm. The addition of the polyproline domain to the polyQ peptide causes a reduction of thinning magnitude with an average thin spot thickness of 3.64 nm. However, the region of thinning becomes broader, spanning almost 9 nm2 (Figure 2c). On the other hand, the addition of the N-17 domain to the polyQ peptide has the opposite effect causing more thinning down to an average thin spot thickness of 3.52 nm, but spanning less than 6 nm2 (Figure 2d). Lastly, the addition of both N-17 and polyproline to the polyQ peptide causes both significant and broad thinning over a range spanning almost 11 nm2 with an average thin spot thickness of

Table 1. Statistical Significance of Average Membrane Thinning peptide

average thin spot thickness (nm)

p-value (statistical difference from control membrane)

KKQ35KK KKQ35P11KK N17Q35KK N17Q35P11K

3.54 3.64 3.52 3.42