Neutralization of HIV subtypes A and D by breast milk IgG from women with HIV infection in Uganda

Summary

Objectives:

Among HIV-exposed infants in resource-limited countries, 8–12% are infected postnatally by breastfeeding. However, most of those uninfected at birth remain uninfected over time despite daily exposure to HIV in breast milk. Thus, we assessed the HIV-inhibitory activity of breast milk

Methods:

We measured cross-clade neutralization in activated PBMC of Ugandan subtype A (92UG031) and D (92UG005) primary HIV by breast milk or purified milk IgG and IgA from 25 HIV-infected Ugandan women. Isotype-specific antigen recognition was resolved by immunoblot. We determined HIV subtype from envelope population sequences in cells from 13 milk samples by PCR.

Results:

Milk inhibited p24 production by ≥50% (dose-dependent) by subtype A (21/25; 84%) and subtype D (11/25; 44%). IgG consistently reacted with multiple HIV antigens, including gp120/gp41, but IgA primarily recognized p24 alone. Depletion of IgG (n = 5), not IgA, diminished neutralization (mean 78 ± 33%) that was largely restored by IgG repletion. Mothers infected with subtype A more effectively neutralized subtype A than D.

Conclusions:

Breast milk from HIV-infected women showed homotypic and cross-subtype neutralization of HIV by IgG-dependent and -independent mechanisms. These data direct further investigations into mechanisms of resistance against postnatal transmission of HIV to infants from their mothers.

Introduction

Breastfeeding accounts for 25–44% of the up to 330,000 cases of mother-to-child transmission (MTCT) of HIV infection annually, particularly in sub-Saharan Africa.^1,2 The risk of postnatal MTCT by breastfeeding is 8–12%,^1,3 despite daily exposure to HIV in breast milk. Breast milk also lowers morbidity and mortality of children from malnutrition, respiratory and diarrheal diseases.^3–5 As a result, current WHO recommendations for infant feeding among HIV-infected mothers in resource-poor nations encourage continued exclusive breastfeeding for up to 6 months.⁶

The primary vehicles for MTCT of HIV infection are cell-free and cell-associated HIV virions found in maternal blood, vaginal fluids and breast milk.⁷ The amount of virus per liter of milk is low, so cumulative exposure of the infant to breast milk over time is more likely to lead to transmission of the virus in mothers that shed more cell-free viral particles in their breast milk.⁸ Both colostrum⁹ and milk^10–13 contain HIV-specific IgG and IgA as well as innate immune constituents, each of which shows inhibition of HIV inhibition in vitro.^14–19 Therefore, we characterized the ability of breast milk antibodies from HIV-infected women in Uganda, where HIV seroprevalence and MTCT are high, to neutralize the predominant local HIV subtypes A and D. We also characterized the inhibitory role of antibodies by isotype in cross-subtype neutralization.

Materials and methods

Study subjects

We tested milk from a random sample of 25 of 150 HIV-infected mothers enrolled in the Pathobiology of Breast Milk study in Kampala, Uganda²⁰ from November 2003–November 2004. These antiretroviral (ARV)-naïve pregnant women were ≥18 years old, ≥28 weeks gestation, and intended to breastfeed for ≥6 weeks. Each received single dose nevirapine at birth. All milk samples had undetectable levels of nevirapine by 4 weeks postpartum,²⁰ and no women received ARV’s postnatally per local guidelines at that time. The median age of women (n = 25) was 23 years (range 20–30). Median values for CD4+ T-cells were 410/μL (range 123–934), plasma HIV RNA 85,992 copies/mL (range <400–750,000) and breast milk HIV RNA 77 copies/mL (range 4–43,363). Mothers whose newborns tested positive for HIV by PCR between birth and six months were termed transmitters and mothers of newborns testing negative were termed non-transmitters. Exclusion criteria and study approval is as previously described.²⁰

Breast milk samples

Breast milk samples were obtained between 4 and 14 weeks postnatally. Mothers manually expressed breast milk after washing with water. Samples were centrifuged at 3000 rpm for 15 min at room temperature, visible lipid removed and lipid-poor milk (heretofore “milk”) aliquoted. Milk was not heat-inactivated. Cells were washed twice with phosphate buffered saline (PBS), re-suspended in 0.5 mL PBS, and frozen at −70 °C.

Neutralization

With modifications of reported assays,²¹ we incubated HIV subtype A (92UG031) or subtype D (92UG005) (AIDS Research and Reference Reagent Program, NIAID, NIH:U-NAIDS Network for HIV Isolation and Characterization) at 50 TCID₅₀ per well with milk diluted 1:4–1:16 in RPMI 1640 (Invitrogen; Carlsbad, CA), 5% IL-2 (Roche, Basel, Switzerland), heat-inactivated 10% FBS (Hyclone, Logan, UT) and gentamicin (Invitrogen) for 30 min at 37 °C. This mixture was added for 2 h to peripheral blood mononuclear cells (PBMCs) activated for 3 days with phytohemagglutinin (PHA)(Sigma, St. Louis, MO), washed and incubated at 37 °C with media changed on day 4. We tested supernatants on day 6 for p24 by ELISA (Perkin Elmer, Waltham, MA). Percent neutralization was calculated as 100 − ([p24 with virus and milk/p24 with virus alone] × 100).

We determined breast milk toxicity by counting viability of PHA-activated PBMCs at 2 h and 4 days by trypan blue exclusion after incubation with milk diluted 1:4 for 2 h at 37 °C and washing. Samples showing toxicity were treated with Lipid Removal Agent (LRA, 2 mg in 0.5 mL milk; Suppelco, St. Louis, MO) for 4 h at 4 °C, centrifuged, and reassayed.

Fractionation and purification of antibodies in breast milk

We prepared purified IgG milk fractions by incubating milk (500 μL) and 350 μL magnetic protein G beads (Miltenyi-Biotec, Auburn, CA) as modified from the manufacturer’s protocol. After collecting the non-IgG flow-through, the antibody-bound beads were washed with radioimmunoprecipitation (RIPA) buffer, low salt buffer and PBS, and IgG was eluted with pH shift buffer (0.1 M Triethylamine and 0.1% triton-X, pH 11.0) and adjusted to pH 7 with 1 M MES (pH 3.0). We restored the 500 μL volume with PBS, buffer exchanged with PBS to avoid toxicity using Amicon ultra-0.5 centrifugal filter devices (Millipore; Billerica, MA).

For IgA, we incubated milk (500 μL) with 400 μL of washed SSL7 agarose (Invivogen, San Diego, CA) in a spin filter (Pierce, Rockford, IL) rocked overnight at 4 °C. The non-IgA flow-through was collected, the antibody-bound agarose washed with buffer (10 mM sodium phosphate, 150 mM sodium chloride, pH 7.2), and IgA eluted with 500 μL of 0.1 M glycine buffer (pH 2.7), adjusted to pH 7 using 1 M tris-HCL (pH 9.0) and buffer exchanged. Purified IgG and IgA milk fractions contained <1% of other isotypes by ELISA using affinity-purified anti-human antibodies (Jackson ImmunoResearch; West Grove, PA)⁵

Antibody-depleted non-Ig milk fractions were prepared by covalently linking purified goat anti-human IgA, IgG and IgM (Southern Biotech, Birmingham, AL) to CNBr-activated sepharose 4B (Sigma). Antibody-linked sepharose (450 μL IgA, 300 μL IgG, 100 μL IgM) was packed into a spin filter, spun and washed 3 times at 4000 g with binding buffer (150 mM NaCl, 0.02% NaN3, and 20 mM Na₂HPO₄, pH 7.4). Milk (500 μL) was added, rocked overnight at 4 °C, centrifuged at 4000 g for 2 min, flow-through collected and buffer exchanged.

Total IgG and IgA in each milk fraction were adjusted to pre-processing levels. Whole and fractionated milk (1–3 μg protein) were resolved with 15% SDS-polyacrylamide gel (Bio-Rad Laboratories; Hercules, CA) and silver stained (Bio-Rad), with purified IgG, IgM, IgA (Jackson ImmunoResearch), lactoferrin and lysozyme (MP biomedicals, Solon, OH) as controls.

Protein immunoblots

The presence of HIV-specific antibodies in milk and purified antibody fractions were determined by immunoblot (ImmuneticsQualiCode HIV-1/2 Kit; Boston, MA) per manufacturer’s instructions for IgG or with anti-human IgA-biotin (1:50, Jackson Immuno Research) with streptavidin-alkaline phosphatase (1:1000, Invitrogen).

Sequencing and subtyping

We amplified a 416 bp fragment spanning the HIV-1 C2–V3–C3 region of the env gene by nested PCR (nucleotides 6959-7375) using the outer primers JA167/JA170 and inner primers JA168/JA169.²² Genomic DNA (DNeasy Blood and Tissue Kit Qiagen, Valencia, CA) was isolated from milk cell lysates of 13 mothers, for which we had cell pellets, collected at 14–24 weeks postpartum. First round amplification using the outer primer pair was carried out at 95 °C for 2 min, 95 °C for 20 s, 48.8 °C for 10 s, and 70 °C for 10 s for 30 cycles, and the second round at 95 °C for 2 min, 95 °C for 20 s, 51 °C for 10 s, and 70 °C for 10 s for 40 cycles, each with 0.5 μL KOD Hot Start Polymerase (Novagen, Gibbstown, NJ). We extracted amplified fragments from a 1.5% agarose gel (MinElute Gel extraction kit, Qiagen) and sequenced using forward primer JA168. HIV subtype was determined using a maximum likelihood tree (PhyML V3.0) the LANL HIV-1 subtype 2010 reference set (http://www.hiv.lanl.gov/content/sequence/NEWALIGN/align.html). All sequences were deposited in GenBank (accession numbers: KC329755–KC329769).

Statistics

We compared differences in demographic variables between HIV-transmitters and -nontransmitters and in neutralization of HIV subtypes A and D by breast milk using unpaired student’s t test and Wilcoxon matched-pairs signed rank test. We tested differences between neutralization in milk and milk fractions by ANOVA and if p < .05, we then analyzed the data by Tukey’s multiple comparison test (Prism 5.0d; GraphPad Software, Inc., La Jolla, CA).

Results

Neutralization by breast milk

Most breast milk samples from 25 women showed dose-dependent neutralization of HIV-1 subtype A or D strains (Fig. 1a). Subtype A strain 92UG031 was more readily neutralized than subtype D. For subtype A, 84% of milk samples supported ≥50% reduction in p24 production compared with 44% with the subtype D strain 92UG005 (p = .0025) (Fig. 1a). Toxicity of milk for target cells, present in 3 of 25 samples and abrogated by LRA treatment, did not relate to inhibition. Neutralization of neither subtype A nor D correlated with HIV RNA levels in plasma or breast milk, but neutralization of subtype A did correlate inversely with CD4⁺ T cell number (p = .0083; R² = 0.266).

Figure 1.

(A) Neutralization of HIV subtype A (92UG031, triangles) and D (92UG005, circles) target viruses by breast milk from 25 Ugandan women with HIV infection. Differences in neutralization of subtype A vs. subtype D were compared using unpaired student’s t test, **p = .002. The right side of the graph shows dose-dependent neutralization of subtype A virus by 6 milk samples at various dilutions. (B) Neutralization of homologous and heterologous HIV target virus subtypes by breast milk. Milk from 9 mothers infected with subtype A virus and 4 mothers with subtype D virus was incubated with target virus subtype A (92UG031, black bars) and D (92UG005, grey bars).

Based on HIV-1 subtype analysis, milk samples from 8 of 9 mothers infected with subtype A showed greater neutralization of a local primary subtype A than a local subtype D virus (p = .01) (Fig. 1b). Neutralizing activity from women infected with subtype D showed similar activity against both subtypes.

Of the 25 milk samples, 19 were obtained from non-transmitting and 6 from mothers who transmitted HIV-1 to their infants in the prenatal (n = 4; HIV PCR-positive at birth-2 weeks) or perinatal period (n = 2; HIV PCR-negative at 0–2 weeks, positive at 6 weeks). No samples were obtained from women who transmitted HIV postnatally through breast milk. Neutralization was similar in transmitters and non-transmitters, but HIV RNA in milk was higher in transmitters (9002 ± 6979 vs. 103 ± 25 copies/mL; p < .03).

Homology among clinical and target virus populations

Consistent with the ability of milk from mothers to neutralize the local subtypes, sequence analysis revealed extensive homology between the target strains and those in milk cell pellets (Fig. 2). This 416 bp envelope segment, one of the two most variable regions in the entire HIV genome, encodes the V3 loop and 2 important conformational CD4 binding sites (D loop and CD4 binding loop, an early target for CD4 binding). These two sites include residues that bind the broadly-neutralizing human monoclonal antibodies IgG b12 and VRC01^23–25 and predicted V3 loop targets for the monoclonal antibody PG16.²⁶ Each subtype A sequence contained the V3 crown motif GPGQ, whereas subtype D isolates shared the sequence GPGR or GPGQ (Fig. 2). Among neutralizing epitopes in the V3 loop crown,^27,28 those sequences having the GPGQ motif, 8 of the 9 subtype A HIV sequences (Fig. 1b) had broader cross-neutralizing activity than viruses with the GPGR motif²⁹ identified in the more resistant subtype D viruses. As expected, our subtype A and D sequences displayed restricted amino acid variation in the CD4-binding region (Fig. 2), particularly at positions 369 and 371 (amino acids D and E), which forms a relatively conserved discontinuous neutralizing epitope and is conserved in most of our isolates.²⁸ Sequencing analysis also shows that of 13 women with known subtype in our study, 9 had the 332 glycan (5 subtype A and 4 subtype D), which is a known N-linked glycosylation neutralizing target, and 66% showed both homologous and heterologous neutralizing activity.

Figure 2.

Sequence homology of the V3 loop and CD4 binding site regions of milk whey infected with HIV subtype A and D viruses. The C2–V3–C3 region of 13 milk whey samples were amplified using nested PCR and HIV subtype was determined using a neighbor-joining tree and the LANL HIV subtype reference set. The HIV subtype A V3 crown motif GPGQ(red; 9 samples), subtype D V3 crown motif GPGR (blue; 4 samples), and CD4 binding region (orange) are highlighted. The underlined amino acids in the CD4 binding region constitute part of a highly conserved discontinuous neutralizing epitope. X = non-synonymous mutations, # = out of frame indel, * = glycan binding motifs.

Neutralization by antibody isotypes and non-antibody milk fractions

Antibodies appeared to support the majority of HIV neutralization in breast milk. Purified antibody isotypes showed distinct contributions to neutralization. Milk from five women infected with subtype A virus supported a mean of 57% neutralization (range 33–99%) of the subtype A virus and 42% (range 13–86%) of the subtype D virus (Fig. 3). The IgG purified from these 5 samples supported the majority of neutralization against subtype A (range 78–99%) and D viruses (range 63–97%). The remaining non-IgG fractions retained some of the original activity against subtype A (range 0–43%) and subtype D (range 0–51%) viruses, but less than that of the IgG fractions (p < .01). Restoration of purified IgG to the IgG-depleted milk largely reconstituted the inhibitory activity to that of intact milk (Fig. 3).

Figure 3.

Neutralization of subtype A and D viruses by milk fractions with subtype A mothers. Milk from five Ugandan mothers infected with subtype A virus was separated into antibody (IgG and IgA) and non-antibody (non-Ig) fractions. Milk fractions were incubated with subtype A (92UG031) and D virus (92UG005). Values represent mean ± SEM. Differences in neutralization between milk fractions were determined using one-way ANOVA and Tukey’s multiple comparison test. *p < .05, **p < .01, ***p < .001.

Among milk samples from women infected with HIV subtype A, purified IgA showed less activity than intact milk (Fig. 3). Indeed, the percent neutralization by purified IgG exceeded that of IgA (p < .05) in milk samples even though the concentration of total IgG was lower than that of IgA (222 ± 60.3 vs. 307.2 ± 70.1 μg/mL, respectively [p = .03]). Both purified IgA and the removal of IgA had only a modest impact on neutralization (Fig. 3).

Innate factors in milk supported only a very limited role in HIV inhibition. After depletion of antibodies, the non-Ig fraction showed 1–8% neutralization of subtype A (1–17% of neutralizing activity of intact milk) (data not shown), despite an abundance of lactoferrin and lysozyme (>850 μg/mL and >10 μg/mL, respectively), factors associated with inhibitory activity.

Antigen specificity of HIV-specific IgG and IgA in breast milk

IgA in all milk samples reacted predominantly with p24 antigen alone (Fig. 4). In contrast, IgG recognized multiple HIV antigens (Fig. 4), including the envelope gp120 and gp41 fusion domains. IgG was more likely than IgA to recognize gp120 (68% vs. 0% of samples, respectively, p < .0001), gp160 (100% vs. 24%, respectively, p < .0001), and gp41 (80% vs. 0%, respectively, p < .0001), with similar trends for p66, p51, and p31. In contrast, the more conserved internal antigen, p24, was recognized at comparable frequencies. Although IgG comprised 40% ± 17% of all milk immunoglobulins (IgA 55% ± 18% and IgM 5% ± 4%), IgG did not mask the reactivity of IgA. Depletion of IgG did not increase the number or intensity of bands recognized by IgA (not shown).

Figure 4.

HIV-specific antibody to HIV proteins in milk of infected Ugandan mothers. Reactivity of milk IgA and IgG with HIV antigens was detected by immunoblot. Yellow = no reactivity, green = weak reactivity, blue = strong reactivity. The ability of milk to neutralize subtype A and D target viruses is shown (grey = ≥ 50% neutralization; right panel), and nested PCR of the C2–V3–C3 region of the env gene was done on 13 of 25 milk cell pellets to determine HIV viral subtype.

Discussion

We have shown that breast milk from HIV-1-infected women in Uganda supports dose-dependent neutralization of two local HIV subtypes A and D in primary human immune cells. Although this inhibition extends to both HIV subtypes, cross-subtype neutralization appeared greater for subtype A than subtype D with the two strained utilized in this study. We confirmed that, although IgA is the predominant isotype in milk,^5,11,30,31 IgG, not IgA, mediates most neutralization. IgG comprises a larger fraction of HIV-specific antibodies³² and reacted in all milk samples with multiple HIV proteins, including gp120 and gp41 proteins, which were recognized by IgA in only a minority of samples (6 of 25; 24%). These results are important because prevention of post-natal HIV transmission among breastfed infants has remained one of the most difficult challenges in controlling the HIV epidemic in children. The problem has been how to limit postnatal HIV transmission while retaining the nutritional and immunologic benefits of milk against the morbidity and mortality from diarrhea and pneumonia in infants at risk.

Protective effects of mucosal IgG may also be mediated through antibody-dependent cellular cytotoxicity (ADCC). Clinically, women with high titers of HIV-specific IgG antibodies in their cervicovaginal lavage fluids have lower genital viral loads than women with specific IgG only in serum, and the majority of these fluids that also have ADCC activity exhibited gp120-specific IgG, not IgA.³³ Similarly, HIV-specific IgG-mediated ADCC titers in milk from women infected with subtype C correlated with the specific-IgG binding responses, as did env-specific IgG binding and neutralization of tier 1 pseudoviruses.¹³ In other settings, the human monoclonal antibody (MAb) IgG 2F5 inhibited transcytosis of HIV across both polarized epithelial cell monolayers and rectal tissue,³⁴ as did gp160-specific IgG from colostrum and milk^9–11 and the human MAb IgG b12.²¹ These data and ours highlight the important functional activity of HIV-specific IgG in a variety of mucosal fluids. The higher levels of local HIV-specific IgG than IgA,^10,11 observed in other mucosal fluids, likely derive from transudation or active transport from serum, and, to a lesser extent, from the predominance of specific IgG-rather than IgA-secreting cells in milk,³⁵ although the IgA cells should be more prominent in the gland tissue itself.³⁵

Although HIV-specific IgA has been detected in a range of mucosal fluids, its reactivity with envelope glycoproteins is quite limited.³² The infrequent reactivity with gp160, gp120 and gp41 in our study is lower than that reported from other mucosal fluids, including breast milk.^13,32,36 These differences may relate to the sensitivity of the detection method (immunoblot vs. ELISA or antigen-coated bead binding). Similar to our results in humans, HIV-specific IgA responses in the mucosa of HIV-infected chimpanzees³⁷ and SIV-infected rhesus macaques is either decreased or absent.^12,38 The predominance of human p24 gag-specific IgA is recapitulated with SIV p27 gag-specific IgA,¹² highlighting the impaired recognition of HIV and SIV envelope by IgA in milk.

Even when HIV envelope-specific IgA was present in milk, we confirm that such purified IgA had limited but detectable neutralizing activity, as previously reported in milk¹³ and other fluids.^{11,12,32,39,40} Similarly, in milk samples from subtype C-infected Malawian women, IgA supported neutralization in only 1 of 35 milk samples¹³ and did not correlate with ADCC, results that are comparable to those with IgA from milk from SIV-infected rhesus macaques.¹² HIV-specific IgA, particularly polymeric IgA derived from a potent human monoclonal antibody (IgG b12), can neutralize the virus in vitro,²¹ but clinical mucosal fluids show low levels and limited activity in vivo. Indeed, HIV-specific IgA has been detected more, rather than less often in milk from transmitting than from non-transmitting women.⁴¹ However, specific IgA and IgG may support different activities in colostrum and milk particularly in their ability to inhibit viral transcytosis.^10,11

We and others have shown that IgA congeners of human monoclonal antibodies that bind the CD4 binding site (IgG b12) and the conserved membrane proximal external region of gp41 (2F5) can substantially inhibit epithelial uptake and transcytosis of HIV as well as transfer from epithelial cells to PBMCs.^21,34 Clinically, mucosal IgA, but not IgG, from cervicovaginal fluids of Cambodian women infected with HIV subtype E blocked binding to the alternative galactosylceramide receptor on epithelial cells and transcytosis.⁴² Thus, mucosal IgA, particularly that derived in vivo rather than in vitro, may have a more prominent impact on local binding, uptake and transfer of HIV by epithelial cells at primary sites of infection, rather than neutralization.

In this context, innate components of breast milk also showed limited neutralizing activity (non-Ig fraction) in this study and others.¹³ However, multiple individual constituents of breast milk, such as lactoferrin, lysozyme, mucin, lipids, secretory leukocyte protease inhibitor (SLPI), bile salt-stimulated lipase, lewis X component, and milk oligosaccharides each show such activity.^{14,16,17,19,30,43,44} Complement components present in breast milk may also affect infectivity or neutralization. Moreover, breast milk from HIV-seronegative women is also capable of blocking transmission of HIV in BLT humanized mice.⁴⁵ Thus, multiple components of breast milk, including neutralization by IgG, likely act in concert by complementary mechanisms to block post-natal transmission of HIV-1.

Preferred targets for inhibition would be shared by multiple strains and subtypes. Our sequencing results suggest that the cross-subtype neutralization we observed with these milk samples infected with subtypes A and D viruses may be associated with binding to a highly conserved, broadly neutralizing epitope in the CD4-binding region of gp120. Residues in the C2, C3 and V5 regions of the HIV CD4 binding site also comprise antibody recognition sites.²⁸ The C3 region also includes several HIV N-linked glycosylation neutralizing targets at positions 332, 332, and 339^46,47 that are associated with production of broadly cross-neutralizing antibodies, activity that is highly dependent on single amino acid changes and glycan binding.⁴⁷

One limitation of this study is that we did not have access to autologous viruses from these patients. Fouda et al. reported that neutralization of autologous subtype C viruses by breast milk was infrequent as was neutralizing activity with subtype C pseudoviruses in cell lines.¹³ However, we used primary human cells due to substantial differences in neutralization of subtype A virus between these cells and the TZMbl cell ine.⁴⁸ Our primary local viruses and target cells pooled from ten healthy U.S. donors, although not from local Ugandans, may better model physiologic events. Finally, we only looked at cross-subtype neutralization with two local strains, one subtype A and one subtype D. Based on these results, we will expand our virus panel to better elucidate the ability of breast milk and its constituents to neutralize more diverse viruses and subtypes.

In summary, the frequency of HIV transmission to infants is unacceptably high without antiretroviral therapy.^1,3 Our findings of high frequencies of neutralization by breast milk, particularly the IgG fraction, are consistent with observations that the majority of infants who are uninfected at birth remain uninfected during breastfeeding despite daily exposure to the virus in breast milk. This inhibition was independent of CD4+ T cell number and levels of HIV RNA in blood or milk. Current work is directed to integrate the ability of breast milk to neutralize HIV in PBMC with that in other targets (e.g., epithelial and dendritic cells). Such immune defenses interact with the levels of cell-free or cell-associated virus in milk, with antibody and innate factors, and genetic and immunologic characteristics of both infant and the founder virus(es) to determine the risk of post-natal infection. Antiretroviral therapy of mothers and prolonged prophylaxis of infants provides substantial protection against MTCT of HIV,⁴⁹ but, to date, only about 60% of eligible women in SubSaharan Africa are receiving one or more drugs in this vulnerable period.² Indeed, a potentially clinically protective role for inducible virus-specific responses at local mucosal sites, including breast milk, has recently been proposed with 3 different vaccine strategies.^50–52 Thus, understanding the range of mechanisms of defense that are or are not elicited by natural infection, and that can be stimulated experimentally, particularly in breast milk and other mucosal sites, will advance our efforts to prevent postnatal MTCT of HIV through breastfeeding from women in resource-limited settings.