|Home | About | Journals | Submit | Contact Us | Français|
While human immunodeficiency virus (HIV) transmission through the adult oral route is rare, mother-to-child transmission (MTCT) through the neonatal/infant oral and/or gastrointestinal route is common. To study the mechanisms of cell-free and cell-associated HIV transmission across adult oral and neonatal/infant oral/intestinal epithelia, we established ex vivo organ tissue model systems of adult and fetal origin. Given the similarity of neonatal and fetal oral epithelia with respect to epithelial stratification and density of HIV-susceptible immune cells, we used fetal oral the epithelium as a model for neonatal/infant oral epithelium. We found that cell-free HIV traversed fetal oral and intestinal epithelia and infected HIV-susceptible CD4+ T lymphocytes, Langerhans/dendritic cells, and macrophages. To study the penetration of cell-associated virus into fetal oral and intestinal epithelia, HIV-infected macrophages and lymphocytes were added to the surfaces of fetal oral and intestinal epithelia. HIV-infected macrophages, but not lymphocytes, transmigrated across fetal oral epithelia. HIV-infected macrophages and, to a lesser extent, lymphocytes transmigrated across fetal intestinal epithelia. In contrast to the fetal oral/intestinal epithelia, cell-free HIV transmigration through adult oral epithelia was inefficient and virions did not infect intraepithelial and subepithelial HIV-susceptible cells. In addition, HIV-infected macrophages and lymphocytes did not transmigrate through intact adult oral epithelia. Transmigration of cell-free and cell-associated HIV across the fetal oral/intestinal mucosal epithelium may serve as an initial mechanism for HIV MTCT.
Epidemiologic data indicate that the risk of genital HIV transmission in adults is substantially higher than the risk of oral transmission (38, 40, 46, 52). However, HIV mother-to-child transmission (MTCT) via the neonatal oral and/or gastrointestinal route is not uncommon and was even less so in the pre-antiretroviral-treatment era (13, 29, 33).
HIV MTCT in the fetus/neonate may occur in utero or during labor from exposure to HIV-containing amniotic and cervicovaginal fluids (24, 27, 32, 35, 42). Furthermore, HIV MTCT may result from breastfeeding milk containing HIV (4, 5, 37, 47, 49, 59). While the rate of HIV MTCT has been reduced to less than 2% with antiretroviral therapy (ART) in developed countries, HIV MTCT in developing African and Asian countries may be as high as 25% to 30% (13, 29). Analysis of HIV transmission in mother/child pairs has shown that the majority of HIV-1 strains transmitted from mother to child are R5 tropic (7–9).
Using a single-layer, polarized epithelial cell model, we recently showed that HIV can traverse both adult and fetal oral epithelia; however, virions that transmigrated through adult epithelial cells were rendered noninfectious, whereas those that passed through fetal epithelial cells remained highly infectious (54). We further found that HIV inactivation by adult oral epithelial cells was mediated by high-level expression of the anti-HIV innate proteins beta-defensin 2 (HBD2) and HBD3. Thus, high-level antiviral innate protein expression may contribute to epithelial resistance to HIV transmission across the adult oral epithelium, in contrast to the fetal oral epithelium, which lacks expression of these innate proteins and allows transcellular passage of infectious virions.
In the current study, we further investigated the mechanisms of HIV transmigration through mucosal epithelia by using ex vivo oral tissue explants. We show that the more highly stratified adult oral epithelium limits viral penetration more efficiently than does the less-stratified fetal oral epithelium. The greater efficiency of HIV transmission across fetal versus adult oral epithelia may reflect a reduced barrier function of fetal epithelia associated with paucistratification. We also show that R5-tropic-HIV-infected macrophages can penetrate into fetal mucosal epithelia from the apical surface, suggesting that this may be one of the predominant mechanisms of transmission of R5-tropic HIV from mother to child (7–9).
One or two fresh biopsy specimens of nonkeratinized buccal mucosae were obtained using 6-mm-diameter biopsy punches from healthy, HIV-seronegative volunteers (age range, 30 to 41 years) who had no inflammation in the oral cavity. Each biopsy specimen was cut into two or more pieces and used for propagation of tissue explants. Fetal buccal, oropharyngeal, and small intestinal (jejunal region) tissue explants containing the mucosal epithelium and lamina propria were obtained from fetuses 18 to 24 weeks old that had been subjected to elective termination for nonmedical reasons from HIV-uninfected women. The tissues were placed in a tube with 2 ml of RPMI medium containing 10% heat-inactivated fetal bovine serum, 20 mM HEPES, 100 mM glutamine, 20 μg/ml gentamicin, 200 U/ml penicillin, and 200 μg/ml streptomycin.
To establish polarized organ cultures, adult oral biopsy specimens were used approximately 30 min after biopsy procedures. Fetal oral and intestinal biopsy specimens were used approximately 2 to 3 h after abortion procedures. Explants were placed with the mucosal side facing up in the top chamber of Millicell filter inserts (Millipore) (diameter, 12 mm; pore size, 0.4 μm). The lateral edges of the explants were sealed with 3% agarose, as described previously (11, 30, 31). The orientation of the explants was monitored using a stereomicroscope (Stereomaster; Fisher Scientific).
Infant buccal and tonsil tissues from 4 infant cadavers (newborn, 2 days old, 53 days old, and 3 months old) were used for immunostaining of HIV-susceptible cells. Approval for collection of adult, infant, and fetal biopsy tissues was obtained from the Institutional Review Board at the University of California, San Francisco.
To detect membrane-associated, insoluble tight junction proteins in oral epithelium, the epithelial portion of the mucosa was isolated from the submucosa by the use of a surgical scalpel under a stereomicroscope (Stereomaster; Fisher Scientific). The stratified epithelium was homogenized and extracted with Triton X-100 buffer (1% Triton X-100, 100 mM NaCl, 10 mM HEPES [pH 7.2], 2 mM EDTA) containing a cocktail of protease inhibitors consisting of the following ingredients: phenylmethylsulfonyl fluoride (PMSF) (1 mM), aprotinin (10 μg/ml), leupeptin (10 μg/ml), and pepstatin A (10 μg/ml) (36). The extracts were centrifuged at 15,000 × g for 30 min. The supernatant and the pellet were considered to be the Triton X-100-soluble and -insoluble fractions, respectively. Protein concentrations were measured using the Bradford method, and both soluble and insoluble fractions were mixed with 2× sample buffer (4% sodium dodecyl sulfate-polyacrylamide gel electrophoresis [SDS], 0.75 M Tris [pH 8.8], 20% glycerol, 20 mM dithiothreitol). Proteins were separated using denaturing SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and 16% gels, and claudin-1 and occludin were detected using a Western blot assay and rabbit antisera (Zymed).
Tissues were fixed with 2% glutaraldehyde–4% formaldehyde in 0.1 M sodium cacodylate buffer, pH 7.3. Tissues were postfixed with 1% osmium tetroxide alone or with 1.5% potassium ferricyanide followed by 2% uranyl acetate and then dehydrated with ethanol. Tissues were embedded in Eponate 812 or EmBed 812. Ultrathin sections were stained with 2% uranyl acetate and examined at 120 kV using a JEOL 1400 transmission electron microscope.
Green fluorescent protein (GFP)-labeled HIV-1 virions were produced by cotransfection of 293T cells with proviral DNA of either X4-tropic HIV-1NL4-3 or R5-tropic HIV-181A and an expression vector encoding a GFP-Vpr fusion protein, as described previously (45). After 48 h, the released virions were concentrated by ultracentrifugation at 20,000 rpm in an SW41 rotor for 2 h at 4°C. p24 antigen in the viral stocks was quantified using a p24 enzyme-linked immunosorbent assay (ELISA) and stored in aliquots at −80°C. The R5-tropic HIV-1SF170 and X4-tropic HIV-192UG029 primary clinical isolates were grown in peripheral blood mononuclear cells (PBMCs) from HIV-uninfected individuals. PBMCs were activated with 2.5 μg/ml phytohemagglutinin (Sigma) and 1 μg/ml interleukin-2 (BD Bioscience) for 3 days.
To analyze penetration of cell-free HIV, GFP-labeled X4-tropic HIV-1NL4-3 and R5-tropic HIV-181A virions (100 ng of p24 per explant [approximately 108 virions]) were added to the upper chambers, and filter inserts with explants were incubated at 37°C or 4°C. In parallel experiments, matching tissues were pretreated with 10 μM colchicine for 2 h at 37°C, washed, and then used for virion penetration experiments. To disrupt tight junctions, explants were incubated in culture medium containing 10 mM EDTA for 2 h, extensively washed, and then used for virus penetration experiments. The duration of transcytosis in most experiments was 4 h and for some experiments was 8 or 20 h. Cross-sections of tissues were cut to 7 μm thickness with a horizontal orientation and examined for GFP virus or immunostained for cellular proteins. To amplify the GFP signal of GFP-labeled virions, sections were stained with rabbit anti-GFP antibodies conjugated with fluorescein isothiocyanate (FITC) (Invitrogen) (5 μg/ml). Penetration of GFP virus into the epithelium was confirmed by immunostaining of sections with mouse or goat anti-HIV-1 p24 antibodies (Virostat) (5 μg/ml).
To examine the infectivity of HIV that penetrated into intact and disrupted epithelium, adult buccal biopsy specimens were cut into 2 pieces, which were then used for polarized oriented explants. One of the matching explants was treated with 10 mM EDTA for 2 h, and the second one was not treated and served as a control. In parallel experiments, R5-tropic HIV-1SF170 and X4-tropic HIV-192UG029 primary clinical isolates were then applied to the surfaces of intact and EDTA-disrupted tissue explants at 100 ng of p24/insert. The infectious activity of input R5-tropic HIV-1SF170 and X4-tropic HIV-192UG029 stock viruses was measured using a reverse transcriptase (RT) assay as described previously (21). The RT values of HIV-1SF170 and HIV-192UG029 stock viruses were 982,000 and 588,000 cpm, respectively. After 4 h of incubation, explants were washed with trypsin to remove virions from mucosal surfaces as described previously (50), and tissues were homogenized in 0.5 ml of RPMI 1640 media. Tissue homogenates were centrifuged for 10 min at 2,000 rpm. Supernatant (200 ml) was used for infection of 106 PBMCs, and after 1, 2, and 3 weeks, 200 μl of media from each of the PBMC cultures was tested using an ELISA for HIV-1 p24. The infectivity of HIV-1SF170 and HIV-192UG029 viruses that penetrated into intact fetal buccal and intestinal tissues was evaluated using a similar approach.
For antibody-mediated inhibition of HIV-1 penetration into fetal oral epithelium, the following antibodies were used: rabbit polyclonal anti-galactosyl ceramide (anti-GalCer) (Chemicon) (25 μg/ml), mouse monoclonal anti-heparan sulfate proteoglycan (anti-HSPG) (Seikagaku) (25 μg/ml), mouse monoclonal anti-CXCR4 (pool of clones 44708, 44716, 44717, and 12G5) (10 μg/ml of each), and anti-CCR5 (pool of clones 45531, 45549, and 2D7) (50 μg/ml of each). Antibodies were added to the apical surfaces of polarized oriented fetal oral tissue explants, which were then incubated for 1 h prior to the addition of virus. A parallel set of experiments was performed in which tissues were treated with appropriate isotype controls for the anti-GalCer (rabbit IgG), -HSPG (mouse IgM), and -CXCR4 and -CCR5 (mouse IgG2a and IgG2b) antibodies. The concentrations of isotype antibodies were similar to the concentrations of specific antibodies. After 4 h of incubation, tissues were fixed and immunostained for HIV detection using a pool of human monoclonal antibodies composed of antibodies against HIV gp41 (2F5, D50) and gp120 (2G12), as described by Ganor et al. (15). Quantification of viral penetration was performed by counting epithelial cells containing HIV-specific signals. Inhibition of viral penetration was defined as the percentage of HIV-infected epithelial cells exposed to receptor-specific antibodies relative to HIV-infected epithelial cells exposed to isotype control antibodies. All antibodies to CXCR4, CCR5, and HIV proteins were provided by the NIH AIDS Reagent Program.
To examine cell-associated HIV penetration into oral epithelium, PBMCs from heparinized blood were isolated using a Ficoll-Paque Plus density gradient (Sigma). CD4+ T lymphocytes and CD14+ monocytes were then isolated by positive selection using anti-CD4 and anti-CD14 Microbeads (Miltenyi Biotec), respectively (Miltenyi Biotec). To propagate R5-tropic HIV-1SF170 cell-associated virus, monocyte-derived macrophages were established. To induce differentiation of monocytes into macrophages, purified monocytes were cultured in the presence of 20 ng/ml macrophage colony-stimulating factor for 7 days. To confirm macrophage differentiation, cells were immunostained for CD68, and CD68-positive macrophages were infected with R5-tropic HIV-1SF170 at 20 ng of p24 per 106 cells. To propagate cell-associated X4-tropic HIV-192UG029, CD4+ T lymphocytes were infected with HIV-192UG029 at 20 ng of p24 per 106 cells. Infection was confirmed after 7 to 10 days by detection of p24 released from macrophages or lymphocytes by ELISA. HIV-infected CD68+ macrophages were dissociated from the plastic using enzyme-free cell dissociation buffer containing 5 mM EDTA and a cell scraper, as described previously (25, 26). To detect penetration of HIV-infected macrophages and lymphocytes into oral epithelium, infected cells were incubated with 10 μM carboxyfluorescein diacetate succinimidyl ester (CFSE) for 10 min as described in the manufacturer's protocol (Invitrogen), enabling transmigrated HIV-infected cells to be distinguished from the intra- and subepithelial macrophages and lymphocytes present in the tissues at the time of collection. When CFSE penetrates the cells, its acetate groups are cleaved by intracellular esterases to yield a highly fluorescent green signal. HIV-infected, CFSE-labeled macrophages and lymphocytes were collected, washed, and added to the mucosal surfaces of polarized oriented explants at 106 cells per explant. In some experiments, incubation of HIV-infected cells with tissues was done in the presence of breast milk from HIV-uninfected, healthy, breastfeeding mothers. A pool of whole breast milk from 5 donors was used for experiments. Cells were resuspended in breast milk at a 1:2 dilution and added to the mucosal surfaces of explants. After 4 h, tissue explants were fixed and sectioned, and cross-sections were examined for CFSE-labeled cells coimmunostained with goat anti-HIV-1 antiserum.
Immunostaining of tissue sections was performed as described previously (53, 54). The following antibodies were used for detection of HIV-1 and cellular proteins: mouse anti-HIV-1 p24 (NIH AIDS Research reagent program) (5 μg/ml); goat anti-HIV-1 p24 (Virostat) (5 μg/ml); goat anti-HIV-1 (US Biological) (5 μg/ml); mouse anti-CD3 (BD Bioscience), goat anti-CD68 (R&D), rat anti-CD1a (Biosource), and mouse anti-dendritic cell specific intercellular adhesion molecule 3 (ICAM-3)-grabbing nonintegrin (DC-SIGN) (BD Bioscience) (1 μg/ml of each); and rabbit anti-ZO-1 (1 μg/ml), -occludin (1 g/ml), and -claudin-1 (5 g/ml) (all from Zymed). Secondary antibodies labeled with fluorescein isothiocyanate (FITC), tetramethyl rhodamine isothiocyanate (TRITC), or cyanine 5 (Cy5) were purchased from Jackson ImmunoResearch. The specificity of each primary antibody was confirmed by negative staining with the corresponding isotype control antibody. Cell nuclei were counterstained with TO-PRO-3 iodide (Molecular Probes) (blue). Immunostaining for each experiment was performed from 3 to 10 times. Cells were analyzed using a krypton-argon laser coupled with a Bio-Rad MRC2400 confocal head. The data were analyzed using Laser Sharp software.
For quantitative analysis of cell-free HIV penetration into adult oral epithelia, the sections of adult buccal explants exposed to GFP-labeled HIV were immunostained with anti-HIV p24 and anti-occludin antibodies, and the GFP- and p24-positive (colocalized) cells were counted. To examine penetration of HIV into virus-susceptible cells, sections were coimmunostained for HIV p24 and for CD4, CD68, or CD1a, which are markers for lymphocytes, macrophages, and Langerhans cells (LCs), respectively.
For quantitative evaluation of HIV penetration into fetal oral epithelia, tissue sections were immunostained for CD45 (clone T29/33; Dako), which is a marker of white blood cells, including dendritic cells/LCs, macrophages, and CD3+ lymphocytes. The same sections were costained for HIV p24, which confirms the presence of GFP-labeled HIV in CD45+ immune cells. HIV-GFP+ p24+ CD45+ intraepithelial and subepithelial immune cells in tissue explants were counted under various experimental conditions.
To quantitate HIV-infected immune cells, fetal oropharyngeal explants exposed to GFP-labeled X4-tropic HIV-1NL4-3 or R5-tropic HIV-181A virus were immunostained with antibodies to HIV-1 p24 and the following immune cell markers: CD3 for T lymphocytes, CD68 for macrophages, and CD1a for LCs. HIV-GFP/p24-positive cells expressing immune cell markers within the epithelium and lamina propria were counted.
For quantitative analysis of cell-associated HIV penetration into oral epithelia, sections were immunostained with goat anti-HIV-1 antiserum, and CFSE-labeled HIV-infected and CFSE-unlabeled HIV-infected cells were counted.
The number of HIV-positive epithelial and immune cells in adult and fetal mucosal epithelia was counted in at least 3 sections of each explant and 10 randomly chosen separate fields in each section. Results are presented as the percentage of positive cells or the average number of positive cells per square millimeter. Standard errors of the means of the results determined for positive cells were calculated using Microsoft Excel software.
Comparisons of differences in X4- and R5-tropic virus infection between appropriate target cells and infectivity of tissue-penetrated X4- and R5-tropic viruses in PBMC cells were performed using a cutoff of P < 0.05 with Student's t test.
To determine whether there were differences in the integrity of junctions among adult, infant, and fetal oral epithelia, we analyzed tight junction protein expression in these tissues by the use of a confocal immunofluorescence assay. We also examined fetal intestinal epithelia, since fetal and neonatal intestinal epithelia may also be portals for HIV MTCT. Tissue sections were immunostained for the tight junction proteins zonula occludens 1 (ZO-1) (Fig. 1A), occludin, and claudin-1 (data not shown). Tight junction proteins were detected in multistratified (10- to 30-layer) adult, 9-day- and 3-month-old infant, and 18- to 24-week-old fetal buccal epithelia (Fig. 1A). Tight junctions were also detected in infant tonsil and fetal oropharyngeal epithelia (data not shown). The thicknesses of infant and fetal oral epithelia were comparable. Infant oral epithelia had 3 to 7 layers of stratification, and fetal epithelia had 2 to 5 layers of stratification. However, some areas of infant and fetal oral epithelia had up to 10 cell layers. The monostratified columnar fetal intestinal epithelium was also positive for the tight junction proteins ZO-1 (Fig. 1A) and occludin (data not shown).
By electron microscopy, tight junction structures appeared as highly electron-dense regions between membranes of juxtaposing epithelial cells of the granulosum and spinosum layers of the adult oral epithelium (Fig. 1B, insets). Electron microscopy of fetal oral (Fig. 1C, red arrowheads in the inset) and fetal intestinal (Fig. 1D, red arrowheads in the inset) epithelia also revealed tight junctions.
Western blot analysis of claudin-1 and occludin from adult and fetal oral tissue extracted with Triton X-100 showed that these proteins were predominantly associated with the insoluble fractions of junctional complexes (Fig. 1E). These findings indicate the presence of well-formed tight junctions in both the adult and fetal oral epithelia. Together, these findings demonstrate that adult, infant, and fetal oral and fetal intestinal epithelia all contain well-developed tight junctions that support a barrier function.
To study the penetration of HIV from the surface of the adult epithelium, we mounted adult buccal explants from 11 different donors in agarose media, allowing interaction of HIV with only the mucosal epithelial surface (Fig. 2A). Green fluorescent protein (GFP)-labeled X4-tropic HIV-1NL4-3 and R5-tropic HIV-181A virions were added to the epithelial surfaces of the explants, and explants were incubated at 37°C for 4 h. Tissues were then fixed and sectioned, and viral penetration was examined by detection of GFP. Since the GFP signal was weak, signals were amplified by immunostaining of sections with anti-GFP antibodies.
Migration of GFP-labeled virions into the surface layers of the adult oral epithelium was detected in explants from 7 of 11 donors (63%). Confocal microscopy analysis revealed the presence of GFP-labeled virions within the most superficial 2 to 5 layers of the oral epithelium (Fig. 2B). The HIV-GFP signal could be seen in both small and large particles, which may represent unclustered and clustered virions, respectively. Viral penetration was inconsistently detected using lower concentrations of GFP-labeled virions (i.e., 10 ng or 50 ng of p24 per explant), but HIV-GFP signal was consistently detectable when incubating the epithelium with the viral equivalent of 100 ng of HIV p24 per explant. Immunostaining of tissue sections for HIV-1 Gag showed p24 colocalization with the GFP signals, confirming penetration of GFP-labeled virions into the epithelium (Fig. 2B). Incubation of explants with virus for 8 or 20 h at 37°C did not lead to further penetration of virions into the deeper layers of the epithelium (data not shown). Quantitative analysis of HIV-GFP/p24-positive epithelial cells revealed that about 10% to 25% of cells of the granulosum layers contained virions (Fig. 2C). The penetration of X4-tropic HIV-1NL4-3 and that of R5-tropic HIV-181A viruses into epithelia occurred at similar levels. To confirm the confocal microscopy data, HIV-exposed tissues were examined by electron microscopy, which revealed the presence of virions in the epithelial cells of surface granulosum layers (Fig. 2D).
To determine whether HIV virions that transmigrate across mucosal epithelia can reach intraepithelial and subepithelial HIV-susceptible immune cells, tissues were examined for the presence of GFP/p24-positive virions in intraepithelial Langerhans cells (LCs), macrophages, and T lymphocytes. Confocal microscopy revealed that LCs, macrophages, and CD3+ lymphocytes were mostly localized within the basal and parabasal layers and did not contain GFP/p24-positive virions (data not shown). These findings indicate that virions from the granulosum layers did not reach the HIV-susceptible immune cells of the basal/parabasal layers.
In the above-described experiments (Fig. 1), we showed that adult and fetal oral epithelia have well-developed tight junctions, which may prevent HIV penetration via intercellular spaces. Disruption of tight junctions may therefore allow viral penetration via these spaces. To examine this possibility, we experimentally disrupted the tight junctions of adult buccal epithelia by incubating explants in media containing 10 mM EDTA (Fig. 3). GFP-labeled R5-tropic HIV-181A virions were then added to the mucosal surfaces for 4 h at 37°C, and their penetration was examined. Confocal microscopy of untreated epithelia with intact tight junctions revealed virion penetration limited to the topmost 2 to 5 layers of the epithelial surface, i.e., stratum granulosum (Fig. 3A, left panel). Analysis of ZO-1 (Fig. 3A, right panel) expression in EDTA-treated tissues showed that the tight junctions were disrupted, and virions in these disrupted tissues were detected within the deeper part of the epithelium. Virions were detected in about 65%, 43%, and 10% of cells of the spinosum, granulosum, and parabasal layers, respectively, of disrupted epithelium (Fig. 3B). These data demonstrate that intact tight junctions prevent paracellular penetration across both adult and fetal oral epithelia and that their disruption facilitates virion access between epithelial cells, allowing for entry of HIV into deeper layers of epithelium.
To determine whether HIV that had penetrated into oral epithelium was infectious, we applied R5-tropic HIV-1SF170 or X4-tropic HIV-192UG029 primary clinical isolates to intact and EDTA-disrupted buccal explants propagated from 3 independent adult donors. After 4 h, explants were homogenized, and HIV infectivity was examined in PBMCs. ELISA analysis of PBMCs after 1 week did not detect HIV-1 p24; however, after 2 and 3 weeks, p24 was detected in PBMCs exposed to the supernatants of tissue homogenates from two of the three EDTA-treated explants (Fig. 3C). p24 was not detected in PBMCs exposed to the supernatants of two of the three control tissues not treated with EDTA. A low level of p24 was detected after 3 weeks in PBMCs infected with supernatant of homogenate from one of the untreated tissue explants.
It has been shown that the adult oral mucosa has HIV-susceptible T lymphocytes, macrophages, and Langerhans cells (10, 12, 17, 18, 56, 57). However, the presence of these cells in infant and fetal oral mucosae has not previously been investigated. To detect HIV-susceptible lymphocytes, macrophages, and dendritic cells in infant and fetal oral mucosae, fetal buccal and oropharyngeal tissues and infant buccal and tonsil tissues were immunostained for CD3, CD68, and CD1a. Confocal microscopy analysis showed that both the fetal and infant oral mucosae contained all three cell types (Fig. 4A). T lymphocytes and macrophages were detected predominantly within the lamina propria, and Langerhans cells were present within the epithelium. We also immunostained fetal intestinal epithelium for CD3, CD68, and dendritic cell-specific intercellular adhesion molecule 3 (ICAM-3)-grabbing nonintegrin (DC-SIGN) to identify T lymphocytes, macrophages, and dendritic cells, respectively (Fig. 4A). This staining revealed the presence of all three cell types within the lamina propria of fetal intestine. Quantitative analysis of HIV target cells in fetal and infant oral epithelia revealed that their numbers were comparable between the two epithelia (Fig. 4B to E). However, the number of HIV-susceptible cells in fetal intestinal epithelium was about 2-fold higher than that in fetal oral epithelium.
To determine whether HIV crosses the stratified (2- to 5-layer) fetal oral and monostratified columnar intestinal epithelium, polarized oriented explants from 10 buccal, 18 oropharyngeal, and 6 intestinal tissues obtained from independent fetuses at 18 to 24 weeks of gestation were exposed to GFP-labeled X4-tropic HIV-1NL4-3 and R5-tropic HIV-181A virions at their apical surfaces. Examination of fetal buccal (data not shown) and fetal oropharyngeal and intestinal epithelia incubated with these virions at 37°C for 4 h revealed virus penetration of both the epithelium and the lamina propria (Fig. 5A, left panels). To determine whether the viral penetration was due to endocytosis and transcytosis of virions from the apical surface of epithelia, experiments were performed at 4°C or in tissues pretreated with colchicine, which inhibits HIV endocytosis and transcytosis (3, 34). Penetration of virions into the lamina propria was not observed in fetal tissues incubated at 4°C (Fig. 5A, middle left panels) or in tissues pretreated with colchicine (Fig. 5A, middle right panels).
Transmigration of GFP-labeled virions was detected in 8 of 10 buccal tissues (80%), 16 of 18 oropharyngeal tissues (88%), and 6 of 6 intestinal tissues (100%). To quantitate HIV transmigration through fetal epithelia, we counted cells costained for CD45—a marker of white blood cells, including LCs, macrophages, and CD3+ lymphocytes (23, 61, 62)—and HIV p24. HIV infection of CD45+ immune cells has been shown previously (55). The presence of HIV-GFP+ p24+ CD45+ immune cells in the tissue explants is an indicator of initial viral transmigration across mucosal epithelia. In oropharyngeal tissues, about 1.5 intraepithelial and subepithelial CD45+ cells per mm2 were positive for HIV-GFP/p24 (Fig. 5B, upper panel). The frequency of CD45+ cells positive for HIV-GFP/p24 in the intestinal tissues was about 2 cells per mm2 (Fig. 5B, lower panel). The numbers of HIV-GFP-positive CD45+ cells in tissues exposed to X4- and R5-tropic viruses were similar, indicating that the levels of initial transmigration of X4- and R5-tropic viruses across epithelia were also similar. HIV-GFP/p24-positive CD45+ cells were almost exclusively detected only in the tissues incubated at 37°C, and only minimal levels were detected in tissues incubated at 4°C or in those pretreated with colchicine, which had frequencies of 0.1 to 0.2 cells per mm2 (Fig. 5B).
Electron microscopy analysis of oropharyngeal and intestinal epithelia incubated with R5-tropic HIV-181A virions at 37°C showed that viral particles had penetrated into oral (Fig. 5C) and intestinal (Fig. 5D) epithelial cells from their surfaces.
We have shown that the fetal oral epithelium expresses the HIV coreceptors CXCR4, CCR5, galactosyl ceramide (GalCer), and heparan sulfate proteoglycan (HSPG) (54). To determine whether these coreceptors play a role in HIV penetration into fetal oral epithelia, we incubated polarized oriented explants from two independent fetuses with antibodies against GalCer, HSPG, CXCR4, and CCR5. Penetration of R5-tropic HIV-1SF170 and X4-tropic HIV-192UG029 viruses into these epithelia was then examined. Immunostaining of these tissues for GalCer, HSPG, CXCR4, and CCR5 confirmed their expression in these tissues (data not shown). Quantitative analysis showed that antibodies to GalCer and HSPG inhibited both X4- and R5-tropic HIV penetration into these epithelial cells by about 45% and 35%, respectively (Fig. 5E). However, antibodies to CXCR4 and CCR5 did not significantly inhibit X4- and R5-tropic viral penetration, respectively.
Immunostaining of LCs, macrophages, and CD3+ lymphocytes for GFP-labeled virions in fetal oral tissues exposed to X4-tropic HIV-1NL4-3 or R5-tropic HIV-181A viruses revealed that both viruses were present in these cells (Fig. 6A). These HIV-GFP-positive immune cells were mainly localized within the epithelium and the adjacent lamina propria, near its border with basal cells. Analysis of CD3+ lymphocytes and macrophages for GFP-labeled virions in the intestinal epithelium also showed HIV-infected lymphocytes and macrophages (Fig. 6B).
To quantitate HIV-infected CD3+ T lymphocytes, CD68+ macrophages, and CD1a+ LCs, HIV-GFP-positive cells were counted in three independent oropharyngeal tissues exposed to X4-tropic HIV-1NL4-3 or R5-tropic HIV-181A virus (Fig. 6C, left panel). We found that CD3+ T lymphocytes, macrophages, and LCs were positive for both X4- and R5-tropic HIV. The number of HIV-infected cells was about 0.6 to 0.8 cells per mm2. The differences between X4-tropic and R5-tropic virus infection rates in both CD3+ T lymphocytes and macrophages were not statistically significant. Quantitative analysis of LCs also showed no significant difference between infection rates by the two viruses.
Analysis of intestinal tissues also showed that HIV-infected CD3+ T lymphocytes and CD68+ macrophages were positive for both X4- and R5-tropic HIV (Fig. 6C, right panel). However, the number of HIV-infected cells was about 3-fold higher than it was in oropharyngeal tissues.
To measure the infectivity of HIV entering the fetal oral and intestinal epithelium, we applied R5-tropic HIV-1SF170 or X4-tropic HIV-192UG029 primary viruses to fetal oropharyngeal and intestinal explants propagated from three independent fetal donors. The infectivity of HIV that penetrated into these mucosal epithelia was tested in PBMCs by infecting the PBMCs with supernatants of tissue homogenates. Detection of viral p24 using an ELISA after 1 week showed that the HIV that penetrated into the oral and intestinal epithelia was infectious both at both locations. However, significant differences between R5-tropic and X4-tropic viruses were not observed.
To investigate transmission of cell-associated HIV, CD4+ lymphocytes and CD68+ macrophages were infected with X4-tropic HIV-192UG029 and R5-tropic HIV-1SF170 viruses, respectively, and the cells were labeled with 5- (and 6)-carboxyfluorescein diacetate succinimidyl ester (CFSE) (Fig. 7A). Cells were added to the apical surfaces of adult and fetal buccal epithelia for 4 h. Confocal microscopy analysis of adult oral tissues with HIV-infected CD4+ lymphocytes and CD68+ macrophages showed that no lymphocytes or macrophages were bound to the mucosal surface or penetrated into the epithelium (Fig. 7B).
To determine the integrity of the mucosal epithelia examined here, adult buccal explants exposed to HIV-infected lymphocytes and macrophages were immunostained for the tight junction proteins ZO-1, occludin, and claudin-1. Confocal microscopy revealed that ZO-1 (Fig. 7C), occludin, and claudin-1 (data not shown) were present as rings encircling epithelial cells, a typical pattern for intact tight junctions. These data indicate that incubation of adult oral mucosa with HIV-infected lymphocytes and macrophages for 4 h did not lead to disruption of epithelial junctions.
Analysis of the fetal epithelium incubated with HIV-infected/CFSE-labeled lymphocytes showed that these cells were bound to the surface of the fetal oral epithelium but did not penetrate into the epithelium (Fig. 8A, upper left panel). However, HIV-infected/CFSE-unlabeled cells were detected within the lamina propria (Fig. 8A, upper left panel, inset) in these tissues, suggesting penetration of cell-free virions, which were released from HIV-infected lymphocytes. Notably, HIV-infected macrophages were bound to the mucosal surface of fetal oral mucosa (Fig. 8A, upper right panel) and transmigrated through the epithelium into the lamina propria (Fig. 8A, upper right panel, inset). Quantitative evaluation of penetration of HIV-infected macrophages into the fetal epithelium showed that about 6 HIV-infected macrophages per mm2 penetrated into the epithelium (Fig. 8B). The number of HIV-infected/CFSE-unlabeled cells was about 2 cells/mm2.
Analysis of tight junction protein expression in fetal explants revealed that ZO-1 (Fig. 8C), occludin, and claudin-1 (data not shown) were intact, indicating that incubation of fetal oral mucosea with HIV-infected lymphocytes and macrophages for 4 h did not cause disruption of epithelial junctions.
Since postnatal HIV MTCT occurs during breastfeeding, we examined cell-associated HIV penetration into fetal oral and intestinal epithelia in the presence of breast milk. HIV-infected CD4+ lymphocytes and CD68+ macrophages were resuspended in breast milk and applied to the mucosal surfaces of fetal buccal, oropharyngeal, and intestinal epithelia for 4 h. Analysis of penetration of lymphocytes into buccal (data not shown) and oropharyngeal (Fig. 9A, upper left panel) epithelia revealed that HIV-infected/CFSE-labeled lymphocytes were not present within the lamina propria. However, HIV-infected/CFSE-unlabeled cells were detected within the lamina propria (Fig. 9A, upper left panel, inset) of these tissues, potentially due to the penetration of cell-free virions produced by HIV-infected lymphocytes from the mucosal surfaces of oral explants. HIV-infected/CFSE-labeled macrophages were found to readily transmigrate across the oral epithelium and penetrated into lamina propria (Fig. 9, upper right panel, inset). The number of macrophages that transmigrated through the fetal oral epithelium was about 4 to 5 per mm2 (Fig. 9B).
Notably, penetration of both HIV-infected/CFSE-labeled lymphocytes and HIV-infected/CFSE-labeled macrophages was detected in the intestinal tissues (Fig. 9A, lower panels); however, the rate of macrophage transmigration was about 5- to 6-fold higher than the rate for lymphocytes (Fig. 9B). HIV-infected/CFSE-unlabeled cells were also detected within the lamina propria of intestinal tissues (Fig. 9A, lower panels, insets) exposed to both HIV-infected lymphocytes and macrophages, which could have been due to spread of virus from transmigrated cells and/or from cell-free virus released by apically applied lymphocytes and macrophages.
In a previous study, we showed that experiments detecting HIV transmission across monostratified, polarized adult and fetal oral epithelial cells gave different results. Only virions that passed through the fetal cells—and not those that passed through the adult cells—remained infectious (54). To understand further the potential mechanisms in adult mucosae that prevent HIV transmission, we established a clinically relevant model system, i.e., ex vivo organ tissue explants of fully developed adult and underdeveloped fetal oral mucosal epithelia (Fig. 1). Both the thickness of the fetal oral epithelium and the distribution of HIV target cells within the fetal mucosa were comparable to those seen with infant oral mucosa (Fig. 1 and and4).4). Thus, these fetal oral tissues serve, in particular, as suitable models for the study of HIV MTCT in infants, which plays a critical role in viral transmission during breastfeeding.
In this report, we have shown that the efficiency of HIV transmission depends on the thickness of the oral mucosal epithelium. Penetration of virions into intact oral mucosae was found to occur only through a few layers (1 to 5 layers) of the adult oral epithelium (Fig. 2). Notably, HIV-infected immune cells were not detected in the adult epithelium. This finding indicates that HIV transiting through a few superficial layers of adult multistratified oral epithelium has little chance of reaching HIV-susceptible immune cells, which reside within the basal/parabasal/suprabasal layers of the oral epithelium. However, disruption of the epithelial tight junctions facilitated penetration of virus into deeper parts of the adult epithelium (Fig. 3). Thus, a critical role for tight junctions is apparent in preventing paracellular transmission of HIV.
Penetration of cell-free HIV through intact fetal oral mucosal epithelia with tight junctions suggests that the migration of HIV across epithelia occurs by transcytosis (2, 3, 19, 20, 34, 43, 54). The inhibition of viral transmission at 4°C and by colchicine-induced depolymerization of microtubules (3, 34) also supports transcellular transmission of virus (Fig. 5). However, we cannot completely rule out the possibility of paracellular passage of some virions through microscopically undetectable disruptions in the tight junctions.
Expression of the HIV coreceptors GalCer, HSPG, CXCR4, and CCR5 and their functional role in viral transmission have been demonstrated in endometrial, vaginal, and intestinal epithelial cells (2, 3, 14, 19, 20, 28, 34, 43, 51, 54, 63, 64). Reductions in cell-free HIV transmission through the fetal oral epithelium by antibodies against GalCer and HSPG (Fig. 5) suggest that these molecules also play critical roles in cell-free HIV MTCT via the fetal/infant oral epithelium. However, antibodies against CCR5 and CXCR4 did not significantly affect cell-free viral transmission through this epithelium, suggesting that chemokine receptors do not play important roles in HIV MTCT through fetal/infant oral mucosal epithelia.
HIV that penetrated into the limited layers of intact adult oral epithelium were not infectious, in contrast to virions that penetrated into intact fetal oral epithelium (Fig. 3 and and6).6). These data are consistent with our other recent findings that a high level of expression of the anti-HIV innate proteins HBD2 and HBD3 in adult oral epithelia may inactivate virus during its transcellular transmission (54). Furthermore, lack of these innate proteins in fetal oral epithelia facilitates transmigration of infectious virions (54).
HIV-infected dendritic/Langerhans cells, macrophages, and T lymphocytes are present in fetal oral and intestinal epithelia (Fig. 6). This finding indicates that transcellular passage of infectious virions through the stratified fetal oral and nonstratified intestinal mucosal epithelium can lead to infection of intraepithelial and subepithelial HIV-susceptible immune cells and thereby initiate systemic infection.
Another mechanism by which HIV may traverse epithelia is by migration of HIV-infected cells through the intact epithelium. To test this possibility, we examined transmigration of HIV-infected lymphocytes and macrophages through adult and fetal oral mucosal epithelia. The absence of HIV-infected lymphocytes and macrophages that transmigrated through the adult oral epithelium after 4 h indicates that cell-associated HIV transmission in adults has little chance to initiate infection through the oral route (Fig. 7). In terms of the fetal oral epithelium, R5-tropic HIV-1-infected macrophages but not X4-tropic HIV-1-infected lymphocytes transmigrated across this epithelium (Fig. 8). During these studies, epithelial tight junctions were intact in both adult and fetal mucosa, indicating no disruption of the epithelium.
HIV-infected macrophages may migrate through paracellular spaces of intact mucosal epithelia (1). Similarly, HIV-infected macrophages may transmigrate via the paracellular spaces of polarized endometrial epithelial cells (6). In contrast, HIV-infected T-lymphoid cells (3) do not penetrate the epithelium below tight junctions. These data are consistent with our findings showing that HIV-infected macrophages can transmigrate across fetal oral and intestinal mucosal epithelia.
The breast milk of HIV-infected mothers contains HIV-infected lymphocytes and macrophages (22, 39, 41, 44, 47–49), which may initiate HIV MTCT during breastfeeding. R5 HIV-1-infected macrophages and not X4-tropic HIV-1-infected lymphocytes were the primary vehicle of infection in the presence of breast milk (Fig. 9), suggesting that the HIV-infected macrophages in breast milk lead to MTCT. Thus, cell-associated R5-tropic HIV-1 macrophages may be the most likely cells to transmigrate across fetal oral and intestinal mucosal epithelia. Predominant dissemination of R5-tropic HIV-1 in MTCT has been well documented (7–9, 58, 60). This finding could reflect the greater migratory activity of macrophages over lymphocytes.
In summary, we found that while HIV can readily transmigrate across fetal oral mucosal epithelia, efficient viral transmission through adult mucosal epithelia is unlikely (Fig. 10). This observation could be due to the differences in levels of stratification between fetal/infant and adult oral epithelia, which are paucistratified and multistratified, respectively. Moreover, fetal and infant oral epithelia lack expression of the anti-HIV innate proteins HBD2, HBD3, and SLPI, and the adult oral epithelium expresses high levels of these proteins (54). Thus, the adult oropharyngeal stratified epithelium may have two lines of defense against HIV: (i) a mechanical barrier of stratified epithelia with tight junctions that prevent penetration of virions into the deeper layers of the epithelium, and (ii) antiviral innate proteins that inactivate those virions that penetrate into the first 1 to 5 layers of epithelium. These defense mechanisms can play a key role in reducing HIV oral transmission in adult populations. Lack of these lines of defense in the fetal/infant oral epithelium may facilitate HIV MTCT. Approaches designed to increase the levels of innate defense proteins in infant mucosal epithelia and to inhibit transmigration of HIV-infected cells may be of value in reducing the risk of HIV MTCT in neonates and infants.
We thank Richard Jordan (University of California San Francisco, Departments of Orofacial Sciences and Pathology) for histological evaluation of tissues, Philip Ursell (University of California San Francisco, Department of Pathology) for providing the infant tissues, Larry Ackerman (University of California San Francisco, Diabetic Center) for electron microscopy, and Matthew Petitt for editorial assistance.
This project was supported by National Institutes of Health grants R21 DE016009 and R21 DE021011, the UCSF ARI Carl L. Gaylord Estate fund, and a UCSF CFAR grant (to S.M.T.).
Published ahead of print 28 December 2011