INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Genital secretions are the source of a majority of human immunodeficiency virus type 1 (HIV-1) infections. Culturable HIV-1 (23, 59) and proviral DNA (21, 37) have been detected in the nonspermatozoal mononuclear cell fraction of semen, which includes CD4⁺ lymphoid cells, monocytes, and macrophages (2, 54). Cell-free virions, measured as viral RNA, are also found in seminal fluid (20, 21, 37).

Heteroduplex tracking analyses of PCR-amplified HIV-1 env sequences from donor-recipient sexual transmission pairs revealed distinctions between viral variant populations in plasma, peripheral blood mononuclear cells (PBMC), seminal cells, and seminal fluid (62). In three of five cases, virus in semen mononuclear cells was the most closely related to that transmitted to the recipient, and hence these cells were implicated as a likely vehicle for transmission (62). Drug resistance mutations in HIV populations are also unequally distributed in the blood and semen (29). Genetic differences were also found between proviral variants in peripheral blood and those in genital secretions in one study of women infected with HIV-1 envelope sequence subtypes A and D (41). Distinctions between viral populations in other anatomical sites have also been noted (14, 24, 27).

Early following infection with envelope subtype B virus via homosexual or perinatal contact, intravenous drug use, and blood transfusion, individuals harbor viruses in their blood that are homogeneous over the V3 region of the envelope protein (8, 10, 30, 36, 39, 47, 57, 60, 61). V3 loop sequences are also relatively conserved between individuals early following infection and correspond to a macrophage-tropic or non-syncytium-inducing (NSI) signature sequence (6, 35, 47, 53, 60), despite the fact that the donor may harbor a highly differentiated virus population in the blood (56, 62). A simultaneously greater amount of viral genetic diversity has been reported in the gag gene at these early times (60, 61). This observation led to the hypothesis that selection may occur for the dissemination of a subset of variants with a particular envelope-mediated phenotype, possibly macrophage tropism. With the recognition of the existence of HIV coreceptors, this could also correspond to a tropism for coreceptors of the CC-chemokine receptor class or exclusion of the viruses that utilize CXCR-4 (1, 12, 13, 18). It is also possible that macrophage-tropic HIV variants are more likely to be found in, and thereby transmitted from, the genital compartment, since, in contrast to the balance in the blood, a greater number of monocytes and macrophages are found in semen (2, 54).

In contrast to other modes of transmission, complex HIV-1 envelope sequence populations have been reported in one study of recently seroconverting women heterosexually infected with subtype A or D (41). Sexual transmission from men requires infection of cells in or migrating to seminal fluid or cell-free transfer of virus to this compartment. A comparison of blood and semen viral populations may therefore indicate whether virus in the male genital tract commingles with that of blood or corresponds to a compartmentalized subset or distinct population of variants. In this report, we describe a cross-sectional study of proviral sequences in peripheral blood and nonspermatozoal semen cells from five HIV-1 subtype B-infected individuals. Strong evidence for nonrandom distribution of virus variants between the two compartments and of a more restricted founder population accounting for the semen provirus population was found in all three of the patients who also harbored virus with mutations characteristic of a syncytium-inducing (SI) viral genotype. However, no evidence was found for restriction of the SI genotype from the semen.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Subjects (Table 1). Semen and blood specimens were obtained from patients seen at the Stanford University Medical Center, the Pittsburgh site of the Multicenter AIDS Cohort Study, and the Massachusetts General Hospital. The patients who were the sources of JO and PE samples were infected for an unknown length of time, and each had clinical AIDS at the time of sampling. The JO sample patient had a few Kaposi's sarcoma lesions and a CD4⁺ T-cell count of 467/mm³ and had been taking zidovudine (AZT) for 6 months, whereas the PE sample patient had a CD4⁺ T-cell count of 45/mm³ and had been taking AZT for 36 months when samples were collected. The PE sample patient died within 1 year of sampling, while the JO sample patient was asymptomatic on combination antiretroviral therapy 6 years after sampling. Patient 613 was infected for more than 11 years prior to sampling and was asymptomatic with a CD4⁺ cell count of 419 on the day the specimens were taken. He remained asymptomatic 1.5+ years later and had not been treated with antiretroviral drugs prior to sampling. Patient 064 was infected for 5 years prior to sampling, was asymptomatic with a CD4⁺ cell count of 342 on the day of sampling, and was being treated with AZT and dideoxyinosine. At 2.5+ years later, he remained asymptomatic with a CD4⁺ T-cell count of less than 200. The patient who was the source of MA samples (22, 32) had been infected with HIV for approximately 6 years prior to sampling. He has remained generally asymptomatic with CD4⁺ T-cell counts in the range of 265 to 644/mm³, the last being 402/mm³, determined 5 years following sampling. He had been treated with AZT for 9 months at the time of sampling.

Sample handling and DNA sequencing. Chromosomal DNA was extracted from Ficoll-banded PBMC and nonspermatozoal mononuclear cells (NSMC) (37). In some instances, chromosomal DNA was extracted from whole semen under conditions that permit only lysis of NSMC (20). The C2-to-V5 region of the env gene was amplified by a nested PCR protocol as described elsewhere (11), except that in some instances the sequences complementary to the M13 universal primer sequence included at the 5' end of the second-round primers ES7 and ES8 were replaced with XhoI and SacI linkers, respectively (ES7x and ES8s). The 700-bp PCR products were amplified from 1 µg of starting DNA and then separated and excised from a 0.8% agarose-1× Tris-borate-EDTA gel, purified by the glass bead method (Gene Clean; Bio 101, Vista, Calif.), and cleaved with XhoI and SacI. The DNA was ligated into XhoI- and SacI-linearized pUC21, and subclones containing inserts were sequenced with fluorescent dye-labeled universal primers and an ABI 370 automated DNA sequencer (Applied Biosystems, Inc., Foster City, Calif.).

Sequence analysis. Viral DNA sequences from each patient were aligned by the program CLUSTALW (52) followed by manual adjustment with GDE (50). Genetic distances were calculated with DNADIST from the PHYLIP software package (16, 17) by the maximum likelihood method (15). Neighbor-joining trees (44) were constructed, and a bootstrap analysis (15) using 1,000 bootstrap replicates was performed to assess the support at each of the internal nodes of the trees. Potential sample mix-ups were evaluated as described elsewhere (33).

Nucleotide sequence accession numbers. Sequences were deposited in GenBank under the accession no. U00821-U00822, U00831-U00843, U13381-U13388, and U96502-U96608.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

A 700-bp region corresponding to the second constant region through the fifth variable region of the envelope gene (C2 to V5) was used to evaluate the diversity of HIV-1 proviral DNA sequences in five patients at various stages of disease progression. Sequences derived from each patient belonged to envelope sequence subtype B, and each set clustered as monophyletic groups compared with those from other patients in this study (see Fig. 2) and the positive control sequences used in our experiments. Furthermore, no closely matched sequences were found in our local database or published databases of HIV sequences (33). Hence, there was no evidence of sample mix-up or contamination.

The JO sample patient had clinical AIDS but CD4⁺ T-cell counts of 500 at the time of sampling. Initially, 10 plasmid subclones from his PBMC (JO-B series in Fig. 1) and 4 from his NSMC (JO-S series) were sequenced. The diversity among the NSMC-derived clones was so low (mean, 0.6%; range, 0.3 to 0.8%; Fig. 1a) that we were concerned that provirus resampling had occurred and the divergence noted was due to Taq polymerase misincorporation during PCR (33) rather than representing viral quasispecies diversity in these cells. Hence, to rule out resampling, sequences from 11 molecular endpoints from his NSMC (JO-SE) were also derived. From this sampling, clonal virus populations were evident, as indicated by the bimodal distribution of divergence values between pairs of sequences in this population (Fig. 1b). Phylogenetic tree analysis revealed that 10 semen-derived variants formed a tightly clustered group of sequences distinct from that found in his PBMC (Fig. 2). This cluster consisted exclusively of variants expected to be of the SI phenotype, based on positively charged amino acids encoded at positions 11 and/or 25 of the V3 loop (19, 38), while all other variants from his blood and semen were predicted to be NSI (Fig. 2 and 3). The fact that this cluster was formed from endpoint-derived sequences (e.g., each was derived from a single proviral template) as well as from multiple sequences from plasmids derived from a separate PCR indicates that its generally narrow diversity was a true reflection of that found in the semen proviral DNA. Additional, smaller tight clusters (e.g., JO-SE11, -SE12, and -SE22) were detected in the blood and semen. A comparison of the divergences measured between all pairwise comparisons of sequences revealed a bimodal distribution in each compartment, with the groups being most divergent in the semen (Fig. 4a and b).

View larger version (28K): [in this window] [in a new window]   FIG. 1.   Distribution of divergences between pairwise comparisons of sequences. Provirus populations in the semen or PBMC were in each case sampled by PCR on multiple templates followed by cloning and sequencing of multiple plasmids ("clones") or by endpoint dilution ("EPD") followed by direct sequencing. For each set of sequences, the nucleotide differences between all possible pairwise comparisons were determined, collected into bins of 0.25% divergence intervals (uncorrected for back-mutation), and plotted.

View larger version (24K): [in this window] [in a new window]   FIG. 2.   Neighbor-joining phylogenetic tree of proviral sequences from the blood and semen of five patients. Triangles represent sequences from NSMC; squares represent sequences from PBMC. Open symbols are from direct sequencing of molecular endpoints from PCR; filled symbols represent sequences from cloned PCR product. Cloned semen specimens from the JO sample patient were derived by PCR amplification, cloning, and sequencing of multiple templates from a single PCR initiated with >10 proviruses; those from PBMC were derived from PCR initiated with approximately 10 proviruses. Cloned NSMC and PBMC sequences from the PE sample patient were derived from single PCRs initiated with >10 and approximately 500 proviruses, respectively. Cloned PBMC specimens from the MA sample patient were derived from a PCR initiated with approximately 200 proviruses. Sequences with mutations predictive of a viral SI phenotype are indicated. Numbers at branch nodes refer to the number of bootstrap repetitions (of 1,000) at which the distal sequences grouped together; those occurring at greater than 80% frequency (corresponding to an approximate P value of 0.05) are shown.

View larger version (61K): [in this window] [in a new window]   FIG. 3.   Deduced amino acid alignment over the C2-to-V5 region of HIV-1 env. For each patient, sequences are compared to one sequence from the PBMC. Dots are placed at positions at which individual sequences match that of the reference sequence. Dashes were introduced to maintain alignments. Asterisks indicate the presence of a stop codon. Question marks indicate an unknown amino acid due to an ambiguous base in the DNA sequence. The V3 loop is overlined and underlined at the top and bottom of each alignment, respectively.

View larger version (44K): [in this window] [in a new window]   FIG. 4.   Distribution of divergences between pairwise comparisons of sequences. Proviral sequences from semen or PBMC derived by either endpoint dilution, cloning, or a combination of the two methods were grouped for analysis of compartments in each patient. For each set of sequences, the nucleotide differences between all possible pairwise comparisons were determined, collected into bins of 0.25% divergence intervals (uncorrected for back-mutation), and plotted.

To evaluate the requirement for endpoint dilution to provide accurate measures of viral population diversity, we determined sequences from both plasmid subclones from the PE sample AIDS patient's PBMC (PE-B; n = 13) and NSMC (PE-S; n = 8), as well as from molecular endpoints from both compartments (sequences derived from PBMC endpoints are designated PE-BE [n = 9]; those from NSMC endpoints are designated PE-SE [n = 7]). The distribution of sequences was similar with either the molecular endpoint- or clone-derived variants (Fig. 1). Three clusters were noted in the phylogenetic analysis (Fig. 2): a tight cluster consisting of 16 variants from the semen and 1 from the blood, a more diverse cluster consisting of 8 variants found only in the blood, and a third loose cluster made up of variants from both compartments. As with the JO sample patient, viruses with the SI signature mutations were found as a cluster. However, in contrast to the JO sample patient, the PBMC-specific cluster was the one with the predicted SI phenotype, while all other variants from his blood and semen were predicted to be NSI.

Fifteen HIV plasmid subclones from NSMC proviral DNA (613-S) and eight subclones from PBMC (613-P) from asymptomatic patient 613 were generated and sequenced (Fig. 2). Again, a bimodal distribution of divergences was evident in the virus populations, and again this was most pronounced in the semen proviral population (Fig. 4). Proviral load was not quantitated in these (and patient 064) specimens, and yet the divergence noted between the two- and three-member clusters was above the level expected from Taq polymerase error alone (33). Hence, a fair estimate of quasispecies diversity was likely to have been obtained. Despite the fact that this patient was asymptomatic at the time of sampling, V3 loop sequences suggestive of the SI phenotype were evident in his proviruses. In contrast to the two patients above, however, the SI variants were distributed in both PBMC and NSMC populations (Fig. 2 and 3). Again, the SI-like cluster was phylogenetically distinct, forming the upper cluster of sequences with 93.5% bootstrap support in Fig. 2.

Nine HIV plasmid subclones from NSMC proviral DNA (064-S) and 12 subclones from PBMC (064-B) from asymptomatic patient 064 were generated and sequenced (Fig. 2 and 3). In this instance, however, no clear bimodal distribution was evident in the semen or blood (Fig. 4). This patient was asymptomatic at sampling, and no mutations suggestive of the SI phenotype were found. No tight clustering of clonal outgrowths was noticed in the phylogenetic tree, and the pattern of variant representation was similar in the blood and semen.

The MA sample patient (22, 32) had been infected with HIV for approximately 7 years prior to sampling; he has remained generally asymptomatic with peripheral blood CD4⁺ T-cell counts in the range of 300 to 500/mm³ throughout the course of his infection. Fifteen PBMC variants sequenced after plasmid subcloning (MA-B300 series) sorted into two diverse sequence clusters, with two outlying variants (MA-B311 and -B312). His proviral load in the NSMC fraction of semen was quite low. Thus, to avoid resampling viral templates (34), all 14 sequences derived from the NSMC were obtained by endpoint dilution and then PCR and direct sequencing (MA-SE series). Again, no clear bimodal distribution was evident in the semen or blood (Fig. 4). Phylogenetic analysis revealed a distributed pattern of variant representation, with little evidence of substantial clonal outgrowths and no clustering in one or the other tissue evident (Fig. 2). The deduced amino acid sequences of the V3 loop of each virus suggested an NSI phenotype (Fig. 3). Thus, no distinct differences were found between provirus populations in his blood and semen.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Previous studies have shown that HIV-1 proviral sequences can be nonuniformly distributed throughout the body. Distinct proviral sequence variants have been reported in some patients in PBMC compared to brain tissue (14, 24, 27, 40, 42, 58), cerebrospinal fluid (31, 51), spleen (14), lymph node (3, 46), lung (25), and semen (62). Furthermore, populations can differ between plasma RNA and integrated proviral DNA in both compartments (62). These RNA specimens were not available for the current study. In the current study, we found differential representation of groups of integrated proviral sequences between blood and semen cells in three individuals but not in two others. Several possible explanations exist to account for the nonuniformity we observed. HIV variants may evolve to replicate more efficiently in specific target cell types through selection for particular tropisms, such as enhancer or receptor specificity or other viral properties. Brain-derived isolates show enhanced macrophage replication competence relative to those simultaneously derived from blood (5, 28), possibly reflecting a phenotypic requirement for passage into the brain within infected macrophages. Different patches of epidermal Langerhans cells (45) and adjacent splenic white pulps (7, 9) have also been shown to contain distinct proviral variants. Such minute quasispecies variegation has been postulated to reflect antigenic stimulation of provirus-bearing T cells with resulting local amplification of these particular variants (7). Local amplification may also account for the distributions we noted. Alternatively, the distinct populations may also reflect different half-lives of provirus-bearing cells in the larger context of the constantly evolving quasispecies and/or different immune pressures selecting distinct populations in different compartments. If different turnover rates of infected NSMC relative to PBMC account for the population differences observed, the rates of provirus clearance following antiviral drug combination therapies may also differ for different tissues. Testicles could also be an immunologically privileged site for HIV-1 (4). More macrophages and lymphocytes are present in semen relative to blood (2, 54). Furthermore, the titers of anti-immunoglobulin G in semen are, on average, 1/10 of the titer present in blood, including antibodies against HIV-1 proteins p55, p24, and p17, which are less prevalent in semen than in blood (55). This raises the possibility that viral evolution in semen may be different from that in the blood because of the differences in immune pressure.

Analysis of HIV sequences in genital fluids is critical for study of variants involved in horizontal and vertical transmission. For example, viral populations from the blood, the major source of vaccine and challenge strains in all animal model studies to date, may differ from those in the genital tract in some characteristic, such as immunologic recognition or receptor choice, that would have an impact on attempts to immunize against it. In any case, claims of transmission of what is termed a minor subset of viruses in the transmitter need to be tempered by the recognition that sampling only the blood-borne variants can fail to accurately infer the representation of the transmitted variants in the semen.

If particular characteristics are required for replication and production of virus in the semen, they may be detected as signature sequences found across different individuals as semen specific. Based on signature sequences in the V3 loop, putative SI variants were found only in the blood of one AIDS patient, only in the semen of the second AIDS patient, and in both compartments of a patient who was asymptomatic when tested. Hence, restriction of viruses with the SI phenotype and/or CXCR-4 coreceptor specificity from the semen cannot account for the rare transmission of these strains (60, 61). Amino acid signature sequences have been found in brain-resident proviruses, which may reflect macrophage tropism (6, 27, 43) and perhaps the evolution of specific disease-causing variants (42). Using the VESPA algorithm (26), we were unable to detect a significant tissue-specific signature sequence over the approximately 220 amino acids of env evaluated (data not shown). However, recognition of distinctions between viruses in the blood and those in semen indicates that further analyses should be conducted to determine the origins of these distinctions and their importance for vaccine evaluation.