De Pasquale, Maria Pia; Leigh Brown, Andrew J.; Uvin, Susan Cu; Allega-Ingersoll, Jessica; Caliendo, Angela M.; Sutton, Lorraine; Donahue, Shannon; D'Aquila, Richard T.
Cervicovaginal secretions can transmit HIV-1 vertically and horizontally. After antiretroviral therapy is started, viral load in cervicovaginal secretions may decline more rapidly than does viral load in blood plasma, 1 and drug responses are as durable in cervicovaginal secretions as in blood. 2 However, antiretroviral drug–resistant virus has been identified in female genital tract specimens. 3–5 Heterosexual and vertical transmission of drug-resistant HIV-1 is occurring. 6,7
Although HIV may replicate independently in the male genital tract relative to blood, 8 and virus in semen may lack resistance mutations found in contemporaneous virus in blood, there are fewer comparisons of nucleotide sequences and resistance mutations in HIV RNA in blood plasma versus female genital tract secretions. In a few subjects, some phylogenetic or mutational differences have been noted between blood and cervicovaginal lavage (CVL) fluid sequences. 3 Further characterization is needed to determine whether genetic differences, including drug-selected mutations, occur in female genital secretion–derived virus versus blood as a result of localized replication. If confirmed, monitoring genital tract as well as blood specimens may improve the surveillance for drug-resistant HIV-1.
The possible sources of HIV-1 RNA originating within the female genital tract are also not yet fully defined. Submucosal lymphoid cells are likely to be the major host cells for HIV-1 replication. Viral load is higher in endocervical canal fluid collected by Sno-strips compared with cervicovaginal fluid. 9 This suggests that the endocervix may be the dominant source of virus in CVL fluid. Genetic comparisons of endocervical secretions versus CVL fluid that can address this question have been limited to date.
We undertook this study to evaluate whether drug-selected pol mutations occurred in the female genital tract as well as blood HIV RNA in women treated with antiretroviral therapy who had detectable viral load in both blood and genital tract. We also studied whether HIV pol phylogenies were consistent with localized replication in the female genital tract and the hypothesis that the endocervix was the dominant source of virus in CVL fluid.
MATERIALS AND METHODS
Specimens of blood plasma, endocervical secretions (in some cases), and CVL fluid were collected from 15 nonpregnant HIV-infected women, 11 of whom were treated with antiretroviral therapy. The other 4 had never been treated. These 15 subjects were chosen from a larger group because each had detectable viral load in blood plasma (>400 copies/mL). The drug-treated subjects were selected because they also had detectable viral load in genital tract secretions. Endocervical secretions were not collected from some women enrolled early during the study. Subjects were also selected for the absence of genitourinary tract symptoms or active sexually transmitted disease. They were advised not to have sexual intercourse, douche, or insert any vaginal products for at least 48 hours before the study visit. All women had microscopic examination of cervicovaginal secretions to diagnose Candida infection, trichomoniasis, and bacterial vaginosis. Previously, a low prevalence of gonorrhea and Chlamydia had been found among this clinic population 10; therefore, routine testing was not done for these infections in this study. No cervicovaginal or endocervical specimens were collected during menses.
Specimen Collection and Processing
Plasma was separated from ethylenediamine tetra-acetic acid–anticoagulated blood and stored at −80°C. Endocervical secretions were collected using Sno-strips (Chauvin Pharmaceuticals, Ltd., Essex, England), as described. 9 After inserting a speculum and visualizing the cervix, 3 Sno-strips were gently inserted into the endocervix. They were held in place for approximately 1 minute to allow adsorption of endocervical fluid (about 8 μL per strip). 9 The 3 strips were immediately placed into 250 μL of QRL buffer (Qiagen DNA/RNA extraction kit) and stored at −80°C. CVL was done by instilling 10 mL of saline solution directed gently at the cervical os; the pooled fluid in the vagina was then aspirated after a few seconds. 9 Low-speed centrifugation removed cells from fresh CVL fluids. Frozen aliquots of cell-free CVL fluids were stored at −20°C. In cases where both Sno-strips and CVL fluids were collected during the same examination, Sno-strips were obtained first and carefully removed without trauma, and then CVL fluids were collected, in that order.
The Sema ELISA kit (Humagen Fertility Diagnostics Inc., Charlottesville, VA) was used to test for the presence of semen in CVL fluid. This test can detect <1 part per million of human seminal plasma in a CVL fluid for at least 24 hours after intercourse. 11
HIV-1 Viral Load Assay
Viral load in plasma and genital secretions was determined using the Nuclisens assay (bioMerieux, Inc., Durham, NC). The lower limit of quantitation was 400 copies/mL.
HIV-1 pol Genotyping
Cell-free blood plasma (1 mL) was centrifuged at 30,000 g for 60 minutes at 4°C. HIV-1 RNA was extracted from the pellet using the QIAamp Viral RNA extraction kit (Qiagen, Valencia, CA). HIV-1 RNA from the CVL fluid specimens was extracted either by the RNA/DNA extraction kit (Qiagen) or the Blood RNA extraction kit (Qiagen). Cell-free viral RNA was extracted from Sno-strips stored in QRL buffer by using the RNA/DNA extraction kit (Qiagen). In one case (B048), proviral DNA was extracted from the CVL cell pellet (Gentra, Minneapolis, MN).
An HIV RNA pol gene fragment was reverse transcribed, amplified by polymerase chain reaction (PCR), and sequenced by population-based cycle sequencing using the Viroseq Genotyping system (version 2, Applied Biosystems, Foster City, CA) on an ABI 377 or the TruGene HIV-1 Genotyping kit (Visible Genetics, Toronto, Ontario, Canada) on a VGI Long Read Tower. The presence of contaminating DNA in CVL fluid and Sno-strip specimens was assessed by amplification with β globin and HIV-1 pol primers without in vitro reverse transcription. Proviral DNA was amplified using outer primers 1607 and 4522 in the first PCR and inner primers 1811 and 4335 in the nested PCR as previously described. 12 Genbank accession numbers are AY 359967 to AY 360005 (deposit at Genbank pending at time of manuscript submission).
After alignment, phylogenies were generated using Molecular Evolutionary Genetics Analysis Software, version 2 (MEGA2) (http://www.megasoftware.net). 13 Pairwise nucleotide distances were estimated for all codons using the Tamura-Nei γ-correction method with an α value of 3 and for synonymous codons using the modified Nei-Gojobori method with a transition/transversion ratio of 3 as estimated from the data and the Jukes-Cantor correction for multiple hits. Phylogenetic trees were constructed from the same distance matrices using the neighbor-joining algorithm. HIV-1 subtype A from the Los Alamos database (http://hiv-web.lanl.gov; accession number AF 004885) was the reference sequence. For some analyses, 35 independent sequences from patients with acute HIV infection who had never received antiretroviral therapy were included as background controls.
One consensus sequence from a genital tract specimen (DOT12.cvl) contained multiple ambiguity codes, indicating it was a mixture of 2 variants (not shown). Resolution of these ambiguities into component nucleotides indicated the mixture most likely comprised one variant that was identical to the plasma sequence and one other variant. It was therefore analyzed phylogenetically as 2 sequences: one with identity to the plasma sequence, and a second sequence containing all the variant bases. This approach was previously used to identify a recombinant virus in an acutely HIV-infected individual. 14 Bootstrapping was conducted using 1000 replicates.
Drug Resistance Mutations in Protease and Reverse Transcriptase
Paired blood plasma and genital specimens were obtained at the same visit from 15 women with chronic HIV-1 infection that was first diagnosed >1 year earlier. Eleven of the 15 were being treated with antiretroviral drugs; the remaining 4 had never received antiretrovirals. Both endocervical Sno-strips and CVL fluids were obtained at the same visit from 5 drug-treated subjects. Either an endocervical Sno-strip (n = 2) or a CVL fluid (n = 8) specimen was collected from the other 10 women. For 3 drug-treated subjects, specimens were collected at sequential times during treatment. A total of 15 pairs of blood and genital tract specimens were sequenced from the 11 drug-treated subjects. The characteristics of the study population are described in Table 1. None of the CVL fluid specimens tested positive for the presence of semen by Sema ELISA kit.
Two different methods of sequencing were performed on each of 6 CVL fluid specimens. VGI TruGene (Visible Genetics, Inc.) and ABI ViroSeq v2 kits (Applied Biosystems) were used with corresponding automated sequencers. This was done to confirm the adequacy of using only one of those methods. There was concordance in detection of identical resistance mutations with each method for each of these specimens. Reverse transcriptase and protease sequences were then obtained with only one of the kits (VGI TruGene) from 9 other genital tract specimens.
Among the treated women, resistance mutations were detected in both blood and genital tract sequences (Table 2). Most notably, no subject had a drug-resistant mutant in blood and a wild-type virus in a female genital tract specimen from the same visit (Table 2). Such discordance has been reported to be common in comparisons of viruses in blood and semen and is felt to be due to limited drug exposure in male genital tract tissues. 15,16 Although resistance was evident in both blood and genital secretions in these women, the mutations identified in each compartment were different in 9 of 11 drug-treated women (as marked in Table 2: subjects B062, DOT12, DOT3, G004, G005, G007, G010, DOT8, B043).
Additional changes in resistance-related codons were seen in plasma, but not in the paired genital secretion specimens, in 4 cases. In 3 of these, only a single additional mutation was evident in plasma, compared with the genital secretion specimen. These included PR 54V (G005, first time point), PR 77V/I (G005, third time point), and RT 184V (DOT8, second time point). One specimen pair had several resistance mutations in reverse transcriptase in the plasma specimen (RT 67N, 69N, 70K, 184V, 188V, 219Q) that were not found in the corresponding CVL fluid specimen (G004;Table 2). However, the CVL fluid specimen in that pair had drug-selected protease mutations, as mixtures, that were not present in the plasma specimen: PR 10F/L/V and 84 I/V (G004;Table 2).
Five other specimen pairs from drug-treated women had ≥1 drug-selected mutations in CVL fluid, which were not detected in blood. These included PR 10I/V and 77V/I (G007), RT 103K/N (G010), PR 54I/V, 82V/A, and 90L/M (DOT3 second time point); PR 90M and RT 184V (DOT8 first time point); and PR 46I (B043) (Table 2).
Two other pairs of sequence differed in PR L63P (DOT 12 and B062) and a third pair differed in both PR L63P and RT 101E (DOT3, first time point). Both PR63P and RT 101E may not always be drug selected but can sometimes emerge during drug therapy. Thus, 12 of the 15 paired sequences from 9 of the 11 drug-treated women had at least some difference in ≥1 pol positions that may be drug selected.
Sequences from 2 of the 11 drug-treated women had identical genotypes in plasma and genital secretions. One subject (G011) harbored viruses with the same resistance mutations in both genital secretions and blood. The 2nd (B082) reported nonadherence to drug therapy and had the same wild-type, drug-susceptible variant in both compartments. One of 3 time points from a third subject (G005) had the same sequences across compartments, although there were different mutations in blood and genital tract at 2 other time points. In the longitudinal specimens from G005, mutational differences between compartments were less when drug-mediated suppression of viral load was greater soon after a regimen change (third time point, G005.3, Table 2).
As expected, sequences from drug-naive subjects did not show any primary drug resistance mutations in either plasma or genital secretions (Table 2). Polymorphisms that can also be secondary resistance mutations differed between compartments in 2 of 4 subjects (B074 and B048, Table 2). Silent mutations at positions not associated with resistance were similar in the 2 compartments (not shown). However, all sequences exhibited variations from the consensus NL4-3 (Table 2).
The 5 subjects who had both Sno-strip and CVL fluid specimens obtained at the same visit had similar patterns of resistance mutations in each genital tract specimen (G011, B062, DOT 12, G005, G010;Table 2).
Phylogenetic Analyses of pol Sequences
One drug-treated subject studied in the resistance mutation analysis, B043, was not included in phylogenetic analyses due to lack of reverse transcriptase sequence information from CVL fluid. There was generally closer relatedness among viruses from different compartments within an individual than to viruses from another individual (Fig. 1), confirming that there was no sequence cross-contamination or specimen mix-up. This included analyses in a background of other clinical isolates, which did not alter the branching pattern of the sequences (data not shown).
Phylogenetic analyses were performed on viral sequences from both compartments with two approaches: including all nucleotide substitutions, and including only synonymous substitutions (i.e., mutations not leading to amino acid change in the protein). Differences might arise between these 2 phylogenetic analysis approaches. Similar drug selection pressure in the 2 compartments might lead to convergent evolution with similar nonsynonymous genetic changes, despite differences in synonymous sites suggesting 2 separately replicating populations.
The phylogenetic trees indicated that many drug-treated women had genetic differences between blood and genital tract virus. G004 had the greatest differences between blood and genital tract (Fig. 1). Other samples with substantial branch lengths between plasma and CVL sequences included B062, G005.3, and DOT3.2.
A parsimony-based analysis of the mixture in the DOT12 CVL sample allowed the deduction that this formed a mixture of 2 sequences. One DOT12 CVL sequence was identical to the DOT12 plasma sequence; a second was very distinct (Fig. 1). Phylogenies were generally consistent across the 2 approaches. Sequences from only 2 of the subjects grouped differently in the 2 phylogenetic approaches. In one case (G005), the phylogeny using synonymous sites showed more relatedness between blood and genital tract than was seen in the approach using all sites. In a second case (DOT8), the opposite was seen; greater difference between blood and genital tract was seen using only synonymous sites than when all sites were included.
Genetic distances between blood- and genital tract–derived sequences in the phylogenetic tree paralleled the gradation of differences in resistance mutations determined by sequencing. Sequences with discordant, nonsynonymous changes in primary and secondary resistance mutation sites in blood and CVL fluid also had more divergence based on analysis of synonymous changes (Table 3). Genetic distances at synonymous sites between virus sequences in blood and CVL fluid were lower among the untreated subject (B076) and the drug-treated subjects with concordant mutations across compartments (B082, G011, G005.2) than among the drug-treated subjects with discordance in resistance mutations between blood and genital tract (P < 0.05, Wilcoxon rank sum test;Table 3). The overall ratio of synonymous/nonsynonymous substitutions was high in the HIV pol sequences analyzed in this study (ds/dn = 2.9).
Comparison of Endocervical and Cervicovaginal Lavage Sequences
The tissue source of virus within the female genital tract was studied by examining the relatedness between endocervical and cervicovaginal sequences in those who had both specimens collected on the same visit (B062, G011, G005, G010, DOT12). Each of these 5 subjects had sequences from endocervical secretions that were genetically distinguishable from the CVL fluid–derived sequences. Three of these subjects (G011, B062, and DOT12) had the greatest phylogenetic differences between endocervix (Sno-strips) and CVL sequences (between compartment genetic distances: 4.66–2.26%) (Fig. 1). Virus sequences from endocervical secretions and CVL fluid were more similar to each other in specimens from the 2 other subjects (G010, G005.2) but were still genetically distinguishable (genetic distances were 1.99 and 1.19%, respectively).
Limited data suggest that antiretroviral drugs penetrate well into female genital tract tissues. 17 In contrast, there is evidence for a relative barrier to penetration of some antiretroviral drugs into the testes. 18 Several reports document that HIV-1 pol sequences amplified from cell-free semen RNA often do not contain the resistance mutations found in a blood plasma specimen collected at the same visit. 15,16 In that context, the results of the present study are of note. There was not a single instance of discordant wild-type virus in the genital tract when a drug-resistant mutant virus was found in the blood among the 15 pairs of sequences from antiretroviral-treated women studied here. Further work is needed to define whether antiretroviral drugs penetrate the female genital tract well, as these data suggest.
There were differences in the mutation pattern in sequences from blood and female genital tract in 12 of the 15 paired sequences (Table 2). Differences between blood and female genital tract virus in drug-selected mutations appeared to be greater when blood viral load was less well suppressed in 1 drug-treated subject studied at several time points (G005, Table 2). Paired sequences from drug-naive women did not have primary resistance mutations in either compartment. Although limited sample size precluded rigorous analysis, there appeared to be fewer differences even in polymorphisms/secondary resistance mutations across compartments in the drug-naive subjects than among the drug-treated subjects (Table 2), consistent with the hypothesis that the mutational differences were drug selected. Differences in mutation pattern across compartments in the drug-treated women are hypothesized to be stochastic differences in emergence of mutations under similar drug selection pressure in blood and genital tract in vivo. This may be related to a relatively small genital tract virus population size or to sampling differences. Either explanation suggests that virus replicates locally in the female genital tract under drug selection pressure, to some extent. The observation that genetic distances of synonymous codons were greater in subjects showing discordant changes across compartments in resistance mutations also supports local virus production in the genital tract. The high ratio of synonymous to nonsynonymous mutations in these sequences suggests that this is a robust analysis. Others also previously noted unexpectedly high synonymous substitution rates during drug therapy that substantially reduced viral load. 19 This suggests the hypothesis that even synonymous changes may provide some selective advantage during therapy, perhaps by effects at the level of viral RNA rather than protein.
Others have also presented less extensive sequence analyses suggesting local HIV replication within the female genital tract. 20
We also examined the hypothesis that the endocervix would be the dominant source of virus in CVL fluid. This hypothesis was suggested by the finding in other work that endocervical secretions collected by Sno-strips had much higher viral load than CVL fluids. 9 However, the CVL fluid sequences here were genetically distinguishable from those derived from endocervical specimens, including in 2 subjects with higher viral load in endocervix (144,000 copies/mL in G011; 192,000 copies/mL in B062) than in CVL fluid specimens (2600 copies/mL in G011; 2000 in B062). Semen contamination of CVL fluid did not explain this finding based on the historical information, negative SEMA test results, and the phylogenetic analyses showing clustering of sequences within each individual. Although contribution of blood-derived virus to the viral population in the genital secretions cannot be completely ruled out, a substantial contamination of blood-derived virus was not suggested by red blood cell counts of the CVL fluid. Other data also suggest that blood contamination and plasma transudation are not sources of any substantial amount of the virus in female genital tract specimens, either in CVL fluid or endocervical secretions collected by cytobrush. 20;21 Moreover the method of sampling endocervical secretions in the present study involved less trauma than cytobrush collection used by Hart et al. 21 Further confirmation of this finding with additional specimens might involve more powerful genetic methods to further minimize possible virus population sampling bias, such as analyzing a large number of clones of PCR products or using heteroduplex mobility analyses. If confirmed, the genetic differences between CVL fluid and endocervix virus may suggest that the ectocervix or vagina, or the submucosa under those areas, is a source of a significant amount of virus found in CVL fluid. This is also supported by the observation that HIV RNA has been detected in vaginal secretions of HIV-1 seropositive women who have undergone hysterectomy. 22
In summary, these results are consistent with ongoing, compartmentalized replication under drug selection pressure in the female genital tract when antiretroviral therapy is not successfully suppressive. Cells in the female genital tract other than those localized to the endocervix may replicate HIV-1. Further work will be needed to understand the implications for HIV pathogenesis during therapy and for epidemiologic surveillance for drug-resistant virus.
1. Cu Uvin S, Caliendo AM, Reinert SE, et al. HIV-1 in the female genital tract and the effect of antiretroviral therapy. AIDS. 1998; 12:826–827.
2. Cu-Uvin S, Caliendo AM, Reinert S, et al. Effect of highly active antiretroviral therapy on cervicovaginal HIV-1 RNA. AIDS. 2000; 14:415–421.
3. Si-Mohamed A, Kazatchkine MD, Heard I, et al. Selection of drug-resistant variants in the female genital tract of human immunodeficiency virus type 1-infected women receiving antiretroviral therapy. J Infect Dis. 2000; 182:112–122.
4. Di Stefano M, Fiore JR, Monno L, et al. Detection of multiple drug-resistance-associated pol mutations in cervicovaginal secretions from women largely treated with antiretroviral agents. AIDS. 1999; 13:992–994.
5. Wainberg MA, Beaulieu R, Tsoukas C, et al. Detection of zidovudine-resistant variants of HIV-1 in genital fluids. AIDS. 1993; 7:433–434.
6. Yerly S, Kaiser L, Race E, et al. Transmission of antiretroviral-drug-resistant HIV-1 variants. Lancet. 1999; 354:729–733.
7. Johnson VA, Petropoulos CJ, Woods CR, et al. Vertical transmission of multidrug-resistant human immunodeficiency virus type 1 (HIV-1) and continued evolution of drug resistance in an HIV-1-infected infant. J Infect Dis. 2001; 183:1688–1693.
8. Gupta P, Leroux C, Patterson BK, et al. Human immunodeficiency virus type 1 shedding pattern in semen correlates with the compartmentalization of viral Quasi species between blood and semen. J Infect Dis. 2000; 182:79–87.
9. Reichelderfer PS, Coombs RW, Wright DJ, et al. Effect of menstrual cycle on HIV-1 levels in the peripheral blood and genital tract. WHS 001 Study Team. AIDS. 2000; 14:2101–2107.
10. Cu-Uvin SHJ, Warren D, Klein RS, et al, for the HER Study Group. Prevalence of lower genital tract infections among human immunodeficiency virus (HIV)-seropositive and high risk HIV-seronegative women. Clin Infect Dis. 1999; 29:1145–1150.
11. Keil W, Bachus J, Troger HD. Evaluation of MHS-5 in detecting seminal fluid in vaginal swabs. Int J Legal Med. 1996; 108:186–190.
12. Martinez-Picado J, Sutton L, De Pasquale MP, et al. Human immunodeficiency virus type 1 cloning vectors for antiretroviral resistance testing. J Clin Microbiol. 1999; 37:2943–2951.
13. Kumar S, Tamura K, Jakobsen IB, et al. MEGA2: molecular evolutionary genetics analysis software. Bioinformatics. 2001; 17:1244–1245.
14. Daar ES, Frost SDW, Wong JK, et al. Mixed infection with multidrug resistant and wild type HIV strains in primary HIV infection: early viral rebound suggests loss of immune control. Paper presented at: Ninth Conference on Retroviruses and Opportunistic Infections. Seattle, WA; February 24–28, 2002. Abstract #96.
15. Eron JJ, Vernazza PL, Johnston DM, et al. Resistance of HIV-1 to antiretroviral agents in blood and seminal plasma: implications for transmission. AIDS. 1998; 12:F181–F189.
16. Byrn RA, Zhang D, Eyre R, et al. HIV-1 in semen: an isolated virus reservoir. Lancet. 1997; 350:1141.
17. Kashuba ADM, Min SS, Corbett AH, Rezk N, Cohen MS. Comparison of Protease Inhibitor and Non-Nucleoside Reverse Transcriptase Inhibitor Concentrations in Male and Female Genital Tract. Journal of Investigative Medicine. 2003; 51( 2).
18. Taylor SBD, Drake SM, Workman J, et al. Antiretroviral drug concentrations in semen of HIV-infected men: differential penetration of indinavir, ritonavir and saquinavir. J Antimicrob Chemother. 2001; 48:351–354.
19. Gunthard HF, Leigh-Brown AJ, D'Aquila RT, et al. Higher selection pressure from antiretroviral drugs in vivo results in increased evolutionary distance in HIV-1 pol. Virology. 1999; 259:154–165.
20. Ellerbrock TV, Lennox JL, Clancy KA, et al. Cellular replication of human immunodeficiency virus type 1 occurs in vaginal secretions. J Infect Dis. 2001; 184:28–36.
21. Hart CE, Lennox JL, Pratt-Palmore M, et al. Correlation of human immunodeficiency virus type 1 RNA levels in blood and the female genital tract. J Infect Dis. 1999; 179:871–882.
22. Farrar DJ, Cu Uvin S, Caliendo AM, et al. Detection of HIV-1 RNA in vaginal secretions of HIV-1-seropositive women who have undergone hysterectomy. AIDS. 1997; 11:1296–1297.