It seems that X4/R5 gatekeeping (that is prevention of X4 HIV-1 from infecting and/or disseminating in the human body at the early stages of HIV-1 infection) belongs to a rare class of almost perfect biological phenomena. Among multiple reported HIV-1 transmission events via sexual acts (between males and females or between males) or through intravenous injection, at the early stages of infection R5 HIV-1 was ubiquitously found. Such precision suggests the existence of a near perfect barrier that selects against X4 HIV-1 transmission. Where are these gatekeepers and what are their mechanisms? There is no definitive answer, however we discuss below potential mechanisms and points at which selection may occur according to the different routes of transmission .
Intra-vaginal transmission
World-wide the majority of HIV-1 transmission occurs through heterosexual intercourse. Women have increasingly bourn the brunt of HIV infection mainly through vaginal intercourse and this route of infection has been the one most widely studied in various experimental models. In vaginal infection genital mucosa serves as the first port of entry for HIV-1 and the mucosal barrier is probably one of the gatekeepers not only for X4 HIV-1 but for other HIV-1 variants as well [
91].
In vaginal intercourse HIV-1 is ejaculated with semen and transverse mucus that covers the mucosa of the lower female genital track. To establish infection HIV-1 needs to access and infect target cells (lymphocytes, macrophages, possibly dendritic cells (DC), and Langerhans cells (LCs) in particular) in the local mucosa, and be transmitted to the draining lymph nodes where it undergoes rapid replication before being disseminated throughout the entire body. It is believed that a major site for HIV-1 transition in the female genital tract is the cervix, especially the endocervix and the transitional zone, which are covered by a single-layered columnar epithelium. Such a layer is less protective against HIV-1 than the stratified epithelia of the vagina [
92,
93], reviewed in [
94]. Also, the endocervix, together with the transition zone, contains a high number of potential cellular targets for HIV-1 [
95]. However, HIV-1 genital transmission to women with a congenital absence of a cervix has been reported [
96]. SIV has also been transmitted intravaginally to rhesus macaques after hysterectomy [
95,
97,
98].
The fluid of the lower female genital tract which covers the genital epithelia provides the first potential barrier for the virus on its way to dissemination. Female genital fluids are different in different parts of the genital tract: the vagina is covered by an exudate, which enters through the stratified epithelia that cover this organ. The barrier function of this fluid against HIV-1 has not been thoroughly studied. It is believed that vaginal stratified epithelia provides a significant mechanical barrier to many viruses. Furthermore, the exudate is increased with sexual arousal and therefore its composition may significantly change during sexual intercourse. Nevertheless, virus can penetrate the superficial layers of the stratified epithelium and this may be sufficient to reach superficial Langerhans cells and CD4 T cells, and would be enhanced by any micro or macro lesions in this epithelium.
Higher in the genital tract, the epithelia are covered by true mucus produced by cervical secretory cells. Mucin is the main mucus component and in the endocervix it is mainly a product of two genes: MUC4 and MUC5B [
99]. Mucus can protect underlying epithelia by two mechanisms: decreasing HIV-1 infectivity via various soluble factors present in it and/or by temporarily trapping virions in the protein mesh, thus slowing their movement by several orders of magnitude compared with water [
100‐
103]. Since HIV-1 is a fragile virus and cannot remain at normal temperature outside of cells for a long time, its infectivity may be significantly decreased if mucus slows viral penetration. Moreover, during vaginal HIV-1 transmission the acidity of mucus is decreased because of the mixture with alkaline semen. This mixture is less viscous than pure mucus, and the diffusion rate of virions in it is only 15 times slower than in water [
101]. Nevertheless, this slowing of HIV-1 penetration may be sufficient for significantly reducing HIV-1 infectivity. Since semen contains not only free virions but also lymphocytes that carry HIV-1 [
98], trapping these lymphocytes in mucus may also reduce HIV-1 transmission by these carriers.
Chemical defense against HIV-1 is mediated by mucus soluble factors, in particular by chemokines produced by epithelial cells. In some experimental models epithelia was shown to constitutively produce a CXCR4-binding chemokine, SDF-1, thus selectively reducing X4-HIV-1 transmission [
104,
105]. Also, cervical mucus contains beta defensins that may inactivate HIV-1 on its way to the epithelia (for review see [
106]). Defensins are secreted by epithelial cells under the hormonal control of oestradiol and progesterone [
107,
108]. Some of these defensins are more restrictive against X4 [
58,
109‐
112]. However, others do not significantly differentiate between X4 and R5 HIV-1 [
109,
110]. Microbicidial enzymes, surfactant proteins and complement present in cervical mucus (see[
94]), as well as other as-yet unknown soluble factors observed in proteomics experiments [
113], may contribute to the gatekeeping effect. As a result, mucus and mucins suppress HIV-1 in inhibitory assays [
114]. The two protective mechanisms of mucus may work synergistically, as even temporary trapping of HIV-1 and HIV-1-infected cells provide a longer exposure to the soluble factors present in mucus as well as to anti HIV-1 antibodies that may have been generated in one of the partners.
Additional gatekeeping effects of mucus may be affected by the difference in the surface charges between R5 and X4 HIV-1. The V3 loop in gp120 of X4 HIV-1 has more exposed cationic charge than R5 HIV-1[
115]. In principle, this may result in a stronger binding of X4 HIV-1 to the polyanionic mucin and a preferential clearance of these viruses, or at least impairment of their infectivity. However, dilution of cervical mucus by semen may make this effect negligible. The higher cationic charges of X4 HIV-1 gp120 may also make these viruses more prone to bind to heparin sulphate proteoglycans that cover mucosal surfaces and thus may work as a sink for these viruses [
116,
117]. However, these theoretical considerations regarding the difference between X4 and R5 HIV-1 have to be tested experimentally. Also, it has been shown that agrin, which plays an important role in establishing viral synapses through which HIV-1 can pass from one cell to another, binds preferentially to R5 HIV-1[
118].
Another level of complexity in vaginal HIV-1 transmission is that both mucus and cervical tissue characteristics are not constant but rather undergo changes during the menstrual cycle. A window of infectivity at 7 to 10 days post-ovulation, when the defense mechanisms are at a low level, has been identified [
119].
Viral particles that go through the cervical mucus reach the epithelial layer. The epithelial layer itself seems to be an efficient mechanical barrier against HIV-1 and other pathogens [
95]. Also, epithelial cells of the lower genital tract are not infected by HIV-1
in vivo. Although these cells can express coreceptor molecules, they do not in general express the HIV-1 receptor CD4, The possible exception to this rule is the expression of CD4 by uterine epithelial cells at the proliferative phase of the menstrual cycle [
120]. However, under laboratory conditions infection of epithelial cells has been reported [
121], and a role in HIV-1 transmission has been ascribed to them [
122]. Nevertheless, it is believed that for efficient infection it is necessary for HIV-1 to bypass the epithelial layer. This most likely happens through lesions that commonly occur as a result of various infections and probably environmental factors. Also, microlesions may be generated as a result of coitus [
123]. Experiments with cervical explants and fluorescence-labeled HIV-1 showed that HIV-1 penetrates genital mucosa similarly to inert particles, that is, via the gaps between cells (T. Hope, personal communication). When virus encounters and infects its first natural cellular targets, predominantly lymphocytes beneath or within the epithelial layer, it may be efficiently disseminated in cell-associated form from cell to cell through viral synapses which seems to be more efficient than cell-free virus transmission [
118,
124,
125]. This early stage of HIV-1 transmission when few founding cells are infected is critical for the further dissemination of HIV-1 through the body [
94]. Surprisingly, although R5 HIV-1 readily infects macrophages
in vitro, the first (founding) infected cells seem to be CD4 lymphocytes [
126].
To characterize HIV-1 targets in cervical tissue more thoroughly, it is necessary to apply multi-color flow cytometry. Recently a protocol of cervical tissue dissociation into single cells that retain their antigenic characteristics has been developed [
127], thus enabling a thorough analysis of cervical mucosal lymphocytes using flow cytometry [
128]. It was found that cervical tissue is particularly rich in CCR5-expressing CD4 lymphocytes, which make this tissue potentially more susceptible to R5 HIV-1 infection than lymphoid tissue [
128]. Accordingly, cervico-vaginal explants express a strong gatekeeping function: on average, dramatically more R5 HIV-1 is released from the
ex vivo-inoculated cervico-vaginal tissues than X4. Moreover, in a relatively small fraction of tissues that were capable of replicating X4 HIV-1 a correlate was found: a higher presence of early differentiated CD4 lymphocytes [
128]. The relation of this fraction to the gatekeeping function of cervical tissue remains to be understood.
Earlier, it was reported that CD4+/CCR5+ cells are consistently detected within the stromal papillae that penetrate the epithelial layer; this location probably makes them easy targets for HIV-1 [
129]. Also, it was reported that the levels of CCR5 mRNA in the cervix were up to 10-fold higher than those of CXCR4 mRNA [
130]. However, opposite results have been also reported [
131]. The level of mRNA for surface antigens is not necessarily translated into the level of expressed proteins. It was shown that cervical lymphocytes express not only CCR5 but CXCR4 as well [
128].
Beneath and within the epithelial layers are situated DCs and LCs repectively, which may protrude their dendrites through the layer to the lumen. Normally, these cells capture, and process antigens and deliver them to the draining lymph nodes where they present them to T cells. In HIV-1 infection these cells may bind and deliver HIV-1 from the lumen to the draining lymph node. Although it was firmly established that cervico-vaginal LCs are able to transfer HIV-1 in vitro, it is not clear whether these cells can be productively infected
in vivo[
132]. Nevertheless, they may represent one of the significant elements of gatekeeping as these cells express among other HIV-1-binding molecules, CD4 and CCR5 but not CXCR4 [
133,
134]. Also at various sites these cells may have different levels of expression of various HIV-1 coreceptors, providing gatekeeping for particular viruses. Cells that, upon binding antigens or upon infection with a virus, are capable of moving out of the tissue are collectively called migratory cells [
135]. Such cells consist of a heterogeneous population with very different features.
The entire list of such cells is not yet known. Their infection with HIV-1 in cervico-vaginal tissue was demonstrated using explants (see [
136]). Macrophages do express CCR5 [
137] and are particularly susceptible to R5 HIV-1 infection
in vitro and
ex vivo. Thus, they could also play a significant role as gatekeepers against X4 HIV-1, although in tonsillar explants macrophages are infected by X4 viruses [
138]. However, studies of early events of infection in non-human primates with SIV did not reveal infected macrophages [
126]. Furthermore, TF virus appears to infect macrophages poorly. Whether high expression of CCR5 and early macrophage infection with HIV-1 reflects the situation in humans
in vivo or represent an “artifact” from using isolates with high macrophage tropism in the explants system, remains to be clarified.
Thus, although cervical mucosa performs gatekeeping functions, it is obvious that this performance is not perfect and that X4 HIV-1 can penetrate this barrier, although much less efficiently than R5 HIV-1.
Penile transmission
Globally 80% of men acquired HIV-1 infection from vaginal intercourse, i.e. through the penis (see [
136,
139]). This is also true when infection is acquired through insertive anal intercourse. However, mechanisms of HIV-1 acquisition through the penis are even less understood than for vaginal transmission in women. Nevertheless, it is clear that gatekeeping mechanisms operate for this route of infection, since R5 HIV-1 also starts infection and dominates its early stages in men.
During intercourse the penis comes into close contact with the vagina and is bathed in vaginal fluid. Thus, it is reasonable to assume that all the above-mentioned protective barriers that are associated with female genital tract fluids and that protect cervical epithelia from viral infection, including barriers that are selective for R5 HIV-1, protect penile epithelia as well. As with the vaginal route of infection, genital fluid may be the first “gatekeeping” mechanism in penile infection. In the course of vaginal intercourse, vaginal fluids are mixed with (pre-) ejaculate. The latter increases mucosal pH and also contains various soluble factors that may dramatically change the characteristics of the mucus. Recently, it was reported [
124] that such a mixture, unlike each fluid separately, has an inhibitory effect on HIV-1 infectivity. Whether this effect is different for R5 and X4 HIV-1 has not yet been determined.
After penetrating the mucus or the mixture of mucus and semen, HIV-1 reaches the penile epithelium. What part of the penis is the most vulnerable to HIV-1 transmission? Clinical trials performed over the last several years have shown that circumcision greatly reduces the probability of penile HIV-1 transmission. Thus, the foreskin (or other parts of penis that may be indirectly affected by foreskin removal) seems to play a significant role in transmission of HIV-1. The outer foreskin is heavily keratinized and therefore is considered well protected against pathogen penetration. It was shown [
124] in
ex vivo experiments with foreskin explants that it is indeed the case: HIV-1 does not penetrate the outer foreskin well. In contrast, the inner foreskin and frenulum are covered with a much thinner layer of keratin. Accordingly, in
ex vivo experiments the inner foreskin was shown to be vulnerable to HIV-1 penetration [
124]. However, it was recently shown that keratinization of the outer and inner foreskin is not statistically different, and other mechanisms for the differential permeability of these two parts of the foreskin should be considered [
140].
During erection the foreskin is stretched out, revealing its inner aspect, which becomes accessible both to cell-free HIV-1 in the vagina as well as to HIV-1-infected cells that may be situated on the surface of the female genital tract. Another factor that may make the foreskin an important portal of HIV-1 entry is the abundance of HIV-1 cellular targets. The entire foreskin is rich in various cells that constitute potential HIV-1 targets including: CD4 T lymphocytes, macrophages, and LCs [
133,
139,
141,
142]. In
ex vivo experiments and in autopsied tissues [
139] infection was revealed in CD4 T lymphocytes and in LCs. However, in the outer foreskin these cells are thought to reside beneath the highly keratinized epithelia and are less accessible for HIV-1. In contrast, in the inner foreskin where the keratin layer is thin, LCs are probably the first target cells that HIV-1 encounters. Here LCs are more abundant and nearer the surface and thus cells are more likely to protrude their dendrites through the epithelial layer towards the outer surface.
Foreskin LCs seem to play a significant role in gatekeeping. In experiments with foreskin explants these cells selectively transfer R5 but not X4 HIV-1 to indicator cells. Another gatekeeping process may stem from the fact that the average density of CCR5-expressing cells in the inner foreskin is 10-fold higher than that of CXCR4-expressing cells [
139]. However, the extent of individual variations in this parameter remains to be confirmed in a large group of subjects.
Although keratinized epithelium is highly protective, various abrasions, lesions due to STDs, as well as microtrauma, provide access for HIV-1 to the target cells that reside beneath the surface of the organ. Lesions or microtrauma would also render the outer foreskin and/or shaft of the penis vulnerable to HIV-1 infection, providing access to an abundance of cells expressing CD4, as well as CCR5 and CXCR4. However, even when HIV-1 gets direct access to sub-epithelial layers, R5 HIV-1 seems to find more targets than X4 HIV-1 since on average cells expressing CCR5 are situated in the outer foreskin, glans, and frenulum closer to the surface than cells expressing CXCR4. Also the average density of CCR5-expressing cells in the outer foreskin is higher than that of CXCR4-expressing cells [
139]. All these features of the foreskin clearly indicate that whether HIV-1 enters through the inner foreskin or through defects in the keratin layer into the outer foreskin, the barriers against infection are selective and protect the foreskin against X4 HIV-1 infection more efficiently than R5 HIV-1.
Although the foreskin seems to be an important site of HIV-1 entry, circumcised men also acquire HIV-1 through the penile route. Thus, other sites of entry besides the foreskin exist. The glans penis in both circumcised and uncircumcised men is covered by highly keratinized squamous epithelia and seems to be relatively protected against HIV-1 entry in the absence of lesions or microtrauma. In contrast, the penile urethra is less protected as it is covered by a non-keratinized columnar epithelium that is narrowly stratified at the meatus and is also populated with CD4 T lymphocytes and macrophages [
93]. Both CXCR4 and CCR5 mRNA have been isolated from urethra swabs in equal amounts [
143]. However, it is difficult to translate these data into the relative abundance of CCR5- and CXCR4-expressing cells on the basis of the mRNA measurements.
In summary, on its way to dissemination within the body via the penile route of infection, HIV-1 has to overcome many protective barriers. Some of these barriers can discriminate between R5 and X4 HIV-1 and are higher for the latter.
Gastro-intestinal mucosal transmission
The probability of infection through receptive anal intercourse is much higher than through vaginal or penile intercourse. These data were confirmed in experiments with macaques infected with the same amount of SIV under controlled laboratory conditions [
144]. Thus, it seems that fewer protective barriers exist for this route of HIV-1 transmission. Nevertheless, the gastro-intestinal route of transmission also exhibits a gatekeeping mechanism against X4 HIV-1.
The vulnerability of the colorectum to HIV-1 infection stems from two major factors. First, a single layer of columnar epithelium separates the lumen from the inner layers. This layer is fragile and maybe damaged during intercourse. Also, epithelial cells, although not infectible by HIV-1, may be damaged directly by the virus violating the layer’s integrity [
145]. Second, lymphocytes in colorectal tissue are constitutively activated, providing HIV-1 cell targets that efficiently replicate virus [
90,
146,
147], facilitating its dissemination. This is probably one of the reasons why this tissue tissue is one of the first that is damaged by HIV-1 infection, irrespective of the transmission route [
86,
148]. Although fragile and not providing sufficient mechanical protection, colorectal epithelium provides some biological protection. Among other soluble factors it secretes chemokines, in particular stromal-derived factor 1 (SDF-1) [
104], the natural ligand for CXCR4 that
in vitro selectively suppresses X4 HIV-1 but not R5 HIV-1 infection [
62].
Even with intact epithelia, HIV-1 may be transmitted through the epithelial layer by transcytosis [
105,
149] or transferred by DCs [
150]. Also, more colorectal CD4 T cells express CCR5 than do tonsillar CD4 T cells [
147]. However, colorectal explants can support replication of both R5 and X4 HIV-1, although X4 less efficient, thus the barrier to X4 infection is unlikely to be absolute [
90]. Nevertheless, colorectal tissue
per se seems to be more vulnerable to R5 HIV-1 infection than secondary lymphoid tissue but can be efficiently infected by X4 HIV-1 as well [
90].
HIV-1 transmission through oral sex and also mother-to-child transmission are most likely to occur through the gingeva or tonsils. The latter infection is mediated by swallowing HIV-containing fluids in the birth canal or with breast milk. There is a gatekeeping mechanism operating at the upper gastrointestinal site, and again R5 has an advantage over X4 HIV-1 [
151‐
155]. One of the important defense mechanisms in oral sex is the anti-HIV-1 activity of the saliva [
156]. Like vaginal mucus, saliva can trap HIV-1 [
157]. However, this mechanism does not seem to be selective for R5 and X4 HIV-1. Also, saliva decreases HIV-1 infectivity because of the presence of various soluble factors [
158], including proteases [
159] and defensins (see [
160]). As discussed above, some of the latter suppress X4 more efficiently than R5 HIV-1 providing a basis for another X4 HIV-1 gatekeeping barrier. Vertical transmission mediated by swallowing breast milk may well occur through infection via the tonsils or other lymphoid tissue associated with the Waldeyer’s ring. Indeed exposure of the tonsils to SIV in neonatal macaques is sufficient to establish infection [
161].
Another gatekeeping mechanism may be associated with transcytosis and related to the fact that epithelial cells of the small intestine preferentially express CCR5 rather than CXCR4 [
147]. Exposure of the small intestine may occur if any virus or infected cells can survive acidification of the stomach, most likely in the first hours to days following birth. It is not clear to what extent this potential pathway for HIV-1 penetration of the epithelial barrier can discriminate between R5 and X4 HIV-1 [
105,
149]. In
in vitro experiments it was reported that primary intestinal (jejunal) epithelial cells were able to transfer R5 but not X4 HIV-1 through trancytosis to indicator cells [
162].
A few cases of early detection of X4 HIV-1 in vertical transmission have been reported [
152,
155,
163]. However, in these cases it was not clear whether X4 HIV-1 was actually transmitted or evolved from the earlier-transmitted R5 HIV-1. Establishment of a phylogenic relationship between mother’s and child’s viruses is required to distinguish between the two above-mentioned possibilities. In the absence of such work, it was widely believed that the gatekeeping mechanisms of vertical transmission are as tight as those of horizontal transmission. When the phylogenic analysis of the mother-to child transmitted variants has been performed it was found that X4 variants always evolves from the transmitted R5 HIV-1 [
164].
However, a recent study has been published that demonstrated the transmission of X4 HIV-1 (and dual-tropic R5X4 variants) from five Ugandian mothers to their babies [
165]. As was shown for cases when X4 evolved as a result of R5 HIV-1 mutation, babies with X4 HIV-1 dominance quickly progress to AIDS. In the case referred to above, these babies died earlier than those to whom R5 HIV-1 was transferred [
165].
In conclusion multiple mechanisms for preferential transmission of R5 HIV-1 through the gastrointestinal route have been reported. Although the published data are somewhat controversial, these mechanisms may include: preferential secretion of chemokines that bind to CXCR4 rather than CCR5, the higher level of CCR5 expression, and potentially, preferential transcytosis. However, none of these mechanisms alone seems to explain the high efficiency of the gastrointestinal “gatekeeper” in protection of X4 HIV-1 transmission
Post-mucosal gatekeepers
Following on from the discussion above, although several somewhat efficient gatekeepers exist at the mucosal portals of HIV-1 entry, they are not perfect. Everything we know from studying these gatekeepers indicates that in spite of them, X4 HIV-1 should be capable of entering the human body, although less efficiently than R5 HIV-1. However,
in vivo X4 infection rarely occurs, except in a few recently reported cases of vertical transmission [
165]. Thus, it appears that there are additional post-mucosal gatekeeping mechanisms. This suggestion is supported by the fact that if HIV-1 bypasses mucosal barriers and is delivered directly (intravenously, with a non-sterile syringe needle or with contaminated blood in a blood transfusion) it is again R5 HIV-1 that is transmitted and that dominates early stages of HIV disease.
Since, as described above, R5 HIV-1 was initially thought to be “macrophage-tropic”, macrophages were first considered to be the infected gatekeepers which select for R5 HIV-1 in cases of intravenous HIV-1 transmission when the mucosal barriers are circumvented [
166,
167]. Although macrophages are thought to be an important HIV-1 reservoir, studies of the early stages of HIV-1 infection indicate that lymphocytes (which do not discriminate between X4 and R5), rather than macrophages, are the first HIV-1 targets [
94,
95,
97,
126]. However, macrophages do become infected and these cells along with other antigen-presenting cells (APCs) are less susceptible to cytotoxic T lymphocytes than are infected T cells. This has been clearly demonstrated in animal models [
168,
169]. As a result of this and other factors, infected macrophages survive longer than lymphocytes, disseminating R5 HIV-1 with which they are predominantly infected. Although the tissue explant model seems to be closer to the situation
in vivo than isolated cell cultures, it has its own limitations [
127]. In laboratories, explant models are infected by HIV-1 suspensions, while
in vivo it seems that virus is also disseminated from cell to cell through the viral synapses. Through these synapses R5 HIV-1 is selectively transmitted from DCs to resting CD4+ T cells [
170].
Finally, it seems that
in vivo some systemic factors exist that are more restrictive for X4 than for R5 HIV-1. The first candidate for such a factor is the immune response, including the innate one. Both X4 and R5 HIV-1 induce cytokines, including RANTES and SDF-1 [
171], that bind to the respective HIV-1 coreceptors and may prevent infection by corresponding X4 or R5 HIV-1. In lymphoid tissue explants X4 triggers secretion of RANTES in concentrations sufficient to suppress R5 HIV-1, however R5 infection does not induce sufficient SDF-1 to suppress X4 HIV-1 infection [
171]. Thus, rather than explaining the gatekeeping mechanism, these experiments indicate one of the potential factors responsible for the “switch” of dominance from R5 to X4 HIV-1 at the later stages of the disease.
However, in experiments with rhesus macaques that have been inoculated with both R5 and X4 viruses (SHIV), R5 outcompeted X4 SHIV [
169]. Also, reports have shown that one of the conserved gp120-neutralization epitopes [
172] is cryptic in R5 but is accessible in X4 HIV-1. These and other immune mechanisms may selectively suppress X4 HIV-1 preventing rapid evolution of X4 from R5 HIV-1. Indeed, R5 and X4 HIV-1 variants have been described that differ from each other by only a few amino acids [
173] suggesting they should easily evolve in the absence of any X4 gatekeeping. The
in vivo mechanisms against X4 HIV-1 are so pervasive that in 50% of individuals infected with clade B HIV-1, X4 never evolves and viruses retain their R5 phenotype despite progression to AIDS [
174]. Moreover, although CXCR4-utilizing HIV-1-a variants have been reported for other clades, there is a relative lack of such variants among non-B subtypes (especially of C and D clades) despite aggressive progression of HIV disease [
175].