An isolate of human immunodeficiency virus type 1 originally classified as subtype I represents a complex mosaic comprising three different group M subtypes (A, G, and I)

Abstract

Full-length reference clones and sequences are currently available for eight human immunodeficiency virus type 1 (HIV-1) group M subtypes (A through H), but none have been reported for subtypes I and J, which have only been identified in a few individuals. Phylogenetic information for subtype I, in particular, is limited since only about 400 bp of env gene sequences have been determined for just two epidemiologically linked viruses infecting a couple who were heterosexual intravenous drug users from Cyprus. To characterize subtype I in greater detail, we employed long-range PCR to clone a full-length provirus (94CY032.3) from an isolate obtained from one of the individuals originally reported to be infected with this subtype. Phylogenetic analysis of C2-V3 env gene sequences confirmed that 94CY032.3 was closely related to sequences previously classified as subtype I. However, analysis of the remainder of its genome revealed various regions in which 94CY032.3 was significantly clustered with either subtype A or subtype G. Only sequences located in vpr and nef, as well as the middle portions of pol and env, formed independent lineages roughly equidistant from all other known subtypes. Since these latter regions most likely have a common origin, we classify them all as subtype I. These results thus indicate that the originally reported prototypic subtype I isolate 94CY032 represents a triple recombinant (A/G/I) with at least 11 points of recombination crossover. We also screened HIV-1 recombinants with regions of uncertain subtype assignment for the presence of subtype I sequences. This analysis revealed that two of the earliest mosaics from Africa, Z321B (A/G/?) and MAL (A/D/?), contain short segments of sequence which clustered closely with the subtype I domains of 94CY032.3. Since Z321 was isolated in 1976, subtype I as well as subtypes A and G must have existed in Central Africa prior to that date. The discovery of subtype I in HIV-1 hybrids from widely distant geographic locations also suggests a more widespread distribution of this virus subtype, or at least segments of it, than previously recognized.

FIG. 1

Confirmation of 94CY032.3 (highlighted) as a subtype I representative in the C2-V3 envelope region. Reference sequences for all known group M subtypes were obtained from the Los Alamos sequence database (13), aligned using CLUSTAL W (26), and adjusted manually using the alignment editor MASE (7). Sites for which there was a gap in any of the sequences were excluded from all sequence comparisons. Pairwise evolutionary distances were estimated by using Kimura’s two-parameter method to correct for superimposed substitutions (12). Phylogenetic trees were constructed by the neighbor-joining method (20), and the reliability of topologies was estimated by performing bootstrap analysis using 1,000 replicates (8). Values at nodes indicate the percentage of bootstraps in which the cluster to the right was found (bootstrap values of 80% and higher are shown). Branch lengths are drawn to scale.

FIG. 2

Exploratory tree analysis. Neighbor-joining trees were constructed for a 400-bp window moved in increments of 10 bp along the multiple genome alignment. Trees depicting discordant branching orders for 94CY032.3 (highlighted) are shown. The position of each tree in the alignment is indicated; subtypes are identified by brackets. Numbers at nodes indicate the percentage of bootstrap values with which the adjacent cluster is supported (only values above 80% are shown). Branch lengths are drawn to scale.

FIG. 3

Bootstrap plot analysis to map recombination breakpoints in 94CY032.3. Bootscanning was performed essentially as described previously (9, 10, 22), plotting the magnitude of the bootstrap value supporting the clustering of 94CY032.3 with 92UG037.1 (subtype A in red) in comparison with that of 94CY032.3 and 92NG083.2 (subtype G in green) for a window of 400 bp moved in increments of 10 bp along the alignment. Regions of subtype A or G origin are identified by very high bootstrap values (>80%). The location of eight recombination crossovers is indicated (see text for details). Breakpoint analysis between positions 4200 and 4800 was not possible due to the recombinant nature of 92NG083.2 (shaded) (9). The beginnings of gag, pol, vif, vpr, env, and nef open reading frames are shown. The y axis indicates the percent bootstrap replicates which support the clustering of 94CY032.3 with representatives of the respective subtypes.

FIG. 4

Recombination breakpoint analysis of 94CY032.3 in the vif/vpr region. (A) Diversity plots comparing 94CY032.3 to reference sequences from the database (U455, MN, and ELI) as well as to a recently identified (partial) nonrecombinant subtype G sequence (95ML045) (2). Sequences are color coded. Distance values were calculated for a window of 300 bp moved in steps of 10 nucleotides. The x axis indicates the nucleotide positions along the alignment (gaps were stripped and removed from the alignment). The positions of the start codons of the vif, vpr, and tat genes are shown. The y axis denotes the distance between the viruses compared (0.05 = 5% difference). (B) Neighbor-joining trees depicting the position of 94CY032.3 in regions flanking the breakpoints identified in panel A. Trees were constructed from the genomic regions indicated. Subtypes are identified by brackets. Four sequences from Mali represent subtype G; these are the only available subtype G reference sequences in this region, since all other characterized subtype G viruses contain A fragments (2, 9, 16, 17). Numbers at nodes indicate the percentage of bootstrap values with which the adjacent cluster is supported; only values above 80% are shown. Branch lengths are drawn to scale.

FIG. 5

Inferred structure of the mosaic 94CY032.3 genome. Regions of different subtype origin are color coded. All breakpoints are tentative, since full-length nonrecombinant subtype G and I reference sequences are not available. LTR sequences were not analyzed and are thus shown as hatched boxes.

FIG. 6

Screening of intersubtype recombinants containing regions of uncertain subtype assignment for the presence of subtype I sequences. (A) Schematic representation of the recombinant genomes of MAL and Z321. Fragments of subtypes A, D, and G origin are color coded red, blue, and green, respectively; segments of uncertain subtype assignment are shown in white. Arrows denote regions of unusual sequence similarity with 94CY032.3 as determined by diversity plot analysis (data not shown). (B and C) Phylogenetic analysis of the putative subtype I domains of MAL and Z321 delineated in panel A. Neighbor-joining trees of partial pol (A) and vif/vpr (B) sequences. Horizontal branch lengths are drawn to scale; the scale bar represents 0.1 nucleotide substitutions per site. Vertical separation is for clarity only. Values at the nodes indicate the percent bootstraps in which the cluster to the right was supported; bootstrap values of 80% and higher are shown. Brackets at the right represent the major sequence subtypes of HIV-1 group M. Subtype I clusters are highlighted; trees were midpoint rooted.