Computational design of a β–peptide that targets transmembrane helices

Abstract

The design of β-peptide foldamers targeting the transmembrane (TM) domains of complex natural membrane proteins has been a formidable challenge. A series of β-peptides was designed to stably insert in TM orientations in phospholipid bilayers. Their secondary structures and orientation in the phospholipid bilayer was characterized using biophysical methods. Computational methods were then devised to design a β-peptide that targeted a TM helix of the integrin α_IIbβ₃. The designed peptide (β-CHAMP) interacts with the isolated target TM domain of the protein, and activates the intact integrin in vitro.

The discovery of the β-peptide class of foldamers^1–6 has resulted in the development of β- and mixed α–β peptides capable of binding specific biological targets.^2,7,8 Considerable progress has been made in the design of β-peptides that compete for water-soluble protein-protein interactions, or that bind to the surface of membranes.^2,7–11 However, the design of β-peptide foldamers that target the transmembrane (TM) domains of complex natural membrane proteins has been a formidable challenge, which is addressed in the present paper. Our first objective was to design β-peptides that stably insert in TM orientations in bilayers, and adapt biophysical methods to demonstrate their orientations. Next, computational methods were devised to design a β-peptide that targeted the α_IIb TM helix of the integrin α_IIbβ₃. The designed peptide indeed interacts with the isolated target TM domain of the protein, and activates the intact integrin in vitro.

We have chosen the well characterized platelet integrin α_IIbβ₃ as a model system for our design.¹² The α_IIb and β₃ TMs associate in the resting state of the integrin, but this interaction is disrupted when the integrin is activated to bind to its extracellular ligand, fibrinogen. Thus, disruption of the α_IIb–β₃ TM helical interaction – either by introducing disruptive mutations^13–16 or by addition of α-peptides that compete for this interaction^17–20 – leads to integrin activation. Our goal here was to design a β-peptide that binds specifically to the α_IIb TM domain. This endeavor not only tests and extends our fundamental understanding of the mechanism of TM helix-helix interaction, but takes an important step towards the goal of designing non-natural molecules that specifically target TM regions of proteins in a sequence-specific manner.

Previously we developed the CHAMP method (computed helical anti-membrane protein) to enable computational design of α-peptides that target TM helices of natural proteins.¹⁸ The original CHAMP method uses the crystallographic database of membrane proteins to define interacting TM helical pairs, whose backbones are compatible with the target sequence and that serve as starting points for a sidechain repacking algorithm. As there are no experimental structures of TM β-peptides, let alone TM β-peptides interacting with α-peptides, it was necessary to discover an optimal pose of the β-peptide backbone onto the α-peptide, as well as to design a sequence of the β-peptide that stably interacts with the α-peptide in this pose. We began with the previous model of a complex of the α_IIb with a successfully designed CHAMP peptide to guide the selection of the β-CHAMP sequence. In our previous work,¹⁸ the CHAMP peptide targeted a GXXXG motif in the α_IIb sequence. The exposed backbone atoms of the Gly residues in this motif are recognized by a similar motif in the CHAMP peptide, resulting in inter-helical backbone-backbone interactions that help drive helix-association in membranes. Therefore, we positioned a β-peptide poly-homo-Gly (hGly) helix against the GXXXG of the α_IIb helix, using a grid search and the CHARM force field to determine the optimal position of the β-helix against the α-helix.

There are two common helix types for β-amino acids: while the 14-helix is most stable for β³-substituted amino acids in polar solvents, preliminary studies showed the co-existence of a 12-helical conformation.²¹ Thus, both were considered in the design of the CHAMP peptide. In each case, multiple poses of the poly-hGly backbone were exhaustively sampled to discover the optimal backbone orientation. The sequence of the β-peptide was then outfitted with sidechains using a packing algorithm that evaluates various sidechains in low-energy rotamers to obtain combinations that optimize the geometric complementarity between the β-peptide and the α-helical target.²² The β-peptide positions that directly contacted the mainchain could not sterically accommodate a sidechain, and hence remained hGly. The design was completed by inclusion of Trp and Lys residues near the termini to properly orient the peptide in the bilayer. The initial design focused on the 14-helix and the backbone docking restraints induced an (hGly-X-X)₃ motif with the hGly residues lined along one face of the helix. Somewhat surprisingly, this same motif shows a good fit for the 12-helix; the major difference is that the hGly residues line up along the helix in the 3-fold screw of the 14-helix, but spiral around the helix in the 12-helix. This difference leads to different predicted packing angles in the complex.

To test the designed binding mode, two control peptides were prepared. We disrupted the GXXGXXG motif by mutating the central β³-homo-glycine (hG14) to β³-homo-isoleucine (β-CHAMP G14I) and scrambled the β-CHAMP sequence (β-CHAMPscr). All β-peptides were synthesized using optimized microwave-assisted solid phase peptide synthesis techniques outlined in the Supporting Information, TM α-peptide was synthesized using a previously established protocol.²³ The final sequences are shown in Table 1. Our protocols allow us to consistently obtain high purity products in high yield.

Table 1.

Sequences of the designed β-CHAMP peptides

^*One letter code refers to the corresponding β³-amino acid

The CD spectrum of β-CHAMP was measured in trifluoroethanol (TFE), dodecyl phosphatidylcholine (DPC) micelles, and in phospholipid vesicles composed of a mixture of palmitoyl oleoyl phosphatidylcholine (POPC) and palmitoyl oleoyl phosphatidylglycerol (POPG) (7:3) (Figure 2). In DPC micelles and phospholipid vesicles there were minima at 204 nm and 206 nm, respectively, and a maximum at ca. 220 nm, indicative of the 12-helical structure. In TFE, the β-CHAMP CD spectrum suggested an equilibrium between 12- and 14-helical structures, consistent with increasing destabilization of the 12-helix with increasing solvent polarity.^21,24 Control β-CHAMP peptides followed the same pattern, showing higher 14-helical content in TFE.²¹ Also, when the β-CHAMP mixed with the α_IIb TM peptide under conditions in which they formed a 1:1 complex (Figure 2), the spectrum was well described as a mixture of the TM α-helical and the 12-helical form of the β-peptide, strongly suggesting that the β-CHAMP binds its target in a 12-helical conformation.

Figure 2.

CD spectra of β-CHAMP in DPC micelles (blue circles, 40 μM peptide, 10 mM DPC, 10 mM phosphate buffer, pH 7.4), POPC/POPG (7:3) vesicles (green circles, 167 μM total lipid, 1:100 peptide:lipid), and in TFE (red squares, 40 μM peptide). CD spectrum of β-CHAMP in the presence of 1 equiv. of α_IIb-TM (black diamonds, DPC micelles, 40 μM total peptide). The line represents linear combination of the 12-helical spectrum of β-CHAMP and the CD spectrum of α_IIb-TM in equal proportions.

To determine the orientation of the β-CHAMP peptides in phospholipid bilayers, they were examined in hydrated multilayers by attenuated total internal reflectance IR. (Figure 3, Figure S2, Supporting Information). The order parameter of the amide I′ band (1646 cm⁻¹) of β-CHAMP was 0.83, indicative of the helix being aligned close to the membrane normal.^25–28

Figure 3.

FT-IR/ATR spectrum of β-CHAMP in POPC/POPG (7:3) lipid membranes (1:80 peptide:lipid ratio). The blue and red traces represent light polarized parallel and perpendicular, respectively, to the plane of incidence.

The binding of the β-CHAMP peptides to their target α_IIb-TM was studied using analytical ultracentrifugation (AUC) in density matched DPC micelles. In order to exclude the contribution of α_IIb TM absorbance we labeled the N-termini of the β-peptides with a 2,4-dinitrophenyl (DNP) group. Analysis of the sedimentation equilibrium profile at 356 nm allows one to monitor the radial distribution of the β-peptide, which is sensitive to its association with the intended target. In the absence of the α_IIb TM peptide, the designed β-CHAMP and the control peptides sedimented with an apparent molecular weight close to that expected for the corresponding monomers. A small but significant increase in the molecular weight observed for the β-CHAMP suggests a small degree of homo-dimerization, which is likely due to the GXXGXXG motif, as both β-CHAMP G14I and β-CHAMPscr showed no evidence of self-association. In the presence of the α_IIb TM, β-CHAMP sedimented with a molecular weight consistent with a 1:1 complex with the target, while β-CHAMP G14I and β-CHAMPscr displayed no association with the α_IIb TM (Table 2, Figures S3–S8, Supporting Information).

Table 2.

Association of DNP-labeled β-peptides with α_IIb-TM in DPC micelles as measured by AUC

Peptide	Monomer MW, Da	MWobs in absence of αIIb TM peptide, Da	MWobs in presence of αIIb TM peptide, Da
β-CHAMP	3420.56	4380 ± 61	6737 ± 71
β-CHAMP G14I	3476.67	3542 ± 172	3905 ± 106
β-CHAMPscr	3445.81	3446 ± 147	3542 ± 172

Having established interaction of the β-CHAMP with the α_IIb TM domain, we explored whether β-CHAMP can disrupt interaction of the α_IIb and β₃ TM domains enabling the activated α_IIbβ₃ to bind fibrinogen. Activation of α_IIbβ₃ was monitored by optical trap-based single-molecule rupture force spectroscopy. Full-length α_IIbβ₃ was isolated from human platelets (as described in Supporting Information) and the rupture forces between the surface-bound α_IIbβ₃ and fibrinogen were measured after incubation of the integrin with β-CHAMP or β-CHAMPscr. The β–peptides were flanked with polyethylene-glycol groups on the N-terminus to enhance their solubility. The cumulative probabilities of α_IIbβ₃ binding to fibrinogen in the presence of β–peptides are shown in Figure 4. In order to assess non-specific background binding, the experiments were also performed in the presence of the α_IIbβ₃-specific Fab antibody fragment, abciximab, that prevents interaction of fibrinogen and the integrin. β-CHAMP almost completely activates α_IIbβ₃ in the absence of the antibody fragment and has essentially no activity in the presence of abciximab. The integrin-activating effect of the control peptide β-CHAMPscr is substantially lower. Mn²⁺, previously shown to activate α_IIbβ₃,²⁹ was used as a positive control and solvent (TFE) was used as a negative control. The rupture force spectroscopy results were also confirmed by transmission electron microscopy data (Figure 5) clearly showing opening of the α_IIbβ₃ structure (a morphological equivalent of activation) as a result of an α_IIb TM helix interacting with β-CHAMP. These results that show strong and selective interaction of the β-CHAMP with the α_IIb TM domain are in good agreement with the AUC data.

Figure 4.

Rupture force spectroscopy data on interaction of the isolated surface-bound α_IIbβ₃ and fibrinogen in the presence of the β-CHAMP peptides (5 μM). Negative control represents trifluoroethanol, positive control - 2 mM Mn²⁺. The difference between the β-CHAMP data (column 1) and the β-CHAMPscr data (column 3) is statistically significant (p=0.008).

Figure 5.

Transmission electron microscopy of purified α_IIbβ₃ in “closed” (A) and “open” (B) forms, corresponding to inactive and active conformations, respectively. Magnification bar = 30 nm. The bar graph shows fraction of the open structure in the presence of the β-CHAMP peptides (5 μM). Negative control represents trifluoroethanol, positive control - 2 mM Mn²⁺.

In conclusion, these studies indicate the feasibility of designing β-peptides to target TM helices of natural proteins. This required the design of a β-peptide that spanned the bilayer, which was accomplished by inclusion of a block of apolar residues sufficiently long to span the bilayer, as well as inclusion of Trp and Lys residues near the headgroup region of the bilayer. The TM orientation was established using polarized IR spectroscopy. Specific interaction with the α_IIb TM was computationally designed by; 1) placing Gly residues at regular spacings along one face of the 12-helix; 2) optimizing the interaction of the 12-helix with α_IIb TM; and 3) optimizing the placement of other sidechains in the β-CHAMP to allow favorable van der Waals contacts with the α-helical target. The interaction was shown to be specific using variants of the β-CHAMP sequence. These studies provide a well-defined and automated approach to design β-peptides that recognize membrane targets.

Figure 1.

Design of β-CHAMP. a) Docking of a CHAMP peptide on α_IIb TM helix (cyan), the α-carbons of Gly residues are shown as red spheres.¹⁸ b) The position of the interacting motif on an α-helix (GX₃GX₃G ); c) a 14 helix (GX₂GX₂G); d) a 12-helix (GX₂GX₂G); e) final sequence of β-CHAMP modeled as a 12 helix, C₃ atoms of the GX₂GX₂G are shown as red spheres.