A number of discordant couples, in whom the man is HIV positive and the woman is HIV negative, wish to have children. To conceive, they must abandon condom-protected intercourse and risk HIV transmission to the woman and subsequently to her child. Although the quantity and quality of the virus present in the semen are pivotal in sexual transmission of HIV-1, transmission has been documented after only one sexual exposure . HIV is present in semen as free virus in the seminal plasma and as cell-associated virus, but whether spermatozoa are infected with HIV remains controversial [2-10]. It is also not clear whether the level of HIV in semen may vary with time, and how the amount of virus in the semen correlates with that in plasma.
Semprini et al.  have conducted a programme of assisted conception for discordant couples since 1989 in a bid to minimize any risk of HIV transmission. The assumption in this programme is that spermatozoa are not the major HIV reservoir in unfractionated semen, and therefore the spermatozoa are washed clear of seminal plasma and non-spermatozoa cells (NSCs) by gradient centrifugation and ‚swim-up‚, before their use for artificial insemination. ‚Sperm washing‚ may reduce the amount of virus present in the spermatozoa sample, or even eliminate it completely, but this has only been evaluated by the insensitive method of antibody detection of viral antigen on the processed sperm. Artificial insemination timed with ovulation may also minimize the number of exposures of the uninfected woman to potentially infected material. To date, Semprini et al.  have reported more than 1000 insemination attempts for 350 couples, which resulted in almost 200 pregnancies with no seroconversions.
Before undertaking a programme of ‚sperm washing‚ and insemination for HIV discordant couples, we have evaluated whether ‚sperm washing‚ effectively reduced the level of HIV present in the spermatozoa sample to be used for insemination. We provide evidence that ‚sperm washing‚ significantly reduces both free-virus and cell-associated virus levels.
Materials and methods
Semen samples were obtained from 11 HIV-1 positive (CD4 cell count 93-786ml-1) homosexual men attending the St Stephen‚s Clinic, Chelsea and Westminster Hospital, London, UK. Control semen was obtained from HIV-negative donors attending the fertility clinic at the Chelsea and Westminster Hospital and from laboratory workers. All patients consented to participate in the study in accordance with the Chelsea and Westminster Hospital Ethical Committee.
Semen samples were donated by masturbation. Blood samples were collected into ethylenediamine tetra-acetic acid (EDTA) vacutainers from the HIV-positive patients on the same visit. The samples were processed within 2h of donation. Plasma and peripheral blood mononuclear cells (PBMC) were prepared from the blood samples by gradient centrifugation over Histopaque (Sigma, Poole, Dorset, UK). Unfractionated semen samples were diluted 1:4 in medium (Earl‚s balanced Salt solution (EBS) (Sigma), supplemented with 1mM pyruvic acid (Sigma) and 1mM lactic acid (Sigma) and pH was adjusted to 7.2 using hydrochloric acid). A fraction of the diluted whole semen was stored at -70°C. Seminal plasma, spermatozoa and seminal NSCs were prepared by differential gradient centrifugation over 80 and 40% Medicol Medium Isotonic (Medi-Cult, Redhill, Surrey, UK). The cells from both interphaces and the pellet were washed twice with medium. After the final wash the spermatozoal pellet was overlaid with 3ml of medium and incubated at 37°C, 5% CO2 for 20min to allow motile spermatozoa cells to swim up. The supernatant containing motile spermatozoa was collected. In some experiments, the spermatozoa fraction at the 40/80% interface was also collected. This fraction should contain mostly dead spermatozoa, however, when treated in the same way as the pellet from the gradient, a number of viable ‚swim-up‚ spermatozoa were recovered. The pellets remaining after the motile sperm were recovered were also kept for viral load assays. All samples for viral load assays were resuspended in NucliSens lysis buffer and stored at -70°C.
HIV RNA viral load
Viral load assays were performed using a commercial assay, NucliSens (NucliSens HIV-1 RNA Quick‚s test; Organon Teknika, Cambridge, UK), which has previously been reported to be a more reliable assay for the evaluation of viral load in semen samples compared with Amplicor Monitor RT-PCR (Roche Molecular Systems, Branchburg, New Jersey, USA) . The assay was performed on 100μl of blood plasma, 100μl of seminal plasma (diluted 1:4), 100μl unseparated semen (diluted 1:10), 1-10¥105 spermatozoa, or 1-10¥105 seminal NSCs, according to the manufacturer‚s instructions.
HIV proviral DNA
DNA from PBMCs (106 cells) was extracted using Tri Reagent (Sigma). The extraction was performed according to the manufacturer‚s instructions. The DNA pellet was air-dried and dissolved in 100μl of nuclease-free water (Biogenesis; Bournemouth, UK). The concentration of DNA was determined by spectrophotometry. NucliSens lysis buffer was used for DNA preparation from whole semen, spermatozoa and NSCs. The proviral DNA from these sources was amplified in a nested polymerase chain reaction (PCR) as outlined by Simmonds et al. . The PCR reaction consisted of Stoffel buffer, 1mM dNTP, 3.75mM MgCl2, 1μM primer, 2 units of Ampli-Taq DNA polymerase (Perkin-Elmer; Warrington, Cheshire, UK). 0.2μg of DNA template was added per 20μl reaction mix. The DNA was amplified with primers for envelope, designed specifically to amplify a highly conserved region of the HIV-1 envelope region, and b-globin primers were used as positive control primers and as a single haploid copy gene to estimate the cellular genome copy number when testing the efficiency of the PCR reaction.
First round b-globin primers were 5‚ GGTTGGCCA ATCTACTCCCAGG 3‚, and 5‚ GCTCACTCAGTG TGGCAAAG 3‚. Second round b-globin primers were 5‚ ACACAACTGTGTTCACTAGC 3‚ and 5‚ CAACTTCATCCACGTTCACC 3‚. First round env primers were 5‚ TCAGGAAGGGGACCCAGAAATT 3‚ (7317-7336) and 5‚ GATCCCATAGTGCTTCC TGCTGCT 3‚ (7795-7818). Second round env primers were 5‚ GGGGAATTTTTCTACTGTAAT 3‚ (7361-7381) and 5‚ CTTCTCCAATTGTCCCTC ATA 3‚ (7645-7665). Nucleotide numbers are taken from the HXB2 HIV sequence. An initial pre-PCR denaturation was performed at 94°C for 5min followed by 40 cycles of 94°C for 1min, 50°C (for first round) or 45°C (for nested PCR) for 1min, 72°C for 1min. The PCR products visualized on a 1.5 % agarose (Sigma) gel in the presence of ethidium bromide (Sigma). The samples were scored as positive or negative. A negative result was scored only if the sample was negative in three separate experiments.
Spermatozoa or NSCs from HIV-negative individuals were washed in fluorescence-activated cell sorter (FACS) buffer (phosphate-buffered saline (PBS) plus 2mM EDTA plus 0.5% fetal calf serum (FCS)). A total of 105 cells in 100μl were incubated on ice for 10min before adding 10μl of antibody and incubating on ice for a further 45min. The cells were washed twice in FACS buffer and, where appropriate, the second layer antibody or streptavidin-phycoerythrin (PE) were added to the cells and incubated and washed as above. The cells were then analysed on a FACScalibur (Becton Dickinson, Oxford, UK) using Cellquest to acquire and analyse the data. Single-colour staining was performed using the following antibodies: CD4-PE (clone Q4120) and CD14-PE (UCHM-1) (Sigma), CCR5-FITC (45502.111), CXCR4-biotin (44708.111) plus streptavadin-PE, and isotype controls (R+D; Abingdon, Oxon, UK), polyclonal sheep anti-human spermatozoa (Chemicon International Inc., Harrow, Middx, UK), donkeyanti-sheep IgG-fluorescein- isothiocyanate (FITC) (Sigma) and CD4-PE (clone MT310) and isotype control (Dako, High Wycombe, Bucks, UK).
Whole semen was separated into plasma, live ‚swim-up‚ spermatozoa and NSCs. The ‚swim-up‚ reduced the sperm count, by up to one log compared with that in the whole ejaculate (as reported by other authors ). The level of viral RNA was compared with that in whole semen and in blood plasma. The aim was to examine whether ‚sperm washing‚ significantly reduced the viral load in the spermatozoa to be used for insemination, and also to determine where the virus was located. Table 1 illustrates that in all semen samples the ‚swim-up‚ procedure reduced the amount of virus in the spermatozoa to lower than the detection limit (LDL; ranging from 20 to 80 copies per sample tested). HIV RNA in the non-swim-up spermatozoa or 80/40% interface (non-motile) was not detected (data not shown). The NucliSens assay incorporates three standard calibrators into each sample before RNA extraction, which demonstrates efficient RNA extraction and amplification for every fraction tested. The calibrators amplified with the same efficiency in all samples tested (with the exception of patient 6 in the spermatozoal fraction), and therefore there was no evidence to suggest that there was any variation in the extraction of RNA or the amplification between fractions tested. We found no indication that the spermatozoa, even dead or non-motile, were reservoirs of HIV RNA. Four out of 11 samples were positive for HIV RNA in the NSCs fraction, all of whom had significant viral loads in the whole semen sample. The data in Table 1 indicates that the main viral reservoir in semen is the seminal plasma, although the NSCs may also carry virus.
It is possible that washed spermatozoa contain latent virus in the form of integrated proviral DNA. Therefore each of the seminal fluid cellular fractions and PBMCs were tested for HIV-1 DNA by PCR. The efficiency of the DNA PCR was tested by preparing DNA from a defined number of OM10.1 cells (AIDS Reagent Project, NIBSC, Potters Bar, Hertfordshire, UK); each cell carries a single copy of HIV proviral DNA. By end-point titration (logarithmic) of the DNA, using b-globin and the HIV env-primers, it was demonstrated that these products amplified with the same efficiency, and that the PCR could detect an input of 1 to 10 b-globin or HIV env sequences, i.e. 1-10 cells. Ten of the 11 patients had proviral DNA present in the PBMCs, unfractionated semen and NSCs. Patient 4 did not have any detectable proviral DNA present in PBMCs or NSCs, although the unfractionated semen sample was weakly positive on two occasions. Also patient 4 did not have any detectable viral RNA in any of the compartments studied (Table 1). All the spermatozoa analysed were negative for proviral DNA (Fig. 1).
The final approach was to determine whether the spermatozoa have the potential to be infected, or whether virus could simply attach to the cell surface, by evaluating whether spermatozoa expressed CD4 or the HIV co-receptors CCR5 or CXCR4. The data illustrated in Fig. 2 are representative of the profiles obtained from five healthy volunteers. The ‚swim-up‚ procedure did yield very pure populations of spermatozoa cells (Fig. 2(g)). Figure 2(h and j) demonstrates the absence of CD4 and CCR5 on the spermatozoa cells. Two different clones of anti-CD4 antibodies were used (data shown for Q4120). Low levels of CXCR4 expression were observed (Fig. 2(i)) on the spermatozoa. Whether this level of CXCR4 would be sufficient to allow X4 viruses to bind must be confirmed in vitro, although the lack of expression of CD4 and CCR5 on the spermatozoa confirm that these cells are unlikely to be readily infectable. Within the NSCs a population of cells exists with intense expression of CD4 (Fig. 2(b)). These cells may be lymphocytes or monocytes as shown by the expression of CD14 (Fig. 2(e)). Within the NSCs low levels of CXCR4 were detected compared with the isotype control (Fig. 2(c)), but more significantly, there was a distinct population of CCR5 expressing cells (Fig. 4(d)). This may indicate the compartmentalization of CCR5-utilizing viruses within the semen.
Controversy surrounds whether spermatozoa may be infected with HIV [2-10]. This is of particular importance in the evaluation of ‚sperm washing‚ as a viable risk-reduction option for HIV-discordant couples wishing to have children. Several groups have reported that HIV may attach to and infect the spermatozoa via a CD4-like molecule [4,6,7]. We did not demonstrate the presence of CD4 on the spermatozoa using two different clones of anti-CD4 antibodies. A number of other groups have reported similar findings [15,16]. Neither HIV RNA or DNA was detected in any spermatozoa fraction of any of our subjects. Our data demonstrate that ‚sperm washing‚ may achieve substantial reductions in the level of virus present, although because of our sample size it is not possible to guarantee that the LDL would be achieved with all patients. Our data is in concordance with clinical data from Semprini et al. , who reported that no HIV infection has occurred after more than 1000 inseminations with washed sperm. A study by Lasheeb et al.  indicated that ‚sperm washing‚ reduced HIV viral load in semen, although the spermatozoa were separated after freezing the semen sample, which resulted in cell lysis, allowing RNA contamination of the cellular compartments.
Semen was donated from patients with a range of viral loads in peripheral blood (LDL to 400000 copies ml-1). Not all these patients shed virus into the semen but we believe they may be representative of the patients who will wish to take have a ‚sperm-washing‚ service provided, many of whom will have low viral loads as a result of therapy . Six individuals had HIV RNA present in the seminal plasma, four of whom had viral RNA present in the NSCs. Nine individuals had detectable proviral DNA in the NSCs. The main viral reservoir in the semen was the seminal plasma and NSCs (Table 1). The NSCs represent a very heterogeneous population containing immature germ cells, leukocytes and epithelial cells. The leukocyte fraction contains CD45 lymphocytes, CD68macrophages, CD3 T cells (both CD4 and CD8) and CD103T cells, which are found in the epithelium and lamina propria . The main reservoir for HIV appears to be the lymphocyte and macrophage populations, but not the immature germ cells .
Poor correlation was found between the viral load in the semen and that in peripheral blood (Table 1). For example, patient 5 had 380000 copies ml-1 in blood and LDL in semen but patient 9 had only 1200 copies ml-1 in blood and 56000 copies ml-1 in unfractionated semen. Other groups report similar findings [18,19]. The shedding of virus into the semen may be intermittent, with the level of seminal virus increasing with factors such as decreased CD4 cell count and asymptomatic genital tract infection [17,20]. Therefore because the presence of HIV in the plasma is not always correlated with its presence in semen, the detection of virus in plasma is probably not a good surrogate for infectivity as a result of heterosexual transmission. These data also suggest that blood and semen may represent two distinct compartments in terms of HIV replication. Indeed Byrn et al.  cloned and sequenced the protease gene from the semen and blood of two individuals and found that the virus present in the semen and blood represented two distinct HIV families. Furthermore, the majority of individuals seroconvert with macrophage-tropic (M-tropic) or CCR5-using viruses (R5 viruses) [22,23], despite such species representing a minor viral variant in the blood of the transmitter . It has been postulated that the selection of the R5 viruses takes place at the mucosal surface, because it is the Langerhans cells (LC) that are first to become infected , and these cells only express the CCR5 receptor . It is also possible, however, that the semen is a major reservoir for the CCR5-utilizing viruses. This hypothesis is supported by phenotypic and genotypic analyses of viral isolates from blood and semen [24,27], which found predominantly M-tropic virus present in the semen regardless of the tropism of the virus found in the periphery. We have shown that the NSCs express CCR5, but negligible CXCR4, suggesting that this may contribute to the selection of R5 viruses within the semen. As sexual transmission accounts for the vast majority of HIV transmission it is of pivotal importance to determine whether the virus within the semen represents a distinct compartment of viral replication and hence viral phenotype. An effective prophylactic vaccine must confer protection against the phenotype shed in genital secretions. This may be of critical importance when considering non-clade B viruses, which appear to be much more easily transmitted heterosexually. Little is known of how the virus in blood and semen are related in non-clade B infections.
‚Sperm washing‚ as a risk-reduction programme has been a controversial area. Mandelbrot et al.  believe the risk of transmission is sufficiently low that HIV-discordant couples may attempt natural conception. They report a follow-up of 92 HIV-negative women with HIV-positive partners. Most couples had received pre-conceptual counselling on the risk of transmission, and were advised to pinpoint ovulation to reduce the risk. Two women seroconverted at 7 months of pregnancy and another two post-partum, with seroconversions restricted to couples with inconsistent condom use. Perhaps advising individuals that they may abandon condoms to conceive may encourage intermittent condom use. Our data suggest that isolating the sperm for insemination would remove such a risk.
Antiviral therapy has a marked effect on virus shedding in the semen, as treatment-induced changes in the viral load in the blood are generally reflected by corresponding changes in the viral load in the semen . A further risk-reduction exercise might thus be to encourage all HIV-positive men who wish to participate in such a programme to consider commencing potent antiretroviral therapy, but the teratogenicity of many of the drugs currently in use has not been completely evaluated.
Our data suggest that the primary reservoir for HIV RNA in semen is the seminal plasma and NSCs, and that the level of viral RNA could be reduced to LDL in the washed spermatozoal fraction. Second, all the spermatozoa fractions analysed for latent proviral DNA in this study were negative, whereas the unfractionated semen (10 out of 11 patients) were positive. We have failed to detect the presence of CD4 or CCR5 on the spermatozoa, and the expression of CXCR4 on the spermatozoa was low, suggesting that these cells are not likely to be readily infected by HIV. Therefore, we provide evidence to suggest that ‚washing sperm‚ reduces the amount of HIV present, and hence may reduce the risk of HIV transmission. Taken together with the clinical data from Semprini at al.  we would promote ‚sperm washing‚ as an efficient and simple procedure to reduce the risk of HIV transmission in HIV-discordant couples wishing to have children.
1. Clumeck N, Taelman H, Hermans P, Piot P, Schoumacher M, De Wit S: A cluster of HIV infection among heterosexual people without apparent risk factors. N Engl J Med
2. Quayle AJ, Xu C, Mayer KH, Anderson DJ: T lymphocytes and macrophages, but not motile spermatozoa, are a significant source of human immunodeficiency virus in semen. J Infect Dis
3. Baccetti B, Benedetto A, Burrini AG, et al
.: HIV particles detected in spermatozoa of patients with AIDS. J Submicrosc Cytol Pathol
4. Baccetti B, Benedetto A, Burrini AG, et al
.: HIV-particles in spermatozoa of patients with AIDS and their transfer into the oocyte. J Cell Biol
5. Scofield VL, Rao B, Broder S, et al
.: HIV interaction with sperm [letter]. AIDS
6. Bagasra O, Farzadegan H, Seshamma T, Oakes JW, Saah A, Pomerantz RJ: Detection of HIV-1 proviral DNA in sperm from HIV-1-infected men. AIDS
7. Bagasra O, Freund M, Weidmann J, Harley G: Interaction of human immunodeficiency virus with human sperm in vitro . J Acquired Immune Defic Syndr
8. Gobert B, Amiel C, Tang JQ, Barbarino P, Bene MC, Faure G: CD4-like molecules in human sperm. FEBS Lett
9. Lasheeb AS, King J, Ball JK, et al
.: Semen characteristics in HIV-1 positive men and the effect of semen washing. Genitourin Med
10. Nuovo GJ, Becker J, Simir A, et al
.: HIV nucleic acids localize to the spermatogonia and their progeny. Am J Pathol
11. Semprini AE, Levi-Setti P, Bozzo M, et al
.: Insemination of HIV-negative women with processed semen of HIV-positive partners [see comments]. Lancet
12. Semprini AE, Fiore S, Pardi G: Reproductive counselling for HIV-discordant couples [letter; comment]. Lancet
13. Dyer JR, Gilliam BL, Eron Jr JJ, Grosso L, Cohen MS, Fiscus SA: Quantitation of human immunodeficiency virus type 1 RNA in cell free seminal plasma: comparison of NASBA with Amplicor reverse transcription - PCR amplification and correlation with quantitative culture. J Virol Methods
14. Simmonds P, Balfe P, Peutherer JF, et al
.: Human immunodeficiency virus-infected individuals contain provirus in small numbers of peripheral mononuclear cells and at low copy numbers. J Virol
15. Wolf H, Anderson DJ: Male genital tract inflammation associated with increased numbers of potential human immunodeficiency virus host cells in semen. Andrologia
16. el-Demiry MI, Hargreave TB, Busuttil A, James K, Chisholm GD: Identifying leucocytes and leucocyte subpopulations in semen using monoclonal antibody probes. Urology
17. Vernazza PL, Gilliam BL, Flepp M, et al
.: Effect of antiviral treatment on the shedding of HIV-1 in semen. AIDS
18. Liuzzi G, Bagnarelli P, Chirianni A, et al
.: Quantitation of HIV-1 genome copy number in semen and saliva [letter]. AIDS
19. Liuzzi G, Chirianni A, Clementi M, et al
.: Analysis of HIV-1 load in blood, semen and saliva: evidence for different viral compartments in a cross-sectional and longitudinal study. AIDS
20. Xu C, Politch JA, Tucker L, Mayer KH, Seage III GR, Anderson DJ: Factors associated with increased levels of human immunodeficiency virus type 1 DNA in semen. J Infect Dis
21. Byrn RA, Zhang D, Eyre R, McGowan K, Kiessling AA: HIV-1 in semen: an isolated virus reservoir [letter]. Lancet
22. van‚t Wout AB, Kootstra NA, Mulder-Kampinga GA, et al
.: Macrophage-tropic variants initiate human immunodeficiency virus type 1 infection after sexual, parenteral, and vertical transmission. J Clin Invest
23. Roos MT, Lange JM, de Goede RE, et al
.: Viral phenotype and immune response in primary human immunodeficiency virus type 1 infection. J Infect Dis
24. Zhu T, Wang N, Carr A, et al
.: Genetic characterization of human immunodeficiency virus type 1 in blood and genital secretions: evidence for viral compartmentalization and selection during sexual transmission. J Virol
25. Spira AI, Marx PA, Patterson BK, et al
.: Cellular targets of infection and route of viral dissemination after an intravaginal inoculation of simian immunodeficiency virus into rhesus macaques. J Exp Med
26. Zaitseva M, Blauvelt A, Lee S, et al
.: Expression and function of CCR5 and CXCR4 on human Langerhans cells and macrophages: implications for HIV primary infection. Nat Med
27. Kroodsma KL, Kozal MJ, Hamed KA, Winters MA, Merigan TC: Detection of drug resistance mutations in the human immunodeficiency virus type 1 (HIV-1) pol gene: differences in semen and blood HIV-1 RNA and proviral DNA. J Infect Dis
28. Mandelbrot L, Heard I, Henrion-Geant E, Henrion R: Natural conception in HIV-negative women with HIV-infected partners [letter] [see comments]. Lancet