Comparison of predicted scaffold-compatible sequence variation in the triple-hairpin structure of human imunodeficiency virus type 1 gp41 with patient data

Abstract

It has been proposed that the ectodomain of human immunodeficiency virus type 1 (HIV-1) gp41 (e-gp41), involved in HIV entry into the target cell, exists in at least two conformations, a pre-hairpin intermediate and a fusion-active hairpin structure. To obtain more information on the structure-sequence relationship in e-gp41, we performed in silico a full single-amino-acid substitution analysis, resulting in a Fold Compatible Database (FCD) for each conformation. The FCD contains for each residue position in a given protein a list of values assessing the energetic compatibility (ECO) of each of the 20 natural amino acids at that position. Our results suggest that FCD predictions are in good agreement with the sequence variation observed for well-validated e-gp41 sequences. The data show that at a minECO threshold value of 5 kcal/mol, about 90% of the observed patient sequence variation is encompassed by the FCD predictions. Some inconsistent FCD predictions at N-helix positions packing against residues of the C helix suggest that packing of both peptides may involve some flexibility and may be attributed to an altered orientation of the C-helical domain versus the N-helical region. The permissiveness of sequence variation in the C helices is in agreement with FCD predictions. Comparison of N-core and triple-hairpin FCDs suggests that the N helices may impose more constraints on sequence variation than the C helices. Although the observed sequences of e-gp41 contain many multiple mutations, our method, which is based on single-point mutations, can predict the natural sequence variability of e-gp41 very well.

FIG. 1.

Frequency of sequences found in patient sequence set as function of number of substitutions per sequence for N sequences (A) and C sequences (B). The origin of the patient from which the HIV-1 isolate was obtained is indicated: E.U, Europe; U.S, United States; A.F, Africa.

FIG. 2.

Percentage of observed substitutions for three sequence sets that were predicted to be fold compatible by FCD. (A) Triple-hairpin structure. (B) N core. (C) Percentage of expected substitutions at thresholds of 2 and 3 kcal/mol, considering a set the same size as the patient sequence set but randomly sampled from the full-sequence set for the N helices.

FIG. 3.

Preference factors computed at various minECO threshold levels (x axis) for the triple-hairpin (A) and N-core (B) structures in the patient sequence, validated-sequence, and full-sequence sets. The preference factor, defined in the text, describes to what extent the observed sequence variation is specifically explained by the FCD.

FIG. 4.

Percentage of residues compatible with FCD for triple-hairpin structure as a function of the distance between each of the sequences (full set) and reference sequence 1AIK. This distance corresponds to the number of substitutions relative to the 1AIK sequence. The minECO threshold used was 3 kcal/mol.

FIG. 5.

Number of predicted amino acids at each position for minECO threshold of 3 kcal/mol. The classes of permissiveness are defined by thresholds, indicated by the dashed horizontal lines. Very permissive regions (number of predicted substitutions higher than or equal to 10) are marked by black bars. White bars represent the conserved regions (number of predicted amino acid substitutions is lower than 3).

FIG. 6.

Relative entropy plot computed on FCD of triple hairpin of 1AIK. The arrows highlight the residues in the cavity. The numbers superimposed on this plot correspond to the number of different amino acid types observed in the patient sequence set.

FIG. 7.

Percentage of observed substitutions for the validated sequence set that are predicted to be fold compatible by the FCD in the trimer of the hairpin structure. The residues are partitioned into the following groups: residues involved in the N-N (A) and N-C (B) interfaces (10, 51), N-helix residues not implied in such interfaces (C), residues of the C helices (10, 51) (D) and the other residues (E), buried residues (ASA ≤ 25 Å) (F), and those exposed to solvent (G). The minECO threshold used was 5 kcal/mol.