Evolution rescues folding of human immunodeficiency virus-1 envelope glycoprotein GP120 lacking a conserved disulfide bond

Abstract

The majority of eukaryotic secretory and membrane proteins contain disulfide bonds, which are strongly conserved within protein families because of their crucial role in folding or function. The exact role of these disulfide bonds during folding is unclear. Using virus-driven evolution we generated a viral glycoprotein variant, which is functional despite the lack of an absolutely conserved disulfide bond that links two antiparallel beta-strands in a six-stranded beta-barrel. Molecular dynamics simulations revealed that improved hydrogen bonding and side chain packing led to stabilization of the beta-barrel fold, implying that beta-sheet preference codirects glycoprotein folding in vivo. Our results show that the interactions between two beta-strands that are important for the formation and/or integrity of the beta-barrel can be supported by either a disulfide bond or beta-sheet favoring residues.

Figure 1.

Local reversions in gp120 improve viral replication. (A) Schematic of gp120 with the five conserved domains (C1–C5) and five variable domains (V1–V5). The location of the V4 base disulfide bond is indicated (gray sphere). Disulfide bonds and sites for N-linked glycosylation are indicated based on Leonard et al. (1990). Sites for N-linked glycosylation are shown. (B) Time scale of gp120 evolution. The presence of the revertants R1 (C385A/C418A/T415I), R2 (C385A/C418V/T415I), and R3 (C385V/C418V/T415I) as confirmed by sequencing is indicated on the timeline. (C) Sequences of the V4 loop and flanking regions of wt, mutant, and revertant viruses. No mutations were found outside this region. (D) The SupT1 T-cells were infected with wt, mutant, and revertant viruses and viral spread was measured for 14 d. () Inset, gp120 contents of virus. The gp120 amounts were standardized for CA-p24 input, and the gp120 contents of mutants in the respective fractions are given as percentages of the wt gp120 contents (arbitrarily set at 1). The results are representative for results from at least three independent experiments.

Figure 2.

Reversions restore virus infectivity. TZM-bl reporter cells (∼17 × 10³; confluence of 70–80%) were infected with 5.0 ng of mutant or wt virus in the presence of SQV in a 96-well plate, and luciferase activity was measured after 48 h. The amino acids at positions 385, 418, and 415, respectively, are given below the graph. The relative infectivities were as follows: AA (mut), 7.5%; AAI (R1), 10.1%; AVI (R2), 25.6%; VVI (R3), 68.1%; AV, 20.2%; VA, 58.2%; VV, 54.8%; ALI (FG1), 1.9%; and LAI (FG2), 19.1% (*p < 0.05, **p < 0.01, ***p < 0.005).

Figure 3.

Reversions restore gp120 folding. HeLa cells were infected with VVT7 and transfected with plasmids encoding wt gp120, mut, revertant (R1, R2, or R3) or two additional mutants FG1 and FG2. Cells were pulse labeled for 5 min and chased for the indicated times. Cells were lysed and gp120 was immunoprecipitated from lysates. Immunoprecipitates were deglycosylated with endoglycosidase H and analyzed by either reducing or nonreducing 7.5% SDS-PAGE. Folding intermediates (IT), the native form (NT), the reduced state from which the signal peptide was cleaved off (Rc) or not (Ru) are indicated. In addition, secreted gp120 was immunoprecipitated from the culture supernatant after 8 h of chase and directly analyzed by reducing SDS-PAGE (bottom). Percentages of NT = native (nonreduced), cleaved = Rc (reduced), and secreted gp120 are given below lanes of chase samples. Percentage of secretion is relative to wt.

Figure 4.

wt and R3 viruses show similar sensitivities for conformational antibodies. Single cycle infection experiments in TZM-bl cells containing a luciferase reporter construct under control of the HIV-1 LTR were carried out in the presence of escalating concentrations of CD4-IgG2 (A) and monoclonal antibodies b12 (B) and 2G12 (C). The luciferase activity in the absence of inhibiting reagents was set at 100%.

Figure 5.

Reversions change dynamics within a six-stranded β-barrel in the outer domain of HIV gp120. (A) Structure of HIV gp120 core and the β-barrel structure in the outer domain. The four-stranded bridging sheet, which joins the inner and outer domains, is indicated. The β-barrel structure in the outer domain is boxed and shown in detail on the right (see Supplemental Table 3 for definitions). The individual β-strands are labeled and colored in red, yellow, and green, corresponding to the pairs that are bridged by disulfides C296/C331, C385/C418, and C378/C445, respectively. Positions of mutations studied here (C385, C415, and T418) are labeled. The β-barrel of gp120 was defined as follows based on the structural alignment (Chen et al., 2005) (residue numbers in the structure of CD4-bound HIV-1 gp120, PDB code 1G9M, and residue numbers in unliganded SIV gp120, PDB code 2BF1 in parentheses): β12, residues 291-297 (305-311); β13, 330-334 (342-346); β16-β17, 374-387 (342-346); β19, 413-422 (426-435); and β22, 443-450 (456-463). The β-barrel structures in the unliganded SIV gp120 and the CD4-bound HIV-1 gp120 deviate marginally with a backbone Cα RMSD (44 residues, excluding the GAG linker between β12 and β13) of 0.17 nm, as calculated by ProFit (http://acrmwww.biochem.ucl.ac.uk/software/profit/). (B) Relative sums of interstrand backbone–backbone hydrogen bond occurrence among various variants τ^Total (variant). The occurrence of interstrand backbone–backbone hydrogen bonds across a β-strand pair j and k of individual variants τ^j−k (variant) is defined as a normalized sum of the occurrence of individual interstrand backbone–backbone hydrogen bonds i within the β-strand pair j-k of the β-barrel with respect to that of wt, τ^j−k(variant) ≡ Σ_iτi^j−k(variant)/Σ_iτi^j−k(wt) × 100% (data not shown). The sum of all interstrand backbone–backbone hydrogen bonds within the β-barrel τ^Total (variant)—those between strands 22-12, 12-13, 13-19, 19-17, and 17-16 (see Figure 5A)—is used for comparison in the text. The exact percentages relative to wt are mut, 81.1; R1, 90.7; R2, 88.4; R3, 90.9; FG1, 86.4; and FG2, 83.8. (C) Relative atomic positional RMSF of individual residues (filled gray circles) and overall average of those in the β-barrel structure (open circles). A reference line of wt (100%) is drawn (dashed line) for comparison. (D) Snapshots of selected hydrophilic residues around the mutation sites that are involved in interstrand side chain–side chain hydrogen bonding. The structures are sampled every 100 ps during the 5- to 10-ns trajectories. All structures were fit onto the backbone atoms of these residues. Oxygen, nitrogen, and carbon atoms are colored in red, blue, and green, respectively.

Figure 6.

Absolute positional RMSF for individual residues within the β-barrel. Residues 385, 418, and 415 are labeled in bold and underlined.

Figure 7.

Virus replication and folding of the non–β-branched fill-gap mutants FG1 (C385A/C418L/T415I) and FG2 (C385L/C418A/T415I). (A) Sequences of the V4 loop and flanking regions of wt, mutant, and revertant viruses. The original mutations are indicated with gray circles, the mutations and reversions with black letters. (B) SupT1 T-cells (50 × 10³) were infected with 2.5 ng CA-p24, and viral spread was measured for 14 d. Inset, Gp120 and CA-p24 contents in virus were measured by ELISA. The gp120 amounts were standardized for CA-p24 input and the gp120 contents of mutants in the respective fractions are given as percentages of the wt gp120 contents (arbitrarily set at 1).