In vitro selection and characterization of human immunodeficiency virus type 1 variants with increased resistance to ABT-378, a novel protease inhibitor

Abstract

ABT-378, a new human immunodeficiency virus type 1 (HIV-1) protease inhibitor which is significantly more active than ritonavir in cell culture, is currently under investigation for the treatment of AIDS. Development of viral resistance to ABT-378 in vitro was studied by serial passage of HIV-1 (pNL4-3) in MT-4 cells. Selection of viral variants with increasing concentrations of ABT-378 revealed a sequential appearance of mutations in the protease gene: I84V-L10F-M46I-T91S-V32I-I47V. Further selection at a 3.0 microM inhibitor concentration resulted in an additional change at residue 47 (V47A), as well as reversion at residue 32 back to the wild-type sequence. The 50% effective concentration of ABT-378 against passaged virus containing these additional changes was 338-fold higher than that against wild-type virus. In addition to changes in the protease gene, sequence analysis of passaged virus revealed mutations in the p1/p6 (P1' residue Leu to Phe) and p7/p1 (P2 residue Ala to Val) gag proteolytic processing sites. The p1/p6 mutation appeared in several clones derived from early passages and was present in all clones obtained from passage P11 (0.42 microM ABT-378) onward. The p7/p1 mutation appeared very late during the selection process and was strongly associated with the emergence of the additional change at residue 47 (V47A) and the reversion at residue 32 back to the wild-type sequence. Furthermore, this p7/p1 mutation was present in all clones obtained from passage P17 (3.0 microM ABT-378) onward and always occurred in conjunction with the p1/p6 mutation. Full-length molecular clones containing protease mutations observed very late during the selection process were constructed and found to be viable only in the presence of both the p7/p1 and p1/p6 cleavage-site mutations. This suggests that mutation of these gag proteolytic cleavage sites is required for the growth of highly resistant HIV-1 selected by ABT-378 and supports recent work demonstrating that mutations in the p7/p1/p6 region play an important role in conferring resistance to protease inhibitors (L. Doyon et al., J. Virol. 70:3763-3769, 1996; Y. M. Zhang et al., J. Virol. 71:6662-6670, 1997).

FIG. 1

Structure of ABT-378. Ph, phenyl.

FIG. 2

Sequence analysis of the protease coding region from HIV-1 passaged with ABT-378. The amino acid sequence of the protease coding region from clones derived from 13 different passages is indicated. The fraction of clones containing each unique protease sequence is indicated on the right. The top line shows the protease sequence of the wild-type pNL4-3 clone. Identity with this sequence at individual amino acid positions is indicated by dashes.

FIG. 3

Backbone diagram of the dimeric HIV-1 protease. The backbone trace of the dimeric HIV-1 protease is denoted by the thin line. A model of ABT-378 bound in the active site is shown in thick lines at the center of the figure. The carboxy and amino termini of the protein are denoted by C and N, respectively. The eight residues which are commonly mutated during in vitro selection with ABT-378 are indicated by the spheres on both symmetry-related chains of the protein. Three of these residues lie within the active site of the protease (residues 32, 47, and 84), while the other five residues lie outside of the active site (residues 10, 16, 46, 69, and 91).

FIG. 4

Appearance and frequency of p7/p1 and p1/p6 cleavage-site mutations during in vitro selection with ABT-378. A schematic diagram of the gag and pol open reading frames is shown at the top of the figure. The p7/p1 and p1/p6 cleavage sites are indicated by the arrows. During in vitro selection with ABT-378, the p7/p1 cleavage site was altered from AN/F to VN/F (P₂ residue Ala to Val). The p1/p6 cleavage site was altered from F/L to F/F (P₁′ residue Leu to Phe). For the viral passages indicated on the left-hand side of the table, the fraction of clones containing each cleavage-site mutation is shown. No mutations were observed at any of the other cleavage sites (3). MA, matrix; CA, capsid; TF, transframe protein; PR, protease.

FIG. 5

Panel of full-length HIV-1 DNA clones. Amino acid sequences for the wild-type (wt) pNL4-3 p7/p1 and p1/p6 cleavage sites are indicated above the schematic diagram of the p7/p1/p6/protease gene region, while the sequences of the mutated p7/p1 and p1/p6 cleavage sites are shown underneath. In addition, the wild-type sequences for the eight common amino acids which undergo mutation during in vitro selection with ABT-378 are indicated. Beneath the diagram, the protease (prot) sequences for the panel of HIV-1 DNA clones are shown. Clones were named to reflect the initial viral passage in which that particular protease sequence was observed. The presence of the p1/p6 mutation (mut) is designated by m1 in the name of the clone, while the presence of the p7/p1 mutation is designated by m2 in the name of the clone. Clones indicated by the asterisks failed to generate infectious virus in either MT-4 cells, CEM cells, or PBMCs and are therefore not included in Table 4.

FIG. 6

Model of the HIV-1 protease active site and p7/p1 substrate. The left-hand side of the figure (A, C, and E) depicts schematic diagrams of the active-site residues 32, 47, and 84 present in the wild-type protease (A), P16 virus (C), and P17 virus (E). The approximate outline of the S₂ pocket is indicated by a red arc, underneath which a schematic of the corresponing substrate P₂ residue present at each stage is shown. The right-hand side of the figure (B, D, and F) depicts atomic representations derived from three-dimensional molecular modeling of the interactions shown in the matching left-hand side of the figure and present in the wild-type protease (B), P16 virus (D), and P17 virus (F). Active-site residues 32, 47, and 84 of the crystal structure of the HIV-1 protease complexed with the inhibitor MVT-101 (Protein Data Bank entry 4HVP) are shown in green, and the boundary of the S₂ pocket is delineated by the light-green–black surface. A hexapeptide model of the p7/p1 substrate (Gln-Ala-Asn-Phe-Leu-Gly), based on the MVT-101 inhibitor structure, is shown with brown carbons, blue nitrogens, and red oxygens. A brown transparent surface over the side chain of the substrate P₂ residue is shown. The side chains of the active-site aspartate residues (Asp 25 and Asp 125) are also shown near the Asn-Phe scissile bond of the substrate.