Design of an engineered N-terminal HIV-1 gp41 trimer with enhanced stability and potency

Abstract

HIV fusion is mediated by a conformational transition in which the C-terminal region (HR2) of gp41 interacts with the N-terminal region (HR1) to form a six-helix bundle. Peptides derived from the HR1 form a well-characterized, trimeric coiled-coil bundle in the presence of HR2 peptides, but there is little structural information on the isolated HR1 trimer. Using protein design, we have designed synthetic HR1 peptides that form soluble, thermostable HR1 trimers. In vitro binding of HR2 peptides to the engineered trimer suggests that the design strategy has not significantly impacted the ability to form the six-helix bundle. The peptides have enhanced antiviral activity compared to wild type, with up to 30-fold greater potency against certain viral isolates. In vitro passaging was used to generate HR1-resistant virus and the observed resistance mutations map to the HR2 region of gp41, demonstrating that the peptides block the fusion process by binding to the viral HR2 domain. Interestingly, the activity of the HR2 fusion inhibitor, enfuvirtide (ENF), against these resistant viruses is maintained or improved up to fivefold. The 1.5 A crystal structure of one of these designs has been determined, and we show that the isolated HR1 is very similar to the conformation of the HR1 in the six-helix bundle. These results provide an initial model of the pre-fusogenic state, are attractive starting points for identifying novel fusion inhibitors, and offer new opportunities for developing HIV therapeutics based on HR1 peptides.

Figure 1.

Thermal unfolding transition of HR1 oligomers as measured by circular dichroism. Measurements were made in phosphate buffer at a peptide concentration of 10 μM for T865Δ (●), T865ΔAA (■), T865RSV (○), and T865RSV_AA (□).

Figure 2.

Apparent molecular weight determined as a function of loading concentration for T865 (◆), T865ΔAA (■), and T865RSV_AA (□). Molecular weights were determined by fitting the data at each concentration to a single ideal species model and averaging the results from different rotor speeds and wavelengths. Errors in molecular weight determinations are <10%.

Figure 3.

Thermal unfolding transition of HR1/T649 complexes as measured by circular dichroism. Measurements were made in phosphate buffer at a peptide concentration of 10 μM plus 10 μM T649 for T865Δ (●), T865ΔAA (■), T865RSV (○), and T865RSV_AA (□).

Figure 4.

Binding of HR2 peptide T649 to the trimeric T865ΔAA as determined by surface plasmon resonance at 25°C in HBS-P buffer on a Biacore 3000 instrument. Biotinylated T865AA was immobilized onto the surface of a streptavidin-coated sensorchip (Sensorchip SA) at a level of ∼500 response units (RUs), and T649 was injected across the surface at a rate of 20 μL/min at concentrations of 250 nM to 7.8 nM (top panel). The bottom panel shows the result of a global fit to the 7.8 nM data assuming a 1:1 (Langmuir) binding model. The measured data are denoted by the diamonds (for clarity, not every point is shown) and the fit is represented by the line. The results of the global fit agree well with the average of the individual fits at low concentrations but global analysis failed at higher concentrations, presumably due to mass transport limitations. Kinetic parameters are listed in Table 2.

Figure 5.

HIV antiviral activity of HR1 peptides determined using a MAGI assay. Percent inhibition is determined as a function of peptide concentration for T865 (◆), T865AA (◇), T865Δ (●), T865ΔAA (■), T865RSV_AA (□), and T649 (×).

Figure 6.

Ribbon diagram of the T865RSV_AA trimer. Residues 540–545 are not visible, and residues 546–549 are not part of the helical structure.

Figure 7.

Superposition of T865RSV_AA and N34 from the structure of the HR1/HR2 complex (Chan et al. 1997). Arg-557 was used for the alignment of the N-terminal end of the peptides and Trp-571 was used to align the deep pocket region. For clarity, not all side chains are shown. The ∼10° bend observed in the helix of T865RSV_AA is also highlighted. The angular bend is an estimate based on the angle between line segments through the helical axis of the N- and C-terminal halves of the peptide. The alignment and the figure were done with the program PyMOL (DeLano Scientific).