Maternal IgA2 Association with Colostrum Microbiota Contributes to Bacteria Establishment During Intestinal Colonization in the Neonate.

Background: Microbiota colonization during labor and through the rst meals contributes to immune maturation and development of the newborn. Mother provides probiotics and prebiotics factors through colostrum and maternal milk to shape the rst neonatal microbiota. Previous works have reported that immunoglobulin A (IgA) secreted in colostrum is coating a fraction of maternal microbiota. Methods: Thus, to better characterize the IgA-associated microbiota, we used ow cytometry coupled with 16S rDNA gene sequencing (IgA-Seq) in human colostrum and neonatal feces. We identied IgA-bound bacteria (IgA+) and characterized its diversity to elucidate possible role of IgA subclasses during neonatal bacterial colonization of the colon. Results: We found that IgA2 in the colostrum has an active role during microbiota colonization. Colostrum IgA2 is mainly associated with Bidobacterium, Lactobacillus, and Bacteroidetes genera. This association seems to give these bacteria an advantage during their establishment since metabolic pathways related to epithelial adhesion and carbohydrate consumption are enriched within fecal IgA2+ microbiota. Association with specic bacteria could be explained since IgA2 recognizes common antigens expressed on surfaces among bacteria genera. Conclusions: Our data suggest a specic targeting of commensal bacteria by IgA2 revealing a specialized function of IgA microbiota association during neonatal intestinal colonization during the rst days of life. effect

At the same time, human colostrum is the primary source of maternal IgA. Once transferred to the newborn, IgA plays a crucial role in gastrointestinal tract protection through immune exclusion against pathogens [15]. However, it was shown that a signi cant fraction of commensal bacteria are associated with IgA [16] in homeostatic conditions. These antibodies recognize and associate with resident microbiota in acini, duct, and nipple skin during colostrum ejection. The transit of maternal IgA through the newborn gastrointestinal tract and its recovery in the newborn's feces have been documented [17].
IgA in the gastrointestinal tract neutralizes pathogenic microorganisms and directly regulates microbiota composition [18,19,20], suggesting that maternal milk IgA-microbiota impacts bacteria intestine colonization in the newborn [21,22]. This paradoxical function of the IgA has been poorly analyzed. Previous studies in mice have suggested a possible explanation for this seemingly dual function. In one hand, IgA produced by T-dependent responses (against protein antigens) is associated with potential pathogens, while on the other hand IgA made by T-independent stimulation (mainly against carbohydrates antigens) recognizes species of commensal microbiota [23,24].
Human IgA has two subclasses, IgA1 and IgA2, with well-de ned properties and structures [25]. In a recent study, Sterlin and colleagues described that most commensal IgA-decorated bacteria present in adult fecal samples are bound to the IgA2 subclass. These data suggest that this subclass would play a role in regulating commensal bacteria under homeostatic conditions in the intestinal tract [26]. Ladjeva et al. described an individual variability in speci city between colostrum IgA1 and IgA2 subclasses. IgA1 recognizes protein antigens (bovine gamma globulin, lactoglobulin, tetanic toxoid, Streptococcus mutans antigens I and II, and in uenza virus vaccine). In sharp contrast, IgA2 was associated with Escherichia coli lipopolysaccharide (LPS), Streptococcus pneumoniae polysaccharide, Haemophilus in uenzae type b hemagglutinin [27]. Work from our laboratory found that speci c IgA1 levels of antigen increase in women with prior vaccination with tetanus toxoid (protein antigen), while antigen-speci c IgA2 increased, predominantly, in mothers with pneumococcal polysaccharide vaccination [28]. However, the microbiota distribution associated with IgA subclasses in colostrum and their role in transferring maternal microbiota to the newborn has not been described.
In this work, we analyzed colostrum from mothers and fecal samples from neonates (before and after feeding with colostrum) to compare microbiota composition. Our results demonstrated speci c targeting of commensal bacteria by IgA2 and revealed a specialized IgA-associated microbiota function during the newborn's intestine colonization during the rst days of life.

Design and type of study
This study is an analytical, observational, prospective, and longitudinal. The population was selected through a simple random process. Clinically healthy women were recruited at Hospital Regional 1° de Octubre (HR 1° Oct) Instituto de Seguridad y Servicio Social para Los Trabajadores del Estado (ISSSTE). All the activities detailed in this protocol have the approval of the Research and Bioethics Committee of HR 1° Oct (090201/14.1/086/2017). Sociodemographic, anthropometric, and clinical data (education, occupation, maternal and gestational age, and type of delivery) were recorded. Two to three mL of colostrum, and 0.5-2 g fecal samples, from one hundred eighty mother-newborn binomials were collected.
The sample size was determined by a non-probabilistic convenience sampling, considering new delivery mothers between November 2017 and July 2019.
Selection criteria Inclusion criteria. Women between 25 and 30 years old, with a full-term pregnancy (38-42 weeks of gestation), due to physiological delivery (F), who has regularly attended their medical care to monitor their pregnancy, with the delivery of the unique product and agreeing to participate in this project.
Exclusion criteria. Women unable or unwilling to breastfeed, who reported chronic degenerative disease under drug, hormonal, or antibiotic treatment were omitted. Women who reported alcohol and smoking during pregnancy or decided not to continue in the study, o reported an acute illness in the last three weeks before delivery were also omitted.

Sampling
Before delivery, patients who agreed to participate in the study received informed consent, which they read and signed following the Helsinki treaty, General Health Law, and Hospital ethics committee. They were given a questionnaire, which they lled out with their general data and clinical information. Data reported by patients were corroborated with their medical history les from the hospital archive. Samples were taken, under aseptic conditions, with sterile material and quali ed medical personnel. Meconium was obtained during labor and in the rst evacuation of newborns. Colostrum samples were obtained following the procedures described in the milk extraction technique protocol in the lactation clinics of Children's and General Hospitals, within the rst hours postpartum after cleaning the nipple and areola area with physiological saline solution and sterile gauze. Special care was taken to ensure infants were exclusively breastfed during their permanence in the hospital. Two stool samples from the neonates were requested: before and after three days the neonates were fed by their mother. As a control of maternal milk factors transference process, pasteurized maternal milk samples from the Hospital's Milk Bank were used. This pasteurized milk (Pasteurized) is a supplementary feeding option recommended when breastfeeding is not feasible; for example, ablactation, maternal disease, or impediment due to drug intake. This pasteurized sample is prepared with volumes of donor mothers less than seven days postpartum in Hospital Milk Bank, which is pasteurized (63° C for 30 min) and delivered to the neonate [29]. After treatment, a 2 mL sample was taken. Finally, sampling of meconium and feces from a group of neonates fed with pasteurized milk was included (n= 8) according to the pediatrician's indications.
Samples were kept at 4° C for their transfer to the laboratory, to be processed within the rst 20 minutes after sampling.

Phase separation
For separation, the methodology applied in this work was previously reported by Bunker et al. [23]. Brie y, 500 mg meconium/stool samples were taken and resuspended in 2 mL of sterile phosphate buffer saline (PBS) (Cat# P4417-100TAB, Sigma-Aldrich®, St Louis, Missouri, USA). Colostrum, meconium, and stool samples were treated with protease inhibitor cocktail (Cat #P8465, Sigma-Aldrich®) 1% with 0.01% sodium azide (NaN 3 ) (Cat# S2002, Sigma-Aldrich®) following manufacturer speci cations. Samples were centrifuged at 400 g for 10 minutes at 4° C, and three phases were obtained: a) a top lipid phase, b) a middle aqueous phase, and c) cellular components in the bottom. The aqueous phase was separated, trying not to drag the surface phase's remains or the cellular components. These top and bottom phases were discarded. The aqueous phase was centrifugated at 8,000 g for 30 minutes at 4° C (Allegra X-22 Series Beckman Coulter® USA centrifuge). Two phases were obtained, supernatant and pellet. The supernatant was stored in aliquots at -70° C for the determination of free immunoglobulins by ELISA. The pellet was processed to separate the bacteria associated with IgA subclasses (IgA1+ and IgA2+) or free fraction (IgA-).

Separation of bacteria into IgA1+ and IgA2+ fractions
Pellets were processed to separate bacteria fractions according to their association with IgA subclasses, as previously reported [23]. The pellets were incubated for fteen minutes at room temperature with 0.25% BSA, 5% Calf Fetal Serum (CFS) (Cat# 26140, Thermo Fisher®), and 2 mM Ethylenediaminetetraacetic acid (EDTA) (Cat# 6381-92-6, JT Baker®) in PBS. Subsequently, they were washed ve times with PBS and resuspended in a nal volume of 1 mL. Total sample volume was separated into four sterile tubes: a) total bacteria, b) IgA-bacteria, c) IgA1+ bacteria, and d) IgA2+ bacteria. Separation of the IgA-associated bacterial fractions was executed with biotinylated antibodies to human IgA subclasses: anti-human IgA1 (Cat# ab99796, Abcam®) at a dilution of 1:2500 and antihuman IgA2 (Cat# ab128731, Abcam®) at a dilution of 1:2,000. Subsequently, streptavidin complexes associated with Allophycocyanin-Cy7 (APCy7) (Cat# 405208, Biolegend®, San Diego, CA, USA) or Two aliquots of 10 µL from each bacterial fraction were separated and resuspended in 100 µL of PBS to: a) quantify bacteria using a Bacteria Counting Kit (#Cat B7277, Thermo scienti c® CA, USA) and b) evaluate viability using a Bacterial Viability Assay Kit (Cat# ab189618, Abcam®, Cambridge UK); in according to the supplier´s instructions. In brie y, for quanti cation protocol, 1 µL of SYTO® BC bacteria stain to each aliquot and incubated at 37°C, for 5 minutes and washed three times with PBST. Finally, 10 µL of the microsphere suspension was added to the stained cell preparation, for 10 min; washed with PBST for three times and analyze in ow cytometer. For viability assay, 3 µL of a mixture with equal volumes of SYTO® 9 and propidium iodide (PI) was added for each milliliter of the bacterial suspension, incubated at room temperature in the dark for 15 minutes. Finally, mixture was washed three times with PBST and analyze in ow cytometry.

Enzyme protease kinetic assay
To demonstrate that loss of IgA1 signal was due to bacterial enzymatic degradation activity, we used human IgA1 protein (Cat# ab91020, Abcam®), at 10 mg/mL or human IgA2 protein (Cat# ab91021, Abcam®), at 12 mg/mL diluted in PBS. These proteins were to interact with 100,000 IgA1+ bacteria fraction or 200,000 IgA2+ bacteria fraction isolated from human colostrum; at 4°C (as temperature control) or 37 °C and with or without protease inhibitor cocktail (Cat #P8465, Sigma-Aldrich®); for two hours. At each 20 minutes interval, we took an aliquot and were frozen until the moment of ELISA quanti cation assay, as described below. Spectrophotometer (Cat# ND-1000 Thermo Scienti c®). DNA quality was evaluated by electrophoresis in 0.5% agarose gel, using Tris-Borate-EDTA (TBE), at 90 Volts for 50 min. The gels were stained with 0.80 μL of Midori Green dye advanced (Cat# CLS-1909-G Nippon Genetics® Europe, Dueren, Germany) dilution 1:15. The gels were documented by Chemidoc TM XRS+ System Molecular Imager Gel (Cat# 170-8070 Bio-Rad®, Hercules CA, USA). As positive control we used 10 ng/uL of DNA extracted from an Escherichia coli strain (E coli SK100 19p EJ01).
Ampli cation of the V3 region of 16S rDNA An approximately 281 base pair (bp) fragment corresponding to the hypervariable region 3 (V3) of the bacterial 16S ribosomal DNA (16Sr DNA) was ampli ed from each DNA sample. Universal primers amplify the V3 polymorphic region of the rDNA gene, V3-341F (CCTACGGGAGGCAGCAG) and V3-518R (ATTACCGCGGCTGCTGG), were used, containing the adapter and barcode sequences for multiple samples massive sequencing. Ampli cation was carried out by end-point polymerase chain reaction (PCR) in Applied Biosystems® 2720 Thermal Cycler (ThermoFisher Scienti c®), working with a nal DNA quantity of 50 ng per reaction. Bands of interest were analyzed with the documentation system Molecular Imager® for expecting a fragment size of 281 bp. Libraries were sequenced on the Ion Torrent PGM platform (Life Technologies®, Thermo Fisher Scienti c®) using Chip 318 (Life Technologies®, Thermo Fisher Scienti c®), as previously described [7]. As positive control we used a Microbial Community Standard (#Cat D6300 Zymo Research®, Irvine CA, USA).

High-throughput DNA sequencing
Samples with speci c barcodes were mixed, puri ed, and the nal concentration of DNA in the library was quanti ed. Amplicons size in the library was veri ed using the Bioanalyzer equipment. For templates preparation, emulsion PCR spheres with Ion Sphere TM Quality Control Kit (Cat# 4468656 Life Technologies®, Thermo Fisher Scienti c®) were conditioned, and massive PCR reactions were implemented in the Thermocycler system. Likewise, the spheres' quality was veri ed with the sample sequenced using Ion OneTouch TM system equipment (Cat# 4474779 Thermo Fisher Scienti c®) and Qubit 2.0 uorometer Massive sequencing (Cat# Q33327 Thermo Fisher Scienti c®). Finally, samples' sequencing was performed by Ion Torrent PGM TM Sequencer equipment (Cat# A25511 Thermo Fisher Scienti c®).

Analysis of sequencing data
Data were processed using Torrent Suite v4.4.3 software (Thermo Fisher Scienti c®) to exclude polyclonal and low-quality sequences. The reads were assigned to the source sample employing the DNA primers system. The quality control of sequences was performed with FastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc), and all readings were trimmed to 200 nt in length with Trimmomatic v0.36 (http://www.usadellab.org/cms/index.php?page=trimmomatic).
Readings that passed the quality tests were exported as FASTQ les that were later transformed into FASTA les and were analyzed with QIIME pipeline software version 1.9.1. [31,32] (http://qiime.org/). Sequences were grouped into Operational Taxonomic Units (OTUs); the taxonomic assignment was carried out by aligning the sequences against the 16S rDNA Greengenes database v13.8 (http://greengenes.lbl.gov/). Microbial diversity was calculated using alpha diversity (within samples) and beta diversity (among samples). Alpha diversity was estimated using different indices: observed OTUs, Chao's index, Shannon's index, and Simpson's index. These indices were determined using phyloseq and ggplot2 packages in the R environment (v3.4.4). For beta diversity, dissimilarity was estimated using UniFrac analysis (weighted and unweighted). A two-dimensional scatter plot was generated using principal coordinate analysis (PCoA). ANOSIM and Adonis statistical tests were applied. The relative abundance of the bacterial groups identi ed in the samples was compared by the Wilcoxon rank test using the SPSS software (version 14.0). Data analysis (group comparison) was performed by Student's t-test or by U Mann-Whitney. Values of p <0.05 were considered signi cant. Statistical operations were performed using the SPSS program (version 14.0). Linear discriminate t analysis (LDA) effect side algorithm (LEfSe v1.0) was used to identify statistically signi cant taxa in different groups of samples. We used LDA Fisher's linear discriminant to estimate each taxon's effect size between groups. For LDA analysis of bacteria in maternal milk samples and IgA2-associated bacteria in neonate feces fed with colostrum; thresholds of signi cance of 2.0 and 3.5 were used, respectively. PICRUST v1.1.1 (Phylogenetic Investigation of Communities by Reconstruction of Unobserved States) was used to predict functional metagenomes based on 16S rDNA gene data with the Kyoto Encyclopedia of Genes and Genomes (KEGG) orthologue classi cation database in hierarchy level 3 pathways. Statistical analysis of Taxonomic and Function software (STAMP v2.1.3) was used to determine signi cant differences in OTU's abundance and metabolic pathways. We performed a Source Tracker analysis to predict the origin of OTUs in each neonatal stool sample using corresponding human colostrum as a potential source to estimate the proportion of bacteria present in fecal samples attributable to the human milk. This analysis was made in the QIIME platform using Source Tracker (v0.9.5) software [33].

Carbohydrates microarrays
To identify IgA subclasses reactivity against bacteria carbohydrates, we used Glycan Array 100 for 8 sample kits (Cat# GA-Glycan-100-1 RayBiotech® Norcross GA, USA). Slides were rinsed with PBS and blocked with 1% BSA in PBST containing 0.025% NaN 3 , at room temperature for 30 min. Each subarray was stained with 40 µL of each supernatant sample in 1:9 dilution with PBST containing 0.05% NaN 3 and 1% BSA. The slide was incubated in a humidi ed chamber at room temperature for 90 min, and the slide was washed ve times with PBST. The slide was then incubated with 40 µL of secondary mouse antibody anti-human IgA2 coupled with FITC (Cat# 9140-02 SouthernBiotech® Birmingham AL, USA) dilution 1:2500. The slide was incubated in a humidi ed chamber with light protection at room temperature for 30 min and washed ve times, as previously described. Fluorescence signals were detected with Chemidoc TM XRS+ System Molecular Imager Gel (BioRad®).

Statistical analysis
Data were reported as mean ± standard deviation (SD) or frequencies and percentages (%). Two-tailed Student's t-test or Mann-Whitney U test was assessed to compare groups using SPSS v23.0 software (SPSS, Inc). ANOSIM and Adonis were used for category comparisons of phylogenetic distance matrices (UniFrac). Linear regression was used to know the relationship between microbiota diversity as the dependent variable and maternal and neonatal age included as covariates; p ≤ 0.05 was considered statistically signi cant. The Benjamini−Hochberg (BH) correction method was used to estimate the false discovery rate (FDR) and lter the data where a q-value ≤ of 0.05 was considered statistically signi cant.

Results
The bacterial microbiota in the human colostrum are differentially decorated by IgA subclasses.
We analyzed percentages of bacteria associated with IgA subclasses from total bacteria present in human colostrum samples (Fig. 1a) (Fig. 1b). Data showed that IgA1 decorates about 9.67 % of the total bacteria present in colostrum, 21.40 % are decorated with IgA2, and 5.09 % are linked to both subclasses (Fig. 1b); demonstrating that IgA subclasses have a selective recognition with colostrum bacteria.
To evaluate the colostrum IgA-microbiota association effect during bacteria establishment in neonate's intestine, we characterized and compared IgA subclasses association with bacteria in maternal colostrum (Colostrum) versus pasteurized milk (Pasteurized) samples (Figs. 1c -1e). The data demonstrated that IgA2 is more abundant than IgA1 in colostrum (Fig. 1c). At the same time, IgA2 is more frequently associated with bacteria surfaces (Fig. 1d). In comparison, IgA subclasses in pasteurized milk samples were signi cantly reduced. Milk pasteurization denatures antibodies affecting concentration and ability to recognize antigen [35]. We observed fewer bacteria with IgA-associated on their surfaces in pasteurized milk samples (Fig. 1d) and, in general, a signi cantly reduced number of bacteria present in the samples (Fig. 1e). Consequently, IgA1 + and IgA2 + bacteria fractions in the pasteurized milk samples were not detected. These results strongly suggest that fresh, non-pasteurized milk is a source of IgA-decorated bacteria.
Some bacteria have divergent recognition between IgA subclasses in the human colostrum.
Bacterial microbiota relative abundances were determined at genus and family level in the fractions according to their association with IgA subclasses. Twenty bacterial genera were predominantly in colostrum, including, Clostridium, Bacteroides, Pseudomonas, Bi dobacterium, Kaistobacter, Corynebacterium, and the Enterobacteriaceae family (Fig. 2a).
As previously mentioned, bacteria fractions showed distinctive genera between IgA + subclasses fractions in human colostrum (Fig. 2a). This sharp divergence in the composition between IgA subclasses fractions composition was con rmed by LEfSe analysis (Fig. 2b). LEfSe (Linear discriminate t analysis effect size) algorithm was used to identify signi cant taxonomic differences in the microbiota of the groups of samples studied. LEfSe Kruskal-Wallis rank-sum test was used to detect signi cant characteristics of different classes, the Wilcoxon rank test, and linear discrimination analysis (LDA) to estimate the effect size. For analysis between the IgA subclass fractions in colostrum samples, it was found that six bacterial genera Clostridium, Streptococcus, and Staphylococcus associated with IgA1 and Bi dobacterium, Lactobacillus and Bacteroides associated with IgA2, were signi cantly different between fractions (Fig. 2b). Using PCoA analysis, it was determined that, although individual samples share composition between IgA fractions, most of them presented differences between them depending on IgA subclass recognition (Fig. 2c). These data demonstrate a divergent recognition of microbiota by IgA subclasses in human colostrum.
IgA2 is predominantly found decorating bacteria in newborn feces of breastfed-children.
The next step was to evaluate the microbiota present in newborn meconium and stool samples according to its association with IgA subclasses (Fig. 3a-c). As expected, bacteria associated with the IgA subclasses were not found in meconium samples (the rst evacuation before breastfeeding) (Fig. 3a-b).
Breastfeed infants had the highest levels of IgA subclasses (Fig. 3d), IgA-associated microbiota (Fig. 3e), and bacterial quantity (Fig. 3f) compared to pasteurized milk-fed infants (Fig. 3d-f). Free IgA1 concentration in feces decreased in proportion to what was found in colostrum; however, IgA1-associated microbiota was neither found in breastfed-infants (Fig. 3c) nor pasteurized milk-fed samples (Fig. 3e-f). Perhaps the presence of IgA1-speci c bacterial proteases in addition to intestinal lumen conditions may explain this reduction [36,37]. IgA2 would be more resistant to these conditions. We developed a kinetic enzyme assay to explain the observed differences. The data indicated that only 20 minutes of exposition to the bacteria from either colostrum or feces is enough to decrease in 40% the amount of IgA1 ( Supplementary Fig. 1). Then, as a possible explanation, IgA1 from the colostrum is degraded more rapidly by bacteria and enzymes in the newborn intestine. For that reason, it is not detected decorating the bacteria recovered from feces of breastfed infants.
Thus, the results indicated that in stool samples from neonates fed exclusively with colostrum during the rst three days of life, a percentage of the recovered bacteria were exclusively associated with IgA2 (Fig. 3f). In contrast, no bacteria decorated with IgA were found in pasteurized milk-fed infants, similar to meconium samples.
IgA2 + microbiota has a role on bacterial colonization of the neonate's intestine.
Our results found that meconium has a more signi cant presence of the genera Pseudomonas, Staphylococcus, Clostridium, Streptococcus, and the Enterobacter family. In contrast, pasteurized milk samples showed a greater abundance of Clostridium and Pseudomonas genera. Finally, stool samples of children fed with colostrum (Clostridium, Pseudomonas, Bacteroides, Bi dobacterium, Kaistobacter, Lactobacillus, and Enterobacteriaceae family) in comparison with pasteurized milk (Clostridium, Pseudomonas, and Enterobacteriaceae family) showed signi cant differences in their bacterial composition (Fig. 4a).
Observed OTUs, Chao1, Shannon, and Simpson indexes were used to analyze Alpha diversity within each sample. Total bacterial fraction showed higher bacterial richness and diversity compared with IgA subclasses (86% vs. 81%, p = 0.03 for the nonparametric Wilcoxon rank test) (Fig. 4b). Regarding the Shannon index, higher values were found for stool samples of children fed with breast milk and colostrum samples, in contrast to formula and meconium samples (Fig. 4b). As expected, meconium and pasteurized milk samples had the highest dominance indexes. For the rest of the samples, no statistically signi cant difference was found.
As a result, 24 taxonomic groups were signi cant, 6 for IgA2 + and 18 for free bacteria in feces (Fig. 4c). However, a comparison between IgA2 fraction samples from colostrum and breastfed children's feces demonstrated they were very similar. According to PCoA analysis, the data showed that meconium has a very characteristic composition. Fecal bacteria composition from children breastfed for three days is more similar to the fraction of IgA2 + bacteria in colostrum, in comparison with samples from children fed with pasteurized milk (Fig. 4d). To estimate what portion of the fecal microbiota was colostrum, we used Source Tracker analysis. The analysis suggested that 70 % of intestinal microbiota had its origin in the colostrum. This microbiota is composed of Clostridium, Bi dobacterium, and Lactobacillus genera ( Supplementary Fig. 2). The remaining 30 % of the microbiota s characterized by bacteria from other sources like the human perianal zone [38]. These results strongly suggest that colostrum is an essential source of neonate bacteria and its association with IgA2 regulates the microbiota establishment in the newborn intestine.
The predicted microbial metabolic pathways among IgA2 + bacteria are enriched in intestinal colonization and active metabolism.
Finally, a diversity analysis of the association of microbiota with predicted metabolic pathways was implemented by statistical tool PICRUSt) to analyze which metabolic pathways may be involved by the relative abundances of bacteria present in the samples and examine whether there are disparities in bacterial composition between IgA2 + fractions of colostrum and feces of neonates exclusively fed with breast milk. PICRUSt analysis identi ed nine functional metabolic pathways and showed that, despite their different inter-and intra-individual variation (alpha and beta diversity) in intestinal microbiota composition from neonates and colostrum, they have particular functional pro les. In general, IgA2 + bacteria in feces from breastfed infants showed a signi cantly increased abundance of metabolic pathways. These pathways were overrepresented in feces samples from infants fed with colostrum (p < 0.04) compared to infants fed with formula. Among them, bacterial invasion of epithelial cells, folate biosynthesis, and carbohydrate metabolism had the most signi cant difference (p = 1.34e-4). In the same way, transcription machinery, fatty acid metabolism and pentose-glucoronate interconversions metabolic pathways were underrepresented in the same fractions (6.76e-3) (Fig. 5). All these data suggest that IgA2 association could help to shape microbiota on neonate during bacteria colonization, providing an advantage to associate speci c genera enriched with folate biosynthesis and carbohydrate metabolism to establish in the mucosal intestine.
Bacteria association with IgA2 is dependent on glycans/protein recognition.
Since our results indicated an enrichment of metabolic pathways associated with folate and carbohydrates synthesis, we decided to evaluate whether, as previously published [34,39], the IgA subclasses presented differential recognition on antigens depending on their chemical nature. Our data showed IgA2 a nity for carbohydrates like N-acetylglucosamine (D-GlcNAc), D-Mannose (D-Man), and alpha-D-Mannose (a-D-Map) which are expressed on the bacterial external membrane or wall cell (Fig. 6a). Both colostrum IgA subclasses recognize SHMT, present in both isolated fractions (IgA1 + and IgA2+) (Fig. 6b). These data suggest that the recognition of IgA by bacteria in colostrum is the product of a polyreactive effect to common antigens among different bacterial genera, similar to what occurs in the intestine [34].

Discussion
This work provides information about the role of colostrum IgA2 in microbiota colonization during the postnatal stage. We corroborated evidence that colostrum is one of the primary sources of bacteria, among other sources that colonizes the neonatal gastrointestinal tract from the rst days of life.
According to previous studies [28,40], we found that IgA subclasses show a differential recognition between bacteria groups in agreement with previous reports [26]. The predominance of IgA2 could be a consequence of IgA1 degradation by bacterial enzyme activity. Moreover, by lower microbiota density in colostrum, improving a more selective association with most immunogenic antigens on bacteria surface [41]. From these associations, our data suggest that bacteria decoration with colostrum IgA2 has more active participation during colostrum microbiota neonatal colonization.
We described that meconium microbiota was characterized by the Enterobacteriaceae family and Pseudomonas genera, with enrichment of Clostridium, Staphylococcus Streptococcus, which origin is still under discussion [42]. During vaginal birth, the neonate is colonized by maternal microbiota from the urogenital tract and perianal zone, characterized by Clostridium, Pseudomonas, Bi dobacterium, Lactobacillus, and Enterobacteriaceae. The mother provides colostrum feeding during the rst three days, Lactobacillus, Bacteroides, Pseudomonas, Bi dobacterium, and Propionibacterium decorated with IgA2 (IgA2+).
According to our results, breastfeeding cooperates to shape a neonatal intestinal microbiota increased in the Enterobacteriaceae family and Pseudomonas, Bacteroides, Bi dobacterium, Lactobacillus, and Propionibacterium; associated with IgA2 (IgA2+). Meanwhile, Paracoccus, Phyllobacteriaceae, and Clostridium remain without IgA association (IgA2-). This observation was corroborated when we compared our data with fecal samples from neonates fed with pasteurized milk. This milk had low content and diversity of microbiota, dominated by bacteria that can resist pasteurization processes like Clostridium and Pseudomonas [43]. This low-quality milk shapes an intestinal microbiota dominated mainly, by Clostridium, Pseudomonas, and Enterobacteriaceae, among others; associated with problems in neonate's health [44]. These results strongly suggest that colostrum provides and helps to modulate microbiota colonization.
We found that IgA2 decoration is driven, mainly, by glycans recognition on bacteria surfaces [27,28,45,46]. Theoretically, this IgA has a medium/low a nity to recognize antigens. This association with IgA different bacterial communities may approach and bind to the mucus layer [47]. Association with IgA may promote a tolerogenic response against these bacteria [48,49]. Finally, IgA may also generate niches with commensal bacteria communities in the intestine [50]. Thus, IgA may allow controlled colonization by non-pathogenic microbiota [51], improving their intestine establishment [52,53].
On the other hand, Enterobacter, Propionibacterium, and Pseudomonas were bacterial genera that, only after three days of feeding with colostrum, were found to be associated with IgA2 in stool samples. These data suggest that IgA2 recognizes other bacterial genera already residing in the newborn gut. Through this interaction, soluble maternal IgA2 could help to regulate the composition of the microbiota in situ.
Finally, some bacterial genera such as Paracoccus, Phyllobacteriaceae, and Clostridium, found in the newborn's intestine, were not associated with IgA2. These data suggest that speci c bacterial genera colonize independently of their association with maternal IgA2 or could be part of a temporal microbiota.
Then, Paracoccus, Phyllobacteriaceae, and Clostridium could take advantage of the IgA2 + bacteria's metabolic activity previously established in the intestine for their late colonization. This result also suggests a much more dynamic interaction between the different bacterial genera that inhabits the intestine in the rst moments of life [54]. However, more studies are needed to understand this phenomenon.
All these data suggest that IgA2 cooperates in shaping neonatal microbiota. These rst transitory bacteria arriving at the intestine have some advantage over other bacteria settling late during colonization. Bacteria in the neonate's intestine produce metabolites that engineer conditions for mucosal immune system maturation and the next waves of microbiota establishing during the ablactation [55].
During the rst round of colonization, the metabolites produced may regulate the subsequent settlement of bacteria in the following days of life. According to our data, the most widely represented metabolic pathways are folate synthesis and carbohydrate metabolism. Okay S et al. determined that, in the mouse intestine, bacteria that expressed SHMT enzyme, which regulates folate metabolism, were recognized by high-a nity IgA. This interaction shapes microbiota composition in mice [34]; however, this issue is not well understood in humans.
In according with our results, bacteria IgA2 + at three days after breastfeeding, microbiota is represented by bacteria under early adaptation process. This step is overrepresented by carbohydrate consuming and bacterial invasion, among others, meanwhile, a slow growth ratio related by machinery transcription, fatty acid metabolism and pentose-glucoronate interconversions. Bi dobacterium, Lactobacillus and Propionibacterium genera represent metabolic pathways related with latency growth phase during the rst three days of colonization [56]. These genera need to adapt at new intestinal conditions and consume carbohydrates to adapt and generate physicochemical conditions in small microbial niches [57]. After that, these bacteria must enter to logarithmic growth phase after ve days of adaptation, increasing transcription and DNA and fatty acid synthesis.
This work's main limitation is that bacteria found in feces do not necessarily represent intestinal microbiota composition. Feces have become the reference of most bacteria studies because of is a noninvasive methodology, and sampling is easy and convenient. Previous studies in animal models had demonstrated an essential variation between samples [58]. Perhaps, fecal samples could be more similar to intestinal microbiota composition in neonates because before delivery is a relatively sterile mucosa [59]. However, more studies are necessary to clarify if feces analysis truly represents intestinal microbiota in the newborn. Of course, breastfeeding is only one of the many microbiota sources during the rst days of life. According to our results, IgA2 plays an active role in the maternal microbiota's colonization process in the newborn's intestine. However, functional studies and animal models must verify that this association has biological relevance and a bene cial effect on the newborn's health. This work was focused on IgA2 role during bacteria colonization. Besides, IgA1 could have a more active activity in shaping neonate microbiota during defense against pathogens and stimulating other bacteria's colonization. Future studies in animal models are necessary due to the experimental limitation described above.
The present work contributes to the knowledge of human milk and neonatal stool microbiota in a healthy population and supports the idea of mother-neonate transmission through exclusive breastfeeding.

Conclusion
IgA2 shapes neonatal microbiota composition in the intestine by associating colostrum bacteria through glycans surface recognition to provide them an evolutionary advantage during colonization, at least three days. These genera describe metabolic pro les to induce and control the subsequent genera colonization, suggesting an active role of this IgA2-microbiota association beyond the rst days of life.     IgA1+ bacteria (right) from human colostrum, titled with consecutive numbers. SDS-PAGE gels were applied to either western blot to con rm that each protein's comparable amounts were loaded. Data are representative of two or three independent biological experiments.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.