Infant B Cell Memory Differentiation and Early Gut Bacterial Colonization Free

The gut microbiota starts to establish directly after birth and provides a strong stimulus for the development of the immune system, including induction of IgA Abs that are secreted into the lumen (1–3). This process is characterized by class switching of B cells from IgM to IgA. Thus, the early recognition of bacteria and the subsequent modulation of B cell responses is an important process that promotes normal mucosal homeostasis. Several studies using germ-free animal models support the pivotal role of bacterial signals for B cell development. Germ-free animals have small Peyer’s patches and mesenteric lymph nodes that lack germinal centers (4), and display reduced numbers of T cells and IgA-producing cells in the intestinal lamina propria (5). Whether the gut microbiota has an impact on B cell maturation in humans is not clear, although it has been shown that Pakistani infants living in the poor areas have an accelerated maturation of the salivary IgA system compared with Swedish infants (6). In fact, in Swedish infants, the gut microbiota takes longer to establish and is characterized by a lower diversity and slower strain turnover rate than in infants in developing countries (7, 8). Whether colonization with particular bacterial species or groups is especially important for B cell development and activation during early childhood remains to be elucidated.

B cells are characterized by the cell surface expression of CD20 (9), but include several phenotypically and functionally different subgroups. CD5⁺ B cells arise early in ontogeny and have been shown to be the major B cell population in cord blood and during the first years of life, after which they decrease in numbers (10). The exact role of CD5⁺ B cells in humans is not completely understood. It has been shown that they have fewer somatic hypermutations than CD5⁻ B cells and predominantly produce low-affinity polyreactive IgM Abs that cross-react with bacterial Ags, including LPS (11, 12). This suggests that CD5⁺ B cells contribute to a first line of defense against bacterial infections. In mice, the expression of CD5 identifies the specific B cell linage termed B1 (13, 14). Recently, a small subset of circulating CD20⁺ B cells that specifically expressed CD27 and CD43 was described in cord blood and in adults (15). Approximately 75% of these cells also expressed CD5, and they displayed all the functional characteristics of murine B1 cells. The Ag-dependent development of naive B cells toward memory B cells is a process that is not yet completely elucidated. In humans, CD27 is regarded as a memory B cell marker and its expression correlates with the presence of somatic hypermutations in the Ig genes (16, 17). However, as showed by Griffin et al., CD20⁺CD27⁺ B cells represent a heterogeneous subgroup that includes both non–class-switched B1 cells (CD27⁺CD43⁺) and “true” memory B cells (CD27⁺CD43⁻), although the latter make up the vast majority (15).

In this study, we examined whether early intestinal bacterial colonization patterns were associated with the counts and proportions of CD20⁺ B cells, as well as CD5⁺ and CD27⁺ cells within the B cell population, in a prospectively followed cohort of 65 Swedish infants, using multivariate factor analysis. We demonstrate that children who are colonized early with Escherichia coli and/or bifidobacteria have higher numbers of CD27⁺ B cells at 4 and 18 mo of life than infants who lack these species in the early microbiota. Thus, our results suggest that the bacterial colonization pattern early in infancy may affect B cell maturation later in childhood.

Cord blood samples from newborn children (n = 41) and peripheral blood samples from children at 3–5 d (n = 52), 1 mo (n = 53), 4 mo (n = 56), 18 mo (n = 60), and 36 mo of age (n = 55) were obtained from a prospective newborn/infant cohort. The study included 65 healthy Swedish infants (33 boys and 32 girls), born at term (≥38 gestational weeks) in rural areas in southwest Sweden, who were followed to investigate the relation between intestinal bacterial colonization pattern and maturation of the immune system. Ninety-seven percent (63/65) of the children were exclusively or partially breast-fed during the first 6 mo of life. Twelve percent (8/65) of the children were treated with antibiotics in the first 6 mo of life. Peripheral blood was also obtained from 15 healthy adult volunteers, 23–58 y old, who were divided into two groups according to age, that is, individuals younger (n = 8) or older than (n = 7) 35 y of age. All blood samples were collected in preservative-free heparinized tubes. Informed consent was obtained from the parents and volunteers, and the study was approved by the Human Research Ethics Committee of the Medical Faculty, University of Gothenburg, Göteborg, Sweden.

Phenotypic characterization of the circulating B cells by flow cytometry was performed within 72 h after venipuncture. Initial experiments showed that this storage time did not affect the numbers or proportions of lymphocytes compared with fresh blood samples (data not shown). All analyses were performed with whole blood samples (50 μl/tube), which were incubated with the respective mAbs for 20 min, 4°C in the dark. The following mAbs were used: PerCP-conjugated anti-CD20 (clone L27, 1/45; BD Bioscience, Erembodegem, Belgium), allophycocyanin-conjugated anti-CD5 (clone UCHT2, 1/90; Pharmingen, San Diego, CA), and FITC-conjugated anti-CD27 (clone L128, 1/20; BD Bioscience). Next, RBCs were lysed with FACS lysing solution according to the manufacturer’s instructions (BD Bioscience). Thereafter, leukocytes were washed twice and suspended in FACS buffer. The following mAbs were used for analysis of FOXP3⁺ regulatory T cells (T_regs) within the CD4⁺CD25⁺ T cell population: PerCP-conjugated anti-CD4 (clone SK3, 1/45; BD Bioscience), allophycocyanin-conjugated anti-CD25 (clone 2A3, 1/90; BD Bioscience), and after completed cell surface staining, the cells were intracellularly stained with PE-conjugated anti-FOXP3 (clone PCH101; 1/9; eBioscience, San Diego, CA). Fixation, permeabilization, and staining for FOXP3 were performed with the kit PE anti-human FOXP3 Staining Set (eBioscience) according to the manufacturer’s instructions. Allophycocyanin- and FITC-conjugated mouse Ig G₁ isotype controls (clone ×40, allophycocyanin IgG₁ 1/90 and FITC IgG₁ 1/20; BD Bioscience) were used as references when gating on the CD5⁺ and CD27⁺ populations within the CD20⁺ lymphocyte population. As previously described (18), the gate for FOXP3 was set by comparing the expression of FOXP3 within the CD25⁻, CD25⁺, or CD25^high populations with each other and the isotype control for FOXP3 (PE-conjugated rat IgG2a, 1/9; eBioscience). All samples were analyzed on a FACSCalibur (BD Bioscience) equipped with CellQuest Pro software or on a FACS-Canto II (BD Bioscience) equipped with FACSDiva software. All flow cytometry data were analyzed with the FlowJo software (Tree Star, Ashland, OR). To be able to estimate the absolute numbers of cells in various B cell subsets from our samples, we used the TruCOUNT assay according to the manufacturer’s instructions (BD Bioscience). Fifty microliters of undiluted whole blood was incubated with PerCP-conjugated anti-CD45 Abs (clone 2D1, 1/10; BD Bioscience) in TruCOUNT tubes. After 15 min of incubation at room temperature in the dark, 450 μl lysing solution was added, followed by another 15 min of incubation. The samples were then analyzed with flow cytometry. The lymphocytes were defined and further gated as low in the side-scatter scale and high in CD45. Another dot plot was created to identify the beads with an FL1 versus FL2 plot where the beads were defined as having high FL1 and FL2 properties. The following formula was used to determine the numbers of lymphocytes per microliter of blood: events of lymphocytes/events of beads × number of beads per TruCOUNT tube/blood volume. At the last step, we multiplied the number of lymphocytes in each sample with the fraction of CD20⁺ cells to get the numbers of B cells, and thereafter with the fraction of cells expressing the particular surface marker of interest.

Fecal samples, which were obtained at 1, 2, 4, and 8 wk of age, were cultured quantitatively for major groups of aerobic and anaerobic bacteria, as previously described (18). In brief, staphylococci were isolated on Staphylococcus agar and identified as Staphylococcus aureus or coagulase-negative staphylococci with the coagulase test. E. coli and other enterobacteria (non-E. coli enterobacteria) were isolated on Drigalski agar (Media Department, Clinical Bacteriology, Sahlgrenska University Hospital, Göteborg, Sweden) and identified with the API20E biotyping system (bioMérieux Industry, Marcy l’Etoile, France), whereas enterococci were isolated on Enterococcosel agar, in which they hydrolyze esculin (BD Diagnostics, Stockholm, Sweden). Among the anaerobes, Bacteroides were isolated on Bacteroides Bile Esculin agar (BD Diagnostics), and clostridia were isolated from alcohol-treated samples and identified with the RAPID ID 32A system (bioMérieux Industry). Bifidobacteria were isolated on Beerens agar (Media Department) and identified by a genus-specific PCR. Putative lactobacilli were identified as Gram-positive unbranched rods growing on Rogosa agar (BD Diagnostics).

The relations between early intestinal bacterial colonization at 1, 2, 4, and 8 wk (x-variables; including Bacteroides, bifidobacteria, clostridia, lactobacilli, E. coli, other enterobacteria [non-E. coli], enterococci, coagulase-negative staphylococci and S. aureus) and the numbers of B cells that express CD20, CD5, or CD27 in blood samples collected at 4, 18, and 36 mo of age (y-variables) were investigated with multivariate data analysis (SIMCA-P+ software, version 12; Umetrics, Umeå, Sweden). Projection to latent structures by means of partial least squares (PLS) and orthogonal PLS (OPLS) were implemented to correlate the data matrices X and Y to each other in linear multivariate models. The quality of the analyses was assessed based on the parameters R2 (i.e., how well the variation of the variables is explained by the model) and Q2 (i.e., how well a variable can be predicted by the model). In the OPLS, the importance of each X variable to Y is represented by column bars. Jackknifing was used to calculate SEs displayed as an error bar on each column (representing the 95% confidence interval). Univariate analyses were performed with Kruskal–Wallis test followed by Dunn’s multiple-comparison test, two-tailed Mann–Whitney U test, and Spearman’s rank correlation test (GraphPad Prism; GraphPad Software, La Jolla, CA) as described in the figure legends. A p value ≤0.05 was regarded as being statistically significant (*p < 0.05, **p < 0.01, ***p < 0.001).

The surface expression of CD5 and CD27 was examined within the CD20⁺ lymphocyte population (i.e., B cells) based on the gating strategy demonstrated in Fig. 1A. We analyzed these cells both in terms of absolute counts (Fig. 1B–D) and proportions (Fig. 1E–G), in cord blood and peripheral blood obtained from children at the ages 3–5 d, 1 mo, 4 mo, 18 mo, and 36 mo, and from adults.

As shown in Fig. 1B and 1E, the numbers and proportions of CD20⁺ B cells in the circulation did not change during the first days of life. Thereafter, this cell population increased considerably, peaked at 4 mo of age, and subsequently decreased in an age-dependent manner. The counts of CD5⁺ B cells displayed a similar pattern as that observed for the total B cell population (Fig. 1C). The proportion of B cells that expressed CD5 was rather high already at birth (40%) and significantly increased up to 1 mo of age, and at this time point the majority (69%) of the B cells were CD5⁺. Thereafter, this population clearly decreased with age (Fig. 1F). The numbers of B cells expressing the memory marker CD27 increased more slowly than the counts of CD20⁺ and CD5⁺ B cells, and were found to reach a plateau between 18 and 36 mo of age (Fig. 1D). In adults, the numbers of CD27⁺ B cells were comparable with that observed in children at 4 mo of age. However, as shown in Fig. 1G, the percentage of B cells expressing CD27 was low during the first months of life but increased significantly between 4 and 18 mo, and continued to increase in an age-dependent fashion. We found no differences either in the numbers or in the proportions of the CD20⁺, CD5⁺CD20⁺, or CD27⁺CD20⁺ B cells between children treated or not treated with antibiotics in the first 6 mo of life (data not shown). Moreover, no sex differences in the numbers or in the proportions of the CD20⁺, CD5⁺CD20⁺, or CD27⁺CD20⁺ B cells were seen, except for the proportion of CD5⁺CD20⁺ at birth when boys had a significantly higher fraction of these cells compared with girls (data not shown).

It is clear that the gut microbiota has beneficial effects on B cell functions in mice (4, 19), but it is not known whether the intestinal bacterial colonization pattern has an impact on B cell maturation in humans. Thus, in this study, we investigated whether any particular bacterial species or group colonizing the intestine during the first 8 wk of life was associated with the numbers of total CD20⁺, CD5⁺, and CD27⁺ B cells in the circulation at 4, 18, and 36 mo of age.

We investigated the relation between the early intestinal bacterial colonization patterns at 1, 2, 4, and 8 wk of age (x-variables) and total CD20⁺, CD5⁺, and CD27⁺ B cell counts at 4, 18, and 36 mo of age (y-variables), using multivariate factor analysis. Projection to latent structures by means of PLS was performed to determine the influence of each x-variable on the overall model, that is, variable influence on projection (VIP). VIP values can be used to discriminate between important and unimportant predictors. x-variables with VIP values ≥1.0 were included in the final PLS loading scatter plot (Fig. 2A). The scatter plot projected E. coli and bifidobacteria at all time points and CD27⁺ B cells at 4 mo of age on the right side of the y-axis, indicating a positive association between colonization with these bacteria and high numbers of CD27⁺ B cells at 4 mo of age. S. aureus colonization at 2, 4, and 8 wk of age, in contrast, were projected on the opposite side, indicating an inverse relation to high CD27⁺ B cell counts. These association patterns are further depicted in the OPLS loading column plot (Fig. 2B). A similar bacterial association pattern was observed for the proportion of CD27⁺ B cells at 4 mo of age (data not shown). Clostridia at 1, 2, and 4 wk, enterobacteria (non-E. coli) at 4 wk, and lactobacilli at 4 wk were also projected close to CD27⁺ B cells at 4 mo (Fig. 2A), but these variables contributed less to the model because the smaller the bar and the larger the error bar in the OPLS loading column plot, the smaller and less certain is the contribution (Fig. 2B).

The univariate analyses showed that children colonized with E. coli or bifidobacteria at 4 and 8 wk displayed significantly higher numbers of CD27⁺ B cells at 4 mo of age compared with noncolonized children (Fig. 2C, 2D). In contrast, the numbers of CD27⁺ B cells tended to be less numerous in children colonized with S. aureus at 8 wk compared with children not colonized with such bacteria (Fig. 2E). Regarding CD20⁺ and CD5⁺ B cells at 4 mo of age, none of the bacterial species or groups examined was found to be associated to the circulating numbers of these B cell populations, neither in multivariate (Fig. 3A, 3B) nor in univariate analyses (data not shown). Taken together, these results suggest that children who establish an early intestinal flora characterized by E. coli and/or bifidobacteria have higher numbers, whereas children colonized with S. aureus tended to have lower numbers of CD27⁺ B cells at 4 mo of age relative to noncolonized children.

Next, we examined the association between bacterial colonization patterns and the counts of different B cell populations at 18 and 36 mo of age. As shown in Fig. 4A, both E. coli and bifidobacteria colonization were found to be positively associated to high numbers of CD27⁺ B cells at 18 mo in multivariate factor analysis. However, the contribution of the x-variables in the OPLS model at 18 mo was less predictive than that obtained at 4 mo of age (Q2 = 0.003 at 18 mo; Q2 = 0.10 at 4 mo). In univariate analyses, children colonized with E. coli (at 4 and 8 wk) and bifidobacteria (at 8 wk) displayed significantly higher numbers of circulating CD27⁺ B cells than children not colonized by these bacteria (Fig. 4B, 4C). Moreover, at 18 mo of age, different association patterns between bacterial colonization and the numbers of CD20⁺ and CD5⁺ B cells were observed compared with those obtained at the age of 4 mo (Fig. 5A, 5B and Fig. 3A, 3B, respectively). The bacterial colonization pattern associated to high counts of CD20⁺ B cells at 18 mo of age was similar to that related to high numbers of CD27⁺ B cells at the same age, that is, positively associated to E. coli and bifidobacteria (Fig. 5A and 4A, respectively). High CD5⁺ B cell counts in 18-mo-old children were positively associated to early colonization with lactobacilli and bifidobacteria during the first 2 wk of life (Fig. 5B). Univariate analysis showed that children who were colonized with E. coli at 4 wk or lactobacilli at 2 wk had significantly higher numbers of CD20⁺ or CD5⁺ B cells, respectively, in the blood at 18 mo of age relative to children not colonized by these bacteria at the same time points (p ≤ 0.05, data not shown). Finally, at 36 mo of age, no associations were found between early intestinal bacterial colonization patterns and the numbers of B cells expressing CD20, CD5, or CD27 (CD20: R2 = 0.29 and Q2 = −0.02; CD5: R2 = 0.27 and Q2 = 0.03; CD27: R2 = 0.13 and Q2 = −0.25; data not shown). In conclusion, children colonized with E. coli and/or bifidobacteria during the first 8 wk of life exhibit significantly higher numbers of B cells expressing the memory marker CD27 compared with noncolonized children both at 4 and at 18 mo of age. Early colonization with S. aureus, in contrast, was associated to low numbers of CD27⁺ B cells at 4 mo of age.

Some S. aureus strains produce enterotoxins that act as T cell superantigens, and experimental studies demonstrate that S. aureus superantigens induce T_regs (20, 21), which have been shown to suppress B cell proliferation and Ig responses (22, 23). In this study, however, no relation between the proportion of FOXP3⁺ T_regs, within the CD4⁺CD25⁺ T cell population, and the counts (data not shown) or proportion of CD27⁺ B cells at either 4 or 18 mo of life was found (Fig. 6A, 6B).

In this study, we investigated whether early intestinal bacterial colonization patterns were associated to the numbers of CD20⁺, CD5⁺, and CD27⁺ B cells later in infancy in a prospectively followed newborn/infant cohort. For the first time, to our knowledge, we demonstrate that gut bacterial colonization pattern may affect B cell activation and maturation also in humans because early colonization with E. coli or bifidobacteria was associated to higher numbers of B cells that expressed the memory marker CD27 at 4 and 18 mo of life. Colonization with S. aureus in the first weeks of life, in contrast, was related to lower numbers of CD27⁺ B cells at 4 mo of life.

The gut microbiota starts to establish directly after birth and provides a strong stimulus for the development of the immune system (4, 19). Bacteria and bacterial products may cross the intestinal epithelium through several pathways. In mouse models, dendritic cells have the capacity to open up tight junctions and capture Ags using their dendrites (24, 25). In both human and mouse, the small intestine contains lymphoid aggregates that are overlaid by M cells, a unique epithelial cell type specialized for transepithelial transport of macromolecules, particles, and microorganisms (26). Moreover, bacteria that reach high populations in the gut also cross the immature gut barrier and stimulate immune cells via a process termed translocation (27, 28). Later in childhood, however, the development of secretory IgA Abs will hinder translocation; thus, much of the contact between gut bacteria and the immune system will be prevented. In accordance with this, our results show that the presence of E. coli, bifidobacteria, and other cultivable bacteria predicted the variation in numbers of CD27⁺ B cells to a higher degree at 4 relative to 18 mo of life, and the early bacterial colonization pattern was not at all related to the variation in numbers of memory B cells at 36 mo of age.

B cells are activated by bacterial Ags that bind to the BCR, and B cells also respond to bacterial signals in an innate fashion as they express multiple pathogen recognition receptors, including TLRs. Still, the expression of different pathogen recognition receptors on B cells needs to be further examined in humans, but the expression of TLRs appears to be closely linked to the differentiation stage of the B cell (29–31). Microbial Ags that trigger these receptors can thus directly stimulate the B cells to produce Abs and cytokines. LPS, a cell wall component of E. coli and all other Gram-negative bacteria, activates immune cells that express the LPS receptor TLR4. However, in contrast with mouse B cells, human naive B cells do not express TLR4 (29, 30, 32). In this study, we demonstrate that children who were colonized with E. coli during the first weeks of life had higher numbers of B cells that expressed the memory marker CD27 relative to children not colonized by this bacterium. Because of lack of TLR4 expression on human B cells, it is unlikely that E. coli LPS per se is the factor that expands the number of memory B cells. Moreover, Gram-negative bacteria including Bacteroides and Enterobacteriaceae, other than E. coli, were found to be unrelated to high number of CD27⁺ B cells. Finally, early colonization with bifidobacteria, a Gram-positive bacterium that colonizes the majority of the children during the first weeks of life, was also associated with increased numbers of CD27⁺ B cells in the children at 4 and 18 mo of life. Instead, early colonization by E. coli and bifidobacteria, which may reflect successful acquisition of maternal gut bacteria during delivery, could be the trigger of B cell activation. Indeed, the classical infantile bacterial colonization pattern includes E. coli and bifidobacteria among the first bacteria that colonize the large bowel (33, 34).

In parallel with improved sanitary conditions and general cleanliness, the bacterial exposure of the Western infants has been reduced, and apparently this has led to changes in the composition of the gut microbiota since the 1970s (35, 36). As a result, S. aureus, which is primarily considered as a skin bacterium, has emerged as one of the first colonizers of the infantile gut of Swedish infants (37, 38) and colonize ∼70% of the children during the first 8 wk of life (39). The staphylococcal strains can persist for several months in the intestine, but the population counts decrease with age (39), reflecting their poor ability to replicate in a more mature and complex gut flora.

In this study, colonization with S. aureus was associated to lower numbers of circulating CD27⁺ B cells at 4 mo of age compared with children not colonized with such bacteria. Because presence of intestinal S. aureus might reflect a gut microbiota of low complexity unable to suppress staphylococcal growth, the low numbers of memory B cells in these children could be because of lack of other bacterial species or groups that are normally present in a highly complex microbiota. Moreover, S. aureus enterotoxins have been found to induce T_regs in both human and mice (20, 21), and these cells can suppress B cell activation and Ig responses (22, 23). However, we found no relation between the proportion of T_regs and proportion of CD27⁺ B cells in the periphery at either 4 or 18 mo of life. Still, the S. aureus strains isolated in this study need to be further investigated for presence or absence of genes encoding various enterotoxins because these could be differently related to the B cell memory differentiation in the infants.

In accordance with others, we show in this article that the numbers of CD5⁺ B cells increased during the first 4 mo of life and thereafter decreased in an age-dependent manner (10). In contrast with the counts of CD27⁺ B cells, we found no relation between the numbers of CD5⁺ B cells and the early bacterial colonization patterns at 4 mo of age. This indicates that the bacterial colonization patterns may be specifically associated to B cell memory differentiation in early infancy.

In conclusion, children who acquired E. coli and bifidobacteria early in life had higher numbers of circulating CD27⁺ memory B cells in the circulation later in childhood relative to children not colonized by these bacteria. In contrast, children who harbored S. aureus in the gut microbiota displayed lower circulating numbers of memory B cells. Thus, our results suggest that the early gut bacterial colonization pattern may indeed affect B cell maturation also in humans.

We thank study nurses Anders Nordberg and Helen Andersson at Skaraborgs Hospital, Skövde and Lidköping, respectively. We highly appreciate the skillful technical assistance of Kerstin Andersson, Inger Nordström, Eva Ågren, Ingela Kinell, and Yolanta Bonislavska, along with the staff at the Clinical Immunology Laboratory of the Sahlgrenska University Hospital. Finally, we thank all the families who took part in the study.

The authors have no financial conflicts of interest.