With informed consent, interviews and blood samples were obtained from 26 HIV-seropositive Lebanese citizens (17 men and 9 women) living in Lebanon at the time of the study. They were referred (in 1996) from both urban and rural areas to the Medical Center of the American University of Beirut or the Saint George Hospital in Beirut, Lebanon.

HIV-1 antibodies were discriminated from HIV-2 antibodies by a synthetic slot immunoblot composed of recombinant viral proteins and synthetic peptides (Genelavia Mixt, Peptilav 1-2, HIV Blot 2.2-Diagnostic Biotechnology and New LAV Blot II, Sanofi Diagnostic Pasteur, France).

Double-stranded viral DNA of nested polymerase chain reaction (PCR)-amplified env(C2-V3), pol (protease-prot), and gag (p24 fragment) from peripheral blood mononuclear cells were direct sequenced (12,13). Comparative phylogenetic analysis of two or three independent viral regions enables identification of a possible mosaic viral genome (2). The sequences were aligned by use of the CLUSTAL multiple-sequence alignment program (14). After regions containing gaps were eliminated, the aligned sequences were analyzed with the maximum likelihood method by using the fastDNAml program, which uses randomized data input and global rearrangement (15). Additionally, the neighbor-joining method (PHYLIP package version 3.5c [16] or TREECON software [17]) was used, with or without bootstrapping. The stability of the tree's topology was tested by pruning, which consists of removing one species from the alignment and rerunning the phylogenetic analysis. SIV-cpz sequences were used as outgroup. All specimens were also screened for HIV-2 DNA by nested-PCR amplification of viral HIV-2 prot gene. The outer PCR primers were DP20: 5'GACAGAGGACTTGCTGCA; nucleotide position: 2068-2083, HIV-2_ROD, and DP21: 5'GGCCATTGTCTCAGTTTTGG; nucleotide position: 2435-2454. The inner primers were DP26a: 5'CACCTCAATTCTCTCTTTGGA; nucleotide position: 2082-2102; and DP27: 5'TAGATTTAATGACATGCCTAA; nucleotide position: 2360-2380. Analysis based on HIV-2 pol region is highly specific and allows classification of HIV-2 sequences into phylogenetic subtypes (3,18,19).

Figure 1

Figure 1. Phylogenetic classification of HIV-2 sequence from Lebanese patient LE25 (arrow) basing on the prot gene (GenBank accession no. AF026912). The tree was generated by the maximum-likelihood method. Numbers at the branch...

Of 26 patients, 25 were HIV-1 PCR-positive; the remaining patient (LE25) was HIV-2 PCR-positive and was classified as having subtype B infection (Figure 1). No dual infections with HIV-1 and HIV-2 strains were found.

Figure 2

Figure 2. Phylogenetic classification of HIV-1 sequences from Lebanese patients denoted with LE prefix (GenBank accession nos. AF025691- AF025745). The trees were constructed on the basis of the DNA sequences of prot (a),...

The comparative phylogenetic analysis of 25 HIV-1 specimens identified 23 cases with the same subtype assignment for the env and pol sequences and classified 22 as known subtypes A (n = 10), B (n = 10), C (n = 1), D (n = 1) (Figure 2). In contrast, the remaining specimen (LE7) consisted of distinctively divergent pol and env sequences (Figure 2a, b) as they failed to cluster with known standard HIV-1 sequences, including divergent env subtypes I and J (Figure 2c). Intragene recombination within pol and env sequences of specimen LE7 was not observed when nucleotide signature patterns in HIV-1 subtypes were compared. Thus, this strain might represent a potential novel HIV-1 subtype. However, gag sequence of this virus grouped into HIV-1 subtype A, and this association was supported by the bootstrap value of 90% (Figure 2d). Taken together, these results indicate that LE7 viral genome has a mosaic pattern involving elements of an unclassifiable-divergent strain and subtype A (unclassifiable^env/ unclassifiable^pol/A^gag). In addition, potential recombinant viruses combining viral env, pol, and gag from different subtypes were identified in two other specimens. Two known recombinant patterns, G^env/A^pol and A^env/D^pol/A^gag, have been found in patients LE24 and LE9, respectively. However, a potential recombinant is sometimes difficult to recognize, especially between subtypes B and D viruses, which are relatively closely related; specimen LE22 illustrates such a situation. The protease gene sequence, the most outside branch in this clade, grouped weakly with subtype B (Figure 2a). This association, however, was not stable by analysis that used pruning techniques and not supported by bootstrap analysis (only 48% value as opposed to 72% without this sequence). These data suggest that it is unlikely that LE22 prot is subtype B, and because both env and gag sequences clustered into subtype D, we have classified this sample as subtype D.

To better understand the transmission dynamics and temporal introduction of these HIV variants, we analyzed patients' data (demographics, behavioral risk, modes of transmission, and travel histories) (Table).

Of eight male patients with subtype B HIV-1 infection, six traveled to Europe, the Americas, Asia (Thailand), or different locations in the Middle East. All reported sexual contacts abroad either with commercial sex workers or homosexual or bisexual men. Of the remaining two patients, one traveled selectively to Africa and had sexual contacts in Côte d'Ivoire and Nigeria, whereas the other reported no travel outside Lebanon and had no sexual contacts with commercial sex workers. In contrast, all nine male patients infected with non-subtype B HIV-1, including three mosaic viruses, traveled to Africa or the United Arab Emirates and reported sexual contacts with commercial sex workers.

The behavioral risks of the female patients differed substantially from those of the male patients. All female patients were married, and none reported high-risk activities or contacts with commercial sex workers. Moreover, all female patients (except one [LE12], who had acquired HIV-1 infection from blood transfusion) reported that their spouses were HIV-infected and traveled abroad with them or alone.

To determine if the female patients acquired HIV infection from their spouses, we analyzed available epidemiologic data on the spouses of three female patients LE17, LE19, and LE13. The female patients LE17 and LE19 were spouses of patients LE18 and LE23, respectively. These two couples, LE17/LE18 and LE19/LE23, were infected with HIV-1 subtype A and subtype B viruses, respectively (Figure 2b). In further detailed pairwise analysis of viral DNA isolated from the husband and wife, a small nucleotide divergence of 3.1% within env and of 0.6% for both prot and gag (13,20) showed that in the first couple viral strains were closely related. These data are concordant with epidemiologic findings, which strongly indicate that the husband was the source of infection because of his high-risk sexual activities with commercial sex workers during travels to the United Arab Emirates. His wife did not travel outside Lebanon and did not report activities that would have placed her at risk for HIV-1 infection. In the second couple (LE19/LE23), however, only weak relation between viral strains was observed (nucleotide divergence of 9.8% within env and 2% within prot). This could be due to many factors including 1) the time that elapsed from infection; 2) transmission of different quasispecies to the spouse; 3) detection of only major quasispecies by direct sequencing, which differed in the couple; and 4) a different source of HIV-1 infection in the sex partner. The latter hypothesis was not supported by epidemiologic data, which indicated that only the husband was involved in high-risk sexual activities in and outside Lebanon. The third female patient, LE25, was infected with HIV-2 subtype B strain. Her husband, who died of AIDS-related complications, had traveled to Côte d'Ivoire, a region in which this HIV variant is endemic (21). These data suggest that very likely the female patients acquired infections from their HIV-infected spouses.

The molecular and epidemiologic data indicate multiple HIV-1 introductions to Lebanon. First, most male patients were infected abroad with the HIV-1 strains prevalent in the geographic region they visited. Second, the topology of HIV-1 env subtype A demonstrated that viral sequences of patients who had sexual contacts with persons from West Africa clustered in the branch with marker sequences from Côte d'Ivoire. In contrast, HIV-1 sequences of patients who reported contacts with persons from the United Arab Emirates (LE9 and LE18) clustered in another branch with Ugandan/Kenyan marker sequences (Figure 2b). Independent branching of HIV-1 env sequences of subtype A from West and East Africa had been observed (22). Third, the structures of two mosaic viral genomes and epidemiologic data of patients suggested that they were infected with potential recombinant viruses, which had been identified in two geographic regions. Briefly, patients LE24 and LE9, who reported sexual contacts with commercial sex workers in Nigeria or the United Arab Emirates, respectively, have been infected with mosaic viruses G^env/A^pol and A^env/D^pol/A^gag. Such recombination patterns were found in West-Central Africa and East Africa, respectively (23,24). Finally, analysis of env sequences from Lebanese patients revealed a relatively high intra-subtype genetic diversity within subtype B (average 14.9%, range 4.2% to 25.6%) and subtype A (15.5%, range 2.5% to 28.9%. Epidemiologic data and a low prevalence of HIV-1 infection suggest multiple introductions (rather than long existence) of genetically diverse HIV-1 strains to Lebanon.

The results presented in this report provide new information on HIV genetic variation in Lebanon. The identification of HIV-2 and 6 env HIV-1 subtypes, including three distinct HIV-1 mosaics, confirms high heterogeneity of HIV variants in Lebanon. Given the limited number of samples tested, finding both HIV-2 and multiple HIV-1 subtypes is remarkable and indicates that the distribution pattern of HIV variants resembled that of Cyprus or the Philippines rather than that of Thailand or India. These data indicate that the spread of complex HIV-subtype distribution patterns involving a large number of distinct HIV-1 and HIV-2 strains to countries without endemic HIV may be more common than previously thought.

The HIV-2 infection detected in this study, the first documented in Lebanon, reiterates the importance of appropriate screening methods for detecting both HIV-1 and HIV-2 to protect the blood supply. Our findings provide preliminary information for monitoring the spread of HIV variants in Lebanon and developing subtype-specific subunit vaccines.