The first molecular studies of env sequences diversity issued from infected individuals, adults as well as children, have shown that most of acute/recent infections are characterized by the presence of a highly homogenous genetically-restricted virus population in contrast to the high genetic diversity observed several years later at time of chronic infection 52 53 54 55 56 57 58 59 60 61 62 . These observations led rapidly to the assumption of a substantial bottleneck in virus transmission (Figure 2A). Recently, the use of single genome amplification (SGA) of HIV-1 plasma viral RNA obtained from acutely infected adult individuals, allowed the identification of transmitted/early founder viruses, and the precise estimation of their diversity 63 64 65 66 67 . These studies showed evidence of infection by a single virus in ~80% of heterosexuals and ~60% of HIV-infected men who have sex with men 63 64 66 67 . In contrast, the frequency of multiple variants transmission was found to be higher among intravenous (IV) drug users (~60%), including one subject who was infected by at least 16 variants 65 . The high frequency of mutiple-variants transmission in IV drug users could be due to the absence of a mucosal barrier to virus transmission and higher virus inocula. To date, such large studies have not been conducted on samples from children born to infected mothers. However this question has been considered using different molecular approaches (Table 1) (Figure 2B) 61 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 . Data vary from one study to another but all together, comparing the viral population in mother-child pairs, they showed a homogeneous genetically-restricted population in the majority of infected infants (156 of 235; table I), suggesting a selective transmission of some viral variants during MTCT (Figure 2B). The lack of consensus among these studies may be due to the fact that many studies compared only a limited number of mother-child pairs and did not address the route of transmission (in utero, perinatally or postpartum through early breasfeeding). In addition, the ages of the infants at time of sample collection varied considerably among studies.

* Time of the first PCR or coculture positive collected sample after birth.

PN: perinatally; IU: in utero; PP: postpartum through breastfeeding.

Focusing on transmission through breastfeeding, three recent studies have shown that there is no or very limited viral compartmentalization between milk and blood, suggesting that breast milk viruses are typical of circulating viruses 90 91 92 . As with other routes of MTCT, a genetic bottleneck that restrict the number of variants transmitted through breastfeeding to a single or a small number of variants was reported 82 .

Several studies examined the molecular characteristics of the envelope glycoproteins of transmitted viruses that might explain the selective process. Two studies of Wolinky’s group, in which MTCT route was not defined, reported that the potential N-linked glycosylation site (PNGS) proximal to the first cysteine of the V3 loop (N295) was strikingly absent from the infant’s sequence populations but present in the majority of the maternal sequence sets 61 68 . This observation has not been found by other investigators 69 70 73 74 . Studies performed on sequences encoding the full-length gp120 reported shorter variable regions and/or fewer PNGS in clade A and C viruses transmitted from mother-to-infant, mainly perinatally or early postpartum 81 86 88 . In contrast, we and others did not find altered gp120 length or PNGS number in the clade B and CRF01-AE viruses transmitted perinatally from mothers to their infants 85 87 . Interestingly enough, similar observations were reported during or shortly after heterosexual transmission of HIV-1. Transmitted viruses of subtype A and C showed a more compact and less glycosylated gp120 but this molecular property was not observed for sexually transmitted viruses of clade B 55 93 94 . This highlighted potential differences in the biology of the different subtypes, regardless of their transmission mode. Although we did not find shorter gp120s or fewer PNGS in maternally transmitted CRF01-AE viruses, we however found a limited number of PNGS, N301 in V3 and N386 in C3, that seemed to be conserved in all infected infants but were not uniformly present in their mothers, suggesting that they may confer an advantage on the virus to be transmitted 85 . By comparing functional properties of pseudotyped viruses expressing envelopes carrying or not N301 and/or N386, we confirmed that these two PNGS may play a role in resistance to autologous sera 95 . Interestingly, the N-linked glycan at position 301 was recently shown to be important for HIV-1 neutralization by several new broad and potent monoclonal NAbs of the PGT series (PGT125-128, PGT130 and PGT131) 96 . Similarly, Moore et al. recently reported in two HIV-1-infected adults the apparition of the N-linked glycan at position 332 targeted by PGT128, through immune escape from earlier strain-specific antibodies 97 . It could be possible that NAbs drive viral selection during MTCT, leading to variants with conserved glycan motifs that would confer a selective advantage.

MTCT offers a unique opportunity to explore the putative role of NAbs in restricting or preventing infection, especially when the transmission occurs in presence of high levels of passively acquired maternal IgG, i.e. during the perinatal and early breastfeeding periods. Conflicting results have been obtained concerning the role of maternal NAbs in reducing MTCT of HIV-1. Early studies, each relatively small, showed that non-transmitting mothers had more frequently detected and/or higher titers of autologous NAbs than transmitting mothers, suggesting a role for maternal NAbs in reducing MTCT 98 99 100 101 . Supporting this model, a few studies have suggested that transmitted viruses are escape variants resistant to neutralization by maternal antibodies 80 81 82 102 103 . However, other studies did not confirm these findings 88 89 95 104 105 106 . The observed discordant results may be due to small sample sizes, disparate maternal and infant sample collection time points, and a lack of identification of the route of transmission in several studies (in utero, in absence or in presence of only low levels of IgG, versus perinatally or early postpartum, in presence of high levels of IgG).

Because transmitted variants were found to be resistant to neutralization by their own mother’s plasma in several studies, we previously hypothesized that protective antibodies might be those with a broad neutralizing activity. To test this hypothesis, we conducted three large studies, i.e. two studies in Thailand 107 108 , and one study in French patients 109 , in which we compared the breadth and levels of NAbs in sera of transmitting and non-transmitting mothers, using panels of heterologous primary isolates of different clades. Our data did not support an association between the breadth of HIV-1 neutralizing activity and a lower rate of MTCT of HIV-1, whether transmission occurred in utero or perinatally (the infants were not breast-fed). Similar results were obtained by others in Kenyan infants of HIV-1 positive mothers 110 . All these data clearly indicate that infants who acquired broad and potent NAbs able to neutralize heterologous HIV-1 variants of different subtypes from their mothers did not show a reduced risk of infection. However, our studies suggested that particular HIV-1 variants might be indicators of NAbs associated with in vivo protection. Indeed we found that NAbs toward a CRF01-AE isolate, MBA, were significantly associated with a lower rate of MTCT in Thailand. Similarly, statistically significant higher frequencies or titers of NAbs toward several strains were observed in non-transmitting mothers in our French study when the clade B-infected mothers or non clade-B-infected mothers were analyzed separately 109 . Collectively, these data suggest that particular HIV-1 variants may provide a measure of protective antibodies depending of the population. These HIV-1 variants may be particular strains on which epitopes targeted by protective NAbs would be particularly well exposed in contrast to those of non-protective antibodies. The identification of these strains would be dependent of the HIV-1 subtype of the studied population suggesting that the neutralizing response specificities might vary between viral subtypes. Conducting large studies on homogeneous populations (in terms of geographical origin, viral clade, maternal viral loads and antiretroviral prophylaxis) to limit confounding viral and/or host factors would be necessary to identify such indicator strains.

The targeted epitope(s) of the neutralizing response associated with protection was not explored in studies cited above. To date, only one recent study conducted in a cohort of South African patients addressed this question by focusing on the putative role of maternal gp41-specific antibodies passively transferred to newborns 111 . An association of antibodies to the membrane-proximal external region (MPER) of gp41 with virus neutralization and protection was reported. This observation needs to be confirmed by other studies. More particularly it should be extended to other important conserved epitopes using various techniques, such as serum absorption with recombinant Env proteins or functional neutralization assay with various Env-pseudotyped viruses, that have been recently developed to dissect the neutralizing activity present in human sera 112 113 114 . These epitopes, most of them being of conformational nature, were identified thanks to the isolation of human monoclonal broadly neutralizing antibodies (HuMoNAbs) 96 115 116 117 118 (Figure 3). They are conserved regions of the HIV-1 envelope glycoprotein such as the CD4 binding site (CD4bs) on gp120, quaternary structure dependent epitopes on the V1-V2 variable loops of gp120, glycan-dependent epitopes involving the V3 region of gp120, and the MPER of gp41. A fine comparison of the specificity of the neutralizing response between transmitting and non-transmitting mothers by such extensive serum mapping using Env proteins or Env-pseudotyped viruses derived from HIV-1 strains indicator of protective antibodies should help to define more precisely if some antibody specificities correlate with protection in the MTCT context.

In the absence of antiretroviral prophylaxis, HIV is transmitted via breastfeeding to only 10-15% of infants, despite daily milk exposure. This suggests that breast milk may contain antiviral immune factors protecting most infants from mucosal HIV infections. HIV-1 envelope-specific antibody responses have been detected in milk. However, despite a predominance of secretory IgA in milk, the magnitude of the HIV-1 envelope-specific IgA activity is lower than that of envelope-specific IgG activity 119 120 121 122 . Similarly, the predominant SIV envelope-specific antibodies in milk from lactating rhesus monkeys are of the IgG isotype 123 . This is in accordance with what has been observed in other mucosal compartments, such as genital tract and saliva 119 120 124 125 126 127 128 . A few studies explored the putative role of envelope-specific antibodies present in breast milk of infected mothers to prevent infant infection. Quantitative studies of HIV or SIV envelope-specific binding antibodies did not reveal any difference between transmitting and non-transmitting mothers 129 130 131 . However, the neutralizing activity of these antibodies was not analyzed in these studies. Two recent studies made extensive comparisons of the neutralizing and antibody-dependent cellular cytotoxicity (ADCC) responses in breast milk and plasma from African HIV-infected lactating women 122 132 . Although lower in magnitude, the neutralizing and ADCC activities in milk were directly correlated with those in plasma and were primarily mediated by plasma-derived IgG antibodies. Similarly, SIV-specific IgA antibodies had limited neutralization potency compared to SIV-specific IgG in SIV-infected rhesus monkeys 123 . This suggests that IgA-mediated neutralizing and ADCC responses did not play a major role in preventing MTCT transmission via breastfeeding. However, the capacity of milk IgA to block HIV-1 transcytosis across epithelial cells was not studied. Such a protective role, recently reported for preventing vaginal and rectal infections in animal models, cannot be excluded 133 134 . The preventing role of IgG responses in mothers’ milk is difficult to explore due to the presence in newborns of high physiological levels of passively acquired maternal IgG. One study reported a higher magnitude of milk IgG-mediated ADCC in HIV-1-infected women who did not transmit HIV to their infants postnatally than in transmitters 132 . This suggests that an efficient vaccine should be able to induce a protective humoral response that could be transferred also through breastfeeding.