INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The transmission of human immunodeficiency virus type 1 (HIV-1) from mother to child is believed to occur mainly during delivery (intrapartum) and later through breast feeding (14). The percentage of infected children decreased markedly, from 15 to 35% to 0.8 to 3%, when pregnant women were given zidovudine (AZT) early in pregnancy and when their children were delivered by cesarean section (33, 34, 53). Early in utero transmission is a rare event (2). The evidence for in utero transmission of HIV includes identification of the virus in fetuses aborted in the first, second, and even third trimesters (35, 37, 40, 54). Comparison of maternal HIV-1 variants and virus populations detected in neonates at birth indicated that only some viruses are transmitted to the fetus and suggested that a selection occurs (1, 48, 52, 61). Although maternal HIV may enter the fetal blood directly through microbreaches of the placenta (9), studies on blood and placenta specimens obtained from a cohort of untreated, HIV-1-positive pregnant women and infants have shown HIV-1 sequences within the placenta of all seropositive pregnant women, irrespective of the virological status of the infants (20, 41, 64). In addition, only a limited number of maternal viruses were detected in the placenta (41), suggesting that in utero transmission is controlled by subtle regulatory mechanisms related to characteristics (structural and/or functional) of the viral variants and/or related to the development/organization of placenta cell subpopulations.

The placental chorionic villi are constituted by several cell subpopulations, including a trophoblast layer (cyto- and syncytiotrophoblast) forming the interface between maternal and fetal blood. These epithelium-like cells are organized as a tight monolayer (39). Cytotrophoblasts lying inside the villus gradually form a syncytium during pregnancy, which becomes exposed to maternal blood. Neighboring cells are connected by tight junctional complexes that actively participate in the trophoblast barrier function by precluding the passive paracellular transport of even small macromolecules. Exchanges between the mother and the fetus occur by transcytosis, a pathway of membrane trafficking specific to polarized epithelial cells (44). During this vesicular transcellular transport, the transcytosed cargo remains enclosed in transcytotic vesicles, precluding contact with the host cell cytoplasm, and is released intact at the opposite pole (29, 44). Maternal blood immunoglobulins G (IgGs) are carried across the placenta by receptor-mediated transcytosis into fetal blood (24, 36). Trophoblastic cells also form the first efficient barrier to protect the fetus from maternal infection, but its efficiency depends on the pathogen (28, 38, 55, 56). Data concerning the permissivity of purified primary trophoblasts or trophoblastic cell lines to cell-free HIV infection remain controversial (18, 32, 43, 47, 62, 63). All these studies were done using trophoblast target cells that were not organized as a functional, polarized trophoblast barrier.

We have therefore established an in vitro experimental model, based on our previous study on HIV interaction with simple epithelia, that mimics the interaction of primary HIV isolates from infected mothers and the trophoblastic barrier. The barrier is represented by BeWo cells (46) that were initially isolated from a choriocarcinoma and are now widely used as a model for trophoblast cells. These cells form a polarized, tight monolayer when cultured on polycarbonate filters in a two-chamber system (12, 24).

We found that cell-free virus does not cross the BeWo monolayer either by transcytosis or following BeWo cell infection. In contrast, the contact between HIV-infected peripheral blood mononuclear cells (PBMCs) and the apical side of the trophoblast layer leads to two events. The first and rapid event consists of HIV-1 particles budding at the contact site (19), followed by their transcytosis across the trophoblastic barrier without viral replication. The second event is the fusion of HIV-1-infected PBMCs with trophoblastic cells, resulting in virus replication and the production of virus progeny toward the basal pole of the trophoblastic monolayer. Maternal virus variants are selected following both transcytosis and fusion-induced virus replication. The comparison of the in vitro-translocated virus quasispecies derived from infected mothers with those transmitted to their children in vivo evidenced that in the case of in utero transmission, the child's virus was indistinguishable from the maternal, transcytosed, predominant variant. Our data suggest therefore that our in vitro system is at least partially reflecting the initial steps of in vivo transmission of HIV-1 in utero.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Cells. The human choriocarcinoma BeWo cell line (ATCC CCL 98) (46), obtained from the American Type Culture Collection (Rockville, Md.), was maintained in Dulbecco's modified Eagle medium containing 25 mM glucose, supplemented with 20% heat-inactivated fetal calf serum (GIBCO BRL; Gaithersburg, Md.), 20 mM glutamine, 25 mM HEPES (pH 7.4), and antibiotics (50 IU of penicillin/ml and 50 µg of streptomycin [GIBCO BRL]/ml in 5% CO₂-95% air) (12). Venous blood samples were taken from healthy blood donors, and PBMCs were isolated by Ficoll-Paque (Pharmacia Biotech, Uppsala, Sweden) centrifugation and used either as effector cells or as indicator cells or to amplify viral isolates. Phytohemagglutinin (PHA)-stimulated PBMCs were cultivated in RPMI 1640 medium supplemented with 10% heat-inactived fetal calf serum (GIBCO BRL) and 10 U of interleukin-2/ml. T-cell (CD3⁺/CD14) and monocyte/macrophage (CD3/CD14⁺) populations were separated by negative selection using Dynabead columns as recommended by the manufacturer (Dynal, S.A., Oslo, Norway).

Viral isolates and HIV-1-infected PBMCs. Primary virus isolates were recovered from PBMCs of HIV-1-infected, pregnant women after their informed consent to participate in the study (41, 51). None of them was under antiretroviral therapy. Viral stocks were obtained after two passages of the primary isolates on PHA-stimulated PBMCs. Two HIV-1 subtype A isolates (1329 and 1379) were both of the non-syncytium-inducing (NSI) (R5) phenotype derived from Cameroonian nontransmitting mothers. Subtype B viruses were from Italian mothers: two were NSI (R5) (A113 and 115) and three were syncytium inducing (SI) (X4) (A245, A196, and A204). A204, A115, and A196 were from mothers who transmitted the virus to their child, either in utero (A204) or intrapartum (A115 and A196). The other viruses were from nontransmitting mothers (1329, 1379, A113, and A245). Virus stocks were snap frozen in aliquots and were stored at 80°C. The infectious titer of each virus stock was determined by limiting dilution assays on PHA-activated PBMCs, as previously described (49), and was expressed as 50% tissue culture infective dose (TCID₅₀) per milliliter. Either cell-free viruses or infected PBMCs were used in the in vitro model. Infected PBMCs used as effector cells were prepared as follows: 10⁶ PBMCs were inoculated with a dose of 100 TCID₅₀ of virus previously treated with 200 U of RNase-free DNase per ml (15 min at room temperature). The infected PBMCs were extensively washed and maintained at 37°C for 7 to 10 additional days, until virus production was maximal, as determined by HIV-1 p24 antigen content in cell culture supernatant. Uninfected PBMCs or T-cell-line CEM cells chronically infected with laboratory-adapted HIV-1 strains (HIV-1 LAI subtype B and HIV-1 NDK subtype D) or HIV-2 strain (HIV-2 ROD subtype type A) (59) were used as controls.

In vitro model to study HIV translocation across polarized BeWo trophoblastic monolayers. We investigated the mechanisms responsible for the passage of HIV across the trophoblastic barrier, using the in vitro model shown in Fig. 1. This model involves three distinct cell types: (i) effector cells consisting of HIV-1-infected PBMCs (replaced in some experiments by cell-free virus); (ii) target cells, a tight, polarized monolayer of BeWo trophoblastic cells; and (iii) indicator cells, uninfected PBMCs supporting replication of infectious HIV-1 particles that had been transcytosed through or were released from infected BeWo monolayers.

View larger version (22K): [in this window] [in a new window]   FIG. 1.   Experimental system for monitoring HIV translocation across the polarized trophoblastic monolayer. The inoculum, cell-free HIV or HIV-infected PBMCs (effectors), is applied at the apical pole of target trophoblastic BeWo cells. Both transcytosis across and productive infection of target cells are monitored in the basolateral chamber after addition of indicator PBMCs.

Nonpolarized BeWo cells. To study the effects of HIV-1 on unpolarized trophoblastic cells, BeWo (5 × 10⁵ cells/well) were cultured on plastic support (6-well plates [Costar Corp.]) for 36 h to reach 50 to 70 percent confluency.

Inoculation of BeWo target cells and virus detection. The experimental protocol used to study the interaction of HIV (cell-free particles or HIV-infected effector PBMCs) with the BeWo cell monolayer is shown in Fig. 1. We studied the interaction of HIV-1-infected effector PBMCs with the apical surface of BeWo cell monolayers by adding 2 × 10⁶ effector PBMCs infected with HIV-1 primary maternal isolates (clade A or B) to the apical chamber where they were in contact with trophoblastic BeWo target cells (6, 7). To study cell-free virus interaction with BeWo, these target cell monolayers were inoculated with cell-free HIV-1 by adding virus isolates (100 TCID₅₀) to the apical or basolateral chamber or to unpolarized BeWo cells. Fresh media were added into the opposite chamber. In some experiments, BeWo cells were treated with AZT (10 µM; Sigma); cells were incubated for 2 h before contact with cell-free virus and during all the following steps. As a control, PBMCs (10⁶ cells) infected with cell-free HIV-1 were treated with AZT under similar conditions.

Morphological detection of fusion between HIV-1-infected PBMCs and trophoblastic cells. Effector HIV-1-infected PBMCs were loaded with a fluorescent probe, CT green (0.5 µM), for 30 min at 37°C and were placed in the apical chamber in contact with the trophoblastic BeWo target cell monolayer for 2 h 30 min. The BeWo cell monolayers were fixed, and nuclei were fluorescently labeled with Hoechst 3345 (5 µg/ml) for 15 min. The cell monolayers were then mounted in Mowiol (7). The redistribution of the CT green from effector HIV-infected PBMCs to BeWo target cells indicated fusion. This was analyzed by confocal microscopy (Bio-Rad 1024 with a 60× optical lens [Nikon] using argon/krypton and UV lasers) as previously described (7). Consecutive sections were 0.5 µm apart. Images corresponding to the vertical projection of at least 20 consecutive optical sections were processed using the double-labeling laser Sharp software. CT green appears yellow and nuclei appear red after merging of two sections recorded at the same level in each channel.

Quantitation of fusion. Cell fusion was measured by fluorescence resonance energy transfer using compatible dyes, CT green (excitation wavelength, 488 nm; emission wavelength, 524 nm) and CT red (excitation wavelength, 540 nm; emission wavelength, 572 nm) (31, 60). Briefly, each of the fusion partners was loaded with one of the two dyes, effector HIV-1-infected PBMCs with CT green and target trophoblastic BeWo cells with CT red (0.5 µM) by incubation for 30 min at 37°C. Effectors were added to the targets as described above. Fusion caused the formation of a syncytium and the mixing of the two dyes in sufficiently close contact to allow resonance energy transfer to occur from CT green to CT red following CT green excitation. Cells were then detached from their support by using trypsin, washed in phosphate-buffered saline, and fusion quantified by flow cytometry after CT green excitation at 488 nm and collection of the CT red signal at 572 nm (31, 60), using the Elite flow cytometer and Elite software (Coulter, Marseilles, France). Results are expressed as the percentage of target cells (index of fusion) that underwent fusion.

Analysis of virus phenotype and coreceptor usage. The phenotypes of maternal viruses were analyzed before and after transcytosis or replication. The NSI/SI phenotype was identified using the standard test involving MT2 T-cell lines (30, 42, 50). Briefly, MT2 cells (5 × 10⁵) were incubated for 1 h with the sample for analysis, washed extensively, and cultured for up to 15 days. The cultures were analyzed by visual inspection for cytopathic effect (syncytium formation), and infection was quantified by measuring HIV-1 p24 antigen in culture supernatants by enzyme-linked immunosorbent assay. The chemokine receptor usage of a sample was determined by using human glioma CD4⁺ U87 cell lines stably expressing CCR5 (R5) or CXCR4 (X4) (22, 23). Briefly, cells were seeded at a low concentration (10⁴/well) in 96-well plates, incubated for 48 h, and inoculated with 200 µl of the test sample for 1 h. The plates were washed extensively to remove the virus inoculum and cultured in fresh medium. Cultures were observed daily for cytopathic effect (syncytium formation). Supernatants were collected from each well and tested for HIV-1 p24 antigen.

DNA extraction and PCR amplification of HIV-1 tat and V3-V5 env fragments. Total DNA was extracted from the three cell types involved in the in vitro model: (i) effector-infected PBMCs, (ii) infected target BeWo cells immediately or at the indicated times after contact with effector-infected PBMCs (or with cell-free virus) and extensive washes, and (iii) from infected indicator PBMCs. The cell pellets were resuspended in a lysis buffer (0.01 M Tris-HCl [pH 7.4], 0.1 M NaCl, 0.01 M EDTA, 1% sodium dodecyl sulfate) containing RNase (100 µg/ml) and proteinase K (100 µg/ml). DNA was then extracted with phenol, phenol-chloroform, and chloroform-isoamylalcohol; precipitated with ethanol; and suspended in sterile water for PCR amplification.

HMA. Comparative genetic analysis of HIV samples was performed by heteroduplex mobility assay (HMA) (21). Each amplified gene product in the V3-V5 env region was mixed with a homologous PCR fragment from a reference virus DNA corresponding to maternal HIV-1 subtype A or B. Heteroduplexes were separated by electrophoresis on a 5% polyacrylamide gel (using a 30% acrylamide-0.8% bisacrylamide stock), stained with ethidium bromide (1 µg/ml), and examined. Patterns of HIV-1 env quasispecies detected in the different fractions (Ap, apical inoculum [representative of maternal virus]; BW, BeWo monolayer after 2-h-30-min contact with infected PBMCs; PI, produced by BeWo cells after infection; Tr, transcytosed; or Ch, pair child viral isolate) were analyzed together for direct comparison.

PCR and direct sequencing of gp120-V3 region. Sequences specific for HIV in the V3 region of the gp120 gene were PCR amplified (49, 52), first with the outer primers (JA 9-12). One-tenth of the product was further amplified using a set of inner primers (JA 10-53). The amplified product was directly used for solid-phase sequencing. Briefly, DNA fragments obtained by amplification with inner primers were purified by immobilization of biotinylated primers on streptavidin-coated magnetic beads (Dynabeads M280-streptavidin; Dynal). The sequencing reaction was performed with fluorescein-labeled primers using a commercial kit (AutoRead; Pharmacia). The reaction products were then loaded on a 6% polyacrylamide gel in an automated laser fluorescent sequencing apparatus (ALF; Pharmacia LKB).

Sequence analysis. V3 nucleotide sequences were aligned using the CLUSTAL W program with minor manual adjustments. Phylogenetic trees were generated by the neighbor-joining and Fitch-Margoliash distance methods implemented in the PHYLIP package, and their reliability was estimated from 100 bootstrap replicates. Phylogenetic trees were also constructed by maximum parsimony and maximum likelihood.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

BeWo cells are resistant to cell-free virus infection and transcytosis, regardless of their polarity. We first investigated the permissivity of target BeWo trophoblastic cells to cell-free HIV, as cell-free HIV is generally assumed to be the main vector of transmission and is present in maternal blood at the trophoblast interface. Trophoblast monolayers were infected with cell-free HIV stocks (100 TCID₅₀). No evidence of transcytosis from the apical to the basolateral side or productive infection by cell-free HIV was observed, since no HIV-1 p24 antigen was detectable in the basolateral medium when analyzed directly or when indicator PBMCs were inoculated with these HIV-1 p24 antigen-negative supernatants and cultured up to five additional days (Table 1). These results were similar regardless of the virus subtype, coreceptor usage, SI/NSI phenotype, or transmitted or nontransmitted status. There was no detectable HIV-1 p24 antigen in the supernatant from the apical chamber when the polarity of the infection was reversed and when virus isolates were added to the basolateral side (not shown). Similarly, no significant HIV-1 p24 antigen was detected in the culture supernatant of unpolarized BeWo cells inoculated with cell-free virus (not shown). In contrast, BeWo monolayers treated with EGTA (10 mM) to alter the integrity of their tight junctions showed a paracellular transport of the virus, together with the leveling of medium in both chambers (not shown) (6).

View larger version (61K): [in this window] [in a new window]   FIG. 2.   Infection of polarized and unpolarized BeWo trophoblastic cells by cell-free virus. Media conditioned with uninfected PBMCs (Mock), viral isolate 115 (A), or 204 (B to E) were inoculated either to the apical (Ap) or to the basolateral (BL) side of a trophoblastic BeWo cell monolayer (as indicated), to unpolarized BeWo cells (D), or to uninfected PBMCs (E). Cells were treated (+ AZT) or not treated ( AZT) with AZT. Five days after contact, DNA was extracted and HIV-1 tat sequences were PCR amplified, resolved on an agarose gel next to a PCR-amplified HIV-1 tat sequence from HIV-1-infected-PBMCs used as a positive control (+), and transferred onto membranes before hybridization with a 32P-labeled tat A20 probe. Membranes were exposed to Kodak film for autoradiographic visualization. -Actin PCR-amplified sequences are represented in panel F. Each bar represents a triplicate (or a duplicate for HIV-1-infected PBMCs) of PCR-amplified sequence for each sample.

Cell-to-cell contact between HIV-infected PBMCs and BeWo cell monolayers induces both transcytosis of infectious virus and HIV replication. Since cell-free viruses did not cross the BeWo monolayer, we investigated whether cell-to-cell contact between effector HIV-1-infected PBMCs and polarized BeWo target cells could result in virus transcytosis, as was previously reported when epithelial cells were used as the target (6). In contrast to cell-free virus, HIV-1 p24 antigen was detected in the supernatant of the indicator PBMCs cultured for 5 days after being in the basolateral chamber during inoculation (Table 1). As described in detail in Materials and Methods, the paracellular passage of virus could be excluded since no effector apical HIV-infected PBMCs loaded with a fluorescent dye (CT green) were found in the basolateral chamber throughout the experiment. In contrast, opening tight junctions with EGTA resulted in the detection of HIV-infected fluorescent cells in the basolateral medium, presumably due to a paracellular transport. Furthermore, examination of the monolayer by confocal microscopy has confirmed that effector HIV-infected cells did not bypass tight junctions.

HIV-1 replication correlates with fusion between infected PBMCs and BeWo cells. To study whether fusion between infected PBMCs and BeWo cells contributes to HIV-1 replication, we used HIV-1-infected PBMCs loaded with a fluorescent probe (CT green) placed on the apical side of trophoblastic BeWo target cells. At the end of the cell-to-cell-contact followed by extensive washes, the incubated monolayers were fixed and nuclei were fluorescently labeled to visualize the cells, and the two fluorescences were analyzed simultaneously by confocal microscopy. We observed a redistribution of the fluorescent CT green dye from the apical effector HIV-infected PBMCs to trophoblastic BeWo target cell cytoplasm (Fig. 3A, panels b to c), indicating a fusion between effector and target cells. The fusion occurred with HIV-infected effector cells, regardless of the isolate subtype (A or B) or the SI or NSI phenotype (Fig. 3C, panels a to d). Furthermore, the fusion was observed with total PBMCs (Fig. 3A and C), as well as with CD3⁺/CD14 lymphocytes (Fig. 3B, panel a) or CD3/CD14⁺ monocytes (Fig. 3B, panel b) but in all cases only when infected (Fig. 3A, panel a).

View larger version (73K): [in this window] [in a new window]   FIG. 3.   Fusion of HIV-1-infected cells with a trophoblast monolayer. PBMCs infected by HIV-1 (A, panels b to d), SI (C, panel c), or NSI (C, panels a, b, and d) clade A (C, panels a and b) or B (C, panels c and d) from transmitting (C, panel c) or nontransmitting (C, panels a and d) mothers, as well as HIV-1-infected T lymphocytes (CD3+/CD14) (B, panel a), HIV-1-infected monocytes/macrophages (CD3/CD14+) (B, panel b), or uninfected PBMCs (A, panel a) were loaded with a fluorescent probe and inoculated at the apical side of a trophoblastic BeWo monolayer for 2 h 30 min. The trophoblastic BeWo monolayer was then fixed, the nuclei were labeled, and the two fluorescences were analyzed by confocal microscopy. Fusion between effector and target cells causes redistribution of the dye in the effector cell to the target cell cytoplasm. BeWo cell nuclei appear red; CT green-loaded HIV-1-infected effector cells appear green.

View larger version (52K): [in this window] [in a new window]   FIG. 4.   Quantification of fusion between HIV-1-infected effector cells and trophoblastic BeWo target cell monolayer. The fusion partners, effector HIV-1-infected PBMCs and trophoblastic BeWo monolayers, were loaded with the appropriate fluorescent dyes, and effector cells were then added at the apical side of the trophoblastic BeWo monolayer for 2 h 30 min. Fusion resulted in the cell cytoplasm mixing, leading to sufficiently close mixing of the two dyes such that energy transfer between them could occur. The fluorescence energy transfer was quantified by flow cytometry, by exciting the dye in the effector PBMCs (, 488 nm) and measuring the fluorescence emitted by the dye in the target BeWo cells (, 572 nm). Results are expressed as a fusion index (percentage of target cells that had undergone fusion) using PBMCs infected with HIV-1 clade A or B (HIV + PBMCs), HIV-1 (IIIb)- and HIV-2 (ROD)-infected CD4+ T-cell lines, and uninfected PBMCs (n.i.) as effector cells.

Replication but not transcytosis induces selection of HIV viral variants in HIV coreceptor usage. We next investigated whether maternal viruses were under a selective pressure when crossing the BeWo cell barrier, as indicated by previous studies on placenta specimen ex vivo (41). The markers of selective pressure that we first tested were the HIV coreceptor usage (X4 or R5) for entry into CD4⁺ T target cells and the biological phenotype (SI or NSI). These markers were analyzed for virus recovered from each fraction of the system: the inoculum after transcytosis and after replication. The coreceptor usage coincided with the expected biological phenotype on MT2 cells: in all cases, X4 with SI (X4/SI) and R5 with NSI (R5/NSI). After contact of BeWo cells with total PBMCs and CD3⁺/CD14 cells, the original predominant biological phenotype of HIV maternal isolates was conserved for the transcytosed viruses (Table 4). Interestingly, viruses derived from the fusion of effector cells with BeWo target cells always showed a strict X4/SI phenotype, even when effector PBMCs were infected with viruses with a predominant R5/NSI phenotype. In contrast, the contact with BeWo cells of CD14⁺/CD3 cells (mostly monocytes and macrophages) purified from X4/SI- or R5/NSI-infected effector PBMCs preferentially selected and propagated viruses with the R5/NSI phenotype (Table 4, apical results). As for unfractionated effector PBMCs, the fusion of CD14⁺/CD3 cells that were isolated from effector PBMCs infected with predominant X4/SI viruses resulted in the production of X4/SI viruses, whereas the fusion of CD14⁺/CD3 cells purified from predominant R5/NSI-infected effector PBMCs resulted in the production of R5/NSI viruses (Table 4).

HIV translocation across trophoblastic BeWo monolayer results in enrichment of HIV minor variants. To further examine whether maternal viruses were under selective pressure during passage across BeWo monolayers, the genetic diversity of viruses isolated from the different compartments was analyzed by HMA. We used viral isolates from a nontransmitting mother (indicated by the lack of infection of her child 18 months after birth), one set of matched mother-and-child viral isolates (probably infected in utero, since the child was infected at birth), and one set of matched mother-and-child virus isolates (probably infected at delivery, since the child was diagnosed as infected 1 month after birth). Each of the maternal isolates was used to infect effector PBMCs and was incubated with BeWo target cells as described above. Virus quasispecies derived from these infected effector PBMCs (Fig. 5, tracks Ap), from the BeWo target monolayer following contact and hence fusion with these infected effector PBMCs (Fig. 5, tracks BW), from indicator PBMCs infected either by transcytosed viruses (Fig. 5, tracks Tr) or by viruses produced after fusion between infected effector PBMCs and target BeWo cells (Fig. 5, track PI), and finally from PBMCs isolated from the matched infected child (Fig. 5, tracks Ch) were PCR amplified and compared by HMA. Figure 5 showed three representative cases of such comparative analyses.

View larger version (85K): [in this window] [in a new window]   FIG. 5.   Selection of HIV quasispecies by a BeWo trophoblastic barrier. A given amount (150 to 250 ng) of each PCR-amplified product was mixed with an equivalent PCR product derived from a reference plasmid, A1 or B1, that corresponded to the mother's HIV-1 subtype. PCR amplifications were performed on DNA extracted from the inoculum used apically (Ap) corresponding to PBMCs infected with HIV-1 maternal isolates, the trophoblastic BeWo monolayer after contact with the inoculum (BW), viruses derived from infection of the BeWo monolayer after one passage on basolateral indicator PBMCs (PI), transcytosed virus (Tr), or infected PBMCs isolated from the pair-infected child (Ch). Heteroduplex (HE) and homoduplex (HO) patterns obtained on 5% acrylamide gels were analyzed after ethidium bromide staining (upper gels). The quality and quantity of PCR-amplified V3-V5 env products were analyzed after resolution on a 2% agarose gel (middle gels). Not transmitted, maternal viral isolate 1329; Transmitted in utero, viral isolates 204 from a transmitting mother (Ap) and her child (Ch) infected in utero; Transmitted intrapartum, viral isolate 196 from a transmitting mother (Ap) and her child (Ch) infected intrapartum.

Viral variants selected during transcytosis are closely related to in utero-transmitted viruses. We next analyzed the V3 gp120 env sequences of the HIV-1 variants in the fractions analyzed in Fig. 5. The major V3 gp120 env sequence detected in the PBMCs of the only in utero-infected child studied was closely related, if not identical, to the major V3 sequences of transcytosis-derived virus. In contrast, the major V3 env sequence detected in the fusion-derived virus was closer to the maternal PBMC-derived virus (Fig. 6). Such homology was not observed in the case of intrapartum transmission, where the major V3 sequences from the child were related to but distinct from those derived from both transcytosis and fusion.

View larger version (19K): [in this window] [in a new window]   FIG. 6.   Phylogenetic tree of HIV-1 V3 env nucleotide sequences obtained by direct sequencing. The V3 env nucleotide sequences were compared with the following known sequences of children infected with subtype B virus: C190, C204, C196, C136, C145, C115, C130, or C201. Intrapartum transmission refers to mother-child pair 196. In utero transmission refers to mother-child pair 204. The children's samples were collected when the infants were 1 month old. C190 was used as an outgroup, and FR.LAI was used as a subtype B reference. The Fitch-Margoliash tree is shown. The results were the same for all trees in all essential aspects. For the sake of clarity, only bootstrap values above 60 are indicated.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The human syncytiotrophoblast is a highly polarized epithelium-like layer responsible for regulating maternal-fetal exchanges (39). During pregnancy, the apical side of this barrier is in contact with infected maternal blood which contains both cell-free HIV and HIV-infected cells. We have used an in vitro model to examine the first steps that could restrict the in utero transmission of HIV-1. This model uses trophoblast-like BeWo cells that form a tight, polarized monolayer in a two-chamber culture system. The apical side of the trophoblastic barrier, when in contact with HIV-1-infected PBMCs or cell-free HIV-1 isolated from seropositive mothers, mimics the maternal blood-placenta interface, whereas the basolateral side may be considered the first compartment of the fetal side. We first demonstrated that cell-free viruses cannot cross the BeWo monolayer by transcytosis from the apical to the basolateral side or by target cell infection and replication. This is consistent with previous results showing that intestinal or endometrial epithelial cells grown as a tight monolayer do not allow transcytosis of cell-free virus (6, 7), as well as with studies with primary trophoblasts or trophoblastic cells that appeared to be resistant to HIV infection (18, 19), although the cells used in these previous studies were not polarized. The reason for this resistance is not yet clear. The study reporting that BeWo cells could be infected by cell-free virus showed a poor productive infection limited to few HIV-adapted laboratory strains (43). In the present study, primary HIV isolates were used rather than laboratory strains. Virus sequences were occasionally detected in BeWo cells after contact between cell-free virus and the apical or basolateral surface of trophoblastic cells, but this detection was never coupled with a detectable production of virions. No viral sequence was detected in AZT-treated cells, which is consistent with a recent study showing that trophoblasts isolated from the placentas of AZT-treated, pregnant, HIV-positive women are not infected (57). The inability of cell-free HIV to transcytose or to infect BeWo cells cannot be ascribed to factors related to cell polarity, as identical results were obtained when cell-free virus was applied either apically or basolaterally or when we attempted to infect unpolarized BeWo cells. This contrasts with findings for other viruses that infect and replicate in epithelia in a polarized manner, such as vesicular stomatitis virus or canine parvovirus (infecting predominantly through the basolateral membrane) (3, 27) and measles virus or simian virus 40, which preferentially infects via the apical membrane (5, 15).

In contrast to cell-free HIV, HIV-infected PBMCs induced HIV translocation across the trophoblastic barrier. Contact between HIV-infected PBMCs and trophoblastic cells is required for the release of infectious viruses into the basolateral chamber both by transcytosis and after fusion between effector PBMCs and target trophoblastic cells. Trancytosis of infectious HIV across the trophoblastic barrier occurs shortly after cell-to-cell contact, as expected for such a process (6). In contrast, replicating viruses were detected later in the basolateral side, 3 days after effector-target cell contact and fusion of the two cell partners.

The molecular mechanisms governing HIV binding to, entry into (47, 62, 63), and transcytosis across trophoblastic cells remain unclear. Some authors have suggested that entry is independent on any cell surface molecules known to be involved in HIV-1 entry, including CD4 (62, 63). Supporting this theory, CD4 is not detected on BeWo cells by immunofluorescence assay when they are cultured as a polarized monolayer (M. Bomsel, unpublished data), even though CD4 has been detected on fresh primary trophoblastic cells from early placenta (63) and to a certain extent from term placenta (43). Alternative receptors for HIV envelope glycoproteins, such as the glycolipid galactosyl ceramide, a marker of the apical surface of epithelial cells, might be involved in HIV transcytosis and/or entry into trophoblastic cells. The observations that HIV transcytosis across epithelial cells is impaired by a monoclonal antibody specific for galactosyl ceramide (6) and that galactosyl ceramide can be detected at the BeWo apical pole (Bomsel, unpublished) are in favor of this possibility.

As adhesion molecules may act as a cofactor for HIV fusion (4, 45, 47, 58), they may also be involved in the fusion of primary infected cells with trophoblasts and/or the transmission of the virus. Previous reports on the impact of adhesion molecules (26) and of human leukocyte antigen (HLA) class I (16) and class II molecules (10) on the infectivity of HIV particles are consistent with this hypothesis. We find that both lymphocyte (CD3⁺/CD14⁾ and monocyte/macrophage (CD3/CD14⁺) effectors infected with SI or NSI viruses undergo fusion with trophoblastic target cells. In contrast, not only uninfected PBMCs but also chronically infected T-cell lines do not significantly fuse with trophoblastic cells. Thus, in addition to viral envelope glycoproteins, host cell molecules, including adhesion molecules, may participate in fusion by being recruited into the plasma membrane microdomain where HIV envelope glycoproteins bind (25). Such host cell HIV fusion cofactors may be absent or inactive in the chronically infected cell lines that we used. In support of this hypothesis, a previous study showed that choriocarcinoma-derived trophoblasts cannot be infected by chronically infected HIV-1 monocytic and lymphocytic cell lines with impaired adhesion capacity (8). However, the lack of infection by cell-free HIV suggests that the capacity of infected cells to fuse with and infect target BeWo cells is mediated by host molecules distinct from those involved in virus-cell fusion.

Virus transmission resulting from cell-to-cell contact might also be independent of cell fusion. It might also be a fusion process between budding particles released from the surface of infected PBMCs and BeWo.

Both CCR5 and CXCR4 coreceptors are detected by immunohistochemistry at the apical surface of BeWo monolayers (Bomsel, unpublished), as reported for early primary trophoblasts (63). Viruses derived from cell-to-cell contact exhibited an X4 phenotype, irrespective of the initial predominant R5 or X4 phenotype of the virus propagated in effector PBMCs. This may be due to either a preferential fusion of X4-infected PBMCs with the BeWo monolayer or by a preferential replication of X4 viruses after fusion.

As expected, R5 viruses are preferentially selected in CD14⁺/CD3 cells derived from X4-infected PBMCs (Table 4). However, the fusion of these cells with BeWo cells resulted in the release of X4 virus. These data indicate that, despite the preferential tropism of R5 viruses for CD14⁺/CD3 cells, X4 minor variants are preferentially selected after cell-to-cell contact and fusion of the two cell types. These results further support the hypothesis proposed above of either a preferential fusion of X4-infected cells with BeWo cells or a preferential replication of X4 viruses in target cells. Finally, our results also demonstrated that both fusion of R5-infected cells (Fig. 3) and replication of R5 viruses (Table 4) are not restricted in BeWo cells.

Previous studies have shown that maternal HIV-1 variants are selected during mother-to-child infection (1, 48, 52). The data presented here suggest that a selection may first occur at two levels: during transcytosis and after trophoblast fusion with the effector cells resulting in productive infection. In the only case of in utero transmission that we studied up to now, the predominant transcytosed viral variant detected was genetically identical to the predominant env variant found in the infected child. This suggests that transcytosis may be partly involved in controlling a complex series of events which may restrict the transmission of HIV in utero. However, further investigations of additional cases of in utero and intrapartum transmissions are needed to confirm this hypothesis.

In summary, we presented several lines of evidence that our reconstitution system of a trophoblast barrier in vitro is a suitable model for dissecting the first steps leading to a restricted passage of HIV-1 in utero. The trophoblast barrier is not permeable to cell-free virus, but the contact between this barrier and infected cells results in the transcytosis of infectious virus and in the fusion between effector and target cells. Infected cells, either T cells or monocytes, fuse efficiently with the trophoblastic barrier and lead to the release of infectious virus into the basolateral side. The fusion process clearly induces a selection of viral variants, as reported in studies on in utero transmission in vivo (41). In addition, in vitro transcytosis of infectious virus may be relevant to the processes leading to in utero transmission of HIV.

Our genetic and phenotypic analysis of viruses crossing the trophoblast barrier by transcytosis or replication after fusion suggests that the selection of maternal viruses within the placenta may indeed involve multiple steps. The initial selection within trophoblastic cells is likely to be regulated by factors in the placental microenvironment, such as cytokines, chemokines, hormones, and maternal antibodies (as illustrated in Fig. 7). Other placental cells, such as macrophages (Hofbauer cells), which are present underneath the trophoblast in vivo, may also play a role in the selective process. Our in vitro model is a very useful tool to approach the selective forces within the placenta that lead to control of HIV-1 transmission in utero.

View larger version (43K): [in this window] [in a new window]   FIG. 7.   Model to approach the first steps of in utero transmission of HIV. Step 1: HIV-1-infected mononuclear effector cells representative of maternal blood cells are in contact with the apical side of the syncytiotrophoblast. This contact can initiate two events (steps 2 to 3 or steps 4 to 5). The first one is a fusion of infected effector target cells (step 2a) or a fusion between virions freshly released from infected effector cells (step 2b) and syncytiotrophoblast target cells, resulting in both cases in a productive infection (step 3), with viral progeny released into the basolateral fetal side (step 6). The second event is transcytosis of the virus (steps 4 and 5) and release of infectious virus into the fetal side (step 6). Maternal viruses translocated to the fetal side by either pathway (step 2 or 4) undergo a first series of selection followed by a further selective process (step 7) involving other placental cells, such as macrophages (Hofbauer cells), monocytes, or CD4+ T cells in the fetal side. Such selective forces may lead to the emergence of viral species that may or may not be transmissible to the fetus. Hormones, chemokines, and cytokines, in the placental microenvironment, could also regulate the two translocation pathways of cell-associated virus (step 8). In contrast, no translocation of infectious cell-free virus across the trophoblast barrier is detectable (step 9). However, infectious cell-free virus complexed with maternal specific IgG in the maternal blood should also be considered for a potential IgG receptor-mediated transcytosis pathway to reach the fetal side of the trophoblast layer (step 10).