Human Milk-Derived B Cells: A Highly Activated Switched Memory Cell Population Primed to Secrete Antibodies¹

Breast milk is a part of the maternal-mucosal immune system, having important implications for infant development (1, 2). Breast feeding provides multiple soluble factors directly involved in infant mucosal defenses. Milk secretory IgA, nonabsorbed oligosaccharides, lactoferrin, and cytokines may contribute to the maturation and efficiency of the newborn immune system (3, 4, 5). While transiently compensating for the relative inefficiency of neonatal host defenses, breast milk may also represent a source of transmission of infectious agents, including HIV (6, 7, 8, 9), human T-lymphotropic virus (10), and CMV (11). Breast milk also contains leukocytes that remain largely uncharacterized but may have a protective role against neonatal infections (7). Although the presence of specific Abs against viral or bacterial Ags in breast milk have been previously demonstrated (12), the existence, features, and functions of B lymphocytes remain largely unexplored in this compartment.

In the course of the immune response, Ag-specific B lymphocytes are generated in germinal centers, wherein naive B cells undergo clonal expansion, V-region gene mutations, and differentiation into either Ig-secreting cells (IgSCs)⁵ or resting memory B cells (13, 14, 15). Inductive sites for mucosal immunity are constituted by regional MALT that contributes to the generation of IgSCs and memory B cells (16). Subsequently, memory B cells have the capacity to respond rapidly to previously encountered Ag by means of differentiation in plasma cells secreting high-affinity Abs. Two main subsets of memory B cells have been shown to differ in both phenotype and function (17, 18). CD27⁺IgD⁻ B lymphocytes are for the most part class-switched B lymphocytes expressing surface IgG or IgA molecules. CD27⁺IgD⁺ B lymphocytes are non-class-switched B cells expressing IgM⁺ and are involved in the response to T-independent Ags (19, 20). After being primed to become memory B cells or IgSCs, they recirculate through the secondary lymphoid organs to target effectors sites, including peripheral lymph nodes, spleen, tonsils, and also effector sites of mucosal immunity such as the lamina propria of various mucosae and the stroma of exocrine glands, including the lactating mammary gland (17, 21, 22, 23, 24). The MALT structures consist of organized mucosa-associated B cell follicles and larger lymphoid aggregates and are the origin of cells that traffic to mucosal effector sites (16). This concept of an integrated immune response in the MALT implies that following Ag exposure at one mucosal inductive site in the gastrointestinal or respiratory tracts, lymphocytes may colonize distant unexposed mucosal surfaces (9, 25, 26, 27). Specific homing receptors control this selective B cell migration through interactions with tissue-specific vascular addressins (28). In HIV-1-infected lactating women, protection toward breastfeeding transmission of HIV-1 conferred by local immune response remains a matter of controversy (6, 9). Breast milk anti-HIV-1 Abs can be produced locally or transuded from the vascular or tissue compartments.

Breast milk cells (BMC) can be collected easily and may reflect less accessible cells from the mammary gland tissue. Recently, we and others have demonstrated that CD4⁺ and CD8⁺ T lymphocyte populations can be identified and characterized in the breast milk compartment (29, 30, 31, 32). The aim of this study was to explore the phenotype, functions, and origin of B cells purified from breast milk from women with and without HIV-1 infection. Breast milk B cells were strikingly different from their blood counterpart, as the majority displayed a phenotype of IgD⁻ memory B cells. We observed that breast milk B cells had a particular profile of adhesion molecules (CD44⁺, CD62L⁻, α₄β₇^+/−, α₄β₁⁺), suggesting that these cells may originate from the GALT. Most of these cells were spontaneously activated, with a frequency of plasma cell precursors higher than in blood. Plasma cells from the breast milk compartment produced mainly IgG. Finally, we identified breast milk-derived B cells specific to HIV-1 Ags in HIV-1-infected lactating women.

This study was conducted at the Center Muraz, Bobo-Dioulasso, Burkina Faso, and at University Hospital of Montpellier, France. After written informed consent was obtained, 9 healthy lactating women (three from Bobo-Dioulasso, Burkina Faso, and six from Montpellier, France) and 12 HIV-1-infected mothers (from Bobo-Dioulasso, Burkina Faso) provided blood and breast milk. The study was approved by the local Ethical Committee and the National Ethical Committee of the Ministry of Health, Burkina Faso. After babies were fed, milk samples were collected 5–42 days postpartum by bimanual expression directly into a sterile polypropylene tube. Two or three milk samples per woman were collected within a 1-wk period. The median total volume obtained from each participant was 80 ml. All HIV-1-infected women were treated by antiretroviral therapy or had been recently exposed to antiretroviral drugs for prophylaxis of mother-to-child transmission. Breast milk HIV-1 RNA levels were quantified using the Generic HIV Viral Load assay (BioCentric) (33). This is an Agence Nationale de Recherches sur le SIDA-approved test based on a real-time RT-PCR assay by means of the ABI Prism 7000 thermocycler (Applied Biosystems). Median plasma and breast milk viral loads from HIV-1-positive women were below the detection limit level (<300 RNA copies/ml) in both blood and breast milk. The median CD4⁺ T cell count in blood was 324 cells/mm³ (interquartile range (IQR), 252–447 cells/mm³; normal CD4⁺ T cell range, 460-1200 cell/mm³).

Breast milk samples were kept at 4°C and processed within 4 h of collection. The acellular fraction (lactoserum and lipid fraction) was removed by centrifugation at 1200 × g for 15 min. Next, BMC pellets were washed three times in PBS supplemented with 5% FCS and finally suspended in RPMI 1640 medium plus 10% FCS, 2 mM l-glutamine, 100 U/ml penicillin, and 100 μg/ml streptomycin (all reagents from Eurobio). PBMC were obtained by standard histopaque density centrifugation. BMC and PBMC were then analyzed by flow cytometry.

B cells were isolated from BMC and PBMC aliquots by negative selection using a rosetting method as previously described (31). We used an enrichment cocktail containing Abs directed against membrane receptors of human hematopoietic cells (CD8, CD16, CD4, CD36, and CD56) and RBC (glycophorin A). Leucoreduced RBC concentrates (Etablissement Français du Sang, Toulouse, France) were kept at 4°C for at least 15 days to discard residual blood leukocytes before being added to the BMC suspension. One hundred microliters of red blood concentrate was added to 3 ml of BMC suspension. The Ab cocktail crosslinks unwanted breast milk-derived leukocytes to RBC, forming immunorosettes. When centrifuged over the buoyant density medium, rosetted cells were pelleted along with the RBC. This method resulted in elimination of >95% of non-B lymphocytes. The enriched B cells were recovered from the Ficoll-plasma interface, washed three times in PBS/2% FCS and resuspended in RPMI 1640 medium plus 10% FCS, 2 mM l-glutamine, 100 U/ml penicillin, and 100 μg/ml streptomycin.

Cell surface marker expression was investigated using a FC-500 flow cytometer (Beckman Coulter). Several combinations of mAbs conjugated to FITC, PE (PE/RD1), energy-coupled dye (ECD), PE-cyanine 5 (PC5), or PE-cyanine 7 (PC7) were used for multicolor staining. The positive thresholds for cell surface markers were set using isotypic controls. mAbs and matching isotype controls were used at the working dilution recommended by the manufacturer (10–20 μl of mAbs added to 100 μl of the test sample).

Lymphocytes were identified by CD45 (mouse anti-human IgG1 FITC or ECD conjugated, clone J.33) staining vs side light scatter. The different populations were identified by phenotypic features using different combinations of the following markers: CD3 (ECD conjugated, clone UCHT1), CD4 (PC7 conjugated, clone SFCI12T4D11), CD8 (PC5 conjugated, clone B9.11), CD16 (PE conjugated, clone 3G8), CD19 (PC7 conjugated, clone J4.119), CD56 (PC7 conjugated, clone N901). All of these Abs were mouse anti-human IgG1 and were provided by Beckman Coulter. B lymphocyte subsets were characterized by their distinct phenotypes: CD27⁻IgD⁺IgM⁺ naive B cells, CD27⁺IgD⁺IgM⁺ and CD27⁺IgD⁻ memory B cells using the following mAbs: anti-CD27 (mouse anti-human IgG1 PC5 conjugated, clone 1A4CD27; Beckman Coulter), IgD (polyclonal rabbit anti-human F(ab′)₂, FITC conjugated), and IgM (polyclonal rabbit anti-human F(ab′)2, PE conjugated) from Dako.

B cell activation was assessed by CD38 (PE or PC7 conjugated, clone LS198-4-3), CD86 (PE conjugated, clone HA5.2B7), and CD95 (PE conjugated, clone 7C11) surface marker expressions (all three anti-human IgG1 Abs were provided by Beckman Coulter). A gating strategy enumerated the larger sized B lymphocytes, as B cells size increased throughout plasma cell terminal differentiation (34). Among B lymphocytes, large-sized cells bearing CD19⁺, CD20^low/−, CD27^high, and CD38^high phenotypes were identified as plasmablasts or plasma cells, and CD19⁺, CD20^low/−, CD27^high, and CD138⁺ phenotypes were identified as plasma cells (35) (mouse monoclonal anti-CD20 IgG2a FITC conjugated, clone HRC20, and mouse monoclonal anti-CD138 IgG1 PE conjugated; both from Beckman Coulter,). The threshold for CD38^high analysis was established using the CellQuant CD38/CD8 kit for quantization of CD38 cell surface expression (BioCytex) in accordance with the manufacturer’s instructions and as previously described (36). The CD38^high threshold level was defined as 22,000 CD38 binding sites per cell.

The B cell homing receptor profile was investigated using conjugated Abs directed against adhesion molecule integrins: CD49d (anti-α₄ chain Ab, FITC conjugated, clone HP2/1), β₁ chain (anti-CD29 Ab, PE conjugated, clone 4B4LDC9LDH8), L-selectin (anti-CD62L Ab, ECD conjugated, clone DREG56), CD44 (anti-CD44 Ab, FITC conjugated, clone J.173); the four Abs were mouse anti-human IgG1 Abs provided by Beckman Coulter, and β₇ chain (rat anti-human IgG2a, clone FIB504) was from BD Biosciences. The α₄ integrin chain (CD49d) is associated with β₇ chain or β₁ chain (37, 38), and it forms either α₄β₇, a receptor for the mucosal addressin cell adhesion molecule-1 (MAdCAM-1), or α₄β₁ (also called very late Ag-4, or CD49d/CD29), a receptor to VCAM-1.

Enriched B cells were stimulated by CD40L-transfected CDw32L mouse fibroblasts (given by K. Tarte, Rennes University, Rennes, France), plus IL-2 and IL-10, to induce the in vitro differentiation into plasmablasts and plasma cells secreting IgA, IgG, or IgM (39). The ELISPOT plates (Immobilon-P flat-bottom 96-well plates; Millipore) were coated with mouse monoclonal anti-human H chain γ, α, or μ Abs (Tebu-Bio) as previously described (40). Following 5 days of culture, 5 × 10² breast milk or blood B cells were seeded in each well of the nitrocellulose ELISPOT plates. Alkaline phosphatase-conjugated anti-human H chain Abs and p-NBT chloride (BCIP (5-bromo-4-chloro-3-indolyl phosphate)/NBT; Sigma-Aldrich) were used to obtain insoluble blue stainings that are the fingerprint of IgSCs.

To enumerate spontaneous IgSCs by ELISPOT assay, enriched B cells from breast milk and blood were seeded on the nitrocellulose plate for 18 h, without addition of activation factors, at concentration of 5 × 10³ B cells/100 μl in each well.

IgG, IgA, and IgM anti-HIV-1 ASCs were enumerated by an HIV-1-specific ELISPOT assay (41). Briefly, enriched B cells from HIV-1-infected patients were polyclonally stimulated as described above. After 5 days of cell culture, 1 × 10⁴ breast milk or blood B cells were seeded in each well of the nitrocellulose ELISPOT plates and incubated for 18 h. The optimal concentrations of anti-human Ig and HIV conjugates have been previously determined by titration (40, 41). A dilution of anti-human H chain γ (IgG), α (IgA), or μ (IgM) Abs (Tebu-Bio) at 0.5 μg/100 ml PBS (pH 7.2) for coating and a 2–4 mg/ml dilution of p24- and gp120-conjugated peptides were used. After extensive washings with PBS, 75 μl of purified HRP-labeled HIV-1 peptides (0.5 mg/ml) were added for 1 h. These peptides mimic the immunodominant epitopes of the HIV-1 envelope glycoproteins and nucleocapsid recombinant proteins (kindly provided by Dr J. F. Delagneau, Bio-Rad, Marnes-la-Coquette, France). After additional washings, 3-amino-9-ethyl-carbazol (Sigma-Aldrich) was added to each well and insoluble red precipitates appeared within 5–10 min. The wells were then washed with distilled water to stop the reaction.

Comparisons of B cell frequency, phenotype, and Ig secretion were made between breast milk and blood compartments by the nonparametric Mann-Whitney U test using StatWiew 5.0 software. A p value of <0.05 (two-tailed) was regarded as significant.

Breast milk lymphocytes were enumerated by flow cytometry based on side light scatter/CD45 expression. Lymphocyte numbers ranged from 450 to 17,000 cells per ml (median, 4800 cells/ml; IQR, 720–11,500 cells/ml). The distribution of lymphocyte subpopulation is given in Table I (see also supplemental Fig. 1).⁶ T cells consisted of the majority of breast milk lymphocytes, while B cells represented a minority (median, 3.8%; IQR, 2.3–8.0%). No significant difference was found in breast milk lymphocyte populations between HIV-1-infected and uninfected women.

Naive and memory B cell subsets were explored on the basis of their phenotype (Table II, Figs. 1 and 2, and supplemental data Fig. 2). Naive B cells were largely underrepresented in breast milk, whereas they predominated in peripheral blood. Thus, breast milk B cells mostly consisted of memory B cells, including >70% IgD⁻CD27⁺ memory B cells in both healthy controls and HIV-1-infected women. In contrast, IgD⁺IgM⁺CD27⁺ memory B cells were infrequent in breast milk, whereas they represented almost half of the total memory B cell population in blood.

The proportions of IgG, IgA, or IgM memory B cells were evaluated by examining γ-, α- or μ-H chain SCs following 5 days of polyclonal activation (Fig. 3, A and B). IgG-SCs represented the main isotype in breast milk-derived B cells from both HIV-1-infected and uninfected women.

Analysis of adhesion molecule expression was performed on memory B cells derived from breast milk and blood (Table II and supplemental Fig. 3). By contrast with blood B cells, fewer breast milk B cells expressed CD62L. Also contrasting with blood B cells, breast milk B cells expressed more frequently the β₇ integrin chain. Conversely, almost all B lymphocytes expressed the α₄ and β₁ integrin chains both in breast milk and blood. B cells from either compartment expressed the CD44 receptor with a similar frequency.

Spontaneous B cell activation was investigated by morphologic (cell size), phenotypic (surface activation markers), and functional (cells producing Abs spontaneously) criteria. Surface expression of the activation marker CD38 was higher on breast milk B cells than on circulating B cells in both HIV-1-infected and uninfected women (Table II). We observed a higher frequency of large-sized B cells in breast milk compared with circulating B cells (median, 67.2% vs 8.0%, respectively, p < 0.001; Fig. 2, A and B). Cell surface analysis of BMC confirmed that a large proportion of breast milk-derived B cells display a plasma cell phenotype (CD19⁺, CD20^low/−, CD27^high, CD138⁺), whereas this phenotype was infrequent in blood B cells (Fig. 2, C and D, and Table II). A higher number of total spontaneous (IgG plus IgA plus IgM) SCs was observed in breast milk than in blood (median, 8600 vs 540 IgSCs/10⁵ B cells, respectively (p = 0.0079) in 5 healthy controls, and 9200 vs 860 IgSCs/10⁵ B cells, respectively (p = 0.0071) in 10 HIV-1-infected women). This indicated that IgSCs constituted a significant proportion of breast milk-derived B cells. In breast milk, IgG was the most frequently observed isotype among spontaneous IgSCs (Fig. 3, C and D).

Enriched B cells from breast milk and blood were activated with CD40L-expressing CDw32L mouse fibroblasts, in addition to IL-2 and IL-10. Resting memory B cells acquired a phenotype of plasmablast after 5 days of culture (data not shown) and secreted γ-, α-, or μ-H chain Ig as controlled by ELISPOT assays. To enumerate anti-HIV-1 ASCs in breast milk and blood, we performed an ELISPOT assay based on glycoproteins and nucleocapsid recombinant proteins of HIV-1. Anti-HIV-1 ASCs were found in three of eight breast milk samples and in six of eight blood samples tested from HIV-1-infected women (Fig. 4). Breast milk-derived anti-HIV-1 ASCs produced mainly IgG, whereas IgA anti-HIV-1 ASCs were found in only one subject.

Breast milk investigation is central to determining the immunological role of this biological fluid involved in both mucosal defense of breast-fed infant and viral transmission. Breast milk T lymphocytes have been previously described (29, 30, 31, 32, 42, 43), but the B cell populations remain poorly investigated. In this study, we analyzed the phenotype, functions, and origin of B cells purified from breast milk to improve knowledge of the mucosal immune response in women with and without HIV-1 infection. Our results demonstrate that in contrast to blood compartment 1) naive B cells remain largely underrepresented in breast milk, and 2) breast milk B cells mostly consist of IgD⁻ memory B cells. Hence, contrary to the IgD⁻ memory B cells, IgD⁺ memory B cells may not be recruited in breast milk.

Most breast milk B lymphocytes showed the phenotypic hallmark of activated cells with high levels of CD38 expression and large size. This is consistent with the lower expression of complement receptors observed on breast milk B cells, which might indicate that these cells could be plasmablasts or plasma cells (44). Moreover, studies in humans and animals reported that plasmablasts and plasma cells constitute a substantial fraction of the B cell population in the mammary gland (45, 46, 47).

We enumerated a low number of spontaneous IgSCs in peripheral blood from HIV-1-infected patients. These data are in agreement with previous results showing that spontaneous Ig secretion in blood is driven mainly by HIV-1 Ag and dramatically decreases under effective antiretroviral therapy (48, 49). In contrast, the persistence of the high level of spontaneous Ig secretion in breast milk from HIV-1-infected patients on successful antiretroviral therapy suggests that other factors than HIV-1 Ag are involved in the activation of breast milk-derived B cells. The B cell activation observed may be dependent on interactions with the immediate tissue environment. Breast milk contains soluble factors such as CD14 and oligosaccharides that may stimulate B cell activation and differentiation (50, 51). Consequently, a significant proportion of B cells from breast milk undergo terminal plasma cell differentiation and thus express CD138 and a high level of CD27 receptors.

We obtained resting memory B cells from breast milk samples that were able to secrete Ig following polyclonal activation. Using the same activation method, anti-HIV-1 ASCs were enumerated in HIV-1-infected lactating women, indicating the presence of B cells specific to HIV-1 Ag in the breast milk compartment. The nondetection of anti-HIV-1 ASCs in five out of eight patients may be the result of an insufficient number of B cells recovered from breast milk samples. Further explorations of Ag-specific memory B cells in mucosal tissues involved in interhuman transmission of virus remain an interesting challenge. Indeed, studies have shown that some individuals who are highly exposed to HIV but who remain persistently HIV seronegative have mucosal neutralizing HIV-1-specific Abs (52, 53, 54).

The mammary gland is an effector site of the mucosal immune system. Memory B cells and plasmablasts that colonize the mammary gland late in pregnancy originate from other mucosal areas where they have been exposed to Ags (23, 24, 25). Thus, mucosal immune response is viewed as an integrated response, since B cells migrate from MALT to mucosal effector sites for subsequent extravasation and terminal plasma cell differentiation (16, 28, 55). Nevertheless, different homing-molecule profiles between NALT and GALT have been observed, suggesting that mucosal B cells are strongly compartmentalized (55, 56, 57, 58). We were interested in investigating whether breast milk B cells may originate from the mammary gland or from the systemic compartment as a result of the intensive vascularization that takes place during lactation. Breast milk is enriched in activated memory B cells that are distinct from those circulating in the blood and bear a particular profile of mucosal adhesion molecules (α₄β₇^−/+, α₄β₁⁺, CD44⁺, CD62L⁻). This suggests that most likely those cells originate from the mammary gland. The presence of the α₄ and β₇ integrin chains and the lack of CD62L provide evidence that the homing pattern of breast milk B cells is similar to those of GALT B cells (58, 59, 60). Our results are consistent with previous observations indicating that the mammary gland contains ligands of both α₄β₁ and α₄β₇ that are VCAM-1 and MAdCAM-1, respectively. In contrast, other nonintestinal mucosae such as the salivary glands or bronchus express only VCAM-1 (28). Our observations confirm that the breast milk compartment may be more closely associated with intestinal mucosa than with upper respiratory mucosal sites.

Lactating mammary gland is primarily colonized by B cells and IgA-SCs (45, 46, 47), whereas T cells are at least 10-fold more frequent than B cells in the breast milk compartment (29, 31, 32). Moreover, within the breast milk lymphocyte population the proportion of IgA-SCs does not exceed IgG-SCs. Thus, T and IgG B lymphocytes attracted to this site may be able to cross the mammary epithelium and enter milk secretions in contrast to IgA-SCs that remain sessile in the mammary gland. The mucosae-associated epithelial chemokine CCL28 may be a key regulator of B lymphocyte migration and retention in the mammary gland (61, 62). CCL28 is up-regulated during lactation and is linked to the CCR10 receptor expressed on IgA-SCs in the mammary gland. Conversely, IgG-SCs or IgM-SCs are not attracted by CCL28 in vitro and thus do not accumulate in the mammary gland (61, 62). Consequently, >90% of Igs secreted in breast milk are IgA (3). In breast milk samples of both HIV-1-infected and uninfected women, we observed that IgG-SCs predominate over IgA-SCs. The extravasation of B cells that cross the mammary epithelium and enter milk secretions may entail a negative selective process operating at the mammary alveolar or ductal epithelium. It can be hypothesized that IgA⁺ memory B cells are anchored more tightly than corresponding IgG ⁺ cells in the mammary gland tissue. A weaker attraction between chemokine receptors such as CCL28 and IgG memory B cells may exist, allowing trafficking from subepithelial mammary tissue into breast milk. This suggests that breast milk-derived B cells may originate from the same mucosal site that IgA-SCs homed in the mammary gland but are submitted to distinct adhesion molecules.

Because of their low concentration, breast milk plasmablasts and plasma cells probably contribute only a small part of Abs secreted in breast milk. However, the daily consumption of large volumes of milk during the breastfeeding period implies ingestion of important quantities of B lymphocytes. Hence, a breast-fed child consumes on average 670 ml of milk per day (between 8 days and 2 mo of life). Considering a concentration as low as 80 viable B lymphocytes per ml, as many as 50,000 B cells, including resting B lymphocytes, plasmablasts, and plasma cells, would be ingested per day. Because colostrum and transitional milk contain greater quantities of cells, this value would be elevated 5-fold during the first week of life. The functional role of these B cells in breast milk remains unclear. Studies in animal models indicated that leukocytes from milk may infiltrate the tissues of intestinal tract, enter in mesenteric lymph nodes, and confer immunity to the recipient. Thus, adoptive transfer of immunologically active maternal B cells through breast feeding may be a mechanism for preventing viral or bacterial transmission via breast milk. However, this has not been demonstrated in humans despite some attempts (63, 64, 65).

Ag-specific memory B cells in breast milk have never been investigated so far. We identified and characterized these cells in breast milk. Our study demonstrates that breast milk contains predominantly switched memory B cells, plasmablasts, and plasma cells. Breast milk B cells bear characteristics that differ significantly from circulating B cells. These cells may originate from the GALT and migrate to breast milk through the mammary tissue. Breast milk B cells provide an opportunity for the exploration of a mucosal immune area involved in newborn protection against infections but also in transmission of milk-born viruses.

We thank the Human Milk Bank of the University Hospital of Montpellier, mothers who kindly provided breast milk samples for the study, Dr. Gael Petitjean for sample collection, and Alana Forster for carefully reading the manuscript. We are grateful to the French Agency for AIDS Research (ANRS) and particularly to Brigitte Bazin, Claire Rekacewicz, and Jean-François Delfraissy for their constant encouragement and support.

The authors have no financial conflicts of interest.