|Home | About | Journals | Submit | Contact Us | Français|
Helix-helix interactions are fundamental to many biological signals and systems, found in homo- or hetero-multimerization of signaling molecules as well as in the process of virus entry into the host. In HIV, virus-host membrane fusion during infection is mediated by the formation of six helix bundle (6HB) from homo-trimers of gp41, from which a number of synthetic peptides have been derived as antagonists of virus entry. Using a yeast surface two-hybrid (YS2H) system, a platform designed to detect protein-protein interactions occurring through a secretory pathway, we reconstituted 6HB complex on the yeast surface, quantitatively measured equilibrium and kinetic constants of soluble 6HB and delineated the residues influencing homo-oligomeric and hetero-oligomeric coiled-coil interactions. Hence, we present YS2H as a platform for facile characterization and design of antagonistic peptides for inhibition of HIV and many other enveloped viruses relying on membrane fusion for infection, as well as cellular signaling events triggered by hetero-oligomeric coiled coils.
Many proteins naturally fold into varying degrees of helix bundles, which are indispensible to normal physiology and to the onset of diseased states1. Of the well-known examples are viral membrane fusion proteins like influenza virus hemagglutinin (HA) and HIV-1 envelope glycoprotein gp41. HIV infection critically depends on the attachment and fusion of virus to host cells through the gp120/gp41 complex. The extracellular domain of gp41 consists of a fusion peptide, an N-terminal heptad repeat (NHR), a loop region, a C-terminal heptad repeat (CHR), and a membrane-proximal external region (MPER) (Figure 1a). Binding of gp120 to CD4 and subsequent interaction with co-receptors (CCR5 or CXCR4) lead to the dissociation of gp120 from gp41 and insertion of the viral gp41 fusion peptide into the target cell membrane, which then forms a prehairpin intermediate. Fusion of viral and cellular membrane is provided by the formation of a gp41 six helix bundle (6HB), a conformation described by three CHR packing in an antiparallel manner to a central three-stranded NHR coiled coil2–3 (Figure 1b, c). gp41 prehairpin intermediate is transiently accessible to viral fusion inhibitors derived from NHR or CHR, named as N or C peptide accordingly2,4–8 (Figure 1c, d). T20 is one of the early C peptides developed for HIV fusion inhibition9–10. CP32M, a rationally designed C peptide, extends to the upstream region of CHR compared to T20, and contains mutations to enhance affinity to NHR. These resulted in higher thermostability and greater inhibition of viral infection against diverse HIV strains11. Compared with the inhibitors derived from the C peptides, there are fewer based on the N peptides due to their low solubility. 5-Helix (5H) peptide generated by linking 6HB peptides in tandem without the last C peptides is used to mimic N peptide, and is shown to be highly stable and potent in viral inhibition12. Other N-peptide derivatives have been based on fusion to synthetic trimerization sequences13. While much interest has been focused on developing synthetic inhibition peptides, natural peptides circulating in the blood have been reported to possess a capacity to inhibit virus fusion14.
Developing biochemical assays to examine the potency of inhibitors of 6HB formation have been hampered by low solubility of N- and C-peptides and by their tendency to aggregate15. As a facile and quantitative platform to study 6HB formation and to aid in the design of antagonistic peptides, we used yeast surface two hybrid (YS2H) system and expressed gp41 N- and C-peptides on yeast cell surface. In YS2H, a system designed to express a pair of proteins for measuring protein-protein interactions16–17, one protein is fused to yeast agglutinin and is thereby displayed on yeast surface as the ‘bait’, while the other protein is secreted in soluble form and serves as the ‘prey’ (Figure 2a, b). In the presence of interaction, two proteins associate with each other within the secretory pathway, and the prey is captured on the cell surface by the bait. The affinity of the interaction between the prey and the bait can be quantitatively estimated from the relative abundance of short tags fused to bait and prey, typically measured by flow cytometry17. This system has previously been used to study heterodimeric interactions; however, its applicability to study higher order, complex oligomerizations has not been demonstrated.
In order to reconstitute gp41 6HB complex on the yeast surface, we expressed N36 and C3418, peptides derived from regions in NHR and CHR, respectively (Figure 1d, Figure 2). N36 and C34 are known to exclusively form a 6-helix bundle complex devoid of any aggregates3,18. Surface expression of N36, fused to Aga2 (a subunit of agglutinin) as the ‘bait’, was confirmed by antibody binding to Flag tag appended to the C-terminus of N36. With C34 co-expressed as the ‘prey’, antibody binding to Myc tag indicated that C34 was captured by N36 (denoted as N36/C34 to refer to a pair of Aga2-fused bait and soluble prey; Figure 2b–c). N36/C34 association was further recognized by conformation-specific monoclonal antibodies (mAb) of gp41: NC-119 and D5020 (Figure 2c). Although the presumed epitope for D50 is at the C-terminus of CHR (Ile642-Lys665)20, yeast cells expressing C34 alone (C34/-) were not recognized by D50, indicating that C34 without N36 did not adopt a conformation existing in the context of 6HB. With N36 and C34 swapped for the role of bait and prey (i.e., C34/N36), similar levels of binding of tag-specific antibodies and mAbs NC-1 and D50 were achieved (Figure 2c). We have chosen to anchor N peptide as a fusion to Aga2 and secrete C peptide as itself, a configuration more suitable for evaluating the potency of C peptide-based antagonists.
To validate that in our YS2H system N- and C-peptides were assembled into the conformation of 6HB previously seen in solution and crystals3,18,21, we performed Western blot on N36/C34 peptides cleaved by TEV protease from the surface of yeast (Figure 3). The staining pattern of the N36/C34 by antibodies against Flag and Myc tags closely matched that of the synthetic peptides (N36+C34) of identical amino acid sequence: one of the major bands from N36/C34 was identical to ~40 kDa band from the synthetic peptides, which have been shown to reconstitute 6HB in solution18. The identity of the 40 kDa band to be 6HB was also corroborated by the staining of mAb NC-1. Compared to the synthetic peptides that formed one major band of 6HB, the N36/C34 assembled in yeast did exhibit higher molecular weight bands stained by both anti-tag antibodies. It can be speculated that during the secretory pathway, not all N36 and C34 peptides would associate with each other to assemble into 6HB but some of N36 may also form a higher order oligomers, which were still able to associate with C34 and retain NC-1 binding. In contrast to N36/C34 cleaved from yeast, N36 alone (N36/-) or N36 with 6HB disrupting mutation I559P (I559P/C34; see below for details) did not produce a presumed 6HB complex band.
In order to examine if YS2H would provide a quantitative readout to a change in hetero-oligomeric interactions, we have introduced point mutations at various positions and measured the binding of antibodies against reporter tags and different oligomerization states of N36/C34 complexes (Figure 4). Physical forces contributing to 6HB or coiled-coil conformation in general are from the combination of van der Waals (vdW) or hydrophobic contacts at the helix-helix interface and electrostatic attraction among the residues positioned outside of the interface. In a homotrimeric coil of NHR, amino acids in the position of ‘a’ and ‘d’ in a helical wheel or heptad repeat form a helix-helix interface at the core (Figure 1b). The residues at ‘e’ and ‘g’ of NHR would normally be those that promote electrostatic interaction, but instead they are mainly hydrophobic so that CHR packs onto NHR to form 6HB. The mutations that disrupt hydrophobic or electrostatic forces would completely or partially disrupt coiled-coil interactions. Mutations at the interface between NHR and CHR (e.g., L556A (at position ‘e’), A558G (‘g’), and V570D (‘e’) of NHR and S649D (‘a’) of CHR) led to a modest (A558G) to almost complete (L556A, V570D, S649D) loss of C34 association with N36 (Figure 4). 5HB-specific mAb D522, which largely recognizes the hydrophobic pocket present at the C-terminal end of NHR, displayed binding to 6HB mutants different from anti-Myc and D50. To cells expressing only N36 (N36/-), D5 showed strong binding (Figure 4), revealing that without an assembly with C34, N36 adopts a conformation seen in the context of gp41. This is an important result that underscores the utility of yeast display, which circumvents the difficulty of studying isolated N-peptide derivatives in solution because of their hydrophobic, aggregation prone nature. The mutations that disrupted a homotrimeric N peptide conformation (e.g., I559P and I573D at position of ‘a’ of NHR), therefore, led to a complete loss of both anti-Myc and D5 antibodies (Figure 4). Such disruption to 6HB formation by I559P was also confirmed by Western blot (Figure 3). L568A, a mutation introduced at position ‘c’ of NHR, least compromised 6HB formation. In contrast to various mutations that led to partial to complete perturbation of 6HB formation, N656L led to an increase in the level of Myc expression (Figure 4), attributed to the introduction of a hydrophobic residue that would form vdW contacts with Ile548 and Val549 (Figure 1d).
After validating the assembly and perturbation of gp41 6HB, we examined the pairing of N36 or full-length NHR (Met530 – Gln590) with the variants of C34, such as T20 and CP32M as antagonistic C peptides, which have been developed for inhibiting HIV entry (Table 1). N36/T20 helix-helix interface is formed with three heptad repeats of α-helix (Figure 1d), too short to produce a strong interaction, and this resulted in marginal binding of anti-Myc antibody and D50. In contrast, NHR/T20, spanning longer than four heptad repeats of α-helix at the interface, resulted in the level of Myc expression comparable to or less than the levels seen in NHR/C34 and N36/C34. The observation that NHR/T20 led to maximum binding with mAb D50 is consistent with the fact that D50 epitope is at the C-terminal end of CHR (Ile642-Lys665), fully included in T20 but only partially present in C3420. When paired with either N36 or NHR, CP32M, a peptide spanning Gln621 to Gln652 of CHR and containing mutations to enhance electrostatic and hydrophobic interactions with NHR, exhibited elevated levels of Myc expression. The lack of mAb D50 binding to N36/CP32M and NHR/CP32M is due to the epitope for D50 being almost absent in CP32M. In YS2H, the levels of reporter tags (Myc & Flag) is directly related by Langmuir equation to equilibrium dissociation constant (KD) for the interaction between bait and prey, given as Flag/Myc = α(1+KD/[prey]), where α is the ratio of Flag/Myc for KD ~ 0. With the estimated values of α = 4 and [prey] = 10 nM in YS2H17, the values of KD were obtained for NHR or N34 binding to C34, T20, or CP32M (Table 1). Although direct measurements of solution affinity for these pairs are not available, the difference in affinity is consistent with reported potency of C-peptide based inhibitors in the order of CP32M > C34 > T2017. The ability to quantitatively measure equilibrium binding constants highlights YS2H as a facile platform for the prediction of the potency of antagonistic peptides.
In addition, YS2H can be adapted as a platform for measuring dissociation kinetics for hetero-oligomeric interactions. To minimize rebinding of dissociated C peptides to N peptides, yeast cells were washed and resuspended in a much larger volume of binding buffer (333-fold dilution of yeast culture) and antibody binding to reporter tags was measured at different time points (Figure 5). Notably, the decrease in antibody binding to Myc followed a biphasic behavior, characterized by a rapid reduction of Myc within the first 10 min, followed by a slower decrease. An initial rapid dissociation of C peptide may be due to a loss of small percentage (~15%) of C peptides that were associated with the N peptides forming non-ideal 6HB complexes, some of which may correspond to higher molecular weight bands detected in Western blot (Figure 3). When a two phase dissociation model (Y = A1×exp(−t/T1) + A2×exp(−t/T2)) was used to fit the data, in agreement with the overall trend of dissociation, the slower kinetics were dominant (A1 ≈ 15 and A2 ≈ 85) for all interaction pairs measured (Table 1). Following a similar trend with equilibrium affinities, a dissociation rate of C34 from NHR (NHR/C34; 16.2×10−6s−1) was 3.5-fold slower than that from N36 (N36/C34; 56.7×10−6s−1), and a dissociation rate of NHR/C34 was 3.9-fold slower than that of NHR/T20 (62.7×10−6s−1). Notably, although CP32M exhibited equilibrium affinity to N36 higher than C34 or T20, a dissociation rate of CP32M from N36 (242.7×10−6s−1) was 4.3-fold faster than that of C34 (56.7×10−6s−1) (Table 1), indicating that CP32M binding to N36 is comparatively dominated by its on-rate.
In summary, we have demonstrated a construction of a pair of α helices in the YS2H system, and the assembly of viral 6HB structure on the yeast cell surface. The equilibrium binding strength between coiled coils within the bundle as well as the kinetics of soluble peptides could be directly and quantitatively characterized. Subtle alterations that resulted from single point mutations perturbing homo-trimeric as well as hetero-oligomeric coiled coils could also be detected in YS2H, which demonstrates its utility as a platform for the design, optimization, and evaluation of antagonistic peptides as drug candidates. Besides HIV, many enveloped viruses including influenza, respiratory syncytial virus, and Ebola virus require helix bundles for membrane fusion during virus entry into the host23. In the past, a lot of effort has been directed towards the de novo synthesis of various types of helix bundles24–28. The assembly of helix bundles or hetero-oligomeric peptides on the yeast surface would allow rapid screening and design of candidate inhibitors, without impediments from often unreliable in vitro protein refolding and costly chemical synthesis.
The work is supported by National Institutes of Health Grants GM090320, Northeast Biodefense Center U54-AI057158 (Lipkin) (MJ) and Department of Biotechnology and Department of Science and Technology, Government of India (RV). We thank NIH AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, NIH for providing of NC-1 and D50 antibodies. We also thank Dr. Joyce of Vaccines Basic Research, Merck Research Laboratories for providing of D5 IgG for our study.