Perinatal Transmission of Major, Minor, and Multiple Maternal Human Immunodeficiency Virus Type 1 Variants In Utero and Intrapartum

Abstract

Previous studies have provided conflicting data on the presence of selective pressures in the transmission of a homogeneous maternal viral subpopulation to the infant. Therefore, the purpose of this study was to definitively characterize the human immunodeficiency virus type 1 (HIV-1) quasispecies transmitted in utero and intrapartum. HIV-1 envelope gene diversity from peripheral blood mononuclear cells and plasma was measured during gestation and at delivery in mothers who did and did not transmit HIV perinatally by using a DNA heteroduplex mobility assay. Children were defined as infected in utero or intrapartum based on the timing of the first detection of HIV. Untreated transmitting mothers (n = 19) had significantly lower HIV-1 quasispecies diversity at delivery than untreated nontransmittting mothers (n = 18) (median Shannon entropy, 0.711 [0.642 to 0.816] versus 0.853 [0.762 to 0.925], P = 0.005). Eight mothers transmitted a single major env variant to their infants in utero, and one mother transmitted a single major env variant intrapartum. Four mothers transmitted multiple HIV-1 env variants to their infants in utero, and two mothers transmitted multiple env variants intrapartum. The remaining six intrapartum- and two in utero-infected infants had a homogeneous HIV-1 env quasispecies which did not comigrate with their mothers' bands at their first positive time point. In conclusion, in utero transmitters were more likely to transmit single or multiple major maternal viral variants. In contrast, intrapartum transmitters were more likely to transmit minor HIV-1 variants. These data indicate that different selective pressures, depending on the timing of transmission, may be involved in determining the pattern of maternal HIV-1 variant transmission.

Worldwide, human immunodeficiency virus type 1 (HIV-1) infection is spreading rapidly in women of childbearing age (15). In the United States alone, more than 100,000 women are currently infected with HIV-1 (31) and 2,000 infants are born to HIV-1-positive mothers each year (33). The epidemic of HIV-1 infection in women of childbearing age represents a serious threat to children, because more than 95% of infected children acquire HIV-1 from their mothers. Zidovudine (ZDV) treatment of infected mothers during gestation and at delivery and of their infants during the first 6 weeks of life has been shown to decrease the risk of perinatal transmission in treatment-naive, non-breast-feeding women by more than 70% (3, 8, 36). However, several factors, including advanced maternal disease status, the spread of antiretrovirus-resistant virus, noncompliance with therapy, and lack of an alternative to breast-feeding in certain populations, contribute to the continued spread of HIV-1 infection from mothers to their infants.

One major area of controversy in the field of pediatric HIV-1 infection has been whether selective transmission of a specific, homogeneous maternal viral population to the infant occurs in utero or intrapartum. In 1992, Wolinsky et al. published data indicating that the transmitted strain was a minor variant of the maternal viral pool (40). He and others have used such data to suggest the presence of selective pressure in determining which HIV-1 variants are transmitted (1). Data from other studies however, have shown a more random pattern of transmission of multiple and/or major maternal HIV-1 variants (22, 25). The lack of consensus among these studies may be due to the facts that each study compared only a few mother-infant pairs and the results were not correlated with the timing of transmission. Therefore, questions remain regarding the selective transmission of maternal HIV-1 variants in both in utero and intrapartum perinatal HIV-1 infection.

The constant evolution of HIV-1 in vivo results in the presence of numerous genetic strains referred to as quasispecies. HIV-1 variants arise in vivo due to the high mutation rate of reverse transcriptase and a short generation time and in response to diverse selective pressures (30, 37). The level of HIV-1 env gene diversity in vivo has been inversely correlated with the rate of disease progression in infected adults and children and provides an indication of the presence of selective pressures exerted by the host immune response (16, 39). The pressure exerted by maternal autologous neutralizing antibody (20) and cell-mediated immunity (28, 38) could result in the selective transmission of HIV-1 escape variants to the infant. However, the lack of a strong maternal neutralizing antibody response may also be correlated with increased viral load and, therefore, with the risk of random transmission (14).

In order to determine if selective pressure is indeed involved in in utero and/or intrapartum perinatal HIV-1 transmission, we conducted a genetic analysis of maternal and perinatally transmitted HIV-1 strains. Heteroduplex mobility analysis (HMA) (12) was used to determine the degree of HIV-1 envelope gene diversity in HIV-1-infected, transmitting and nontransmitting mothers at delivery. HIV-1 env region sequences in infected mother-infant pairs were compared by HMA to determine the degree of genetic relatedness between the maternal viral quasispecies and perinatally transmitted variants. Our results support the presence of selective pressure in perinatal HIV-1 transmission but suggest, however, that different pressures may be involved in determining which maternal env variants are transmitted in utero and intrapartum.

(This work was presented in part at the Seventh Conference on Retroviruses and Opportunistic Infections, San Francisco, California, January 2000.)

MATERIALS AND METHODS

Patient information and samples. (i) Mothers.

The 59 seropositive mothers studied were monitored as part of a prospective study of maternal-fetal HIV-1 transmission conducted by the Los Angeles Pediatric AIDS Consortium between May 1989 and August 1995. The mothers were chosen as study participants based on sample availability, including at least one pre-term (mean, 3 ± 2) and one time-of-delivery sample. Mothers were also chosen based on the availability of samples from their infants from within 48 h of delivery and sufficient clinical follow-up of both the mothers and their infants. Of a total of 116 HIV-1-infected mother-infant pairs monitored throughout the study period, 57 (13 transmitters and 44 nontransmitters) were excluded based on a lack of infant samples within the first 48 h following delivery (n = 23), lack of appropriate maternal samples (n = 20), and/or lack of sufficient clinical follow-up (n = 14). Samples were collected from patients with informed consent under the approval of the institutional review boards at each site participating in the study.

A total of 22 women received ZDV therapy during pregnancy and/or delivery as part of a clinical trial or according to current clinical guidelines. Treated women received oral ZDV (500 mg/day) prenatally beginning in the second or third trimester and as a constant infusion (2 mg/kg loading dose followed by 1 mg/kg/h) during labor and delivery.

(ii) Infected infants.

Samples were collected within 48 h of birth and at 2, 4, 6, 8, 10, and 12 weeks of age from 23 infants defined as infected following at least two positive HIV-1 cocultures from peripheral blood at two separate time points and confirmation of seropositive status beyond 15 months of age. According to the current working definition, infected infants with positive HIV-1 cocultures and/or PCRs (coculture/PCRs) within 48 h following birth were defined as infected in utero, whereas those with negative coculture/PCRs within 48 h of birth and with subsequent positive coculture/PCRs were defined as infected intrapartum (6). Two infected infants born to ZDV-treated mothers received ZDV during the 12 weeks of follow-up for this study. No infants were breast-fed.

Methods. (i) Sample preparation.

Ficoll-Hypaque density gradient centrifugation was used to prepare peripheral blood mononuclear cells (PBMCs) from heparinized (HIV-1 coculture) or EDTA (PCR and flow cytometry)-treated samples. Mononuclear cells were washed with normal saline twice and enumerated, and cells not used immediately were stored under liquid nitrogen. PBMC DNA and viral RNA in plasma were prepared from maternal and infant EDTA-treated samples by using QIAmp DNA and QIAmp viral RNA kits (Qiagen, Inc., Valencia, Calif.).

(ii) Quantitative PCR.

Quantitative HIV-1 RNA PCR was performed on maternal plasma samples by using the AMPLICOR HIV-1 MONITOR assay according to the manufacturer's instructions (Roche Molecular Systems, Somerville, N.J.). Samples in which HIV-1 RNA was not detected were assigned a value of 50 RNA copies/ml in the statistical analysis. Quantitation of HIV-1 DNA in PBMCs was performed by PCR, using previously published methods (13). Samples with copy numbers of less than 10 were assigned a value of 5 HIV-1 DNA copies/μg of PBMC DNA in the statistical analysis.

(iii) HIV-1 culture.

Quantitative PBMC cocultures were performed in 24-well plates according to the NIH/NIAID Clinical Trials Group consensus protocol (2, 13). Negative HIV-1 cocultures were assigned a value of 0.1 HIV-1-infected PBMC/10⁶ cells in the statistical analysis. Quantitative plasma cultures were performed on fresh (n = 49) and frozen (n = 10) plasma samples according to the NIH/NIAID Clinical Trials Group consensus protocol (2, 13). Negative plasma cultures were assigned a value of 1 tissue culture infective dose (TCID)/ml of plasma in the statistical analysis.

(iv) Flow cytometry.

Flow cytometric analysis of PBMCs was performed as previously published (18).

(v) HMA.

Reverse transcription of HIV-1 RNA purified from EDTA-treated plasma was carried out by using random hexamers from the GeneAmp RNA PCR kit according to the manufacturer's instructions (Perkin-Elmer Cetus, Emeryville, Calif.). In order to optimize the sensitivity of viral variant detection and to ensure proper sampling, the starting HIV-1 RNA and DNA copy numbers were determined by the methods listed above for each specimen used in the HMA (11). First-round nested PCRs consisted of ≤1 μg of PBMC DNA (starting copy number of 30 to 40) or cDNA derived from ≤200 μl of plasma (starting copy number of 500 to 1,000), 1.7 mM MgCl₂, 0.2 μM concentrations of each primer in 50 μM KCl, 10 mM Tris (pH 8.3), 200 μM concentrations of each nucleoside triphosphate, and 2.5 U of Taq DNA polymerase (Perkin-Elmer Cetus) in a final volume of 50 μl. Amplifications were carried out in a DNA Thermal Cycler 9600 (Perkin-Elmer Cetus) for 35 cycles consisting of 10 s at 95°C, 10 s at 62°C, and 30 s at 72°C. The first-round PCR primer pair was RedE1 (5′ ATTATGGGGTACCTGTGTGG 3′, HXB2R position 5886) and RedE2 (5′ CTGCACCACTCTTCTCTTAG 3′, HXB2R position 7288) (32). For patients with fewer than 20 copies of HIV-1 DNA/μg of PBMC DNA, multiple first-round PCR amplifications were performed and the independent runs were combined. The first-round PCR products were concentrated in QIAquick PCR columns (Qiagen), and the entire purified PCR product was used in the second-round PCR amplification in order to prevent the loss of any sequence diversity. Second-round PCRs were carried out as described above, using primer pair RedE3 (5′ TAGGCCAGTAGTATCAACTC 3′, HXB2R position 6523) and RedE4 (5′ GACTTCTCCAATTGTCCCTC 3′, HXB2R position 7195), generating a final PCR product approximately 690 bp in length.

Heteroduplexes were formed by denaturing 6 μl of the second-round PCR product at 95°C for 2 min in 1× annealing buffer (10 mM Tris-Cl [pH 7.5], 100 mM NaCl, and 5 mM EDTA) and quickly cooling the specimens on ice. Loading buffer (10× = 4.2% Orange-G and 25% Ficoll-400) was added, and heteroduplexes were resolved on nondenaturing 5% polyacrylamide gels (acrylamide-bis, 29:1) in Tris-borate-EDTA for 2.75 h at 250 V. Gels were stained with ethidium bromide (0.5 μg/ml), UV illuminated, and photographed.

(vi) Data analysis.

HMA gel photographs were scanned by using Adobe Photoshop (Adobe Inc., San Jose, Calif.) and stored as binary TIFF files. Gel lane scans were performed on a PowerMac computer, using the public domain NIH Image program (developed at the National Institutes of Health and available on the Internet at http://rsb.info.nih.gov/nih-image). Lane scans of equal length (144 pixels) were recorded from immediately below the single-stranded DNA to immediately below the homoduplex. The quasispecies diversity, or sequence heterogeneity, for each sample was estimated by calculating the normalized Shannon entropy using the NIH Image Entropy tool (10). The quasispecies sequence divergence, or percent base pair mismatch in reannealed strands, for each sample was estimated by calculating the median mobility shift (MMS) with the NIH Image Distance tool (10).

(vii) Statistics.

Descriptive statistics are provided as median (25th and 75th percentile). Correlations between individual variables were determined using a Spearman rank correlation coefficient (rho). Maternal, virologic, and immunologic variables between groups were compared using the Mann-Whitney U, χ², and Fisher's exact tests. P values of less than or equal to 0.05 were considered significant.

RESULTS

Patient characteristics.

Samples were available for analysis from 59 HIV-1-infected mothers and their infants. The majority of the infected women enrolled in our study were asymptomatic (85% belonged to CDC class A), with a median (25th and 75th percentile) CD4⁺ T-cell count at delivery of 500 (322 and 687)/mm³. Fourteen mothers transmitted HIV-1 to their infants in utero based on positive HIV-1 coculture and PCR within 48 h of delivery. Nine mothers transmitted HIV-1 to their infants intrapartum based on negative HIV-1 coculture and PCR within 48 h of delivery with subsequent follow-up positives (2 to 6 weeks after delivery). The thirty-six infants born to the nontransmitters in our study were defined as uninfected based on seroreversion by enzyme-linked immunosorbent assay or Western blotting.

HIV-1 quasispecies diversity in mothers at delivery.

The degree of differentiation of the maternal HIV-1 quasispecies was assessed by HMA performed on cell-associated and cell-free virus from transmitting and nontransmitting mothers at delivery. An approximately 690-bp nested PCR product encompassing the V3 to V5 regions of the HIV-1 gp120 gene was amplified from each mother and used to form heteroduplexes. Figure 1 shows representative patterns of HIV-1 homo- or heteroduplex migration observed in 16 HIV-1-infected transmitting and nontransmitting mothers at delivery. The majority of mothers studied showed the presence of multiple, slowly migrating heteroduplexes formed from amplified PBMC proviral DNA at delivery (83%; Fig. 1). Ten mothers (5 transmitters and 5 nontransmitters) with low CD4 counts (median, 190/mm³) had very low levels of proviral DNA env gene diversity as exhibited by the presence of a single, rapidly migrating homoduplex. Of note, all five nontransmitters with a single homoduplex received ZDV during gestation. As seen in Fig. 1, plasma-derived viral RNA showed a somewhat lower degree of differentiation than the proviral DNA, with 75% of infected mothers exhibiting slowly migrating cDNA heteroduplexes at delivery.

FIG. 1.

HMA comparison of cell-associated and cell-free HIV-1 V3-V5 env gene regions in nontransmitting (A) and transmitting (B) mothers at delivery. ΦX, DNA molecular weight markers (Gibco BRL, Grand Island, N.Y.). PB, PBMC-derived viral DNA PCR products; PL, plasma-derived viral cDNA PCR products. The sizes (in base pairs) of the molecular size markers are located to the right of the gel. SS, single-stranded DNA; CD4, maternal CD4 cell count. ZDV indicates treatment (+) or absence of treatment (−) in nontransmitting mothers. None of the transmitting mothers shown here received ZDV treatment. TIM, in utero (IU) or intrapartum (IP) transmission.

In order to better characterize maternal HIV-1 quasispecies complexity, each HMA gel lane was scanned and analyzed with the Image software package available as freeware from NIH. The Image Entropy tool was used to estimate HIV-1 env gene quasispecies diversity by examining the spread of V3-V5 proviral DNA and plasma RNA homo- or heteroduplex pixel intensities and calculating a normalized Shannon entropy (E) for each gel lane. The normalized Shannon entropy was read on a scale of 0.0 to 1.0, with increasing values indicating increasing numbers of heteroduplex bands. A second analytical examination of quasispecies complexity was performed with the NIH Image software distance tool. This method involves evaluation of heteroduplex electrophoretic mobility shifts which are proportional to the sequence difference between the reannealed strands. The MMS is also read on a scale of from 0.0 to 1.0, with increasing MMS indicating increased sequence divergence present within the quasispecies.

HIV-1 proviral DNA quasispecies entropy correlated well with maternal CD4⁺ cell count (rho = 0.58, P = <0.001) and showed an inverse correlation with maternal plasma HIV-1 RNA levels at delivery (rho = −0.40, P = 0.004). HIV-1 RNA quasispecies entropy among mothers at delivery also correlated with maternal CD4⁺ cell count (rho = 0.45, P = 0.003) and was inversely correlated with maternal plasma HIV-1 RNA levels (rho = −0.40, P = 0.004). The HIV-1 proviral DNA and plasma RNA MMS correlated well with maternal CD4⁺ cell count (DNA MMS/CD4, rho = 0.426, P = 0.003; RNA MMS/CD4, rho = 0.409, P = 0.004) and both showed an inverse correlation with HIV-1 RNA levels in maternal plasma at delivery (DNA MMS/RNA, rho = −0.401, P = 0.005; RNA MMS/RNA, rho = −0.349, P = 0.010).

Figure 2A shows the normalized Shannon entropy for PBMC-derived proviral DNA V3-V5 env region heteroduplexes from each mother in our study at delivery. While the median HIV-1 DNA quasispecies entropy at delivery was lower in transmitters compared to all 36 nontransmitters studied, the observed trend did not reach statistical significance due to the large overlap between groups (Fig. 2A; P = 0.053). The median rate of HIV-1 RNA quasispecies entropy in plasma was lower than the rate of proviral DNA quasispecies entropy among both transmitters and nontransmitters, and the difference between these two groups was not significant (E = 0.595 [0.543 and 0.740] versus 0.692 [0.567 and 0.778], P= 0.283).

FIG. 2.

Maternal HIV-1 quasispecies diversity and divergence at delivery. (A) Quantitation of cell-associated viral env gene diversity as determined by entropy analysis of HMA gel lanes for mothers who did and did not transmit HIV-1 to their infants perinatally. (B) Quantitation of cell-associated viral env gene divergence in transmitting and nontransmitting mothers. Mothers who received ZDV during gestation and/or at delivery are indicated by open circles, mothers who did not receive any ZDV during gestation, labor, and delivery are indicated by closed circles. Horizontal lines indicate the median for each measured variable. ○, ZDV; ●, no ZDV; ___, median.

Figure 2B shows the MMS for PBMC-derived proviral DNA heteroduplexes at delivery from each mother in our study. Similar to our observations for HIV entropy, neither the median HIV-1 DNA nor RNA quasispecies MMS at delivery showed a statistically significant difference when all transmitters and nontransmitters, including those that received ZDV, were compared (data not shown).

Most of the overlap in DNA and RNA quasispecies entropy and MMS occurred between transmitters and ZDV-treated nontransmitters (Fig. 2). ZDV is known to significantly reduce the risk of perinatal transmission, which could confound the interpretation of our results (3, 35). We therefore evaluated the data from all mothers who did not receive ZDV in order to determine the relationship between HIV diversity and perinatal transmission in the absence of treatment. As shown in Table 1, exclusion of ZDV-treated women (18 nontransmittters and 4 transmitters) from the statistical comparisons indicated that untreated transmitters had significantly lower levels of HIV-1 proviral DNA and viral RNA diversity and divergence in plasma than untreated nontransmitters at delivery. Transmitters who did not receive ZDV also had significantly higher levels of HIV-1 RNA in plasma, higher levels of HIV-infected PBMCs, higher levels of viremia in plasma, and significantly lower CD4⁺ cell counts than untreated nontransmitters at delivery.

TABLE 1.

Biologic variables in HIV-1-infected ZDV-untreated transmitting and nontransmitting mothers at delivery

	Transmitters		Nontransmitters		Pa
	Median (25th and 75th percentile)	No.	Median (25th and 75th percentile)	No.
DNA quasispecies entropy	0.711 (0.642 and 0.816)	19	0.853 (0.762 and 0.925)	18	0.005
DNA MMS	0.395 (0.168 and 0.559)	19	0.631 (0.403 and 0.697)	18	0.006
RNA quasispecies entropy	0.619 (0.543 and 0.740)	19	0.759 (0.715 and 0.878)	18	0.007
RNA MMS	0.309 (0.152 and 0.388)	19	0.483 (0.333 and 0.588)	18	0.003
HIV RNA copies/ml	61,960 (28,855 and 140,565)	19	5,757 (1,509 and 14,572)	18	<0.001
HIV DNA copies/μg of PBMC DNA	233 (69 and 326)	19	20 (5 and 26)	18	<0.001
Infected PBMCs/106 cells	25 (25 and 125)	18	1 (1 and 20)	18	<0.001
Plasma viremia (TCID/ml)	10 (1 and 10)	19	1 (1 and 1)	18	0.013
CD4 count/mm3 of blood	398 (227 and 698)	18	726 (595 and 967)	18	0.002

^aP values were calculated by using the Mann-Whitney U test. 

It is of interest that we evaluated the 18 nontransmitters in our study who received ZDV and found they had a significantly lower median CD4 count (451 [204 to 592]/mm³) compared to that of untreated nontransmitters (726 [545 to 967]/mm³, P < 0.001) at delivery. The lower CD4 count observed in these treated nontransmitters was reflected by their low levels of HIV-1 proviral DNA and viral RNA diversities and divergences in plasma, which were not significantly different from values obtained from the transmitters in our study (DNA E = 0.685 [0.548 and 0.849], RNA E = 0.575 [0.441 and 0.685]; DNA MMS = 0.407 [0.164 and 0.545]; RNA MMS = 0.272 [0.147 and 0.306]). Despite their lower CD4 counts, ZDV-treated nontransmitters had HIV-1 RNA copy numbers at delivery similar to those of untreated nontransmitters (11,026 [3,801 and 24,583] versus 5,757 [1,509 and 14,572] copies/ml, P = 0.197) and significantly lower than those of transmitters (65,516 [42,729 and 187,583], P < 0.001). Thus, ZDV treatment reduced levels of HIV RNA in plasma but did not have a significant effect on maternal CD4 count or HIV quasispecies diversity and/or divergence. These results suggest that many of the nontransmitters in our study would otherwise have been at high risk for perinatal transmission had they not received ZDV therapy.

HIV-1 diversity and the timing of transmission.

The results of our study also showed that women who transmitted HIV-1 infection to their infants in utero had significantly lower rates of HIV-1 proviral DNA and viral RNA env gene diversity and divergence in plasma at delivery as well as significantly lower CD4 counts than women who transmitted the virus intrapartum (Table 2). At delivery, mothers who transmitted HIV-1 to their infants in utero were significantly less likely to have multiple distinct HIV-1 proviral DNA env strains as exhibited by the presence of a single rapidly migrating homoduplex than mothers who transmitted intrapartum (5 of 14 versus 0 of 9, P = 0.043; Fig. 1B).

TABLE 2.

Biologic variables in HIV-1-infected transmitting mothers according to the timing of transmission

	In Utero		Intrapartum		Pa
	Median (25th and 75th percentile)	No.	Median (25th and 75th percentile)	No.
DNA quasispecies entropy	0.648 (0.510 and 0.718)	14	0.803 (0.703 and 0.842)	9	0.020
DNA MMS	0.279 (0.080 and 0.427)	14	0.500 (0.385 and 0.618)	9	0.023
RNA quasispecies entropy	0.560 (0.541 and 0.619)	14	0.697 (0.653 and 0.816)	9	0.020
RNA MMS	0.214 (0.107 and 0.343)	14	0.384 (0.299 and 0.462)	9	0.025
HIV RNA copies/ml	146,098 (95,207 and 340,050)	14	54,307 (36,284 and 75,460)	9	0.022
HIV DNA copies/μg of PBMC DNA	327 (149 and 350)	14	184 (107 and 247)	9	0.042
Infected PBMCs/106 cells	75 (25 and 125)	13	25 (15 and 75)	9	0.214
Plasma viremia (TCID/ml)	10 (1 and 10)	14	6 (1 and 10)	9	0.466
CD4 count/mm3 of blood	259 (198 and 628)	13	503 (426 and 759)	9	0.023

^aP values were calculated by using the Mann-Whitney U test. 

The intrapartum transmitters had CD4 counts, proviral DNA and plasma RNA entropy, and MMS levels at delivery similar to those among the ZDV-untreated women who did not transmit in our study (CD4 count = 503 [426 and 759] versus 726 [595 and 967] CD4⁺ cells/mm³, P = 0.231; DNA E, P = 0.136; RNA E, P = 0.304; DNA MMS, P = 0.208; RNA MMS, P = 0.083). In fact, the only significant difference between the intrapartum transmitters and ZDV-untreated nontransmitters was in their median levels of HIV-1 RNA in plasma at delivery (54,307 [36,284 and 75,469] versus 5,757 [1,509 and 14,572] HIV-1 RNA copies/ml of plasma, P = 0.001). These results highlight the major differences between the in utero and intrapartum transmitters who show more similarities with ZDV-untreated nontransmitters except for viral load in plasma.

Pattern of maternal viral quasispecies transmission.

In order to characterize the pattern of maternal HIV-1 quasispecies transmission in utero and intrapartum, PBMCs and plasma-derived HIV-1 env gene PCR fragments derived from the 23 transmitters in our study were analyzed alongside those of their infants. Figures 3 and 4 show representative examples of the maternal-infant HMA comparisons performed in our study. Seventeen mothers transmitted a single HIV-1 env variant to their infants as evidenced by a single infant HIV-1 env homoduplex generated from PBMCs and plasma at their first positive time point (Fig. 3A and C and 4A and C). The remaining six mothers transmitted multiple distinct HIV-1 env strains to their infants as evidenced by the presence of multiple slowly migrating heteroduplexes derived from infant PBMCs (n = 6 of 6) and plasma (n = 2 of 6) at their first positive time point (Fig. 3B and 4B).

FIG. 3.

Comparison of maternal and infant viral env gene HMA pattern in three in utero-transmitting pairs. Time point: T2, second trimester; T3, third trimester; D, delivery; DOB, date of birth; 12 weeks, 12 weeks of age of the infant. Source: M, mother; I, infant. M+I, intersample comparisons of maternal and infant viral env gene strains obtained on the date of birth.

FIG. 4.

Comparison of maternal and infant viral env gene HMA pattern in three intrapartum-transmitting pairs. See the legend to Fig. 3 for an explanation of abbreviations. M+I, intersample comparisons of viral env gene strains obtained from the mother at delivery and from the infant at the time of the first positive HIV-1 test.

Maternal and infant HIV-1 env gene PCR products were also combined and analyzed in order to determine the degree of relatedness between their viral strains (Fig. 3 and 4). Eight mothers transmitted a single major viral envelope gene strain to their infants in utero, and one mother transmitted a single major variant intrapartum, as indicated by the presence of a single infant HIV-1 homoduplex that comigrated with the maternal homoduplex at delivery and the absence of any new mother-infant heteroduplexes (Fig. 3A and 4A). Four mothers transmitted multiple major HIV-1 env variants in utero, and two mothers transmitted multiple major env variants intrapartum (Fig. 3B and 4B). At the first positive time point, the remaining two infants infected in utero and six infants infected intrapartum had a single minor HIV-1 env variant which did not comigrate with maternal bands at delivery (Fig. 3C and 4C).

Table 3 summarizes the pattern of maternal HIV-1 quasispecies transmission among the 23 infected mother-infant pairs in our study. Overall, in utero transmitters were significantly more likely to transmit either single or multiple major HIV-1 env variants than intrapartum transmitters (12 of 14 versus 3 of 9, P = 0.01; Table 3). In contrast, intrapartum transmitters were significantly more likely to transmit a single minor HIV-1 variant to their infants.

TABLE 3.

HIV-1 quasispecies transmission in utero and intrapartum

Maternal env variant transmitted	No. (%) transmitted in utero	No. (%) transmitted intrapartum	Pa
Major	8 (57)	1 (11)	0.025
Multiple	4 (29)	2 (22)
Minor	2 (14)	6 (67)

^aP value was calculated by using the χ² test. 

Viral env gene evolution over time.

HIV-1-infected transmitting mothers and their infants were monitored over time in order to determine the pattern of viral env gene evolution during gestation and shortly after birth. As shown in Fig. 3 and 4, little to no HIV-1 env gene evolution was demonstrated by HMA throughout gestation in the majority of transmitting mothers. Nine transmitters showed noticeable but slight changes in HIV-1 env gene HMA banding patterns over time. One in utero transmitter had a decrease in the total number of HMA bands, indicating a decrease in the number of distinct viral env gene strains present. Eight transmitters (two in utero and six intrapartum) showed an increase in viral env gene divergence throughout the gestation period; however, these changes were small (Fig. 3C and 4C). Based on these nine mothers, both in utero- and intrapartum-transmitted strains matched most closely maternal strains at delivery. However, multiple closely spaced samples from the third trimester were not available to further elucidate the timing of transmission.

In contrast to their mothers, several infected infants showed dramatic changes in both HIV-1 env gene diversity and divergence during the first 12 weeks following delivery. While 8 HIV-1-infected infants monitored from birth through 12 weeks showed constant levels of HIV-1 env gene diversity and divergence (Fig. 3A), 5 infants showed decreases (Fig. 3B and 4B), and 10 showed increases (Fig. 3C, 4A, and 4C). These results indicate that rapid evolution of HIV-1 env gene sequences can occur early in perinatal infection.

Comparison of HIV-1 populations in plasma and PBMCs.

To assess the degree of relatedness between cell-associated and cell-free viral populations, env gene PCR products derived from PBMCs and plasma were combined and analyzed by HMA. In the majority of transmitting mothers (91.3%), the PBMC- and plasma-derived env PCR products comigrated without the formation of novel heteroduplexes on the HMA gels, indicating a high degree of genetic relatedness (data not shown). In 21 of 23 infected infants, the plasma- and PBMC-derived viral env gene populations also showed a high degree of genetic relatedness as indicated by the formation of homoduplexes with or without a small amount of smearing (data not shown).

Two transmitting mothers (one in utero and one intrapartum) in our study had distinct viral populations in their PBMCs and plasma at delivery as evidenced by the formation of new heteroduplexes on HMA gels (Fig. 5). Both of the infants born to these women also possessed distinct HIV-1 env strains in their PBMCs and plasma at their first positive time point (Fig. 5). Cross-mixing of maternal and infant PBMC- and plasma-derived env gene PCR products followed by HMA analysis indicated that the cell-associated virus present in these two infected infants at their first positive time point (one at birth and one at 2 weeks of age) was most closely related to the virus in their mothers' PBMCs at delivery. Likewise, the plasma-derived viral env gene PCR products from these two infected infants were most closely related to the viral env gene products derived from maternal plasma at delivery, suggesting that both cell-associated and cell-free transmission of HIV-1 can occur in utero and intrapartum (Fig. 5).

FIG. 5.

Comparison of cell-associated and cell-free HIV-1 populations in an intrapartum-transmitting mother-infant pair. See the legend to Fig. 3 for an explanation of the abbreviations used. Maternal samples were collected at delivery, and infant samples were collected at the first positive time point (2 weeks).

DISCUSSION

Virologic and immunologic factors associated with an increased risk of perinatal transmission include high maternal virus load at delivery, advanced maternal disease status, and low maternal CD4⁺ cell count (7, 14, 17, 34, 35). In the present study, HMAs of maternal PBMCs and viral PCR products in plasma indicated that HIV-1 env gene diversity at delivery is also associated with the risk of perinatal transmission. Viral env diversity is generated in the presence of host immune pressures, necessitating genetic evolution in order to ensure viral escape. Zhang et al., in a study of adults with primary HIV-1 infection, found no HIV-1 env gene diversity prior to seroconversion (41). Previous studies have also indicated a correlation between host immune control of virus replication, the development of HIV-1 env gene diversity, and a lower rate of disease progression (10, 11, 16, 39). The results of our study provide evidence for the importance of maternal immune control of virus replication, as evidenced by HIV-1 RNA levels and HIV-1 env gene diversity, in prevention of perinatal HIV-1 transmission in utero.

While mothers who transmitted HIV-1 to their infants in utero exhibited lower levels of viral env gene diversity, intrapartum transmitters had diversity levels similar to those of ZDV-untreated nontransmitting mothers. The higher CD4 counts and high levels of env gene diversity observed in these mothers indicate that intrapartum transmission can occur even in the presence of an active maternal immune response. Possible factors that preclude in utero transmission, including the presence of maternal anti-HIV-1 antibodies and cell-mediated immunity, may be overwhelmed by obstetrical events during delivery and may lead to intrapartum transmission (21, 23).

The majority of ZDV-treated women in our study were enrolled prior to the publication of ACTG 076 (8); thus, treatment of these mothers was initiated prior to or during gestation based on maternal clinical parameters. ZDV treatment during gestation decreased HIV-1 RNA copy numbers in maternal plasma to levels similar to those seen in untreated nontransmitters. However, a concomitant significant increase in maternal CD4 count and HIV-1 env gene diversity was not observed. These data indicate that low maternal HIV-1 env gene diversity is not in and of itself sufficient to promote perinatal HIV-1 transmission. Rather, our results emphasize the importance of maintaining low maternal viral loads via either immunologic and/or antiretroviral control strategies in order to minimize the risk of perinatal transmission.

While 78% of transmitting mothers possessed multiple distinct HIV-1 env strains during gestation and at delivery, 74% of their infants were infected with a single env variant at their first positive time point. Even among infants infected with multiple HIV-1 viral strains, the V3-V5 region analyzed by HMA was found to be more homogeneous than that of their mothers at delivery. These data provide strong evidence for the presence of selective pressures, resulting in the transmission of a limited number of HIV-1 quasispecies via both the in utero and intrapartum routes of perinatal transmission. In addition, the difference observed in the pattern of maternal HIV-1 quasispecies transmission according to the timing of transmission indicates that different selective pressures may be involved in determining which variants are transmitted in utero and intrapartum.

Possible selective pressures influencing perinatal HIV-1 transmission include maternal viral phenotype, tropism, coreceptor usage, maternal neutralizing antibody, and/or other immune pressures (19, 24, 27, 38). In the present study, low maternal CD4 counts and low rates of HIV-1 env gene diversity were observed primarily in mothers transmitting major env variants to their infants in utero. Previously, we reported that in utero HIV-1 transmission is also associated with a weak maternal autologous neutralizing antibody response (5). These results, in combination with those of the present study, suggest that poor maternal immunologic control of viral replication can lead to perinatal transmission of major HIV-1 variants.

In contrast, the intrapartum transmitters studied here had lower HIV-1 RNA levels and more HIV-1 env gene diversity, suggesting the presence of a more active antiviral immune response, which may have led to transmission of minor escape variants to the infant. An alternative explanation is that minor variants arise due to outgrowth of a rapidly evolving population within the infant. In fact, among infants infected both in utero and intrapartum with minor maternal HIV-1 variants at their first positive time point, a rapid rate of viral env gene evolution was observed throughout the 12-week follow-up period. Further studies are needed to determine if the rapid rate of HIV-1 env gene evolution observed in these infants is due to the presence of passively transferred maternal antibodies and/or to an active infant antiviral immune response.

We attempted to use HMA analysis to determine the time of in utero versus intrapartum transmission as early or late in gestation. Unfortunately, little to no significant maternal HIV-1 env gene evolution was observed during gestation among most of the transmitters studied. This may be due to the additional amount of immune suppression that occurs during pregnancy or to the presence of a weak antiviral immune response as indicated by low rates of HIV-1 env diversity in the majority of transmitting mothers. While the divergence among HIV-1 env strains increased during gestation in several transmitters, evidence for the generation of novel env strains was extremely limited and heteroduplex patterns indicated that early viral strains persisted throughout pregnancy. Thus, while more frequent maternal sampling might allow further clarification, the small amount of virus evolution observed during pregnancy and the persistence of early strains indicate that it may not be possible to determine exactly the timing of perinatal transmission by using molecular methods (9).

Similar difficulties were encountered in our attempt to identify the source of viral inoculum transmitted to the infant as from maternal PBMCs or plasma. HMA comparisons of PBMC and plasma viral populations in transmitting mothers and their infants indicated that viral env strains in plasma existed as major or minor subpopulations of strains present in PBMCs. Some evidence was found to support the genetic relationships between maternal and infant PBMC viral strains and maternal and infant viral strains in plasma. However, it is not clear if this linkage is due to transmission via both PBMCs and plasma or if transmission of multiple PBMC strains followed by seeding of plasma with replication-competent strains occurred.

Among the six intrapartum-transmitting mother-infant pairs in which infection with a minor maternal HIV-1 strain was found to have occurred, neither maternal PBMCs nor maternal env strains in plasma formed a close match with infant env strains. Previous studies have indicated that mucosal membranes, including the cervix, harbor HIV-1 envelope variants distinct from those found in PBMCs and plasma (26, 29). Therefore, it may be that maternal cervical secretions provided the source of HIV-1 inoculum in these cases; however, we did not have appropriate samples available to test this hypothesis. However, the lack of a good match between the intrapartum-transmitted mother and infant viral strains may also have been due to outgrowth of virus capable of escaping passively transferred antibody and/or with a greater replication potential.

Perinatal HIV-1 transmission is a complex multifactorial process that remains poorly understood. We conducted the first published prospective study of HIV-1 quasispecies transmission in mother-infant pairs with a known definition of the timing of transmission. In contrast to those of several previous reports (1, 25, 40), our data suggest the involvement of multiple mechanisms in perinatal HIV-1 transmission (4) which may differ by the timing of transmission. In addition, the results of the present study indicate that transmission of major, minor, and multiple maternal HIV-1 strains can occur both in utero and intrapartum. Evidence was also found to support the potential importance of different selective pressures during in utero and intrapartum transmission. This paper extends our understanding of factors influencing perinatal transmission both in utero and intrapartum; however, further study of the mechanisms influencing selective transmission of the maternal HIV-1 quasispecies is clearly indicated.