HIV and SIV in Body Fluids: From Breast Milk to the Genitourinary Tract

Abstract

HIV-1 is present in many secretions including oral, intestinal, genital, and breast milk. However, most people exposed to HIV-1 within these mucosal compartments do not become infected despite often frequent and repetitive exposure over prolonged periods of time. In this review, we discuss what is known about the levels of cell-free HIV RNA, cell-associated HIV DNA and cell-associated HIV RNA in external secretions. Levels of virus are usually lower than contemporaneously obtained blood, increased in settings of inflammation and infection, and decreased in response to antiretroviral therapy. Additionally, each mucosal compartment has unique innate and adaptive immune responses that affect the composition and presence of HIV-1 within each external secretion. We discuss the current state of knowledge about the types and amounts of virus present in the various excretions, touch on innate and adaptive immune responses as they affect viral levels, and highlight important areas for further study.

1. INTRODUCTION

HIV-1 is present in many secretions including oral, intestinal, genital, and breast milk. However, most people exposed to HIV-1 within these mucosal compartments do not become infected despite often frequent and repetitive exposure over prolonged periods of time. The levels of HIV RNA and DNA in these secretions are usually much lower than those observed in contemporaneously obtained blood [1–3]. The mechanisms by which the quantity of virus is reduced at mucosal surfaces are unknown and may vary by site. Levels of HIV RNA in mucosal secretions generally increase with local inflammation or infection, but even in these settings the quantity of virus usually remains lower than that in plasma. Levels in secretions generally decrease with antiretroviral therapy (ARV). Unlike plasma and cerebrospinal fluid levels, which are increased in primary infection, HIV-RNA levels in mucosal secretions including saliva, cervicovaginal (CVF) fluid and semen (excluding hyperexcretors), remain unchanged between primary and chronic infection [2, 4].

Mucosal compartments have distinct innate and adaptive immune responses. The role of these responses in curtailing HIV levels is not well understood. Although the quantity of HIV at mucosal surfaces is restricted, the HIV variants are generally genetically similar to those in the blood and undergo limited evolution (see below). Thus, the genetic bottleneck observed during transmission does not appear to be due to the selection of variants at mucosal surfaces [3, 5].

IgA is the predominant isotype in secretions (refer to article “humoral and cellular immune responses in mucosal secretions and tissues”) but does not seem to be a critical factor in limiting HIV [6]. In fact, some studies suggest that IgG may be more important in limiting HIV than IgA, despite much lower levels [7, 8]. High levels of HIV-specific IgA were associated with an increased risk of transmission in the RV144 trial [9, 10].

All mucosal fluids including oral, intestinal, genital, and breast milk contain both cell-free and cell-associated virus [11], though the contribution of each type to transmission remains controversial [11, 12]. Cell-free virus has been the focus of most investigations on HIV-1 transmission. Many studies correlate levels of cell-free virus in the blood, genital fluids, and breast milk with transmission. Antiretrovirals are extremely effective in lowering cell-free HIV RNA levels and preventing most transmission of HIV. However, there are compelling reasons to suggest that cell-associated virus may be important in transmission as discussed in the individual sections below.

In this article, we will present an overview of the amount and type of virus present in external secretions. Other articles in this issue will dive into the complexities of mucosal transmission of HIV-1 with a review of barriers of mucosal entry (Carias, page 4), virologic aspects of mucosal transmission (Ende, page 14), mucosal target cells (Smith, page 28), mucosal and tissue humoral and cellular immune responses (Sabbaj, page 41; Mestecky, page 49), mucosal immunity (Shacklett, page 63), CD4 depletion (Veazey, page 76), role of sex hormones (Patel, page 92), mucosal vaccines (Kozlowski, page 102), and monoclonal antibodies (Anderson, page 123).

2. OROPHARYNGEAL SECRETIONS

2.1. Cell-free HIV in Oral Secretions

HIV levels in saliva are generally quite low, but this notion may underestimate the importance of the oropharyngeal cavity to HIV risk and transmission. HIV is present in both salivary gland tissues [13] and saliva, detected in 42% [14] - 96% [15] of patients by RNA methodologies [14–16]. The restricted quantities of salivary virus vary widely and usually correlate with plasma viral loads. There are only a few reports of salivary viral loads with median HIV RNA of 162 copies/ml (range 0–72,080) in subjects with a median plasma HIV RNA of 14,817 copies/ml (range 167–254,880) [15] and mean salivary HIV RNA of 247 copies/ml (range 0–86,240) compared to 59,260 copies/ml (range 1,034–886,780) in plasma [16]. Viral DNA is detected much less frequently, 21% to 49% of samples, but generally in very low quantities even in severely immunocompromised individuals [17, 18].

Shedding of HIV RNA in saliva increases with increasing plasma viral load, decreasing CD4+ T cell counts, diabetes, and bleeding gingival sites. Salivary viral shedding decreases with antiretroviral therapy [15, 16, 19]. Furthermore, HIV can rarely be cultured from saliva [20–24], suggesting that salivary factors inhibit or destroy the infectious virus.

In contrast to the low levels of HIV RNA in saliva, pharyngeal swabs contain a large amount of virus without the usual quantitative restriction noted in secretions. Median log₁₀ copies/ml are 4.24 for plasma and 4.22 for pharyngeal HIV-1 RNA load [25]. Within subject variability is greater with pharyngeal samples than plasma levels, 0.66 vs. 0.30 log₁₀ copies/ml. Pharyngeal levels of HIV RNA are inversely correlated with CD4 count and decrease with a history of tonsillectomy and ARV, independent of plasma HIV level. Furthermore, HIV has been cultured from 4 of 14 men (29%) with intact tonsils and high pharyngeal HIV RNA levels (>50,000 copies/ml) [25]. This is not surprising as the posterior pharynx contains large lymphoid aggregates, but it is impressive how little infectious virus is present in saliva given the proximity.

Saliva is known to have many factors that neutralize or destroy pathogens. Studies of salivary factors, including mucins, defensins, secretory leukocyte protease inhibitor, lactoferrin, cystatins, and anti-HIV antibodies among others, are associated with a 2 to 5 fold reduction in HIV [21, 23, 26–35]. Contrary to most salivary bacterial and viral responses, IgA and secretory IgA (S-IgA) are not the main reactive antibodies to HIV, although they constitute >95% of the antibodies in saliva. Salivary concentrations of IgA and S-IgA are not increased in HIV-infection and have poor reactivity [36]. In paired serum and saliva samples, only 57% of subjects have anti-HIV specific IgA in the saliva compared to 97% in the serum [37]. Additionally, S-IgA in saliva is restricted to the IgA1 subclass [38]. Surprisingly, despite a 100-fold lower concentration, salivary IgG have significantly greater reactivities [37]. While a 2–5 fold reduction in HIV from salivary factors is significant, it does not explain the almost complete lack of infectious HIV in saliva. Hypotonic disruption appears to be a major mechanism of salivary action. Saliva, at one-seventh the tonicity of normal interstitial fluids, almost immediately lyses ≥90% of mononuclear leukocytes that produce infectious virions. Hypotonic disruption leads to a 10,000-fold inhibition of HIV multiplication in vitro [39], while other salivary factors neutralize any cell-free virus present. However, hypotonic lysis may be overcome in settings where a sufficient volume of isotonic fluid, whether blood or milk, is present in the mouth.

Therefore, while the saliva contains very little HIV, the posterior pharynx contains large quantities of virus that can be infectious. Despite this, only a few cases of oral HIV transmission have been reported: one from a bite of a patient with heavy oral bleeding [40] and a few from orogenital sex [41–44]. The possibility that more cases may occur and go undetected, due to the inability to rule out other more common transmission routes, deserves consideration.

2.2. Cell-associated HIV in Oral Secretions

There are very little data to guide a discussion of cell-associated HIV RNA or DNA in oral secretions. The hypotonicity of saliva makes it unlikely that many lymphocytes or macrophages survive intact to harbor infectious virus except in cases of significant oral bleeding or in isolated regions in the oral cavity. In gingival crevicular fluid, which is often enriched with mononuclear cells, proviral HIV-1 DNA is detected in approximately half of the subjects (17 of 35) [17]. Detection correlates with a higher plasma viral load [17]. In the evaluation of pharyngeal swabs referenced above, cell-associated HIV was neither evaluated nor controlled for in the study [25]. A single report of both unspliced and multiply spliced RNA transcripts detected in archival paraffin-embedded tonsils [45] suggests that cell-associated virus is present in tonsillar tissues.

3. SEMEN

The most common mode of HIV-1 infection is through sexual transmission [46] and semen is the most common vector of HIV-1 delivery into the anogenital mucosal compartments [47, 48]. However, the mechanisms by which semen facilitates HIV-1 infection and the contribution of seminal factors to the variability of viral transmission per sexual act remain controversial [49, 50]. Furthermore, semen and seminal plasma have demonstrated enhancing [51–53], inhibitory [54–57], or no effect [58, 59] on HIV-1 transmission in in vitro and in vivo studies.

3.1. Semen: Facilitator or Inhibitor of HIV-1 Transmission?

The cationic polypeptides in seminal plasma have intrinsic antiviral activity [57]. Furthermore, components of seminal plasma have inhibitory effects on HIV-1 gp120 transmission by DC-SIGN-positive dendritic cells in the anogenital mucosa [54, 56]. On the other hand, semen-derived enhancers of virus infection (SEVI), comprised of prostatic acid phosphatase and semenogelin-derived fragments, have been found to significantly enhance HIV-1 infectivity up to 10⁵-fold in cell culture models in the setting of low viral doses [52, 53, 60]. These amyloid fibrils are thought to enhance HIV-1 infection by promoting the binding of HIV-1 to target cells [61, 62]. Additionally, microbicide antiviral efficacy, with the exception of Maraviroc, was decreased in TZM-bl cell lines when exposed to semen [63]. However, these semen-enhancing effects were not demonstrated in vivo using semen and purified semen components to transmit SIV/HIV-1 in rhesus macaques nor in human cervical and colonic explants [58, 59, 64]. Studies to assess the direct influence of semen and seminal components on HIV-1 transmission across the anogenital mucosa are imperative for microbicide prevention strategies.

3.2. Origin of HIV-1 in Semen

Semen is an alkaline secretion with spermatozoa, immature germ cells, and HIV-susceptible polymorphonuclear leukocytes including macrophages and CD4+ T cells [65]. Semen and pre-ejaculatory fluid contain cell-free virions and seminal-infected leukocytes [47, 66–69]. During acute stages of infection, blood viral load peaks a little over two weeks after infection, while semen viral load is highest nearly a month after infection (4.5 +/− 0.4 log₁₀copies/ml) [4]. Nearly 10 weeks after HIV infection, semen levels reach a nadir and stabilize to chronic levels 4 months post-infection [4]. Most human and non-human primate studies demonstrate a concordance between plasma and semen cell-free and cell-associated viral load, noting that less than 4% of men with serum HIV <400 copies/ml on ARVs have a detectable viral load in semen [70–72]. In cases of concurrent urethritis and genital inflammation, however, low seminal HIV titers (80–2560 copies/ml) are noted in 25% of 83 men with undetectable plasma HIV [73].

The sources of HIV-infected leukocytes and free virions in seminal plasma remain unclear, though likely arise through multiple mechanisms within the male genital tract and from passive diffusion via the blood [74–76]. Phylogenetic analysis of HIV-1 sequences in blood, testis, and prostate tissues revealed distinct HIV-1 populations in testicular and epididymis-derived seminal cells versus prostate-derived seminal plasma as early as 3 months after seroconversion [77]. HIV and SIV infect testicular germ cells [78]. The defective, nonviable, immotile sperm produced by these infected germ cells have been found to contain HIV DNA [79]. Viable spermatozoa, however, typically do not become infected by HIV [67, 80, 81]. In fact, successful inseminations of processed sperm from HIV-seropositive men into their discordant partner have not resulted in any infections [82–84]. Determining the origin of HIV in semen is important for the understanding HIV sexual transmission biology and designing therapies to eradicate HIV from semen.

3.3. Cell-Free Versus Cell-Associated HIV-1 in Semen

Cell-associated viral transmission has not been extensively investigated. Most HIV transmission studies utilize cell-free virus in animal and in vitro models. However, the mechanisms of transmission of cell-associated HIV differ from cell-free virions. Intracellular virions are presumably shielded from local antiviral antibodies and antimicrobial peptides, as well as evade MHC-1 cytotoxic T cells detection of virus-infected cells in the setting of MHC-1 concordance [11, 85–87]. Cell-associated transmission may be implicated in the rare case of seminal HIV-1 transmission with undetectable plasma cell-free virus in the setting of antiretroviral therapy [88]. Furthermore, there are differences in genetic sequences of cell-free HIV versus cell-associated HIV in semen [77]. The genotype of a transmitting partners’ seminal cell-associated HIV-1 has been found to be the same as the infective virus in the recipients with acute seroconversion [89, 90]. More studies are needed to determine the mechanisms and risk factors associated with sexual transmission of cell-associated HIV. Understanding the distinction of cell-associated transmission has implications on developing neutralizing antibodies for prevention strategies targeting both cell-associated and cell-free HIV-1 [91–93].

3.4. HIV-1 Induced Proinflammatory Milieu in Semen

Even at the earliest stage of HIV infection, there are significant immune dysregulations within semen. A proinflammatory cytokine and chemokine reaction in blood and semen occurs early on, and is maintained throughout chronic infection. The cytokine changes are not reversed by therapy or suppression of HIV RNA [74, 94, 95]. Several cytokines in semen - including G-CSF, TNF-α, IFN-γ, and IL-10 – have been associated with T cell activation and virion shedding [96]. Additionally, the presence of HIV-1 causes a depletion of CD4+ and CD8+ lymphocytes in semen that is partially reversed with ARVs [97, 98]. How this proinflammatory milieu activates HIV-1 replication in infected cells is an area of needed investigation.

3.5. Co-Infection

Concomitant Sexually Transmitted Infections (STIs) potentially increase the risk of HIV-1 transmission by creating mucosal ulcerative breaches, increasing genital leukocyte concentrations, and releasing inflammatory cytokines which increase HIV replication [99]. Concurrent infection with various STI pathogens substantially increases HIV RNA quantities and shedding in semen, enhancing HIV-1 transmission [100, 101]. Both non-gonococcal and gonococcal urethritis have been associated with significant increases of HIV-1 RNA in semen, revealing up to a 10-fold increase (15.8 × 10⁴ copies/mL) with concurrent Neisseria gonorrhoeae infection [102, 103]. STI-upregulated HIV-1 RNA concentrations are significantly reduced with antimicrobial therapy [102, 103]. Although not extensively studied, cell-associated proviral DNA in semen is also increased in men with acute STIs [104]. However, the relationship of STIs to viral discordance is not straightforward. A study in chronically HIV-1 infected men who have sex with men (MSM) on ARVs found that low-level replication (plasma HIV concentration 50–500 copies/mL) and high titers of seminal cytomegalovirus (CMV) shedding (> 4 log₁₀ DNA copies/mL) are associated with seminal shedding of HIV in 10% of their subjects (N = 11/114), regardless of the presence of asymptomatic STIs [105]. A better understanding is needed on the role of concurrent STIs on genital HIV shedding as this has broad implications for strategies to reduce HIV transmission. Animal models of STI-SIV co-infection could provide a useful platform for mechanistic studies and preliminary interventional trials [106].

3.6. Antiretrovirals

Antiretroviral therapy has been shown to reduce HIV-1 RNA and DNA in blood and semen, and partially reconstitutes depleted T-lymphocytes [70, 97, 107]. However, even when undetectable in plasma, HIV-1 RNA has been detected in semen of seropositive men [108–114]. In a large series of 304 patients, 6.6% (N = 20/304) of men with undetectable viral load while on antiretrovirals had detectable HIV-1 RNA in semen, ranging from 135–2365 copies/ml. The prevalence of viral discordance remained stable over time despite the evolution of ARV regimens [113]. Cells with HIV proviral DNA can persist in semen for at least 6 months [70, 115] and have been shown to be replication competent in vitro [108]. Shorter duration of therapy was associated with cases of detectable HIV-DNA in semen (median 5.3 vs. 9.2 months below detection level, P = 0.09) [70].

4. RECTAL SECRETIONS

HIV infection from unprotected anal intercourse is associated with the highest risk of sexual HIV transmission [49, 50]. HIV target cells are abundant in the urethra and foreskin of the penis. It is believed that HIV infection occurs when these tissues come into contact with rectal fluid [116]. Circumcision is thought to reduce the risk of HIV acquisition by up to 60% for HIV-negative MSM who primarily take the insertive role during anal sex [117–119]. However, the mechanism of how foreskin removal protects against HIV and other STIs at a structural and cellular level is not well understood.

Rectal epithelial factors have been a focus of a number of investigations on anogenital transmission, but there have been limited studies characterizing HIV-1 load in rectal secretions. It is generally believed that plasma viral burden is a surrogate for HIV concentration in the genital tract and rectal secretions [120, 121]. Factors found to be associated with detection of rectal shedding of HIV DNA include anal-rectal inflammation and presence of anal HPV DNA. Whereas high plasma HIV RNA levels, anal-rectal inflammation, and presence of anal-rectal HIV DNA predicted detection of HIV RNA. Earlier studies associated high plasma viral load (>10,000 copies/mL) with detection of HIV RNA in anal-rectal samples [122, 123]. Reduction in genital HIV-1 shedding occurs in settings of ARV and undetectable plasma viral load [123, 124].

An earlier, well-cited study conducted in the US and Peru suggested that HIV-1 RNA levels in rectal mucosal secretions were higher than either blood or seminal plasma in MSM independent of ARV use. Cell-free and cell-associated HIV from rectal secretions were sampled using Sno-strips in a small cohort of 64 homosexual men of whom 42% were on stable ARV regimen for at least a month. Regardless of ARV, median HIV RNA levels were 3.55 log₁₀ copies/ml in seminal plasma and 4.96 log₁₀ copies/ml in rectal secretions while blood plasma viral loads averaged 4.24 log₁₀ copies/ml. This study modeled that a more than 1 log₁₀ decrease in blood viral load by ARV caused a 0.5 log₁₀ parallel reduction in rectal and seminal viral load [125].

In contrast, a more recent study suggested viral loads in blood and rectal secretions are highly correlated and quantitatively uninfluenced by the presence of STIs [124]. Plasma and rectal viral loads were found to be correlated. Rectal gonorrhea (GC) and chlamydia trachomatis (CT) infections did not increase the rectal viral load. Detectable virus in rectal secretions is found in 38% of 80 HIV-positive men undergoing rectal swab monitoring of STIs. Three quarters of the men were on ARV and had low plasma viral loads. Nearly all (95%) had rectal HPV, 67% had HSV-2, and 39% had positive GC or CT. A threshold plasma viral load greater than 3.15 log₁₀ copies/mL was the only significant factor associated with having detectable virus in the rectum. The presence of rectal GT or CT in the setting of ARV and low plasma viral loads was not found to significantly affect rectal HIV shedding [124]. The rarity of rectal HIV-1 shedding in the setting of undetectable plasma viral loads was also demonstrated in a series of 54 men on ARVs with plasma viral load < 50 copies/ml; only one had detectable HIV-1 RNA in a single anorectal sample [123].

Concurrent presence of STIs has typically been associated with increased genital HIV-1. A meta-analysis of 39 studies, the majority of patients not on treatment, concluded that GC or CT significantly increased the odds of detecting HIV in the genital tract [101]. Additionally, HSV-2 co-infection in persons not on antiretroviral therapy increases the concentration of HIV in the genital tract [101], with subsequent reduction of rectal and plasma HIV-1 levels with valacyclovir treatment [126]. Further studies are needed to definitively assess the degree to which STIs and ARVs affect rectal viral load, and thus HIV transmission risk.

5. CERVICO-VAGINAL FLUIDS

5.1. Cell-free HIV in Cervico-vaginal Fluids

HIV levels in Cervico-Vaginal Fluids (CVF) are usually, but not always, correlated with plasma levels [127–129]. Like other secretions, HIV in CVF increases in the setting of local infections and inflammation [130–132], and often decreases with antiretroviral therapy [128, 133, 134]. HIV-1 RNA can be detected in the endocervix, the ectocervix and the vagina. Levels of the virus at these different sites vary greatly, with more frequent detection in endocervical than vaginal secretions [127] and up to a 10-fold difference in viral loads between the endocervix versus ectocervix [135]. A little more than a third of women have undetectable genital tract viral loads in the absence of therapy [128, 136], and up to 40% of have detectable viral loads intermittently while on suppressive therapy [136]. Genital tract shedding also increases with pregnancy both in women with detectable and undetectable plasma viral loads [137]. Viral loads in female genital secretions range from undetectable to 5.94 log₁₀ [135] with median genital HIV RNAs varying from 3.87 log₁₀ to 5.1 log₁₀ in antiretroviral-naïve subjects [128, 134]. The highest levels seen in women with undetectable plasma viral loads ranged from 5.68–5.81 log₁₀ from the endocervix, ectocervix, and vagina [136]. These levels declined to a maximum of 4.83 log₁₀ in women who had undergone hysterectomies [136]. Therefore, while CVF viral loads are usually correlated with plasma viral load, there is much greater variability in the female genital tract than in the plasma.

In addition to greater inherent variability, cervico-vaginal viral loads may vary by subtype. In one study, median viral loads from cervical wick samples on untreated subjects with subtype C virus was 5.1 log₁₀ versus 4.0 log₁₀ for subjects with subtype B virus, despite a lower frequency (3% versus 17%] of detected sexually transmitted infections in subtype C subjects [128]. It is worth noting that bacterial vaginosis was not evaluated. Additionally, while viral loads are usually restricted in CVF, in this study, untreated subtype C subjects with detectable cervical viral loads at entry had levels equivalent to plasma levels, 5.1 log₁₀ cervical versus 4.9 log₁₀ plasma.

Layers of complexity are added as CVF viral loads vary with menstruation in most [127, 138–140], but not all studies [141–143]. Cervical levels of HIV RNA are lowest at the mid-cycle surge in luteinizing hormone and peak just prior to menses [127, 138]. Similarly, total levels of both S-IgA and IgG are greatest 2 days prior to ovulation [7, 144]. In CVF, IgG is the dominant isotype and is important in virus neutralization. The change in nadir versus peak HIV viral load has been quantified as 0.74 log₁₀ copies/swab in HSV-2 co-infected women [138]. Serum progesterone levels and serum plasma levels of HIV RNA are weakly associated even with limited subject numbers (n=17) [127]. The importance of hormones in HIV transmission is supported by the observation of increased mucosal transmission in women on some (depot medroxyprogesterone acetate), but not all, hormonal contraceptives [145].

Genital tract infections lead to inflammation of the genital mucosa, and increased levels of HIV RNA in local secretions [130–132]. Studies have evaluated the reduction in cervicovaginal HIV RNA with the treatment of co-infections. In subjects with cervicitis, treatment is associated with a median reduction of HIV RNA from 4.05 to 3.24 log₁₀ copies/swab for all episodes of cervicitis, reductions from 3.94 to 3.28 log₁₀ copies/swab in subjects with gonorrheal cervicitis, and 4.21 to 3.19 log₁₀ copies/swab in subjects with chlamydial cervicitis [131]. Subjects co-infected with HIV and HSV-2 on HSV-suppressive but not antiretroviral therapy had reductions in endocervical HIV RNA of 0.31 and 0.35 log₁₀ copies/swab [130, 146] and in CVL viral loads of 0.29 log₁₀ copies/ml [132]. Despite reductions in viral load and genital ulcers, HSV-2 suppressive therapy failed to reduce HIV-transmission in a large clinical trial [147].

Mucosal transmission of HIV is correlated with both plasma and genital viral load. Each log₁₀ increase in genital HIV-1 RNA is associated with a 2.2-fold increase in female-to-male transmission risk with a median endocervical viral load of 3.89 vs. 3.18 log₁₀ copies/swab in transmitters versus non-transmitters [146]. Additionally, either presence of HIV RNA [129] or higher cervicovaginal viral loads in pregnant women is associated with higher rates of perinatal HIV transmission [148]. While there is a clear correlation between cell-free genital viral load and transmission, there are documented mucosal transmissions from females to males with undetectable endocervical viral loads [146]. Furthermore, the presence of HIV-RNA and HIV-DNA in CVF is highly correlated [129]. These observations beg the question, is transmission more closely linked to cell-free or cell-associated HIV-1?

5.2. Cell-associated HIV in Cervico-vaginal Fluids

Both CD4+ T cells and macrophages in CVF can be detected in low quantities from HIV-uninfected [149, 150] and HIV-infected women [151, 152]. Most studies of cell-associated HIV have qualitatively assessed the presence of HIV-1 DNA. HIV-1 DNA is more frequently detected in subjects with a high mean genital HIV-1 RNA [129, 131], making it difficult to determine the importance of cell-free versus cell-associated HIV in transmission.

Like cell-free virus, the prevalence of cell-associated HIV is increased with inflammation and infection [153–156] and rapidly decreases with ARVs [157]. Additionally, the presence of HIV-1 DNA is increased during pregnancy [137], hormonal contraception, and vitamin A or selenium deficiency [158]. In studies where HIV DNA is quantified, median proviral load ranges from 7 to 48 copies per ug DNA (or 10⁵ cells) or 20–3000 copies per lavage [159–166]. When comparing endocervical to vaginal secretions, viral DNA is generally more frequently detected (51–62% versus 14–23% of samples)[127, 158]) and greater in quantity (median 40 versus not detected, maximum 2220 versus 340 copies/ug DNA) [163].

No studies have evaluated HIV DNA by subtype. No significant association has been demonstrated between HIV DNA in cervical secretions and menstruation in a single qualitative study [167] However, since hormonal contraceptive use is associated with detection of cell-associated HIV in genital samples [158, 168, 169], a more careful evaluation of HIV DNA shedding throughout menses is warranted. HIV DNA and cell-free HIV RNA, but not cell-associated RNA transcripts, in CVF are increased in pregnancy [137].

Cell-associated HIV detection is associated with cervicitis, candidiasis and STIs [131, 153–156, 158, 159, 168, 170–173]. Like cell-free virus, detection of cell-associated virus is decreased with treatment of co-infections [170]. In subjects with cervicitis, HIV-1 DNA detection was decreased in cervical secretions from 67% to 42% after effective treatment [131].

In perinatal HIV-1 transmission, HIV DNA levels in cervico-vaginal fluids are associated with transmission [129, 165, 166] and each log increase in HIV DNA is associated with a significantly higher risk of transmission (OR 2.28, p-value = 0.03) [166]. In a study that linked cell-free HIV RNA in CVF and HIV DNA in OPS to risk of mother to child transmission (MTCT), HIV DNA in CVF was excluded from the model since cell-free HIV RNA and HIV DNA in CVF were closely correlated [129]. Therefore, the contribution of cell-associated HIV to mucosal transmission in both sexual encounters and MTCT remains unclear.

5.3. Mother to Child Transmission of HIV-1

Mother-to-child transmission of HIV-1 occurs in utero, intrapartum and through breast milk at rates in chronically infected untreated mothers of approximately 5–10%, 10–20%, and 5–15%, respectively [174]. Intrapartum transmission may occur either through mixing of blood during labor and delivery or through exposure to maternal vaginal secretions. The relative contribution of either mechanism as well as where in the infant the initial infection is established remains unknown. Intrapartum MTCT of HIV-1 is increased in infection, inflammation and greater exposure to virus and decreased by antiretroviral therapy and by Caesarian section as reviewed elsewhere [175].

HIV RNA and DNA can be recovered from oral secretions of uninfected infants immediately postpartum. HIV-RNA can be detected in 41% (N = 9/22) of oropharyngeal aspirates obtained through oral suction from uninfected infants at birth in women receiving no therapy, zidovudine monotherapy or zidovudine/lamivudine dual therapy [176]. Three of these samples had quantifiable HIV RNA at a mean of 3000 copies/ml. Additionally, HIV RNA can be detected in 28% of combined oropharyngeal-gastric aspirates of women mostly on zidovudine prophylaxis at median levels of 126 copies/ml (range 8–1270) [177]. Interestingly, in the later study, detection rates were 28% both in the 69 women who delivered vaginally and the 32 women who delivered by Cesarean section. Viral detection is correlated with maternal plasma viral load and decreased by ARVs [176, 177]. Of three infants with suspected intrapartum transmission, only one had a positive aspirate at birth [177]. In addition to HIV RNA, proviral DNA is detected in oropharyngeal aspirates collected from the mouth of the infant immediately following delivery in 14.7% (N = 33/224) of samples [129]. These are infants that tested HIV-negative at birth suggesting the proviral DNA present was from passing through the genital tract of the mother. In a univariate analysis, HIV-1 DNA in oropharyngeal aspirates was associated with HIV-1 DNA or RNA in cervico-vaginal secretions from the mother near the time of delivery. However, in multivariate analysis, only maternal viral load remained significantly correlated with the presence of HIV DNA in oropharyngeal aspirates. In the same study, 75% (N=9/12) of the infants infected in utero were HIV-DNA positive in the oropharyngeal aspirate samples, but the origin of the HIV DNA, infant or maternal, was not determined [129]. Despite exposure to HIV RNA and DNA in the oropharynx from blood or CVFs at the time of delivery, most infants escape infection.

6. BREAST MILK

The association between breastfeeding and HIV transmission was first reported in 1985 [178], and shortly thereafter HIV was cultured from the cell-free fraction of breast milk [179]. Cases of infant infection after post-partum maternal HIV acquisition, differing rates of HIV infection among breastfed and formula-fed infants, as well as HIV infection in infants whose mothers were HIV negative but had been breastfed by an HIV infected woman provided compelling evidence of breast milk transmission. Avoidance of breastfeeding is a logical and highly effective means of eliminating HIV breast milk transmission. However, breast milk provides critical protection against infectious agents and is central to infant immune development. Thus, in settings with high morbidity and mortality, early cessation or elimination of breastfeeding has no net benefit [178, 180, 181]. Avoidance of breastfeeding does lower the infant HIV infection but this gain is offset by an increase in infant mortality and considerable morbidity [182, 183]. Fortunately, antiretroviral therapy is extremely effective in curtailing breast milk transmission and is important for maternal health. These benefits, as well as the lower cost of providing ARV compared to formula, have led to the recommendation that women in resource-limited settings exclusively breastfeed for 6 months and then continue breastfeeding with complimentary food [184].

Breast milk transmission continues in sub-Saharan Africa where over 90% of pediatric HIV infection occurs. In 2015, approximately half of new HIV infections globally occurred in women; the majority of these infections occur in young women less than 24 years of age. Rates of mother to child transmission in countries where 90% of HIV-infected pregnant women reside are 6% at 6 weeks but increase to 16% after breastfeeding. It is estimated that more than half the new pediatric infections in 2013 were due to breast milk transmission. Post-partum adherence, drug toxicity, antiretroviral resistance and barriers to engagement and retention to care during the prolonged breastfeeding period remain significant obstacles to the elimination of pediatric HIV. In addition, pregnancy and the post-partum period are associated with approximately a 2-fold increased risk of HIV acquisition [185, 186]. Since acute maternal infection doubles the risk of mother to child transmission, infants are extremely vulnerable [187]. Thus, elimination of pediatric HIV transmission will depend on reductions of breast milk transmission.

Cell-free HIV RNA, cell-associated RNA and cell-associated proviral HIV DNA have been detected in breast milk. Most studies have focused on cell-free HIV RNA and very few have reported on cell-associated HIV RNA [188]. Cell free HIV RNA has been detected in breast milk from the majority of women not receiving antiretroviral therapy. Breast milk levels range from less than 50 to over a 1 million copies per ml and mean and/or median levels are below 1,000 copies/ml [3]. Thus, as has been observed with other secretions, levels of breast milk HIV RNA are typically at least 100-fold lower than those observed in plasma. Generally, HIV RNA levels are similar in both breasts but can be discordant and intermittent shedding is not infrequently observed. This variation may be related to changes in mammillary epithelial integrity associated with inflammation and/or breastfeeding practice (see below).

The detection of HIV proviral DNA indicates that HIV infected cells are also present in breast milk. However, exact quantitation is problematic since the number and composition of cells within breast milk is extremely variable. The cellularity of milk decreases markedly throughout lactation. Colostrum contains as many as a million cells per ml but this rapidly drops to less than 1,000 cells per ml. Breast milk contains lipid laden macrophages, polymorphonuclear leucocytes, B and T lymphocytes and shed epithelial cells. Since HIV does not infect all of these cells types, molecular methods using general house-keeping genes to estimate cell number provides a denominator that includes cells that are not infectable with HIV. Nevertheless, HIV DNA quantitation of the cellular fraction of breast milk has been assessed in many studies [189]. Rates of HIV proviral DNA detection in breast milk are highly variable (20 to 80%) depending on the degree of maternal immunosuppression and the sensitivity of the assay. [190–201]. Of interest, HIV antigen secreting CD4 + T cells have been described in breast milk and these cells can produce replication competent HIV to a greater extent than observed in blood [188].

In the absence of antiretroviral therapy, levels of breast milk HIV RNA and proviral DNA are highly correlated and also correlate with systemic levels of plasma viremia. Other major predictors of breast milk viral load include CD4 count, breast inflammation (mastitis, abscess), and breastfeeding behavior (exclusive breastfeeding, weaning). Inflammatory conditions of the breast are relatively common afflicting up to a third of nursing women and do not appear to be more frequent in HIV-infected women compared to HIV-negative women. Mastitis is usually asymptomatic (subclinical) and is defined by elevations in milk electrolytes (sodium or sodium to potassium ratios) and elevation of inflammatory cytokines. Both subclinical and clinically symptomatic mastitis have been associated with higher levels of HIV RNA, but those with the more symptomatic disease have the higher levels [202]. Mastitis is usually unilateral and increases HIV RNA on the affected side by as much as 10-fold without significant changes in the contralateral side [202]. Levels return to baseline or lower after the resolution of symptoms. In contrast, levels of breast milk HIV DNA do not seem to vary with mastitis [191] and over time, but do increase significantly during weaning [203].

Intriguingly, breastfeeding behavior affects the levels of both HIV RNA and DNA. Non-exclusive breastfeeding, i.e. providing the infant with liquids or solids other than breast milk, is associated with significant increases in breast milk HIV RNA. Changes in infant suckling frequency at the time of weaning is associated with more than a 10-fold increase in both HIV RNA and DNA levels, however, RNA levels are affected to a greater extent. Although rates of mastitis are increased at the time of weaning, increases in HIV RNA and DNA are independent of breast pathology and in fact, were most marked in women with no discernible clinical pathology. These changes are believed to be related to changes in the permeability of the mammary epithelium.

6.1. HIV transmission and levels of HIV RNA and DNA

Not surprisingly, levels of both breast milk cell free HIV RNA and proviral DNA are strongly predictive of transmission. The detection and level of breast milk HIV RNA have been consistently reported as conferring a higher risk of transmission independent of maternal plasma viremia and CD4 count. In one study the transmission risk doubled with each log increase in breast milk cell-free HIV RNA. Higher cumulative HIV exposure is also associated with a greater risk of transmission [204]. As discussed above, factors that are associated with increased HIV RNA such as mastitis and weaning also increase significantly the risk of HIV transmission. The role of cell-associated RNA in transmission has not been examined.

Since cell-free HIV RNA and proviral DNA are strongly correlated, it is not surprising that these levels are also associated with transmission risk. However, the relative contribution of cell-free RNA and cell-associated DNA in transmission remains uncertain with conflicting results in epidemiologic studies. Most studies have described stronger associations with HIV DNA. As with other forms of transmission, dissecting the contribution of cell-free versus cell-associated HIV is confounded by the observation that that proviral HIV DNA is not an accurate measure of the number of cells that harbor infectious virus. Similarly, not all cell-free HIV RNA is infectious.

Antiretrovirals effectively suppress breast milk levels of cell-free HIV RNA and cell-associated RNA but do not significantly decrease HIV proviral DNA levels [1]. In clinical trials, maternal antiretrovirals are extremely effective, with transmission rates below 2% in some cohorts. This suggests that cell-free RNA levels may be the major contributor to transmission. The degree to which residual infections are due to adherence, episodic increases in HIV RNA or the persistence of a cellular reservoir is unknown. Further complicating the analysis is that infants usually receive antiretrovirals as prophylaxis for at least some period. Studies comparing transmission rates with maternal antiretrovirals versus infant prophylaxis have not found significant differences but may be underpowered.

7. SIV IN SECRETIONS

Non-human primate models of Simian Immunodeficiency Virus (SIV) infection have added to our understanding of HIV in mucosal secretions. SIV has not been well studied in the saliva of macaques, with 5 [205] and 16 [206] animals assessed in two studies. SIV RNA levels vary but are generally 2–3 logs lower than those in plasma [205, 206]. At peak levels 14 days post-infection, plasma levels and saliva levels were 10⁷ and 10⁵ copies/ml, respectively [205]. The genotypes present in saliva reflect the genotypes in plasma. The virus has not been cultured nor capable of productive infection through co-culture from saliva, suggesting similar inhibition of SIV in macaques as HIV in humans [206]. SIV-specific IgG and IgA responses increase significantly between the 6–8 week and the 28–30 week post-infection time points with a correlating increased inhibitory capacity.

Non-human primate models have proved useful in confirming findings in studies of HIV in genital secretions. Similar to humans, in acutely infected macaques, the presence of SIV in male reproductive organs is detected as early as 2 weeks post-inoculation and persists chronically [71]. Compartmentalization into the rhesus monkey male genitalia occurs after resolution of peak viremia [207]. In SIV-infected macaques, primary infection induces a robust local inflammation with an increase in leukocytes and subsequent infection of CD4+T cells and macrophages within semen [208]. Persistent viral shedding in the setting of 4 months of ARV has also been demonstrated in SIV-infected macaques. Therapy decreases SIV RNA in various male genital tissues, except SIV RNA in urethral macrophages persists and SIV DNA is unaffected [209].

Studies of HIV and SIV through non-human primate models have demonstrated the ability of cell-associated virus to transmit infection at physiologic levels. Both cell-associated and high titers of cell-free HIV-1 are able to transmit HIV after atraumatic insertion into the vaginal cavity of chimpanzees [210] with 8 of 10 chimpanzees acquiring systemic infection after 1–3 exposures to cell-associated virus. Additionally, while early studies of cryopreserved SIV-infected macaque PBMCs failed to infect following vaginal insertion [211], fresh PBMCs at low-inoculum (7–2048 infectious cells/inoculum) can transmit virus following multiple exposures in settings of induced vaginal ulcers [212] and intact vaginal mucosa [213]. Systemic spread of the virus is faster than with transvaginal cell-free SIV infection [214]. Furthermore, SIV-infected spleen cells at 6.7 × 10⁵ viral DNA copies per inoculum, inserted atraumatically into Depo-Provera-treated macaques, can also transmit SIV efficiently with evidence of labeled infected cells in the vaginal lamina propria and draining lymph nodes 21 hours after exposure [215]. In fact, the doses of cell-associated SIV required to infect macaques are close to the levels detected in semen, whereas the levels required for cell-free infection often greatly exceed physiologic levels.

There have been a few studies using the SIV model to understand breast milk transmission. Most of these have characterized the viral dynamics and properties of breast milk virus. These studies have confirmed that cell free virus can result in infant infection and similar to human studies have demonstrated that the viral variants in milk are not distinct from those in the circulation but there can be small foci of reactivation. One of the most significant recent studies compared infant infection in rhesus macaques and sooty mangabeys. Rhesus macaques have rates of infant transmission similar to HIV in humans, while transmission among sooty mangabeys is rare. Viral load levels [SIV RNA, cell-associated DNA), milk immune factors and ex vivo inhibition of SIV did not differ between these monkeys. However, the levels of CD4+CCR5+ T cells were much lower in the sooty mangabey in both blood and tissues. These studies underscore the role of host factors in HIV pathogenesis.

There are presently no studies of rectal secretions in SIV models. One study of transmission found that systemic infection with cell-free SIV through the oral route can be accomplished at levels of virus 6000 times lower than that required through rectal exposure [216]. This raises interesting questions in subjects with multiple exposures including receptive oral and receptive anal intercourse: through which mucosa is infection most likely?

8. UNANSWERED QUESTIONS

Despite thirty years of HIV research, many important questions about HIV in external secretions remain unanswered and deserve further study. Questions that remain relevant to all secretions include the mechanisms by which mucosal surfaces restrict HIV entry, identification of innate and adaptive immune responses, delineating the role of anti-HIV IgA and IgG in secretions and characterization of metabolites and proteins that restrict viral penetration. Curating factors that alter infectivity could provide novel approaches to HIV prevention.

The relative role of cell-free and cell-associated virus in transmission remains unresolved. The spectacular success of antiretrovirals in preventing HIV infection suggests that cell-free viral infection predominates. However, as efforts to eliminate HIV move forward, modalities that target cell-associated virus may be necessary. More studies are needed to determine the mechanisms and risk factors associated with sexual transmission of cell-associated HIV. Understanding the distinction of cell-associated transmission has implications on developing neutralizing antibodies for prevention strategies targeting both cell-associated and cell-free HIV-1 [91–93].

Studies to assess the direct influence of semen and semen components on HIV-1 transmission across the anogenital mucosa are imperative for microbicide prevention strategies. Determining the origin of HIV in semen is important for the understanding of HIV sexual transmission biology and designing therapies to eradicate HIV from semen. Similarly, understanding the origin of HIV in endocervical, ectocervical and vaginal secretions as well as hormonal influence is critical to curbing heterosexual transmission. It is still unknown why cervico-vaginal shedding is greater in individuals with subtype C virus than subtype B virus, and whether this is related to host mucosal factors, microbiome differences in the population, presence of Gardnerella, specific viral or other factors. How a proinflammatory milieu in genital secretions activates HIV-1 replication in infected cells is an area of needed investigation, as is a better grasp on the role of concurrent STIs on genital HIV shedding. These knowledge gaps have broad implications for strategies to reduce HIV transmission.

The mucosal surface of the intestinal tissue consists of a protective mucus layer with immune factors and a unique microbial niche. IgA present in rectal fluid has been demonstrated in vitro to modulate the risk of HIV infection [217–223], however, the role of mucosal anti-HIV IgA in humans needs clarification. Alterations of rectal microbiota and metabolic pathways occur in untreated HIV-infected individuals [224], and may serve as potential therapeutic targets. Proteins in rectal lavage fluid from HIV-seronegative individuals have been identified with anti-HIV activity that limited HIV infection in TZM-bl reporter cells by nearly 50% in vitro [225]. These intestinal mucus factors play an important role in maintaining epithelial integrity, and make rectal fluid a first line of defense against HIV. A better understanding of how the intestinal mucus, its microbiota, and innate immune factors are disrupted in the setting of the HIV-1 exposure is important to possibly prevent or reduce the propagation of HIV to mucosal surfaces and to restore the epithelial integrity of the rectal mucosal barrier.

Our understanding of the role of mucosal microbial communities in maintaining epithelial integrity and modulating the inflammatory responses that fuel HIV infection is in its infancy. This knowledge would help identify those at highest risk as well as new modalities by which to curtail the epidemic. These studies could also inform efforts to reverse the damage done to the gastrointestinal tract. Chronic inflammation remains a major cause of morbidity and mortality even in virologically suppressed HIV-infected persons. Microbial translocation, ongoing covert replication, and irreversible loss of key immune cells in the gastrointestinal tract have all been associated with chronic immune activation.

As efforts to cure HIV move forward, it is critical to define the extent to which mucosal viral populations contribute to the HIV reservoirs that prevent HIV eradication. Mucosal tissues are major targets of HIV infection and antiretroviral penetration can be variable. The latent virus has been described in the lactating breast [226] and the extent to which the gastrointestinal tract is a reservoir is controversial. The testis is an immune privileged site, which serves as a reservoir for leukemia. Whether this is also a sanctuary site for HIV is not established. Defining the viral cellular reservoirs in mucosal compartments is desperately needed if a cure is to be achieved. Thus, although much is known about HIV in external secretions, even more remains unknown, and a better understanding is critical to modulate chronic disease, prevent new transmissions, and hopefully eradicate existing infections.

LIST OF ABBREVIATIONS

ARV

Antiretroviral or Antiretroviral Therapy

CMV

Cytomegalovirus

CT

Chlamydia trachomatis

CVF

Cervicovaginal Fluid

DC-SIGN

Dendritic Cell-Specific Intercellular Adhesion Molecule-3-Grabbing Non-Integrin

DNA

Deoxyribonucleic Acid

GC

Neisseria gonorrhoeae

G-CSF

Granulocyte Colony Stimulating Factor

HIV

Human Immunodeficiency Virus

HSV-2

Herpes Simplex Virus 2

IFN-γ

Interferon- γ

IgA

Immunoglobulin A

IgG

Immunoglobulin G

IL-10

Interleukin 10

MHC

Major Histocompatibility Complex

MSM

Men Who Have Sex With Men

MTCT

Mother-To-Child Transmission

PBMC

Peripheral Blood Mononuclear Cells

RNA

Ribonucleic Acid

S-IgA

Secretory IgA

SIV

Simian Immunodeficiency Virus

STI

Sexually-Transmitted Infection

TNF-α

Tumor Necrosis Factor α