485 THE RAPID ACCUMULATION of mutations and recombination occurs by processes that actively contribute to the evolution of HIV-1 and its genetic diversity. At least 10 distinct genetic subtypes of HIV-1 have been identified on the basis of the phylogenetic analysis of nucleotide sequences in the envV3 and gagp17 regions.1,2 According to an analysis of the 1993 Los Alamos HIV Database, 10% of the reported sequences were V3/p17 recombinants. Intersubtype mosaic forms of HIV-1 have been reported in geographic areas where multiple HIV-1 subtypes cocirculate. While many studies have been focused on phylogenetic analysis of the structural genes,1,2,5 relatively little attention has been paid to the genetic variation in the HIV1 long terminal repeat (LTR) region. Here we have characterized a collection of the LTR sequences from seven different HIV-1 subtypes and intersubtype recombinants representing different geographic origins. HIV-1 DNA was extracted from peripheral blood mononuclear cells (PBMCs) of 28 randomly chosen HIV-1-infected patients (21 African and 7 Swedish patients), as previously described. HIV-1 was subtyped by direct sequencing of the V3 and p17 regions of all of the patients, followed by phylogenetic analyses, as previously described. Full-length HIV genomic sequences (subtypes A±H), as presented in the 1997 Los Alamos HIV Database, were used as reference sequences in the alignments for V3 and p17 regions. For some subtypes unpublished full-length sequences (subtype H: VI991-3, VI997-2; subtype F: BZ126, F9363, ITM850; subtype A: SE8538, SE8131-3, SE8891, SE7253, SE7535; and subtype J: SE91733, SE9280-9) were available. The LTR was amplified by a nested polymerase chain reaction (PCR), using the previously described primers BJLTR1, BJLTR3, and BJLTR4.8 After cloning of the PCR fragments into the pCR2.1 vector (Invitrogen, San Diego, CA) the double-stranded (ds) DNA was used as a template for sequencing. The sequencing reactions were primed with CY5-labeled universal (or reverse) primers, provided in the Sequenase version 2.0 kit (United States Biochemical, Cleveland, OH), according to the protocol of the manufacturer. Multiple LTR clones (3±10 clones per patient) for each of the 28 HIV-1 strains were sequenced from both directions. The LTR sequences (nucleotides 2 382 to 1 113 relative to transcription start) were then aligned using Omiga 1.01 (ClustalW Oxford Molecular, Ltd., Oxford, UK) and DNASIS sequence analysis software (Hitachi Software Engineering America, San Bruno, CA). The LTR sequences presented in this study were aligned with the various subtype LTR sequences from the Los Alamos HIV Database of 1997 (data not shown). The alignment was optimized manually. On the basis of the LTR, p17, and V3 phylogenetic analyses 1 of the samples was classified as subtype A, 10 as B, 6 as C, and 1 as G (Fig. 1A, B, and C, respectively2,9). A discrepant topology was observed in the V3, p17, or LTR phylogenetic tree in 5 of the 28 HIV-1 strains (marked with asterisks in Fig. 1). The HIV-1 strains that gave discrepant clustering in the phylogenetic trees were therefore putatively classified as recombinant strains (SO5549, A/C; GM6139, A/G; IC5381, A/G; GM6452, E/A; and TH6098, E/A). On the basis of the LTR analysis five of the strains were classified as subtype D (UG6476, UG6083, UG4696, UG6357, and UG5609); however, a clear subtype classification based on p17 and V3 (except for patient UG6476) could not be obtained for those strains (marked with double asterisks in Fig. 1B and C). A possible explanation may be that in those strains these regions contain a recombination point so that one part of the sequence belong to one subtype and one part to another, leading to unclear classification of those isolates. Interestingly, all of these strains are from Uganda, where a high frequency of HIV-1 hybrid genomes can be expected owing to the cocirculation of multiple divergent HIV-1 strains, mainly subtypes A and D.1 The LTR analysis revealed a distinction between the LTR region of subtype C and that of all the other subtypes studied (A, B, D, E, F, and G), in that subtype C isolates (except the sample SO5549) contained an additional potential NFk B-binding site (NFk B1) (GGGCGKTCY). Such a unique pattern for the LTR region of HIV-1 clade C has previously been reported from Ethiopia