Interactions between the Gut Microbiome and Mucosal Immunoglobulins A, M, and G in the Developing Infant Gut

Interactions between the Gut Microbiome and Mucosal Immunoglobulins A, M, and G in the Developing Infant Gut Anders Janzon, a Julia K. Goodrich, a Omry Koren, a,b the TEDDY Study Group, Jillian L. Waters, a Ruth E. Ley a a Department of Microbiome Science, Max Planck Institute for Developmental Biology, Tübingen, Germany b Azrieli Faculty of Medicine, Bar Ilan University, Safed, Israel ABSTRACT Interactions between the gut microbiome and immunoglobulin A (IgA) in the gut during infancy are important for future health. IgM and IgG are also present in the gut; however, their interactions with the microbiome in the developing infant remain to be characterized. Using stool samples sampled 15 times in infancy from 32 healthy subjects at 4 locations in 3 countries, we characterized patterns of microbiome development in relation to fecal levels of IgA, IgG, and IgM. For 8 infants from a single location, we used fluorescence-activated cell sorting of microbial cells from stool by Ig-coating status over 18 months. We used 16S rRNA gene profiling on full and sorted microbiomes to assess patterns of antibody coating in relation to age and other factors. All antibodies decreased in concentration with age but were augmented by breastmilk feeding regardless of infant age. Levels of IgA correlated with relative abundances of operational taxonomic units (OTUs) belonging to the Bifidobacteria and Enterobacteriaceae, which dominated the early microbiome, and IgG levels correlated with Haemophilus. The diversity of Ig-coated microbiota was influenced by breastfeeding and age. IgA and IgM coated the same microbiota, which reflected the overall diversity of the microbiome, while IgG targeted a different subset. Blautia generally evaded antibody coating, while members of the Bifidobacteria and Enterobacteriaceae were high in IgA/M. IgA/M displayed similar dynamics, generally coating the microbiome proportionally, and were influenced by breastfeeding status. IgG only coated a small fraction of the commensal microbiota and differed from the proportion targeted by IgA and IgM. IMPORTANCE Antibodies are secreted into the gut and attach to roughly half of the trillions of bacterial cells present. When babies are born, the breastmilk supplies these antibodies until the baby’s own immune system takes over this task after a few weeks. The vast majority of these antibodies are IgA, but two other types, IgG and IgM, are also present in the gut. Here, we ask if these three different antibody types target different types of bacteria in the infant gut as the infant develops from birth to 18 months old and how patterns of antibody coating of bacteria change with age. In this study of healthy infant samples over time, we found that IgA and IgM coat the same bacteria, which are generally representative of the diversity present, with a few exceptions that were more or less antibody coated than expected. IgG coated a separate suite of bacteria. These results provide a better understanding of how these antibodies interact with the developing infant gut microbiome. KEYWORDS gut microbiome, infant, diabetes, immunoglobulins, IgA, IgM, IgG, antibody coating, infant gut development, FACS, host response, immunology, microbial ecology The gut microbiota, the immune system, and their interactions develop in tandem in infancy (1, 2). The immunoglobulin A (IgA) component of breast milk is protective against infection in infants and may also direct the development of the gut microbiota. Citation Janzon A, Goodrich JK, Koren O, the TEDDY Study Group, Waters JL, Ley RE. 2019. Interactions between the gut microbiome and mucosal immunoglobulins A, M, and G in the developing infant gut. mSystems 4:e00612-19. https://doi.org/10.1128/mSystems.00612-19. Editor Jack A. Gilbert, University of California San Diego Copyright © 2019 Janzon et al. This is an open-access article distributed under the terms of the Creative Commons Attribution 4.0 International license. Address correspondence to Ruth E. Ley, [email protected]. A.J. and J.K.G. are co-first authors. Received 27 September 2019 Accepted 7 November 2019 Published RESEARCH ARTICLE Host-Microbe Biology November/December 2019 Volume 4 Issue 6 e00612-19 msystems.asm.org 1 26 November 2019 on January 7, 2020 at TAMPERE UNIVERSITY LIBRARYhttp://msystems.asm.org/Downloaded from IgA is secreted into the gut lumen, where it binds antigens from food and microbiota, thereby excluding them from direct contact with the host epithelial cells (3). At birth, neonates generally have undetectable IgA in meconium (4), and it takes a few weeks for their immune systems to initiate IgA production and secretion into the gut (5). Breastmilk is an important early source of IgA, and breastfeeding is associated with high levels of fecal IgA in infants (4, 6). Planer et al. characterized the fraction of IgA-coated fecal microbiota in infants over the first few years of life and reported differences between breastmilkand formula-fed infants, which may relate to differences in how the microbiota develop in these two groups (7). Although IgA is the dominant antibody in the gut, IgM and IgG are also present. The juvenile gut sees up to5gofsecretory IgA daily, 100-fold less secretory IgM, and 1,000-fold less IgG (8). IgM and IgA are both produced by B cells locally, and the predominant class switching that occurs in B cells of the gut-associated lymphoid tissue is from IgM to IgA. Both are secreted into the gut via the same mechanism (polymeric Ig receptor), IgM as a pentamer and IgA as a dimer. In contrast, IgG is the most common antibody in circulation but can also be transported into the gut via a neonatal Fc receptor (9). Whereas IgA/M are produced in response to luminal microbial epitopes that are sampled by dendritic cells, IgG induction is thought to require crossing of the barrier by antigens, such that IgG is not produced continuously in response to common gut antigens. Based on its similarity to IgA, IgM may be expected to follow similar patterns of microbiota binding, whereas IgG may not. In healthy adults, IgA has been shown to coat a greater proportion of the stool microbiota than IgG or IgM, but whether the diversity of taxa targeted by these antibodies differs has not been reported (10). To gain abaseline understanding of how IgA, IgG, and IgM coat gut microbiota during microbiome development in infancy, here we performed a longitudinal analysis of the fecal microbiome of healthy infants in which we characterized the diversity of microbiota coated with IgA, IgM, and IgG as a function of time and with respect to feeding regimen and antibody levels. RESULTS The small sample set is representative of the larger TEDDY population. We looked for previously reported patterns of microbiota diversity in relation to covariates in order to establish that the small cohort used here (32 subjects sampled longitudinally, unsorted stool) is representative of the larger TEDDY cohort. More information on the study participants is shown in Table 1. Two recent papers have reported on the gut microbiome in the TEDDY cohort: Stewart et al. (11) employed a 16S rRNA gene survey on 903 subjects, and Vatanen et al. (12) used metagenomics with 783 subjects. Our results recapitulate those of Stewart et al. and Vatanen et al. in the following ways: (i) Bifidobacteriaceae and Enterobacteriaceae dominated the infant gut at early time points, while Firmicutes increased in relative abundance later (Fig. 1); (ii) breastfeeding status was significantly associated with between-sample diversity (e.g., beta diversity, as observed from associations between breastfeeding status and several principal coordinates [PCs] from both unweighted and weighted UniFrac principal coordinate analTABLE 1 Characteristics of study participants a Parameter Value for study participant No. sampled Means ⴞSEM Median Range Baby birth weight (g) 31 3,526.34 ⫾80.33 3,505 2,620–4,480 Mother body mass index 32 26.26 ⫾1.08 24.34 17.83–47.3 Gestational age (wk) 32 39.81 ⫾0.23 40 37–42.14 Mother weight gain (kg) 32 15.24 ⫾1.06 14.45 5.91–28.6 Maternal age 32 33.31 ⫾0.99 31.5 22–43 Paternal age 31 34.74 ⫾0.79 34 27–43 a Location, Georgia (USA, n⫽8), Washington (USA, n⫽8), Germany (n⫽8), and Sweden (n⫽8); delivery mode, Caesarian (n⫽8), vaginal (n⫽23), unknown (n⫽1); sex, female (n⫽16) and male (n⫽16). Janzon et al. November/December 2019 Volume 4 Issue 6 e00612-19 msystems.asm.org 2 on January 7, 2020 at TAMPERE UNIVERSITY LIBRARYhttp://msystems.asm.org/Downloaded from ysis [PCoA]; Fig. 2 and 3; see also Table S1 in the supplemental material) after correcting for multiple testing; (iii) the only other factor with a significant association with beta diversity was geographic location (Fig. 4A); (iv) we observed a weak association between antibiotic exposure and beta diversity (unweighted UniFrac PC4; Fig. 4B and C); (v) age had a significant association with microbiome richness (Chao1, Faith’s phylogenetic diversity, observed species, and Gini coefficient; all P⬍10 ⫺10 ). We observed that age and breastfeeding status were associated with microbiome diversity, Q4 (IgG+) Unsorted 100 200 300 400 500 0 25 50 75 0 25 50 75 0 25 50 75 0 25 50 75 0 25 50 75 Age (days) Phylum Relative abundance (%) Phylum Firmicutes Actinobacteria Proteobacteria Bacteroidetes Verrucomicrobia A Q3 (Ig-) Q2 (IgAMG+) Q1(IgAM+) Unsorted 100 200 300 400 500 0 25 50 75 0 25 50 75 0 25 50 75 0 25 50 75 0 25 50 75 Age (days) Family Relative abundance (%) Family Bifidobacteriaceae Lachnospiraceae Enterobacteriaceae Ruminococcaceae Bacteroidaceae Verrucomicrobiaceae B Q1(IgAM+) Q2 (IgAMG+) Q3 (Ig-) Q4 (IgG+) FIG 1 Changes in microbial composition over time. (A and B) Percent relative abundance of the dominant bacterial phyla (A) and families (B) in the unsorted samples (top plot in each panel). Each of the four quadrants is plotted over time. Q1 (IgAM⫹), IgA and IgM both high, IgG low; Q2 (IgAMG⫹), all high; Q3 (Ig⫺), all low; Q4 (IgG⫹), IgG high, IgA and IgM both low. Ig-Coated Bacteria in the Developing Infant Gut November/December 2019 Volume 4 Issue 6 e00612-19 msystems.asm.org 3 on January 7, 2020 at TAMPERE UNIVERSITY LIBRARYhttp://msystems.asm.org/Downloaded from −0.2 0.0 0.2 0.4 −0.3 −0.2 −0.1 0.0 0.1 0.2 0.3 PC1 PC2 100 200 300 400 500 Age (Days) A −0.2 0.0 0.2 0.4 −0.3 −0.2 −0.1 0.0 0.1 0.2 0.3 PC1 PC2 Clinical Center Location Georgia Germany Sweden Washington B −0.2 0.0 0.2 0.4 −0.3 −0.2 −0.1 0.0 0.1 0.2 0.3 PC1 PC2 Delivery mode Caesarian Vaginal NA C −0.2 0.0 0.2 0.4 −0.3 −0.2 −0.1 0.0 0.1 0.2 0.3 PC1 PC2 3000 3500 4000 Baby birth weight (g) D −0.2 0.0 0.2 0.4 −0.3 −0.2 −0.1 0.0 0.1 0.2 0.3 PC1 PC2 20 30 40 Mother BMI E −0.2 0.0 0.2 0.4 −0.3 −0.2 −0.1 0.0 0.1 0.2 0.3 PC1 PC2 HLA Group DR4/4 DR4/3 DR3/3 F −0.2 0.0 0.2 0.4 −0.3 −0.2 −0.1 0.0 0.1 0.2 0.3 PC1 PC2 Breastfeeding status No Yes G −0.2 0.0 0.2 0.4 −0.3 −0.2 −0.1 0.0 0.1 0.2 0.3 PC1 PC2 25 50 75 100 125 Phylogenetic Diversity H FIG 2 Patterns of beta diversity (using unweighted UniFrac) in 32 infant time series with covariates visualized. Shown are the first two axes (PC1 and PC2) from principal coordinate analysis of the unweighted UniFrac distances among the unsorted fecal microbiome samples of 32 infants over the first couple years of life. Points are colored by participant characteristics and phylogenetic diversity. (A) Infant age in days at the time of sampling. (B) Geographic location of the TEDDY-associated primary care center where samples were collected; this includes samples from Germany, Sweden, (Continued on next page) Janzon et al. November/December 2019 Volume 4 Issue 6 e00612-19 msystems.asm.org 4 on January 7, 2020 at TAMPERE UNIVERSITY LIBRARYhttp://msystems.asm.org/Downloaded from corroborating the findings of previous studies (11, 12), and therefore the subset of subjects that we used in subsequent analysis are representative of the larger TEDDY cohort. Stool antibody levels decrease with infant age and are related to breastfeeding status. We observed that levels of IgA, IgG, and IgM (measured by enzyme-linked immunosorbent assay [ELISA] in the same stool samples for which the microbiome was analyzed) were positively correlated to each other across the sample set (Fig. S1). Age (and therefore also microbiome alpha diversity) was negatively correlated with levels of IgA (repeated measures correlation [r rm ]⫽⫺0.45, P⫽6.14 ⫻10 ⫺22 ), IgG (r rm ⫽⫺0.37, P⫽1.13 ⫻10 ⫺14 ), and IgM (r rm ⫽⫺0.23, P⫽3.29 ⫻10 ⫺5 ), although the IgG and IgM correlation with age was not as strong as that for IgA (Fig. S1). No association of geographic location or HLA genotype was observed with any of the immunoglobulin levels in stool. We assessed the relationships between stool IgA levels, age, and breastfeeding status to ask how IgA levels varied over time. We were particularly curious to see if the cessation of breastfeeding correlated with stool IgA levels and if the effect of breastfeeding was age dependent. We observed that IgA levels in stool decreased significantly with age (P⫽1.79 ⫻10 ⫺11 )(Fig. 5 and 6). Furthermore, we observed that breastfeeding was significantly associated with higher stool IgA levels (P⫽0.039), and that this association was not dependent on the age of the infants. These observations corroborate previous reports that IgA levels decrease with age, but that at any age, breastmilk delivers additional IgA to the infant. We next examined the association between stool levels of IgA, IgG, and common OTUs (e.g., those shared by greater than 40% of samples) (Fig. 6). IgA levels were positively associated with several Enterobacteriaceae OTUs and a few Bifidobacteriaceae OTUs. Only one OTU was associated with IgG levels in feces. This OTU belonged to genus Haemophilus (Benjamini-Hochberg [BH] adjusted P⫽1.66 ⫻10 ⫺3 ). IgA and IgM coat many of the same microbial cells. For 8 infants from Georgia, for whom 15 time points sampled over 18 months were available, we sorted cells according to their antibody-coating status using FACS based on their distributions in four quadrants (Qs) (Text S1). The gatings for these four populations and the resulting quadrants are illustrated in Fig. 7A and B. We used heavy-chain-specific antibodies that do not cross-react with other classes of Ig (Text S1); thus, the patterns observed by flow cytometry indicate that many microbial cells were tagged with multiple antibodies. The gating patterns indicate that IgA and IgM coat many of the same cells, since the patterns of coating overlap. IgG-coated cells, on the other hand, fell into three categories: IgAM⫹, those low in IgG and high in IgA/IgM; IgAMG⫹, those highly coated in IgG, IgA, and IgM; Ig⫺, uncoated cells; and IgG⫹, those highly coated in IgG and low in IgA/IgM. We observed that the correlation between IgA and IgM coating was consistently higher than the correlation between IgA and IgG coating across sampling time points (P⬍10 ⫺10 )(Fig. 7C). Specific taxa discriminate total and FACS-sorted populations. We compared the diversity of the sorted microbiota (all Qs combined) to that of the whole microbiota (unsorted) to assess the impact of the FACS process on microbial diversity. We observed that the microbiota composition of the sorted cells (all Qs) is distinct from that of the total microbiome: a combined PCoA analysis of unweighted UniFrac analysis shows clear separation of unsorted and sorted cells along PC2 (Fig. 8A). The unsorted population was richer (Chao1, phylogenetic diversity, and observed species) and exhibited greater evenness (Gini coefficient) than all sorted cells (P⬍10 ⫺10 for all metrics) FIG 2 Legend (Continued) and the United States (Georgia and Washington). (C) Vaginal or cesarean delivery. (D) Baby birth weight in grams. (E) Body mass index (BMI) of the mother before pregnancy. (F) HLA genotypes of the infant were DR4-DQA1*030X-DQB1*0302/DR3-DQA1*0501-DQB1*0201 (DR3/4), DR4-DQA1*030XDQB1*0302/DR4-DQA1*030X-DQB1*0302 (DR4/4), and DR3-DQA1*0501-DQB1*0201/DR3-DQA1*0501-DQB1*0201 (DR3/3). (G) Breastfeeding status at the time of sampling. (H) Faith’s phylogenetic diversity. In all plots colored by a quantitative trait, blue indicates lower values and red indicates higher values. Ig-Coated Bacteria in the Developing Infant Gut November/December 2019 Volume 4 Issue 6 e00612-19 msystems.asm.org 5 on January 7, 2020 at TAMPERE UNIVERSITY LIBRARYhttp://msystems.asm.org/Downloaded from −0.2 0.0 0.2 0.4 0.6 −0.4 −0.2 0.0 0.2 0.4 PC1 PC2 100 200 300 400 500 Age (Days) A −0.2 0.0 0.2 0.4 0.6 −0.4 −0.2 0.0 0.2 0.4 PC1 PC2 Clinical Center Location Georgia Germany Sweden Washington B −0.2 0.0 0.2 0.4 0.6 −0.4 −0.2 0.0 0.2 0.4 PC1 PC2 Delivery mode Caesarian Vaginal NA C −0.2 0.0 0.2 0.4 0.6 −0.4 −0.2 0.0 0.2 0.4 PC1 PC2 3000 3500 4000 Baby birth weight (g) D −0.2 0.0 0.2 0.4 0.6 −0.4 −0.2 0.0 0.2 0.4 PC1 PC2 20 30 40 Mother BMI E −0.2 0.0 0.2 0.4 0.6 −0.4 −0.2 0.0 0.2 0.4 PC1 PC2 HLA Group DR4/4 DR4/3 DR3/3 F −0.2 0.0 0.2 0.4 0.6 −0.4 −0.2 0.0 0.2 0.4 PC1 PC2 Breastfeeding status No Yes G −0.2 0.0 0.2 0.4 0.6 −0.4 −0.2 0.0 0.2 0.4 PC1 PC2 25 50 75 100 125 Phylogenetic Diversity H FIG 3 Patterns of beta diversity (using weighted UniFrac) in 32 infant time series with covariates visualized. Shown are the first two axes (PC1 and PC2) from principal coordinate analysis of the weighted UniFrac distances between the unsorted fecal microbiome samples of 32 infants over the first couple years of life. Points are colored by participant characteristics and phylogenetic diversity. (A) Infant age in days at the time of sampling. (B) Geographic location of the TEDDY-associated primary care center where samples were collected; this includes samples from Germany, Sweden, and the United States (Georgia and Washington). (C) Vaginal or cesarean delivery. (D) Baby birth weight in grams. (E) Body mass index (BMI) of the (Continued on next page) Janzon et al. November/December 2019 Volume 4 Issue 6 e00612-19 msystems.asm.org 6 on January 7, 2020 at TAMPERE UNIVERSITY LIBRARYhttp://msystems.asm.org/Downloaded from (Fig. S2). We applied linear mixed models to identify taxa that were differentially abundant between the unsorted and sorted populations (all Qs). This analysis identified members of the Bacteroidetes,Verrucomicrobia,Gammaproteobacteria, and most Firmicutes as comparatively enriched in the unsorted fraction and Actinobacteria and Alphaproteobacteria as enriched in the sorted fraction (Fig. 8C). The difference in composition between sorted and unsorted fractions likely stems from the FACS process itself; for example, cells belonging to Actinobacteria and Alphaproteobacteria may be less likely to clump than others. Thus, the difference between sorted and unsorted microbiomes introduces the caveat that subsequent analyses with the sorted data are performed on a subset of the microbiome. Antibody-targeted cells exhibit patterns similar to those of the total population. As observed for unsorted cells, the majority of the variation in the sorted cells (PC1 of the unweighted UniFrac PCoA) was explained by age (Fig. 8B). Similarly, as age increased, alpha diversity also increased (Fig. S3). Analysis of variance indicated that FACS quadrant was significantly associated with most of the first 10 PCs from the PCoA (restricted to only sorted 16S rRNA gene data) of both unweighted and weighted UniFrac and all four alpha diversity metrics. Post hoc pairwise comparisons showed that this association was driven primarily by a difference between IgG⫹(high IgG only) and the other three quadrants. This finding might represent an overall diversity difference between the high IgG-only cells and others. However, the number of cells sorted into the high-IgG (IgG⫹) quadrant was significantly lower than those of the other three quadrants (Fig. S3). Indeed, most of the cells coated in IgG were also coated by IgA and IgM and are therefore sorted into IgAMG⫹rather than IgG⫹. This significant difference in cell number between the quadrants could explain the difference in diversity observed between IgG⫹and the other three quadrants. After exclusion of IgG⫹(high-IgG only), we observed some significant associations between the first 10 PCs of the beta diversity PCoAs and the FACS quadrant, as well as several associations with infant age and infant breastfeeding status. The significant associations with unweighted UniFrac were with age (PCs 1, 2, and 3), breastfeeding status (PCs 1, 2, 3, 6, and 7), and FACS quadrant (PCs 7 and 10) (Table S2). Further analysis of the FACS quadrant associations showed that Ig⫺(uncoated) was different from both IgAM⫹(P⫽0.030) and IgAMG⫹(P⫽0.001) along PC7, and that IgAMG⫹ differed from Ig⫺(P⫽0.010) along PC10 (post hoc pairwise comparisons between the quadrants using Tukey’s honestly significant difference [HSD] method to adjust for multiple comparisons). Among the weighted UniFrac PCs, many are associated with FACS quadrant (PCs 1, 5, 7, 8, and 9; post hoc analysis shows that Ig⫺is different from IGAM⫹and IgAMG⫹) and PC1 is associated with breastfeeding status and PC2 with age (Table S2). These results indicate that the diversity of cells targeted by antibodies is influenced by breastfeeding status and infant age. Furthermore, the specific combination of antibodies on the cells is not random. Specific taxa vary in their antibody-coating profiles. We next identified specific OTUs differentially abundant between the quadrants. To identify common OTUs (i.e., OTUs with nonzero sequence counts in greater than 40% of FACS samples examined) with differential relative abundance between quadrants, we performed linear mixed models using each OTU as a response variable. We searched for differences between the coated (IgAM⫹and IgAMG⫹; excluding IgG⫹because of low cell population) and uncoated (Ig⫺) populations. When comparing IgAM⫹to Ig⫺Qs, we observed significant differences in the relative abundances of 80 out of 190 OTUs (Fig. 9 and Table S3). These mostly included OTUs classified as Blautia, which had higher relative abundance in Ig⫺(uncoated) than other Qs. OTUs that were higher in IgAM⫹(IgM and IgA both FIG 3 Legend (Continued) mother before pregnancy. (F) HLA genotypes of the infant were DR4-DQA1*030X-DQB1*0302/DR3-DQA1*0501-DQB1*0201 (DR3/4), DR4-DQA1*030XDQB1*0302/DR4-DQA1*030X-DQB1*0302 (DR4/4), and DR3-DQA1*0501-DQB1*0201/DR3-DQA1*0501-DQB1*0201 (DR3/3). (G) Breastfeeding status at the time of sampling. (H) Faith’s phylogenetic diversity. In all plots colored by a quantitative trait, blue indicates lower values and red indicates higher values. Ig-Coated Bacteria in the Developing Infant Gut November/December 2019 Volume 4 Issue 6 e00612-19 msystems.asm.org 7 on January 7, 2020 at TAMPERE UNIVERSITY LIBRARYhttp://msystems.asm.org/Downloaded from high) were mostly classified as Bifidobacterium, unclassified Enterobacteriaceae, and Ruminococcus gnavus. Similarly, 101 OTUs had differential relative abundances between IgAMG⫹(all high) and Ig⫺(all low), with Blautia being significantly higher in Ig⫺ (Table S4). −0.10 −0.05 0.00 0.05 0.10 51015 Age (months) PC7 Geographic location Georgia Germany Sweden Washington A −0.10 −0.05 0.00 0.05 0.10 51015 Age (months) PC4 Antibiotic exposure No Yes B Has had antibiotic exposure Has not had antibiotic exposure 51015 1800 2400 2600 3000 3800 4300 4400 4600 5000 5400 6400 8700 9600 10300 10600 11300 11800 13200 1600 2000 2900 3500 4000 5200 5600 6000 10500 10800 12700 13500 13800 14800 Age (months) Subject ID Stool Antibiotics Antibiotic class Beta−lactams Cephalosporin Macrolide Sulphamethoxazole + Trimethoprim Trimethoprim NA C FIG 4 Beta diversity measures in relation to age. (A and B) PC4 (A) or PC7 (B) of unweighted UniFrac distances across time points. Lines show the mean PC values ⫾SEM grouped by time point and geographic location (A) or previous antibiotic exposure (B). (C) Plot illustrating the time of stool collection and antibiotic use for each participant. NA, not applicable. Janzon et al. November/December 2019 Volume 4 Issue 6 e00612-19 msystems.asm.org 8 on January 7, 2020 at TAMPERE UNIVERSITY LIBRARYhttp://msystems.asm.org/Downloaded from Cross-validation for IgA-targeted microbiota. We compared the results of our analysis with those of Planer et al., who used FACS followed by 16S rRNA gene sequencing to characterize the IgA coated microbes of 40 twin pairs over the first couple years of life (7). We used the same reference database as Planer et al. to classify sequences; therefore, we could compare OTUs directly. A total of 14 individual OTUs were identified by Planer et al. as being consistently targeted or not targeted by IgA (n⫽5 and n⫽9, respectively). Of these OTUs, two met the same criteria as those used in this study, in that they had a nonzero value in ⬎40% of the samples. Interestingly, both OTUs behaved similarly with respect to antibody coating. Greengenes OTU 365385 (genus Bifidobacterium), consistently targeted by IgA in the Planer et al. study, was proportionally higher in both IgAM⫹(BH adjusted P⫽3.07 ⫻10 ⫺05 )(Fig. 9A) and IgAMG⫹(BH adjusted P⫽1.88 ⫻10 ⫺05 ) than Ig⫺. The other Greengenes OTU detected in both data sets (606927; family Peptostreptococcaceae) was consistently nontargeted by IgA in the Planer et al. study, and similarly, we observed this OTU to have a higher relative abundance in Ig⫺(uncoated) than IgAMG⫹(BH adjusted P⫽0.013) (Table S4). These comparisons indicate that taxon-specific antibody-coating profiles can be generalizable across studies. DISCUSSION Interactions between IgA and the gut microbiome in the developing gut are important for health. Indeed, low levels in IgA coating of the gut microbiome in infants is associated with development of allergies and asthma during childhood (13) and with Crohn’s disease in children (14). Although the dynamics of IgA levels in the infant gut over time, and in response to breastfeeding, are well characterized, the IgA-microbiota coating, and those of the other antibodies in the gut (IgM and IgG), are less well understood. This study provides a baseline view of how levels of IgA, IgG, and IgM correlate with age and other factors over the first 18 months of life in healthy Western children and details how specific taxa are targeted by these antibodies over time in a subset of children. This study included infants from four distinct geographic locations, two in the United States and two in Europe. Infants from the different locations had slightly different microbiota, as previously reported (11, 12). Effects of breastfeeding status on the microbiome and antibody levels, and the decrease in antibody levels with age, were also similar across subjects, regardless of the shift in diversity associated with the geographic location of the infants. We also observed a strong impact of breastfeeding status and age on gut microbial community structure, as previously reported (11, 12). −0.3 −0.2 −0.1 0.0 0.1 0.2 0.3 100 200 300 400 500 Age PC1 2.5 5.0 7.5 Log(IgA in ug) FIG 5 Infant fecal IgA concentrations over the first couple years of life. PC1 from the PCoA of the fecal microbiome unweighted UniFrac distances is plotted against the age of the infant in days at the time of sampling. The points are colored by the log-transformed fecal IgA concentrations; lower concentrations are blue and higher concentrations are red. Ig-Coated Bacteria in the Developing Infant Gut November/December 2019 Volume 4 Issue 6 e00612-19 msystems.asm.org 9 on January 7, 2020 at TAMPERE UNIVERSITY LIBRARYhttp://msystems.asm.org/Downloaded from FIG S4, PDF file, 1 MB. FIG S5, PDF file, 0.2 MB. TABLE S1, XLSX file, 0.01 MB. TABLE S2, XLSX file, 0.01 MB. TABLE S3, XLSX file, 0.1 MB. TABLE S4, XLSX file, 0.04 MB. ACKNOWLEDGMENTS We thank William Melvin and Wei Zhang at Cornell University for their assistance. Special thanks to Timothy Bushnell, Matthew Cochran, and staff at the Flow Cytometry Core at the University of Rochester Medical Center. This work was supported by a National Science Foundation Graduate Fellowship (J.K.G.), Swedish Research Council grant 2011-922 (A.J.), and the Max Planck Society (R.E.L.). The TEDDY study is funded by U01 DK63829, U01 DK63861, U01 DK63821, U01 DK63865, U01 DK63863, U01 DK63836, U01 DK63790, UC4 DK63829, UC4 DK63861, UC4 DK63821, UC4 DK63865, UC4 DK63863, UC4 DK63836, UC4 DK95300, UC4 DK100238, UC4 DK106955, and contract no. HHSN267200700014C from the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK), National Institute of Allergy and Infectious Diseases (NIAID), National Institute of Child Health and Human Development (NICHD), National Institute of Environmental Health Sciences (NIEHS), Juvenile Diabetes Research Foundation (JDRF), and Centers for Disease Control and Prevention (CDC). This work was supported in part by the NIH/NCATS Clinical and Translational Science Awards to the University of Florida (UL1 TR000064) and the University of Colorado (UL1 TR001082). Members of the TEDDY Study Group are listed in the Text S1 in the supplemental material. A.J., R.E.L., and the TEDDY group designed the study. The TEDDY study group collected the samples and associated participant metadata. A.J., J.K.G., O.K., J.L.W., and R.E.L. performed statistical analysis and interpretation of the data. A.J., J.K.G., J.L.W., and R.E.L. drafted the manuscript. All authors approved the final version of the manuscript. We have no competing interests to declare. REFERENCES 1. Honda K, Littman DR. 2012. The microbiome in infectious disease and inflammation. Annu Rev Immunol 30:759–795. https://doi.org/10.1146/ annurev-immunol-020711-074937. 2. Belkaid Y, Hand TW. 2014. Role of the microbiota in immunity and inflammation. Cell 157:121–141. https://doi.org/10.1016/j.cell.2014.03.011. 3. Stokes CR, Soothill JF, Turner MW. 1975. Immune exclusion is a function of IgA. Nature 255:745–746. https://doi.org/10.1038/255745a0. 4. Maruyama K, Hida M, Kohgo T, Fukunaga Y. 2009. Changes in salivary and fecal secretory IgA in infants under different feeding regimens. Pediatr Int 51:342–345. https://doi.org/10.1111/j.1442-200X.2008.02748.x. 5. Rognum TO, Thrane S, Stoltenberg L, Vege A, Brandtzaeg P. 1992. Development of intestinal mucosal immunity in fetal life and the first postnatal months. Pediatr Res 32:145–149. https://doi.org/10.1203/ 00006450-199208000-00003. 6. Bridgman SL, Konya T, Azad MB, Sears MR, Becker AB, Turvey SE, Mandhane PJ, Subbarao P, CHILD Study Investigators, Scott JA, Field CJ, Kozyrskyj AL. 2016. Infant gut immunity: a preliminary study of IgA associations with breastfeeding. J Dev Orig Health Dis 7:68–72. https:// doi.org/10.1017/S2040174415007862. 7. Planer JD, Peng Y, Kau AL, Blanton LV, Ndao IM, Tarr PI, Warner BB, Gordon JI. 2016. Development of the gut microbiota and mucosal IgA responses in twins and gnotobiotic mice. Nature 534:263–266. https:// doi.org/10.1038/nature17940. 8. Haneberg B, Aarskog D. 1975. Human faecal immunoglobulins in healthy infants and children, and in some with diseases affecting the intestinal tract or the immune system. Clin Exp Immunol 22:210–222. 9. Yoshida M, Claypool SM, Wagner JS, Mizoguchi E, Mizoguchi A, Roopenian DC, Lencer WI, Blumberg RS. 2004. Human neonatal Fc receptor mediates transport of IgG into luminal secretions for delivery of antigens to mucosal dendritic cells. Immunity 20:769–783. https://doi.org/10 .1016/j.immuni.2004.05.007. 10. van der Waaij LA, Kroese FGM, Visser A, Nelis GF, Westerveld BD, Jansen PLM, Hunter JO. 2004. Immunoglobulin coating of faecal bacteria in inflammatory bowel disease. Eur J Gastroenterol Hepatol 16:669–674. https://doi.org/10.1097/01.meg.0000108346.41221.19. 11. Stewart CJ, Ajami NJ, O’Brien JL, Hutchinson DS, Smith DP, Wong MC, Ross MC, Lloyd RE, Doddapaneni H, Metcalf GA, Muzny D, Gibbs RA, Vatanen T, Huttenhower C, Xavier RJ, Rewers M, Hagopian W, Toppari J, Ziegler A-G, She J-X, Akolkar B, Lernmark A, Hyoty H, Vehik K, Krischer JP, Petrosino JF. 2018. Temporal development of the gut microbiome in early childhood from the TEDDY study. Nature 562:583–588. https://doi .org/10.1038/s41586-018-0617-x. 12. Vatanen T, Franzosa EA, Schwager R, Tripathi S, Arthur TD, Vehik K, Lernmark Å, Hagopian WA, Rewers MJ, She J-X, Toppari J, Ziegler A-G, Akolkar B, Krischer JP, Stewart CJ, Ajami NJ, Petrosino JF, Gevers D, Lähdesmäki H, Vlamakis H, Huttenhower C, Xavier RJ. 2018. The human gut microbiome in early-onset type 1 diabetes from the TEDDY study. Nature 562:589–594. https://doi.org/10.1038/s41586-018-0620-2. 13. Dzidic M, Abrahamsson TR, Artacho A, Björkstén B, Collado MC, Mira A, Jenmalm MC. 2017. Aberrant IgA responses to the gut microbiota during infancy precede asthma and allergy development. J Allergy Clin Immunol 139:1017–1025. https://doi.org/10.1016/j.jaci.2016.06.047. 14. De Palma G, Nadal I, Medina M, Donat E, Ribes-Koninckx C, Calabuig M, Sanz Y. 2010. Intestinal dysbiosis and reduced immunoglobulin-coated bacteria associated with coeliac disease in children. BMC Microbiol 10:63. https://doi.org/10.1186/1471-2180-10-63. 15. Stoel M, Jiang H-Q, van Diemen CC, Bun J, Dammers PM, Thurnheer MC, Kroese F, Cebra JJ, Bos NA. 2005. Restricted IgA repertoire in both B-1 Janzon et al. November/December 2019 Volume 4 Issue 6 e00612-19 msystems.asm.org 16 on January 7, 2020 at TAMPERE UNIVERSITY LIBRARYhttp://msystems.asm.org/Downloaded from and B-2 cell-derived gut plasmablasts. J Immunol 174:1046–1054. https://doi.org/10.4049/jimmunol.174.2.1046. 16. Wijburg OLC, Uren TK, Simpfendorfer K, Johansen F-E, Brandtzaeg P, Strugnell RA. 2006. Innate secretory antibodies protect against natural Salmonella typhimurium infection. J Exp Med 203:21–26. https://doi.org/ 10.1084/jem.20052093. 17. Benckert J, Schmolka N, Kreschel C, Zoller MJ, Sturm A, Wiedenmann B, Wardemann H. 2011. The majority of intestinal IgA⫹and IgG⫹plasmablasts in the human gut are antigen-specific. J Clin Investig 121: 1946–1955. https://doi.org/10.1172/JCI44447. 18. Macpherson AJ, Uhr T. 2004. Induction of protective IgA by intestinal dendritic cells carrying commensal bacteria. Science 303:1662–1665. https://doi.org/10.1126/science.1091334. 19. Hapfelmeier S, Lawson MAE, Slack E, Kirundi JK, Stoel M, Heikenwalder M, Cahenzli J, Velykoredko Y, Balmer ML, Endt K, Geuking MB, Curtiss R, III, McCoy KD, Macpherson AJ. 2010. Reversible microbial colonization of germ-free mice reveals the dynamics of IgA immune responses. Science 328:1705–1709. https://doi.org/10.1126/science.1188454. 20. Yasui H, Nagaoka N, Mike A, Hayakawa K, Ohwaki M. 1992. Detection of Bifidobacterium Strains that Induce Large Quantities of IgA. Microb Ecol Health Dis 5:155–162. https://doi.org/10.3109/08910609209141310. 21. Cullender TC, Chassaing B, Janzon A, Kumar K, Muller CE, Werner JJ, Angenent LT, Bell ME, Hay AG, Peterson DA, Walter J, Vijay-Kumar M, Gewirtz AT, Ley RE. 2013. Innate and adaptive immunity interact to quench microbiome flagellar motility in the gut. Cell Host Microbe 14:571–581. https://doi.org/10.1016/j.chom.2013.10.009. 22. Peterson DA, Planer JD, Guruge JL, Xue L, Downey-Virgin W, Goodman AL, Seedorf H, Gordon JI. 2015. Characterizing the interactions between a naturally primed immunoglobulin A and its conserved Bacteroides thetaiotaomicron species-specific epitope in gnotobiotic mice. J Biol Chem 290:12630–12649. https://doi.org/10.1074/jbc.M114.633800. 23. Lönnermark E, Nowrouzinan F, Adlerberth I, Ahrné S, Wold A, Friman V. 2012. Oral and faecal lactobacilli and their expression of mannosespecific adhesins in individuals with and without IgA deficiency. Int J Med Microbiol 302:53–60. https://doi.org/10.1016/j.ijmm.2011.08.004. 24. Mantis NJ, Rol N, Corthésy B. 2011. Secretory IgA’s complex roles in immunity and mucosal homeostasis in the gut. Mucosal Immunol 4:603–611. https://doi.org/10.1038/mi.2011.41. 25. Hagopian WA, Lernmark A, Rewers MJ, Simell OG, She J-X, Ziegler AG, Krischer JP, Akolkar B. 2006. TEDDY–The Environmental Determinants of Diabetes in the Young: an observational clinical trial. Ann N Y Acad Sci 1079:320–326. https://doi.org/10.1196/annals.1375.049. 26. Caporaso JG, Lauber CL, Walters WA, Berg-Lyons D, Lozupone CA, Turnbaugh PJ, Fierer N, Knight R. 2011. Global patterns of 16S rRNA diversity at a depth of millions of sequences per sample. Proc Natl Acad SciUSA108(Suppl 1):4516–4522. https://doi.org/10.1073/pnas .1000080107. 27. Caporaso JG, Kuczynski J, Stombaugh J, Bittinger K, Bushman FD, Costello EK, Fierer N, Peña AG, Goodrich JK, Gordon JI, Huttley GA, Kelley ST, Knights D, Koenig JE, Ley RE, Lozupone CA, McDonald D, Muegge BD, Pirrung M, Reeder J, Sevinsky JR, Turnbaugh PJ, Walters WA, Widmann J, Yatsunenko T, Zaneveld J, Knight R. 2010. QIIME allows analysis of high-throughput community sequencing data. Nat Methods 7:335–336. https://doi.org/10.1038/nmeth.f.303. 28. Bates D, Mächler M, Bolker B, Walker S. 2015. Fitting linear mixed-effects models using lme4. J Stat Soft 67:1–48. https://doi.org/10.18637/jss .v067.i01. 29. Lenth R. 2016. Least-squares means: the R package lsmeans. J Stat Soft 69:1–33. https://doi.org/10.18637/jss.v069.i01. 30. Bakdash JZ, Marusich LR. 2017. Repeated measures correlation. Front Psychol 8:456. https://doi.org/10.3389/fpsyg.2017.00456. Ig-Coated Bacteria in the Developing Infant Gut November/December 2019 Volume 4 Issue 6 e00612-19 msystems.asm.org 17 on January 7, 2020 at TAMPERE UNIVERSITY LIBRARYhttp://msystems.asm.org/Downloaded from