Is the male genital tract really a sanctuary site for HIV? Arguments that it is not

Introduction One concern with current antiretroviral therapy for infection with HIV is that certain anatomical sites in the body are considered ‘sanctuary sites'. What is commonly meant is that these sites are less permeable to antiretroviral drugs, allowing the selection or development of drug-resistant strains, and thereby representing an important source of virological failure of therapy [1–8]. The central nervous system is considered such a sanctuary site [4,9–11], and because of the so-called ‘blood–testis barrier’ (BTB), the male genital tract (MGT) is also considered to be such a site [1–8,12,13]. This assumption is very important in view of the sexual transmission of drug-resistant HIV-1, which is rapidly increasing in some areas of the world [14–21]. Following a review of the literature, however, we oppose the opinion that the MGT is a sanctuary site for HIV. Definition of viral compartment, reservoir and sanctuary site The terms ‘compartment', ‘reservoir’ and ‘sanctuary site’ are not well defined and are used interchangeably [1–8]. With respect to the MGT, definitions can be proposed on the basis of virology, pharmacology, immunology and reproductive biology. We use the following definitions here: A viral compartment is an anatomical site in which the virus in untreated patients evolves distinctively from other anatomical sites or the main pool of infected cells, because of differences between the major cell types sustaining viral replication [2]. A viral reservoir is a cell type or anatomical site in which a replication-competent virus persists much longer than in the main pool of infected cells that sustain the infection, and this cell type or anatomical site can replenish the pool of infected cells [7,8]. The main cellular viral reservoir is the pool of latently infected resting CD4 T cells [8]. A viral sanctuary site is an anatomical site that is highly impermeable to (some) antiretroviral drugs, and in which viral replication continues during treatment, thus allowing the development or selection of drug-resistant strains [1–8]. The misleading concept of the blood–testis barrier Conclusions about HIV in the MGT are mainly based on an analysis of semen, 90% of which consists of plasma and 10% of cells. Seminal plasma derives mainly from the seminal vesicles (60%) and prostate glands (30%), and therefore reflects fluids distal to the vas deferens [22–28]. HIV-1-RNA in the seminal plasma derives mainly from the distal part of the MGT, which is supported by the finding that vasectomy has no effect on the seminal plasma HIV-1-RNA concentration [29], and that proper target cells for HIV in the seminiferous tubules are absent [22–24,30,31]. Approximately 90% of leukocytes in semen are derived from the epididymis [12,32–34]. It is well known that many antiretroviral drugs do not reach adequate concentrations in the seminal plasma and as a consequence no truly highly active antiretroviral therapy is achieved within the lumen of the MGT with many regimens (see Table 1) [2–5,7,35,36]. Drugs with a protein binding of less than 90% are the ones with a moderate to good concentration in the seminal plasma, suggesting that diffusion is the primary driving force for drug transport to the lumen of the MGT (see Table 1) [2,3]. However, active cellular drug transport, e.g. by P-glycoprotein, could also be important [3,37–40].Table 1: Characteristics and concentrations in blood and seminal plasma of licensed antiretroviral drugs.On the basis of histological and physiological studies, and the molecular weight of free and protein-bound antiretroviral drugs, it is unlikely that there is an important endothelial barrier to the current antiretroviral drugs in the testicular interstitial tissue (see Fig. 1 and Table 1) [23,61–72]. Therefore, opposing the suggestion by others [62,63], we argue that the testicular extraluminal tissue is not a drug-deprived area and sanctuary site.Fig. 1. Microscopic anatomy of the seminiferous tubules.: (a) Transverse section of several seminiferous tubules. The blood capillaries are within the interstitium. (b) Transverse section of a portion of a seminiferous tubule. The blood–testis barrier is located in the seminiferous tubule between the Sertoli cells. Reprinted from Tortora and Grabowski (editors) [22], with permission from John Wiley & Sons, Inc.The BTB, which is formed by the continuous layer of abluminally located Sertoli cells and the tight junctions between them (see Fig. 1) [22–26,73,74], is relevant for drug concentrations within the lumen of the seminiferous tubules because tight junctions hamper the intercellular diffusion of hydrophilic solutes at least a hundred times more than other intercellular junctions [65]. The BTB is located only in the seminiferous tubules, which is just a fraction of the total length of the MGT of approximately 6 m (Fig. 1) [22–26]. Also approximately 90% of the seminiferous tubule fluid is resorbed in the epididymis [23–25]. Therefore the BTB is not an explanation for the subtherapeutic drug concentrations in the seminal plasma. HIV particles can get into the lumen of the seminiferous tubules, possibly through transcytosis [75–88]. However, within the normal seminiferous tubule leukocytes, the target cells for HIV appear to be absent [22–24,30,31]. Therefore, antiretroviral drug concentrations ‘behind’ the BTB are not relevant, and as a consequence the lumen of the seminiferous tubules normally cannot be a sanctuary site for HIV. In the advanced stage of HIV infection histological changes in the seminiferous tubules occur and this may affect the BTB [13,77,86,89,90]. Virological counter-arguments Despite (expected) subtherapeutic seminal plasma drug concentrations, longitudinal [91–112] and cross-sectional [57,91–94,110,113–130] studies have shown that the virological response (HIV-1-RNA concentration) in the seminal plasma parallels that in blood plasma, and that the more suppressive the regimen, the better and more sustained the response. Such studies also show that compared with blood plasma, the median HIV-1-RNA concentration in the seminal plasma on average is approximately 10-fold lower. Nevertheless, there are some individuals in whom the HIV-1-RNA concentration in the seminal plasma is higher than in the blood plasma. This divergence is difficult to interpret in the absence of details concerning an individual's stage of HIV infection and details on asymptomatic local infection/inflammation (which may have a prevalence of 8% [130]) or leukocytospermia (idiopathic semen leukocytosis), which might influence local HIV production [93,112,114,120,122, 130,131]. With properly suppressive antiretroviral therapy the concentration of HIV-1-DNA in peripheral blood mononuclear cells and in non-spermatozoal seminal mononuclear cells (NSMC) decreases, and it can drop to below the limit of detection in NSMC in contrast to peripheral blood mononuclear cells [96,118,125,132–136]. Furthermore, HIV-2-long terminal repeat (LTR) circular DNA is absent in NSMC when HIV is properly suppressed [115]. This absence indicates the arrest of viral replication [137–139]. The combination of the absence of HIV-2-LTR circular DNA with the presence of proviral DNA [115] suggests that infected cells in semen derive from blood, but become inactivated and unable to sustain HIV replication within the lumen of the MGT. Studies on HIV-1 drug-related mutations have demonstrated that the mutations in blood and semen are homologous [56,57,95,102,104,113,120,125,128,140–142]. Only in three studies did the mutations appear first in the semen, in five out of 46 patients [56, 57,113,120,140]. In one study [120] this was related to an (asymptomatic) local infection/inflammation, reflecting more virus replication during inflammation. Clearly, drug-resistant variants only rarely appear first in the semen compared with blood and (asymptomatic) local inflammation/infection might be a very important factor for the emergence and transmission of drug-resistant virus. Divergent envelope evolution of HIV is not a phenomenon unique for the MGT, and probably merely reflects an adaptation to local circumstances [117,143–149]. Even between the seminal plasma and seminal leukocytes phylogenetic differences have been demonstrated [150]. As different genes are involved, envelope changes (compartmentalization) must not be made synonymous with the development of drug-resistance mutations [95]. Immunological counter-arguments Under normal conditions no leukocytes are found in the lumen and epithelium of the seminiferous tubules, which is understandable given the immunogenicity of the process of spermatogenesis [22–24,30,31]. The ampulla and tail of the epididymis are the sites where the spermatozoa are stored before ejaculation [22–24]. As spermatozoa are very immunogenic, leukocytes have to become anergic. Several immunosuppressive mechanisms may play a role: the likely acidity of the epididymal and prostatic fluid [151–160]; a lack of co-receptor B7 expression and the presence of transforming growth factor beta and complement inhibitors [12,161–164]; the preponderance of CD8 (suppressor) T lymphocytes over CD4 (helper) T lymphocytes within the luminal lining cells from the rete testis up to and including the prostate glands and seminal vesicles [30,31,165–167]; possibly also a preponderance of CD8 over CD4 T lymphocytes in the semen [12, 88,121,164,168–174]; the presence of HIV-1-specific cytotoxic T lymphocytes [174]; the likely high expression of Fas ligand on spermatozoa, thus inducing the apoptosis of cells such as activated lymphocytes [175–177]; and the strong immunosuppressive activity of the seminal plasma and its components spermine and prostaglandins [13,26,163,164,177–183]. The consequence of an immunosuppressive state within the lumen and luminal lining cells of the MGT, especially in the epididymis, is that the local CD4 cells are infected less easily and produce little or no HIV. The immunosuppressive state also implies a high barrier for the development of drug resistance, which is compatible with the finding that drug-resistant mutations only rarely appear first in the semen. Hypothesis concerning the source of HIV in semen The finding that vasectomy has no effect on the seminal plasma HIV-1-RNA concentration [29], the virological response to therapy in the seminal plasma, and the immunological counter-arguments all indicate that it is not the seminal leukocytes but rather the luminal lining cells and/or the blood/extraluminal tissue that are the source of HIV-1-RNA in the seminal plasma (spillover). These two sources are fully exposed to antiretroviral drugs. The size of the HIV particle could be an argument against spillover [184,185]. However, leukocytes also cross cell membranes. Furthermore, some data suggest that residing tissue macrophages could also account for HIV in the seminal plasma, because these cells can have lumen-reaching dendritic processes [186]. A likely consequence of spillover could be that the transmission of HIV and drug-resistant variants via the semen does not occur in cases of proper viral suppression in blood plasma (≤ 50 copies/ml). An exception is possibly a minority of patients in whom circumstances exist that may be associated with (more) local HIV production. Relevant factors in this respect are (a)symptomatic local inflammation/infection, leukocytospermia and advanced-stage HIV infection. The treatment of local inflammation/infection could be an important measure for preventing the development and transmission through semen of drug-resistant virus [93,95,112,114,120,122,130,131,142]. In conclusion, the HIV-1-RNA concentration gradient between blood plasma and seminal plasma, the parallel virological response to therapy in blood plasma and seminal plasma, including the decrease of proviral DNA and the absence of HIV-2-LTR cDNA in NSMC, vasectomy not influencing the seminal plasma HIV-1-RNA concentration, and the intraluminal immunosuppressive state are all strong arguments against intraluminal HIV production in the MGT. We therefore argue that the source of HIV in the semen is very likely outside the lumen of the MGT, i.e. spillover from luminal lining cells and/or the blood/extraluminal tissue, which are fully exposed to antiretroviral drugs. With respect to the definitions of viral compartment, reservoir and sanctuary site, the seminiferous tubule is neither a virological compartment nor a sanctuary site for HIV-1, because of the absence of appropriate target cells. The remainder of the lumen of the MGT is a viral compartment and reservoir for HIV-1 because proviral DNA-containing leukocytes are present, but this part is also not a virological sanctuary site because under normal circumstances replication is probably arrested as a result of the local immunosuppressive state. There is a minority of patients in whom local factors are present that result in (more) HIV production within the lumen of the MGT, and thereby they are probably more prone to the development of drug-resistant HIV in the MGT. Relevant characteristics of these individuals are possibly semen leukocytosis and an advanced stage of HIV infection. This warrants the aggressive screening for and treatment of sexually transmitted infections and local inflammation in such patients.