The Trophoblastic Epithelial Barrier Is Not Infected in Full-Term Placentae of Human Immunodeficiency Virus-Seropositive Mothers Undergoing Antiretroviral Therapy

Abstract

To study the mechanism of the placental barrier function, we examined 10 matched samples of term placentae, cord blood, and maternal blood obtained at delivery from human immunodeficiency virus (HIV)-infected mothers with children diagnosed as HIV negative in Sweden. All placentae were histologically normal, and immunochemistry for HIV type 1 p24 and gp120 antigens was negative. Highly purified trophoblasts (93 to 99% purity) were negative for HIV DNA and RNA, indicating that the trophoblasts were uninfected. Although HIV DNA was detected in placenta-derived T lymphocytes and monocytes, microsatellite analysis showed that these cells were a mixture of maternal and fetal cells. Our study indicates that the placental barrier, i.e., the trophoblastic layer, is not HIV infected and, consequently, HIV infection of the fetus is likely to occur through other routes, such as breaks in the placental barrier.

Maternal-infant transmission of human immunodeficiency virus type 1 (HIV-1) is the primary cause of HIV-1 infection in children. The risk of mother-to-child transmission of HIV ranges from about 15% in Europe (12, 13) to 39% in Africa (35). The AIDS Clinical Trial Group Protocol 076 (ACTG076) demonstrated that a regimen of zidovudine to a selected group of HIV-infected pregnant women during the second and third trimesters of pregnancy along with administration of zidovudine to their newborns reduced the risk of perinatal transmission by two-thirds (7). This study demonstrated for the first time that perinatal HIV transmission can be prevented. However, this treatment is expensive and is not available in many parts of the world, such as sub-Saharan Africa and developing areas of the Americas where more than three-quarters of the perinatally infected children live (36). Available data suggest that most perinatal HIV transmissions occur at or just before delivery (5, 11, 14, 19), but cases of intrauterine transmission (8, 24, 37, 40) and transmission by breastfeeding (41) have also been described. It is important to obtain more detailed information about the timing and the mechanisms of perinatal HIV transmission, because such knowledge is likely to have a significant effect on future intervention strategies aiming at prevention of maternofetal transmission.

The placenta provides a potential barrier between the maternal and fetal circulations, but limited attention has been given to its role in the transmission of HIV. The human placenta is of the villous hemochorial type and consists of a vast array of fetal villi, which are bathed directly in the circulating maternal blood (17). At the end of the pregnancy, the villi represent a surface area of 10 to 14 m² and therefore permit extensive and intimate contact between fetal tissues and maternal blood. The outermost layer of the villi consists of syncytiotrophoblasts, which form a continuous multinucleated epithelium generated from and maintained by an underlying population of mononuclear cytotrophoblast cells. This trophoblastic layer is supported by a basement membrane, which separates it from the mesenchymal cells of the villous core. Within the core are the fetal capillaries and a significant number of macrophages, often referred to as Hofbauer cells, some of which express the CD4 molecule (30). Since trophoblastic cells constitute the external layer of chorionic villi, they are in direct contact with maternal blood which makes them a primary target for maternal blood-borne infections. However, the placental barrier is not complete, and there is evidence that bidirectional traffic of cells, including leukocytes, may occur in human pregnancy (32).

To determine if cells in the full-term placentae of HIV-seropositive mothers are infected, and if so, which cell type is affected, we prospectively collected 10 term placentae from HIV-infected mothers (9 infected with HIV-1 and 1 infected with HIV-2) from delivery wards in the Stockholm area (Sweden) between October 1996 and November 1997. These women were participants in a larger prospective multicenter study which evaluated factors influencing maternofetal transmission of HIV (13). Ethical approval for the study was obtained, along with informed consent from all women. Four mothers were infected with HIV-1 of genetic subtype C, three with subtype A, and two with subtype B (Table 1). Delivery occurred at term in all cases. Elective cesarean section was performed in six cases and vaginal delivery in four cases. One of these ended in emergency cesarean section because of signs of fetal asphyxia. Nine infants, more than 18 months old, have seroreverted and are uninfected. One child died of sudden infant death syndrome at 1 month of age. Virus isolation and DNA PCR using peripheral blood mononuclear cells (PBMC) as well as mesenteric lymph nodes of this child were negative for HIV-1.

TABLE 1.

Clinical, virological, and immunological characteristics of the mothers

Mother	Virus and genetic subtypea	Clinical statusb	CD4 cell count (cells/μl)	HIV-1 RNA levels in plasma (copies/ml)c	Virus isolation and biological phenotyped	Antiretroviral treatmente	Mode of delivery and gestational age (wk)f	HIV status of infantg
1	HIV-1, C	A2	210	37,600	−	AZT, ddI, 3TC	Elective cs, 37	Uninfected
2	HIV-1, C	A2	310	62,300	R5	AZT	Elective cs, 37	Uninfected
3	HIV-2	A1	729	NTh	−	AZT	Vaginal, 40	Uninfected
4	HIV-1, A	A1	927	<500	−	AZT	Emergency cs, 40	Uninfected
5	HIV-1, A	A2	210	<500	−	AZT	Elective cs, 38	Uninfected
6	HIV-1, C	A3	100	102,000	R5X4	AZT, 3TC	Elective cs, 38	Uninfected
7	HIV-1, A	A1	530	5,000	−	AZT	Vaginal, 39	Uninfected
8	HIV-1, B	A1	619	<200	−	AZT, 3TC	Vaginal, 36	Uninfected
9	HIV-1, C	A2	680	<200	−	AZT, 3TC	Elective cs, 37	Uninfected
10	HIV-1, B	A3	270	<50	−	AZT, 3TC, NVP	Elective cs, 37	Unknowni

^aDetermined by phylogenetic analysis of DNA sequences from the V3 domain of the gp120 envelope glycoprotein obtained from uncultured PBMC (23). 

^bClinical status according to the 1993 classification system of Centers for Disease Control and Prevention (7): A1, asymptomatic HIV infection with CD4 T-cell count of ≥500/μl; A2, asymptomatic HIV infection with CD4 T-cell count between 200 and 499/μl; A3, asymptomatic HIV infection with CD4 T-cell count of <200/μl. 

^cLevels of HIV-1 RNA in plasma were determined by the Amplicor HIV-1 Monitor test (Roche, Branchburg, N.J.) according to the manufacturer’s recommendations. 

^dVirus isolation (23) was attempted from maternal PBMC at delivery. −, negative virus isolation; R5, virus isolate using CCR5 as coreceptor; R5X4, virus isolate using CCR5 and CXCR4 as coreceptors. The virus isolates were tested for coreceptor usage on GHOST cells stably expressing CD4 and CXCR4, CCR3, CCR5, Bob, or Bonzo. 

^eAccording to the ACTG076 protocol or the indicated modifications. AZT, zidovudine; 3TC, lamivudine; NVP, nevirapine; ddI, didanosine. 

^fcs, cesarean section. 

^gHIV infection in the infant is confirmed with one positive HIV DNA PCR or one positive virus isolation or both at two different time points. 

^hNT, not tested, because no test is available for HIV-2. 

ⁱSudden infant death syndrome at 1 month of age. 

Isolation of placental trophoblastic cells.

Macroscopic and routine histologic examination of all freshly delivered placentae, extraplacental fetal membranes, and the umbilical cord did not reveal any gross abnormalities in any of the cases. Immunohistochemical examination of formalin-fixed, paraffin wax-embedded tissue did not detect HIV-1 p24 and gp120 antigen. Enriched trophoblastic cell populations were then isolated by using the method of Kliman et al. (20). Briefly, placental tissue was extensively washed with Hanks balanced salt solution (HBSS) (GibcoBRL, Life Technologies), minced, and digested with trypsin (Difco, Detroit, Mich.) and DNase I type IV (Sigma). The resulting cell suspensions were collected and subjected to discontinuous Percoll (Pharmacia) density gradient centrifugation at 1,200 × g at room temperature for 20 min to enrich for trophoblasts. The mononuclear cell fraction was recovered and washed twice with cold HBSS. This enriched trophoblastic cell preparation contained residual contaminating cells, mainly macrophages expressing CD14, granulocytes expressing CD45, endothelial cells expressing CD31, and blood elements such as T lymphocytes expressing CD3 (9).

The purity of trophoblasts was further increased by adding a new immunomagnetic purification step. Thus, highly purified trophoblasts were obtained by removal of contaminating cells from the enriched trophoblast preparations with immunomagnetic beads (Dynabeads M-450; Dynal, Oslo, Norway) coated with different monoclonal antibodies. By successive immunoelimination, we removed Hofbauer cells, monocytes and macrophages (mouse anti-human CD14, TÜK4; Dakopatts AB), and T lymphocytes (mouse anti-human CD3, pure Leu4; Becton Dickinson). Thereafter, in a third step, granulocytic cells (mouse anti-human CD45, T29/33; Dakopatts AB), endothelial cells (mouse anti-human CD31, JC/70A; Dakopatts AB), and the remaining macrophages and T lymphocytes were simultaneously removed.

The purity of the enriched and purified trophoblasts was evaluated by microscopic examination for morphology, as well as by flow cytometry performed after staining with mouse monoclonal antibodies GB25 raised against placental cyto- and syncytiotrophoblasts of the chorionic villi from full-term placentae (15) and GB17 raised against syncytiotrophoblasts of the chorionic villi from term and from first and second trimester placentae (16) and with a rabbit F(ab′)₂ anti-mouse immunoglobulin G fluorescein isothiocyanate-conjugated second antibody (Dakopatts AB). GB25 and GB17 were produced by hybridoma cells obtained following immunization of mice with isolated human term microvilli (15, 16) and were a gift from Gerard Chaouat (Hôpital Antoine-Béclère, Clamart, France). In addition, the presence of contaminating cells was evaluated with flow cytometry using monoclonal antibodies directed against CD3, CD14, CD45, CD31, and CD4 (CD4 BL4 pure; Immunotech, Marseille, France).

The trophoblastic cell preparations consisted mainly of cytotrophoblasts with high expression of GB25 and low expression of GB17 (data not shown). The mean (± standard error of the mean) percentage of trophoblastic cells expressing GB25 in the enriched (undepleted) cell preparations was 88.7 ± 6.6% and that in the purified depleted cell preparations was 95.4 ± 2.2%. Table 2 shows the purity of the enriched and purified trophoblastic cell preparations from the 10 placentae. The levels of contaminating cells, mainly T lymphocytes and macrophages, were monitored in the same way and were between 1 and 10% (Table 2).

TABLE 2.

Flow cytometry analysis of cell surface markers in trophoblast preparations

Monoclonal antibodies to antigens expressed in:	% Positive cells in placenta of indicated mother
	1	2	3	4	5	6	7	8	9	10
Enriched trophoblastsa
GB25c	83	76	NDd	92	89.9	85	88	97	92	95.8
CD3	9.6	2.9	ND	0.5	1.5	2	0.4	1.8	1.82	0.54
Purified trophoblastsb
GB25	96	93	ND	97	93	93	94	99	96.8	97.2
CD3	2.6	1.5	ND	0.3	1.1	1.6	0.2	1.6	1.2	0
CD14	1.2	0.3	ND	0	0	1.2	0.1	1.1	0.1	0
CD31	2.3	0.8	ND	0.3	0.5	0.1	0.2	1.4	1	0
CD45	1.5	1	ND	0.4	0	0.9	0.2	1.8	0.8	0
CD4	2.7	1.6	ND	0.4	0.1	0.9	0.5	3	3.5	3.0

^aEnriched mixed trophoblastic cell preparation after Percoll gradient. 

^bPurified trophoblastic cell preparation after immunomagnetic cell separations. 

^cGB25 is specific for syncytio- and cytotrophoblasts. 

^dND, not done. 

Virus isolation and detection of HIV DNA and RNA.

Virus isolation (23, 39) was attempted from extensively washed chorionic villous tissues of placentae 1, 2, 4, 5, and 6, from fetal membranes of placentae 4, 5, and 6, and from the CD14⁺ fractions of the enriched (undepleted) trophoblastic cell populations of placentae 2, 4, and 5. With the exception of the CD3⁺/CD14⁺ fraction of placenta 1, none of the cultures was virus isolation positive. In view of the very low viral loads and negative virus isolations even from the mothers’ PBMC, it is not surprising that no virus could be recovered from placental cells. This suggested that there is very little, if any virus present in the placentae studied. Therefore, we performed PCR for detection of HIV DNA in the different placental cell preparations as well as the mothers’ PBMC and the cord blood samples (CBMC). For PCR, cells were lysed (4) and nested PCR was performed in duplicate on 10 μl of cell lysate corresponding to 10⁵ cells, using primers JA79-JA82 and JA80-JA81 (1, 25), known to amplify diverse genetic subtypes within group M of HIV-1 (25). For HIV-2-infected patient 3, the primer sets JA47-JA52 and JA48-JA49 (pol) were used (33). The amplimers were visualized by ethidium bromide staining after electrophoresis in a 1.5% agarose gel. For each amplification, a negative control without DNA and a positive control consisting of 10 DNA copies of HIV-1_MN or HIV-2_K135 (3) were diluted into an HIV-negative PBMC lysate corresponding to 10⁵ cells. A minimum of two, but usually more, independent PCRs were performed on cell preparations with an initially negative PCR result. In addition, each sample was tested for the presence of human DNA by using the primer set PCO3-PCO4 detecting β-globin (31).

PCR-positive cell fractions were further analyzed by semiquantitative limiting dilution PCR analysis as described previously (4). Briefly, a fivefold dilution series was prepared from each PCR-positive cell fraction, and at least five independent PCRs were performed on aliquots containing 1 × 10⁵, 2 × 10⁴, 4 × 10³, and 8 × 10² cell equivalents of DNA. The number of HIV DNA copies per 100,000 cells was determined by using the Poisson distribution formula, according to which the precursor frequency (or DNA copy number) is the inverse fraction of the number of cell equivalents required to give 63% positive reactions (4).

The mothers’ PBMC were PCR positive, indicating that the primers used had the ability to amplify the DNA of the virus variants carried by these mothers (Table 3). The corresponding cord blood samples were all PCR negative. The enriched (undepleted) trophoblastic cell preparations from all 10 placentae were PCR positive, whereas all purified trophoblastic cell preparations were regularly negative (Table 3). In addition, the CD3⁺ cell fractions from all except two mothers were PCR positive, whereas most CD14⁺ cell fractions were negative. These results indicated that the HIV DNA sequences detected in the enriched (undepleted) trophoblastic cell preparations originated from contaminating CD3⁺ and CD14⁺ cells, since removal of these cells resulted in the loss of a PCR signal in the purified trophoblast preparation. For HIV-2-positive placenta 3 and for placenta 4, the weak PCR signal was lost during the cell fractionations (Table 3).

TABLE 3.

HIV DNA copies in the mothers’ PBMC and in placental cell fractions

	No. of HIV DNA copies/100,000 cells from indicated placentaa
	1	2	3	4	5	6	7	8	9	10
Mother’s PBMC	2.1	6.9	<1	8.2	10.2	1.6	2.0	8.2	3.3	6.7
Enriched trophoblasts	1.6	2.4	<1	<1	<1	1.1	<1	1.1	<1	<1
CD14-positive cell fraction	2.7b	+c	−	−	−	−	−	3.3	7.0	−
CD3-positive cell fraction	2.7b	2.4	−	−	2.2	5.6	1.8	4.4	4.4	2.7
CD31-CD45-CD3-CD14-positive cell fraction	−	−	−	−	−	−	−	−	2.2	<1
Purified trophoblasts	−	−	−	−	−	−	−	−	−	−
CBMC	NA	NA	NA	−	−	−	−	−	−	−

^aAt least 200,000 cells were tested, numbers indicate HIV proviral DNA copies/100,000 cells. For the purified trophoblastic cell preparations 400,000 cells from two different aliquots were tested. NA, not available; −, negative. 

^bThe CD3⁺ and CD14⁺ cell fraction depleted simultaneously from placenta 1. 

^cThe CD14-positive cell fraction from this placenta was kept in culture for 3 weeks before HIV-PCR was performed, and it tested positive. 

The enriched, but undepleted, trophoblastic cell preparations had lower HIV DNA levels than the corresponding PBMC. Similarly, the enriched (undepleted) trophoblasts had lower HIV DNA copy numbers than the CD3⁺ placental cell fraction in six out of the seven cases for which comparison was possible (Table 3). In fact, in most cases the HIV DNA copy numbers in CD3⁺ placental cells were comparable to those of PBMC. This indicates that the enriched (undepleted) trophoblast preparations carry HIV DNA due to the remaining CD3⁺ cells. The HIV DNA levels in PBMC or CD3⁺ placental cells showed no clear correlation with CD4⁺ lymphocyte counts or plasma HIV RNA levels of the mothers.

The purified (depleted) trophoblastic cells isolated from HIV-1-positive placentae were also screened for the presence of HIV-1 RNA by using reverse transcription-PCR (RT-PCR). Briefly, total RNA was extracted from purified trophoblastic cells using the Nuclisens method according to the manufacturer’s recommendation (Organon Teknika, Boxtel, The Netherlands) and reverse transcribed into cDNA (first-strand cDNA synthesis kit; Amersham Pharmacia Biotec, Uppsala, Sweden). Detection of HIV RNA was performed by using the previously described pol primers JA79-JA82 (1, 25) and was negative for all placentae. The efficiency of this method was tested on blood donor PBMC spiked with a known amount of viral RNA. These experiments showed that 30% of the RNA was successfully extracted, reverse transcribed, and PCR amplified (data not shown).

Genomic microsatellite analysis.

Genomic microsatellite analysis was used to determine the origin of the different cell types isolated from the placentae. By this method, cells from different individuals, such as a mother and her semiallogeneic child, can be distinguished by analysis of hypervariable regions of the human genome. The genotypes of the mother and the corresponding infant were determined by analysis of multiple loci by PCR amplification using ³³P-labelled dATPs for the loci D1S547, D1S1677, D1S1679, D1S2134, and RB20.1 or the 6-carboxyl-fluorescein (FAM)-labelled primers D5S346 (LNS-CA repeat marker) and D4S127 (CAG repeat in the Huntington disease gene) (38). The maternal and fetal cells were informative (heterozygous) for the respective microsatellite markers, so that the alleles could be identified (illustrated in Fig. 1). The different placental preparations were then also subjected to genomic microsatellite analyses of alleles which differed between the respective mother and infant (listed for each sample in Table 4). This allowed us to evaluate if the different cell fractions in the placenta were of maternal or fetal origin. We found that CD3⁺ and CD14⁺ cells from all the placentae were mixtures of maternal and fetal cells as demonstrated by the presence of three different alleles (Table 4). For placentae 3 and 4, there was not enough material from the CD3⁺ and CD14⁺ cell fraction to perform this analysis. Nevertheless, the enriched mixed trophoblastic cell populations from these placentae were shown to be of fetal origin. This finding indicates that microsatellite analysis does not detect small amounts of maternal contamination, i.e., contaminating CD3⁺ and CD14⁺ cells (6).

FIG. 1.

Origin of different cell populations in maternal blood (a), cord blood (b), placental T lymphocytes (c), and placental macrophages (d) in mother 6 using microsatellite analysis with noninformative marker D5S346 (A) and informative marker D4S127 (B). See Table 4, footnote b, for an explanation of the numbers 1, 2, and 3.

TABLE 4.

Origin of different cell populations in the placenta, cord blood, and maternal blood by genomic microsatellite analysis

	Presence of maternal or paternal alleles in indicated placentae and locia
	1	2	3	4	5	6	7	8	9	10
	D1S547 D1S1677 D1S1679 D1S2134	D5S346	D5S346	D1S547 D1S1677 D1S1679 D1S2134	D5S346	RB20.1 D4S127	D5S346	D1S547 D1S1677 D1S1679 D1S2134	D1S547 D1S1677 D1S1679 D1S2134	D1S547 D1S1677 D1S1679 D1S2134
PBMC
Mother	2-3b	2-3	2-3	2-3	2-3	2-3	2-3	2-3	2-3	2-3
CBMC	NA	NA	NA	1-3b	1-3	1-3	1-3	1-3	1-3	1-3
Placental cells
CD14	1-2-3c	1-2-3b	NT	NT	1-2-3	1-2-3	1-2-3	1-2-3	1-2-3	1-2-3
CD3	1-2-3c	1-2-3	NT	1-2-3	1-2-3	1-2-3	1-2-3	1-2-3	1-2-3	1-2-3
CD31-CD45-CD3-CD14	NT	NT	NT	NT	NT	NT	NT	1-2-3	1-2-3	1-2-3
Purified trophoblasts	1-3b	1-3	NT	NT	NT	NT	NT	1-3	1-3	1-3

^aInformative heterozygous markers for genomic microsatellite analysis. NT, not tested; NA, not available. 

^b2-3, maternal alleles; 1, paternal allele; 1-3, child’s pattern; 1-2-3, mixture of maternal and fetal cells. 

^cThe CD3⁺ and CD14⁺ cell fractions were analyzed together in placenta 1. 

In the present study, we show that highly purified primary placental trophoblasts obtained from HIV-seropositive mothers are negative upon PCR for HIV DNA and RNA. Our results demonstrate, for the first time, that villous trophoblasts in term placentae of mothers undergoing antiviral therapy are uninfected. These results are at variance with those of Lee et al. (22), who detected HIV gag sequences by RT-PCR in placental cell preparations from 30 HIV-seropositive mothers, of whom 23 were undergoing antiretroviral therapy throughout pregnancy and/or intrapartum. Lee et al. (22) obtained placental cells after only one round of immunomagnetic cell depletion using a single anti-CD45 monoclonal antibody. The purity of these trophoblast preparations was determined solely by morphology and was estimated to be on average 95.8%. As estimated by microscopy, contaminating leukocytes, macrophages, and granulocytes were present in less than 3%. In contrast, we carried out immunomagnetic cell separations in three sequential steps and used specific monoclonal antibodies for each cell type, such as placental macrophages or Hofbauer cells and blood monocytes (CD14), T lymphocytes (CD3), and residual leukocytes (CD45). In addition, the placental cell preparations were monitored at each step of immunomagnetic cell separation for the presence of trophoblast membrane antigen GB25. The purity of our enriched (through Percoll gradient only) trophoblast preparations was comparable to those of other studies (22, 29), and indeed HIV DNA could be amplified from all these trophoblast preparations. However and importantly, HIV DNA was not detected in our immunomagnetically purified trophoblast preparations. In fact, we were able to identify the HIV-positive cells as CD3⁺ or CD14⁺, that is, belonging to the fraction of contaminating T lymphocytes and macrophages. This is in line with the results of previous studies demonstrating that nontrophoblastic placental cells carry HIV infection in vivo and are susceptible to infection in vitro (2, 26–28). Taken together, the cell fractionation procedures whereby trophoblasts are obtained seem to be crucial (9) and the level of contaminating nontrophoblastic cells will strongly influence the outcome of HIV detection. Our results show that the highly purified trophoblasts are uninfected. This conclusion is supported by the findings of Kilani et al. (18), who found that pure placental trophoblasts resist infection by multiple cell-free primary HIV-1 isolates. In addition, the purified trophoblasts did not appear to express CD4, which is in agreement with previous findings (10, 21). These results are further corroborated by immunohistochemical data, which failed to detect any HIV-1 p24 or gp120 antigens in the examined placental tissues, including trophoblasts. Thus, the trophoblastic epithelium might be a significant barrier to transplacental HIV infection.

Another complicating factor in these types of experiments is the fact that, due to insufficient separation techniques, the nontrophoblastic cell fraction of the placenta consists of a mixture of maternal and fetal cells. Consequently, the demonstration of HIV infection in nontrophoblastic cells may simply be a reflection of the fact that the maternal blood in the placenta contains HIV-infected CD4⁺ lymphocytes and monocytes. In our study, we used two approaches to test the origin of HIV-infected cells. First, the level of HIV DNA in each cell fraction was evaluated in a semiquantitative, limiting dilution assay. For all mothers, PBMC and placental CD3⁺ T lymphocytes carried similar amounts of HIV-DNA, suggesting that the viral DNA was of maternal origin. Second, we determined the origin of cells in the different cell preparations by microsatellite analysis (6). Indeed, in our experiments we could confirm that the different nontrophoblastic placental fractions consisted of mixtures of maternal and fetal cells, whereas the trophoblasts were of fetal origin.

All mothers in our study were receiving zidovudine therapy according to the ACTG076 protocol, and some mothers were in addition receiving other reverse transcriptase inhibitors. Thus, the study population was highly representative of HIV-1-infected mothers in Europe and in the United States where most seropositive women are treated during pregnancy. Zidovudine undergoes activation through phosphorylation within trophoblasts and Hofbauer cells, but at a rate 50- to 100-fold lower than in lymphocytes (34). Thus, we cannot exclude the possibility that the absence of HIV-1 infection in the trophoblasts was influenced by the antiretroviral therapy. Since in our study all children appeared to be uninfected, we cannot formally exclude the possibility that trophoblasts are infected in those pregnancies which result in transmission of HIV infection to the fetus. However, in the study of Menu et al. (29), in which four women were transmitting and eight were nontransmitting, no correlation was found between the frequency of PCR positivity of enriched trophoblast cell preparations and transmission. Nevertheless, immunomagnetic cell depletion for further purification of the trophoblasts was not performed in this study (29).

Taken together, our results indicate that the trophoblastic barrier remains uninfected in full-term placentae of HIV-seropositive mothers undergoing antiretroviral therapy. We suggest that in utero HIV transmission, if at all, occurs at the end of gestation through alternative routes, such as chorioamnionitis with leakage of the virus into the amniotic cavity or trophoblast damage. This knowledge is important for the design of new, simpler intervention strategies aiming at the prevention of mother-to-child transmission of HIV.