INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Accumulated molecular analyses on the genetic diversity of human immunodeficiency virus type 1 (HIV-1) have clarified that this virus family can be classified into group M (major), a rare group O (outlier), and a new group N and that group M can be divided into at least 10 subtypes, designated A through J (19, 21, 28). Recent studies have provided increasing evidence for the importance of recombination in the genetic diversification of HIV-1. It has been shown that 5 to 10% of the HIV-1 strains that have been fully or partially sequenced possess mosaic genomes composed of two different HIV-1 group M subtypes (4, 9, 24). So far, no sequences that are hybrids of group M and group O viruses have been found. In many cases, putative intersubtype recombinants have originated from geographic areas where multiple subtypes are known to be cocirculating, such as Central Africa (1, 4, 8, 27), South America (25), and Southeast Asia (2, 10).

We recently reported the molecular epidemiology of HIV-1 and HIV-2 in Cameroon (29). There is no doubt that a large variety of HIV-1 subtypes and intersubtype recombinants are cocirculating in Cameroon, where almost all the HIV-1 subtypes (A through H), group O, and group N have been characterized (12, 17, 18, 22, 28, 29). In addition, various types of mixed infection, such as between different subtypes of HIV-1 group M, between HIV-1 and HIV-2, and even between HIV-1 groups M and O, were confirmed to occur at a rather high frequency (approximately 10%) in the Cameroonian specimens (29). It seems that the mixed infections with different HIV strains are not as rare as previously thought. Moreover, another in vitro study (16) has shown that mixed infection resulted in biologically active recombinant viruses that spread rapidly. Recombinant viruses appear to have the potential to lead to global pandemics.

We previously reported that one patient, a 22-year-old single Cameroonian woman, was infected with three different HIV-1 strains, a group O virus, a subtype D virus, and an IBNG (A/G)-like recombinant virus (30). Initially, we thought that the third strain was a subtype A virus, but we have found that it was a recently reported IBNG (A/G)-like recombinant virus whose genome contains partial subtype G sequences in a mainly subtype A genome (for details, see Results and Discussion). Here we present the first genetic analysis of HIV-1 "intergroup" (M/O) recombinant viruses from this triply infected patient. We have found multiple segmental exchanges among three HIV-1 strains (O, D, and IBNG) on a single clone in the peripheral blood mononuclear cells (PBMCs) of this individual. Recombinational events such as this case are important because they contribute to the production of new variants of HIV-1.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Subject. The patient studied (cm61) was a 22-year-old (1994) single Cameroonian woman diagnosed with AIDS-related complex. She had been living in a suburb near Yaoundé, the capital city of Cameroon, for several years and was occasionally hospitalized with fever, severe weight loss, and chronic diarrhea. As risk factors for infection, she reported only frequent heterosexual contacts. Sera were tested as described previously (29, 30).

Tissue culture. PBMCs from the patient were separated by Ficoll gradient sedimentation and cocultivated with virus-free human PBMCs from which CD8⁺ cells were removed or with human T-lymphoid cell lines (molt4#8 [14] or M8166 [3]). All the cultures were maintained for several weeks, and the supernatant was monitored for reverse transcriptase activity as described previously (31) or for the presence of p24/p27 antigen (SIV core antigen assay; Coulter, Miami, Fla.).

PCR and cloning of viral genome segments. Chromosomal DNA was extracted from PBMCs by using glass milk powder (Prep-A-Gene DNA purification kit; Bio-Rad, Hercules, Calif.). The genome (corresponding to nucleotides [nt] 4075 to 7113 in HIV-1_LAI) was amplified by nested extra-long PCR (XL-PCR kit; Perkin-Elmer, Foster City, Calif.) with primers HIV-1pol3 (5'-TAAAAGGAGAAGCCATGCATGGACAAGTAGA-3') and M10 (5'-CCAATTGTCCCTCATATCTCCTCCTCCAGG-3') in the first round and unipol1 (5'-AGTGGATTCATAGAAGCAGAAGT-3') and M8 (5'-TCCTTGGATGGGAGGGGCATACATTGC-3') in the second round. Cycling conditions included a hot start (1 min at 95°C), then 10 cycles of denaturation (95°C) for 15 s and extension (68°C) for 5 min, and then 20 cycles of denaturation (95°C) for 15 s and extension (65°C) for 15 min, with 15-s increments per cycle. The XL-PCR products (approximately 3,070 bp) were shortened by use of a kilosequence deletion kit (Takara Shuzo Co. Ltd., Otsu, Japan). Smaller, overlapping fragments containing the pol region (corresponding to nt 4075 to 4362 in HIV-1_LAI) and the env region (corresponding to nt 6567 to 7113 in HIV-1_LAI) of the genome were amplified by conventional PCR and cloned separately with PCR primers as described previously (29). DNA sequencing was carried out by the dideoxy chain termination method with an automated DNA sequencer (373A; Applied Biosystems, Foster City, Calif.). At least 10 plasmid clones were sequenced to obtain the consensus sequence.

Phylogenetic analysis. DNA sequences were aligned by the ODEN program package of the National Institute of Genetics, Mishima, Japan (13). Nucleotide substitutions of all pairs of the sequences were estimated by the two-parameter (15) and six-parameter (11) methods. The phylogenetic tree was constructed by the neighbor-joining method, and its reliability was estimated by 100 bootstrap replications (5).

Distance plots. The query sequences were first aligned with a set of reference sequences representing all the established genetic subtypes of HIV-1 (9). The genetic distance (two-parameter and six-parameter) between selected pairs of sequences was determined by moving a window of 300 bp along the genome alignment in 25-bp increments. The distances were plotted at the midpoint of the 300-bp segments.

Bootscanning. Bootstrap plots (26) were performed on the neighbor-joining trees for a window of 300 bp moving along the alignment in increments of 25 bp. We evaluated 100 replicates generated by the bootstrap resampling for each phylogeny. The percent bootstrap probabilities for each topology were plotted at the midpoint of each window. Breakpoints were assigned to the midpoint of the transitions between segments from different subtypes.

Distribution of informative sites. To localize intragenic crossover points between regions of DNA sequences, the distribution of phylogenetically informative sites supporting alternative tree topologies was inspected. This was done by surveying the informative sites in a four-sequence alignment including the putative recombinant sequence. There are three possible configurations of the informative sites, two of which support the clustering of the putative recombinant with one parental lineage or the other. The distribution of these two types of sites can be tested by determining whether a break placed at any point along the alignment produces a significant difference in the ratio of the two types of sites on each side of the cut, as assessed by a chi-square value with Yates's correction for continuity; the optimum position of the breakpoint can be found by maximizing this value (23, 24).

Nucleotide sequence accession numbers. The nucleotide sequences in this study have been assigned GenBank accession no. U58148 to U58159, AF023085, AF055728 to AF055732, and AF097692 to AF097698.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Serological data and virus isolation. Blood samples were collected from the patient (cm61) in June 1994, June 1995, and June 1996. All three serum specimens (1994, 1995, and 1996) were dually reactive for HIV-1 and HIV-2 by a particle agglutination test and a Western blotting assay. All sera were then discriminated by using a peptide-based assay and judged to be HIV-1 positive. Virus isolation was performed twice (in 1995 and 1996) by techniques routinely used in our laboratory, but attempts to isolate an infectious HIV-1 strain from subject cm61 were unsuccessful.

Phylogenetic analysis of the env and pol regions. The phylogenetic positions of 94CM61, 95CM61, and 96CM61 were first examined in evolutionary trees based on the env and pol gene sequences. The central portion of gp120 including the V3 region was sequenced, and a phylogenetic tree was constructed (Fig. 1A). Three different types of sequences, subtype A, subtype D, and group O, in the same genomic region were found in each of the three sampling years. The presence of three different sequences in one individual clearly demonstrated that this individual had a triple infection with HIV strains of different origins. It should be noted that the sequences of subtype A were tightly clustered with an IBNG strain. This IBNG strain has been recently reported to be a mosaic virus of subtypes A and G (1). Surveys in Nigeria and Cameroon suggest that this virus is more prevalent than the nonrecombinant subtype A (10, 18).

View larger version (73K): [in this window] [in a new window]   FIG. 1.   Phylogenetic trees of HIV-1 sequences. The trees were constructed from part of the env sequences including those in the V3 region (approximately 390 bp) (A) and part of the pol sequence that encodes integrase (288 bp) (B). The results obtained by the six-parameter method are shown. Each of the pol and env sequences obtained in 1994, 1995, and 1996 are boxed, and the pol and env sequences from the M/O recombinants are circled. Subtypes are indicated by brackets. Bootstrap values of key nodes are shown.

Clones and sequences of viral genome segments obtained by XL-PCR amplification. HIV-1 genomic sequences that were obtained in 1995 and 1996 were repeatedly confirmed by independent XL-PCR amplification and molecular cloning. Twelve clones from the pol region to the env region that were nearly 3,000 bp were obtained from uncultured PBMCs collected in 1995, and seven clones from the same region were obtained from uncultured PBMCs collected in 1996. The purified XL-PCR products were digested with several restriction enzymes (BamHI, EcoRI, HindIII, KpnI, SacI, SalI, SphI, and XbaI). Analysis of these digestions indicated that the former 12 clones were of five types (representative clones were 95CM61.20, 95CM61.34, 95CM61.49, 95CM61.55, and 95CM61.56) and the latter 7 clones were of three types (representative clones were 96CM61.2, 96CM61.4, and 96CM61.5). Complete sequences of each of the above five plus three clones revealed seven potential open reading frames corresponding to the partial pol, vif, vpr, 5' tat, 5' rev, vpu, and partial env genes. (After sequencing, the homology between 95CM61.49 and 95CM61.55 turned out to be 98.1%, and these two clones were found to have almost the same mosaic structure.)

Genetic distances and bootscanning. We next determined which segments of the 95CM61 and 96CM61 genomes were derived from the group O, subtype D, or subtype A viruses. Whether a subtype A segment was derived from a typical subtype A virus or from an IBNG (A/G)-like virus, as we presumed above, also needed to be clarified. Distance values were calculated by the two-parameter and six-parameter methods for a window of 300 nt which was moved in steps of 25 nt (2, 10). Since we found no difference between the results obtained by the two methods, only the results obtained by the six-parameter method are shown (Fig. 2A and 3A). These figures show the distance values between 95CM61 (or 96CM61) and 11 other HIV-1 reference strains for each location in the genome. The reference strains are subtypes A (U455), B (LAI), C (C2220), D (NDK), F (93BR020.1), G (92NG083.2), H (90CF056.1), N (YBF30), and O (MVP5180), as well as A/E (93TH253.3) and A/G (IBNG) (9). Note that in the absence of a nonmosaic, full-length subtype G genome (9, 21), we used 92NG083.2, which is known to contain a small A segment in the vif-vpr region, as a subtype G reference, Figure 2A shows the distance plots of four representative clones of 95CM61. The results of 96CM61.2 and 96CM61.4 were quite similar to those of 95CM61.34 and 95CM61.49, respectively. The distance plot profile of the remaining 96CM61.5 was different from those of the preceding four and therefore is shown in Fig. 3A. The genotype that exhibited the parental strain in one segment generally followed the lowest distance in this analysis. Concerning the region from positions 1 to 1175 (the end of the vif region), for example, the distance between 95CM61 (or 96CM61) and MVP5180 was much smaller than the distances between 95CM61 (or 96CM61) and each sequence of the group M subtypes. The same trend was also observed in other clones. This observation indicates that the region from positions 1 to 1175 was derived from a group O-like virus and that a crossover occurred near the end of the vif gene, which is also the 5' beginning of the vpr gene. As is clearly shown in each plot in Fig. 2A and 3A, 95CM61 and 96CM61 genomes could be related to only group O, subtype D, or IBNG (A/G) viruses and other subtypes were genomically rather distant. After showing the short distances of the respective clones to group O in the pol and vif regions, their distances to IBNG narrowed after the 5' portion of vpr but the patterns of distances among the seven clones were not uniform, especially after the 5' tat region. It is emphasized that the distance to IBNG was shorter than to a typical subtype A virus (U455) in the vpr to env regions. These findings suggest that subtype D and IBNG-like viruses were probably parental strains and that various recombinants might have arisen as offspring in this patient in 1995 and thereafter.

View larger version (67K): [in this window] [in a new window]   FIG. 2.   Analysis of mosaic structures of four types of 95CM61. (A) Each of four respective clones (95CM61.56, 95CM61.20, 95CM61.34, and 95CM61.49) was aligned with 11 full-length HIV-1 isolates (MVP5180, YBF30, U455, LAI, C2220, NDK, 93TH253.3 93BR020.1, 92NG083.2, 90CF056.1, and IBNG) representing groups O and N, subtypes A to H, and A/G recombinant, respectively, and the alignment was sectioned into 300-nt segments, which were moved in steps of 25 nt. The genetic distances between 95CM61 and the respective reference strains were calculated by using the six-parameter method. (B) Breakpoints were fine-mapped by using a five-sequence alignment consisting of 95CM61, U455, NDK, 92NG083.2, and MVP5180. Bootstrap values for the node with either subtype A, subtype D, subtype G, or group O out of 100 bootstrap replications built by the neighbor-joining method were plotted. (D) Because IBNG is known to represent a mosaic of subtypes A and G, breakpoints were then fine-mapped by using a four-sequence alignment consisting of 95CM61 and putative parental strains (IBNG, NDK, and MVP5180), and the magnitude of the bootstrap value supporting the clustering of 95CM61 with IBNG (A/G), subtype D, or group O was plotted, respectively. The distance value (A) and the bootstrap value (B and D) for each segment were plotted at the midpoint of the segment. (C and E) A map of the open reading frames in part of the HIV-1 genome is shown. Segments derived from group O (blue), subtype A (red), subtype D (black), subtype G (green), and an IBNG-like virus (pink) were mapped by the results of diversity and bootstrap plots, as well as phylogenetic tree analyses.

View larger version (34K): [in this window] [in a new window]   FIG. 3.   Analysis of mosaic structures of 96CM61.5. See the legend to Fig. 2 for details of the method. (The results for 96CM61.2 and 96CM61.4 were quite similar to those for 95CM61.34 and 95CM61.49, respectively.)

View larger version (34K): [in this window] [in a new window]   FIG. 4.   Phylogenetic relationships of 95CM61 and 96CM61 in the signal peptide and C1 regions (approximately 400 bp). The tree was rooted by using group O and SIVcpzGAB as outgroups. The sequences of 95CM61 and 96CM61 are circled. Subtypes are indicated by brackets. Bootstrap values of key nodes are shown.

Recombination breakpoint analysis based on informative site distribution analysis. Tables 1 and 2 show the results of an investigation into the distribution of phylogenetically informative sites around the observed breakpoints. Breakpoints were inserted at each possible point between adjacent informative sites, and a 2 × 2 heterogeneity chi-square value with Yates's correction for continuity was calculated for the sites to either side of the breakpoint that supports the clustering of the putative recombinant with each of the consensus sequences. Within the precision attainable by this technique, common breakpoints were identified at the end of vif in all the clones (Tables 1 and 2). Other breakpoints in 95CM61.34, 96CM61.2, and 96CM61.5 were also identified by this analysis. As described in the preceding section, it is evident that both 95CM61.49 and 96CM61.4 had a crossover point from an IBNG-like virus to a subtype D-like virus. However, there was no evidence that 95CM61.20 is an "intersubtype" (IBNG-D-IBNG) recombinant. The lengths from the vpr region to the 5' tat-rev region were probably too short and contained no apparent informative sites within these ambiguous sequences (Table 1).

Evidence of "intergroup" recombination. In the present study, to exclude the possibility of PCR-mediated artifacts, the viral clones were verified on two independent XL-PCR amplifications by using sequential blood samples obtained in 1995 and 1996, and essentially the same mosaic patterns for 95CM61.34 and 96CM61.2 and for 95CM61.49 and 96CM61.4 were obtained. Our multiple recombinants are clearly not the products of PCR-mediated artifacts, because it is unlikely for recombinant Tth DNA polymerase to repeatedly generate crossovers at the same positions.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

To clarify the ongoing process of generation and selection of HIV-1 variants, we have used a set of genetic analyses to identify M/O recombinants in this study. First, the presence of three different sequences in the pol and env regions in patient cm61 in 1994 clearly demonstrated that this patient had a triple infection with HIV strains: a group O virus, a subtype D virus, and an IBNG-like virus (Fig. 1). The existence of three different HIV-1 env sequences was confirmed in sequential samples taken 1 year apart: 95CM61 was obtained in 1995, and 96CM61 was obtained in 1996 (Fig. 1A). Nevertheless, in the phylogenetic analysis based on the pol region, we detected only one type (group O) in cm61 in both 1995 and 1996 (Fig. 1B), although the PCR conditions were the same as those used in 1994. At that point, we thought that recombination might have occurred between groups M and O. Thus, we attempted to amplify 95CM61 and 96CM61 by XL-PCR spanning the region from pol to the env. As a result, four different types of genome were obtained from 95CM61 and three types were obtained from 96CM61. Two of the latter were the same as two of those of 95CM61.

When and how these HIV-1 M/O recombinants arose in this individual are of particular interest. Before answering these question, it is important to distinguish between (i) mixed infection of this patient with three genetically distinct HIV strains and (ii) infection with an "intergroup" recombinant HIV strain. It is evident that three different strains (O, D, and IBNG) were present in this patient in 1994, because these three sequences were demonstrated in both the pol and env regions. Nevertheless, we did not obtain any evidence for the presence of such viruses by XL-PCR in patient cm61 in 1995 (95CM61) and 1996 (96CM61). However, phylogenetic analyses based on the pol and env regions revealed that respective sequences of the pol and env regions from the M/O recombinants were essentially identical to those of the pol and env regions, which were obtained by independent PCRs with the specific primers for each region (Fig. 1). The viral population of the pol sequences converged on a single form (form O) in 1995 and 1996, in contrast to a mixture of forms [forms O, IBNG (A/G), and D] in 1994. Moreover, this patient claimed that she had no history of blood transfusion or intravenous drug use. So-called IBNG-like A/G recombinants were recognized much earlier than 1994 (1) and have been common in Cameroon (29). In addition, we could not find any subtype G sequences in the env analysis (Fig. 1A), and the distance plots indicated that the distances for IBNG were continuously shorter than those for subtypes A and G (Fig. 2A and 3A). Based on these facts, we conclude that the patient initially had a mixed infection consisting of group O, subtype D, and IBNG-like viruses and that M/O recombination presumably occurred in this patient after she acquired the mixed infection, although we cannot exclude the possibility of simultaneous transmission with parental and recombinant strains.

Additional evidence supports the derivation of one M/O recombinant from another in the patient. This is based on the analysis of identified breakpoints in four 95CM61 recombinants and three 96CM61 recombinants that were obtained 12 months later. Figures 2D and 3D show the results of bootscanning around the observed breakpoints, and Tables 1 and 2 show the results of the distribution of phylogenetically informative sites around the observed breakpoints. The breakpoints between genetic segments from group O to M were at the 3' end of vif and were identically present in all seven of the molecular clones that were obtained in 1995 and 1996, even though the sequences in the vif-to-vpr region are not similar between groups M and O. A possible explanation for the existence of a common breakpoint in all the recombinants is that the recombinants were derived from a common M/O ancestor. Another possibility is that some selection process occurred after recombinational events between two highly divergent HIV-1 groups. In other words, only selected recombinants in the pool of newly generated recombinant viruses may have survived due to their having a certain breakpoint (such as between vif and vpr) that gave them an evolutionary advantage. Coincidentally, we have found the same crossover point in vif of YBF30, the only fully sequenced strain of HIV-1 group N. This strain is known to be a recombinant of divergent viral lineages within the HIV-1/SIV_cpz group (7). Other breakpoints appeared mostly near the boundaries of the respective genes. The high frequency of recombination that occurs only near the beginning or end of the respective genes seems to reflect a common adaptive strategy for recombination. In addition, it is noteworthy that respective recombinants rather quickly changed their population within the patient. At least four types of recombinants, varying from a rather simple type (O-IBNG) to more complex types of mosaicisms (O-IBNG-D and O-IBNG-D-IBNG), were present in the patient PBMCs. At least two of the four types existed continuously as revealed in sequential blood samples obtained in 1995 and 1996. These analyses strongly suggest that earlier recombinant forms served as templates for the generation of later ones. However, we were unable to isolate an infectious HIV-1 strain from this patient. Further investigations are needed to clarify what kind(s) of rules determines the survival and subsequent population dynamics of recombinant viruses.

With the spread and mixing of various HIV-1 subtypes, as has been shown to occur in this central African country, the opportunity for coinfection or superinfection has been increasing dramatically in recent years, and subsequent recombinants among these strains are very likely to be generated. It should be emphasized that recombination between groups M and O can occur in vivo, even though the homology between these groups is only 65% on average across the genome. Recombination must be viewed as a viable mechanism not only for increasing HIV-1 variation in infected individuals but also for contributing to the generation of new HIV variants. This fact has also important implications for HIV vaccine strategies that are designed to employ live attenuated viruses (6, 32), which potentially could form recombinants with wild-type strains even if the two viruses are quite divergent.