Co-occurence of Secretory Immunoglobulin A-Coated Bacteria in Maternal Gut, Breast Milk, and Infant Gut in Humans


 BackgroundGut microbiota promote and maintain infant health. Vertical transmission of bacteria from the maternal gut through breast milk to an infant is an important source of microbial colonisation in human offspring. However, the causative active/culturable bacteria and mechanisms responsible for their mother-neonate vertical transfer via breastfeeding remain unclear. Secretory immunoglobulin A (sIgA) may mediate this vertical transmission; however, evidence supporting this hypothesis is required. In this study, we aimed to investigate whether sIgA-coated bacteria in the maternal intestine may migrate to breast milk and colonise the infant gut.ResultsMaternal faeces, breast milk, and neonatal faeces were collected from 19 mother-infant dyads during lactation stages specific to colostrum, transitional, and mature milk. sIgA-coated bacteria were enriched using magnetic-activated cell sorting, and live bacteria were cultured in lactic acid bacteria- and gut bacteria-specific medium. 16S ribosomal RNA gene amplicon sequencing showed that microbiota diversity in maternal faeces, breast milk, and infant faeces decreased sequentially from colostrum to transitional milk to mature milk. Significant beta diversity existed between sample types (p < 0.05). However, high similarity was found between sIgA-coated microbiota of the three types of samples at the mature milk stage. Source track analysis showed that sIgA-coated microbiota in breast milk and maternal gut are major contributors of sIgA-coated microbiota in infant gut. Genera with co-occurrence in sample types included Bifidobacterium, Enterococcus, Streptococcus, Lactobacillus, Klebsiella, Escherichia-Shigella, and an unclassified genus of Enterobacteriaceae. Shotgun sequencing of three dyads identified co-occurring species Lactobacillus and Bifidobacterium, including Lactobacillus salivarius, Lactobacillus reuteri, Lactobacillus gasseri, Lactobacillus jonsonii, Lactobacillus oris, Bifidobacterium longum, and Bifidobacterium breve. Conclusions﻿Breastmilk and infant faeces samples showed unique microbial composition and diversity in the three lactation stages. The fractions of sIgA-coated microbiota in maternal faeces, breast milk, and infant gut showed similar bacterial abundance patterns. This study will facilitate development of strategies to adjust aberrant microbial establishment and reduce the risk of disease by providing essential information for effective probiotic administration to the neonate and/or breastfeeding mother.

pioneering studies. The origin of human milk bacteria remains largely unknown. Although the infant's oral cavity and maternal skin may provide microbes to milk, selected bacteria of the maternal digestive microbiota may access the mammary glands via oral and entero-mammary pathways involving interactions with immune cells. Although Streptococcus and Staphylococcus are predominant in human milk [3], a low abundance of gut-associated genera, including that of Bacteroides, Clostridium, Faecalibacterium, Roseburia, and Bi dobacterium, was repeatedly detected in breast milk [4]. However, the causative active/culturable bacterial species/strains and mechanisms responsible for motherneonate vertical transfer of bacteria via breastfeeding remain unclear. Recent research has focused on shared bacterial species/strains between breast milk and infant faeces in mother-infant dyads using microbiological isolation and strain-level genotyping of bacterial isolates [5]. The ndings con rmed that only limited isolated bacterial species were shared by a few mother-infant dyads, and were dominated by Staphylococcus (Staphylococcus epidermidis, S. hominis) and Escherichia/Shigella spp, with lower levels of Bi dobacterium (Bi dobacterium breve and B. longum) and Lactobacillus (Lactobacillus fermentum, L. gasseri, L. plantarum, L. reuteri, L. salivarius, and L. vaginalis) [6]. Rare breast milk bacteria that occur at levels below the detection limit have been shown to colonise the gut in a breast milk microbiotaassociated mouse model [7]. Metagenomic sequencing analysis identi ed the sharing of only B. longum and Enterococcus faecalis across maternal gut and breast milk [8]. Therefore, alternative methods are required to identify additional candidate bacteria that are potentially transferred from maternal breast milk to the infant gut.
The entero-mammary pathway suggests that microbes located in the maternal gut translocate to the mammary glands, and upon milk consumption, colonise the infant gut [9]. Peyer's patches (PPs) in the small intestine and associated microfold cells represent primary sites for uptake and presentation of lumen bacteria [10]. Microfold cells can selectively uptake secretory immunoglobulin A (sIgA)-coated bacteria through speci c immunoglobulin (Ig)A receptors that can be internalised by subepithelial dome PP dendritic cells (DCs) [11] [12]. DCs can carry internalised bacteria to mesenteric lymph nodes and retain small numbers of live commensals for several days [13]. These commensals can be transported through the mucosal lymphatic circulation to other parts of the body, such as the mammary glands [14].
sIgA-coated bacteria in PPs that are closer to DCs suggest that transference to the mammary gland is easier than transference of bacteria to the intestinal lumen [15]. sIgA is also secreted into the breastmilk of lactating mothers and may support the establishment and maturation of the gut microbiota in early life by colonisation promotion [16]. Therefore, sIgA may mediate the vertical transmission of speci c bacteria from mother to child. However, evidence supporting this hypothesis is required.
A mouse model revealed that sIgA level was highest on bacteria from the proximal small intestine and lowest in those from the colon, and sIgA coating was not homogenous, yet speci cally enriched for some microbiota members [17]. Bacteria that demonstrated sIgA coating in the colon were typically also observed in the small intestine [17]. Commensal members of the facultative anaerobic γ-Proteobacteria or Lactobacillus are frequently motile bacteria with the ability to colonise in close proximity to the healthy epithelial surface and exhibit a high degree of sIgA coating [18].
The current study aimed to identify sIgA-bound bacterial genera in maternal faeces, breast milk, and infant faeces in mother-infant dyads in three stages of lactation. 16S rRNA sequencing and cluster analysis were used to verify whether bacterial transfer occurred in an sIgA-coated state. The ndings of this study increases our understanding of neonatal gut development and provides future opportunities for adjusting aberrant microbial establishment and reducing the risk of disease by providing essential information for effective probiotic administration to the neonate and/or breastfeeding mother.

Results
Study sample characteristics Information regarding basic anthropometrics and reproductive history of all subjects at enrolment is listed in Table 1. On average, women were 30 ± 6 years old, weighed 54 ± 8 kg prior to pregnancy, and had 1.3 ± 0.4 children. All infants were exclusively breastfed at 1 month. Most samples were collected from both mother and infant at each time point; however, we were unable to obtain all samples owing to their mishandling by study personnel. Ultimately, 19 paired samples were obtained, including maternal faeces, infant faeces, and breast milk from each of the colostrum, transition, and mature milk stages, and sIgAcoated microbiota of maternal faeces, mature milk, and infant faeces of mature milk stage, in addition to the absence of one maternal faeces, one colostrum, three mature milk, and one infant faecal sIgA-coated microbiota samples (Supplementary Table 1).

Sequencing summary
All samples were sequenced by bacterial 16S rRNA amplicon sequencing, resulting in a total of 184 highquality metagenomes with average counts per sample of 21399. From these samples, three paired samples of maternal faeces, mature milk, and sIgA-coated microbiota samples of mature milk, infant, and maternal faeces were shotgun sequenced, yielding a total of 15 high-quality metagenomes, with an average of 5.34 (±0.13) bases per sample after quality control.

Alpha and beta diversity
As shown in Fig. 1, the bacterial diversity index (Shannon index) of maternal faeces was signi cantly higher than that of breast milk and infant faeces, except for transitional milk (P <0.05). There was also a signi cant difference (P = 0.02) between infant faeces at the colostrum stage compared with that for the mature milk stage. However, no difference was found among milk from different lactation periods, and among sIgA-coated microbiota of different samples. For bacterial richness estimation (Chao1), there were signi cantly higher levels in colostrum (P = 0.0.2) and transitional milk (P <0.001), and lower level in infant faeces of colostrum stage (P = 0.02) compared with maternal faeces, respectively. There were no differences between breast milk and infant faeces samples at different lactation periods and samples of sIgA-coated microbiota.
Potential contribution of the maternal gut and breast milk to infant gut bacterial communities Using Feast software for microbial source tracking, we estimated likely contributions to infant faecal bacterial communities (sink) using rare ed taxon read counts of operational taxonomic units (OTUs) from milk (source), maternal faecal (as a proxy of the maternal colonic bacteria; source), and sIgA-coated maternal faeces microbiota (source), at three lactation periods. For sIgA-coated microbiota, we also studied contributions of the maternal gut (source) and breast milk (source) to the infant gut (sink).
The contribution of maternal faeces to infant faeces microbiota increased from the colostrum to mature milk stage. At the colostrum stage, source proportions were 20%-44% in 22% of mother/infant pairs (Fig.  3A) and changed to a 18%-67% contribution in 50% of mother/infant dyads in the transitional milk stage (Fig. 3B). The contribution increased to 25%-78% at the mature milk stage (Fig. 3C). Breast milk showed a relatively stable contribution during the three stages, ranging from 12 to 86% in 22% of mother/infant dyads. It is evident that sIgA-coated microbiota of breast milk and maternal gut are the major contributors of sIgA-coated microbiota in infant gut, 14~93% of source proportion in approximately 94 mother/infant dyads (Fig. 3D).
Co-occurrence of speci c genera between different sIgA-coated samples.
The unweighted pair-group method with arithmetic means (UPGMA) analysis of sIgA-coated microbiota resulted in three typical clusters (Fig. 4A). The majority of samples of maternal faeces/breast milk, maternal faeces/infant faeces, or breast/infant faeces pairs had similar microbial patterns belonging to cluster G1, accounting for 68% of sample pairs. Clusters G2 and G3, G1 and G2 are speci c samples of maternal and infant faeces, respectively. The abundance distribution of the 30 dominant genera among the three types of samples was displayed in a species abundance heatmap (Fig. 4B). The heatmap revealed several genera that exhibited co-occurrence between samples, including Bi dobacterium, Enterococcus, Streptococcus, Lactobacillus, Klebsiella, Escherichia-Shigella and an unclassi ed genus of Enterobacteriaceae. In addition, Staphylococcus is the dominant bacteria in breast milk and infant faeces; however, it occurs at a very low level in maternal faeces.
We de ned core milk microbiota genera present in at least 90% of individuals with a minimum mean relative abundance of 0.01 % [19], as shown in Fig. 3C. Colostrum is rich in core bacteria and 11 genera.
In the transitional stage, there were seven co-occurrence core genera with colostrum including Streptococcus, Bi dobacterium, Escherichia-Shigella, the three typically core genera of maternal faeces, and Staphylococcus, Klebsiella, Acinetobacter, and Lactobacillus. However, only Staphylococcus, Klebsiella, and Acinetobacter remained in the mature stage. Streptococcus, Bi dobacterium, Staphylococcus and Escherichia-Shigella because core genus of infant gut of corresponding lactation period. At the mature milk stage, infant gut and sIgA-coated microbiota in the mature milk stage, the intestinal ora of infants, and the sIgA-coated microbiota share similar ve core bacteria. Among them, Streptococcus, Bi dobacterium, and Klebsiella are also core genera of sIgA-coated breast milk microbiota. The most prominent feature was that the mother's gut contained 11 speci c core bacteria, suggesting that the infant's gut microbiota is much simpler than that of the adult. Meanwhile, only Escherichia-Shigella and Enterococcus were in sIgA-binding bacteria of maternal faeces.

Different genera and families between sample types
As lactation progressed, Streptococcus abundance in breast milk gradually decreased, while Enterococcus and Bi dobacterium gradually increased. Staphylococcus maintains a considerable superior abundance. Enterococcus and Escherichia-Shigella increased in mature milk at the cost of reduced Lactobacillus (Fig. S1). Edge analysis identi ed several families signi cantly decreased in abundance in mature milk compared with those in colostrum, including Ruminococcaceae, Corynebacteriaceae, Lachnospiraceae, Peptostreptococcaceae, and Microboccaceae (Fig. 4).
In infant faeces, Klebsiella and Escherichia_ Shigella abundance increased and decreased gradually, respectively, during the lactation stages. The abundance of adult-speci c families, Lachnospiraceae and Bacteroidaceae, signi cantly increased in mature milk compared with that in colostrum (Fig. 4). In sIgAcoated microbiota, Lactobacillus and Bi dobacterium became gradually enriched from maternal faeces to breast milk to infant (Fig. S1). Lactobacillaceae in breast milk and infant faeces signi cantly increased compared with maternal faeces (Fig. 5).
Identi cation of co-occurrence species by shotgun sequencing According to the principal component analysis of shotgun sequenced results in Fig, 6A, there was a signi cant difference between breast milk, maternal faeces, and sIgA-coated microbiota. Three different sIgA-coated microbiota showed no separation. In contrast with 16S amplicon sequencing, two additional co-occurrence genera were identi ed by shotgun sequencing, Clostridium and Gardneralla (Fig, 6B). Among the classi ed species, B. longum was the dominant co-occurrence, followed by Bi dobacterium breve. Co-occurrences of Lactobacillus in sIgA-coated bacteria are primarily L. salivarius, L. reuteri, Lactobacillus gasseri, Lactobacillus jonsonii, and Lactobacillus oris. Among them, L. reuteri and L. gasseri were the dominant species in sIgA-coated bacteria in the infant gut.
Kyoto Encyclopedia of Genes and Genomes (KEGG) functional categories shared in metagenomes of sample dyads of sIgA-coated microbiota Eight KEGG pathways (level 4) shared between shotgun sequencing samples were identi ed (Fig. 7). Two pathways were associated with energy metabolism, including sucrose-6-phosphatase, which is involved in starch and sucrose metabolism (K07024) and the LacI family transcriptional regulator involved in maltosaccharide utilisation (K02529). Four pathways were associated with the survival of bacteria in an ever-changing and hostile environment. Among them, the putative ABC transport system permease protein (K02004) is involved in negative regulation of bio lm formation. The putative ABC transport system ATP-binding protein (K02003) can facilitate the acquisition of essential compounds from the extracellular environment. The ATP-binding cassette, subfamily B (K06147), and ABC-2 type transport system ATP-binding protein ABC importers (K01990) evolved the use of multiple mechanisms to transport nutrients across the membrane that aid survival in an ever-changing and hostile environment. Another two enzyme that is widely used in bacteria is the putative transposase K07497 and the ABC-2 type transport system permease protein (K01992).

Discussion
This study compared the membership and composition of microbiomes from maternal milk and faeces, and infant faeces from the same mother-infant dyad at three time points during the lactation period.
Meanwhile, the composition of sIgA-coated bacteria in matched samples of maternal faeces, breast milk, and infant faeces at the mature milk stage were also studied. Samples from different niches showed distinct microbiota structure and core microbiota, as has been shown previously for other subsets of mother-infant pairs [20][21] [22]. The present study provides novel insight into the transmission of bacteria from maternal to infant gut through breast milk. Our ndings provide evidence that sIgA-coated bacteria are shared and potentially transferred from mothers to infants through breastmilk.
In our study, the Shannon index of colostrum was lower than that of maternal faeces; and Caho1 was higher than that of maternal faeces, indicating that colostrum contained a variety of bacteria with low abundance; however, the dominant bacteria were relatively simple. During each lactation period, the alpha diversity of breast milk microbiota was higher than that of the corresponding infant faeces, consistent with other reports [23]. We observed an increase in the Shannon index from the colostrum to mature stage in the infant faecal microbiome, as reported previously [20] [24]. This might be a result of continuous seeding of breast milk bacteria and/or interactions with resident gut microbes, leading to successional community shifts. The beta diversity of the total microbiota does not support the hypothesis that the milk microbiome would most closely resemble that of infant faeces, which concurs with a previous study [22].
The aim of our study was to discover both anaerobic and aerobic, active/culturable, transmissible bacteria that pass from mother to infant via breastfeeding. sIgA-coated bacteria were collected in an anaerobic environment during all steps. Multiple magnetic bead enrichments were performed to verify the purity of sIgA-coated bacteria. The original study focused on sIgA-coated bacteria using bacterial ow cytometry, magnetic-activated cell sorting (MACS), and 16S rRNA gene sequencing (IgA-Seq) to identify IgA-bound bacteria [25]. However, it is possible that some anaerobic bacteria died during bacterial ow cytometry analysis. Our study used only MACS to analyse the composition of sIgA-coated bacteria in mature milk, corresponding neonatal faeces, and maternal faeces. Live bacteria were enriched by lactic acid bacteria and gut microbiota medium (GMM) [26], which enabled the analysis of active bacteria.
Feast-based microbial source tracker analysis showed an increased contribution of maternal faeces and breast milk to infant faeces microbiota from the colostrum to mature milk stage. During the colostrum stage, a large number of bacterial sources were labelled "unknown". The source of these "unknown" bacteria may be the vagina, skin, and intrauterine environment such as the placenta and amniotic uid [27]. There was a high contribution of sIgA-coated bacteria in breast milk or maternal faeces to that of infant faeces. This partially supports the hypothesis that sIgA might mediate vertical transmission of speci c bacteria from the maternal gut to the infant gut through breast feeding. Recent studies revealed that sIgA coats a "diverse but de ned subset of the microbiota" in the gut [28] [29]. This may enhance the colonisation of these bacteria by promoting adhesion and/or nutrient utilisation [16]. These sIgA-coated bacteria may be recognised by sIgA receptors present in the microfold cell in follicular epithelium of Peyer's patch and dendritic cells [11] [12], which may mediate their transport by dendritic cells to the mammary gland [13]. High levels of sIgA are also found in breast milk, with concentrations up to 15 mg/mL in colostrum and ∼1 mg/mL in mature milk, representing >90% of milk antibodies [30], providing the breastfed infant 0.5~1 g/day. Infants are completely reliant on sIgA owing to their immature immune system. Early exposure to passive sIgA in breast milk has lasting bene cial effects by regulating gene expression in intestinal epithelial cells in offspring throughout life [31]. IgA-producing plasma cells in the mammary gland originate from the maternal gut [32]. sIgA speci city in breastmilk is therefore dictated by maternal exposure to commensal bacteria in the maternal gut. This may allow sIgA to coat the maternal gut-derived bacteria in breast milk and help them colonise the infant gut.
We also found that maternal faeces, breastmilk, and infant faeces shared several species of sIgA-coated bacteria including Bi dobacterium, Enterococcus, Streptococcus, Lactobacillus, Klebsiella, Escherichia-Shigella. We previously found that Bi dobacterium, Enterococcus, and Lactobacillus also belonged to the dominant genus in sIgA-coated microbiota of healthy late pregnant women [33]. In addition, although Staphylococcus is the shared bacterium of sIgA-coated microbiota in breast milk and baby faeces, it is not the dominant bacterium of in maternal faeces, which indicates that the species origin is the skin.
However, the shared bacteria of mother and infant samples were not identical to the core bacteria of the corresponding group. This is because no bacteria were commonly shared in all dyads, potentially re ecting inter-individual variability of milk microbiota pro les, consistent with recent research [34]. Nevertheless, increased sharing with prolonged and direct breastfeeding, and strong positive correlations between the milk and gut relative abundances of shared bacteria provide evidence that some degree of bacterial seeding is plausible.
B. longum and B. breve were dominant co-occurring species in sIgA-coated microbiota. A previous study based on real-time polymerase chain reaction (RT-PCR) showed that B. longum was the most commonly found species in breast milk [35]. Some strains persisted in the infant gut while co-existing with the other predominant bi dobacterial species [36]. It was found that the secretory component of sIgA carried glycan residues which can interact with symbiotic bacteria, including Lactobacillus, Bi dobacteria, Escherichia coli, and Bacteroides [37]. Bi dobacteria have been shown to induce high levels of IgA in the gut, which promote the production of sIgA targeting themselves [38]. This was con rmed by a study that revealed that the number of Bi dobacteria and Enterobacteriaceae were high in the IgA-coated fraction of infants [39].
Lactobacillus is another co-occurrence genus in sIgA-coated microbiota in the present study, which is one of the speci c bacterial taxa in human faeces that are highly coated with IgA [25]. Dominant species were L. reuteri and L. gasseri, which are commonly isolated from breast milk [40] [41]. We recently found that L. reuteri is a typical species in the sIgA-coated bacterial fraction of healthy pregnant women [33]. The occurrence of L. reuteri in breast milk was higher in breast milk samples from rural and pastoral areas of China than in those from the industrialised region [33]. Strains isolated from the sIgA-targeted fraction of the former improved intestinal epithelial barrier function in mice [42].
In KEGG functional categories shared in metagenomes of sample dyads of sIgA-coated microbiota, the identi ed pathway was related to sucrose metabolism, maltosaccharide utilisation, and that related to the ABC transport system. A recent study revealed that the ABC transporter is a key genetic factor for fucosyllactose utilisation in Bi dobacteria [43]. ABC-type transporters are also involved in the transport of B vitamins by gut commensal bacteria [44]. This suggests that co-occurrence bacteria can acquire essential compounds from the extracellular environment to survive in the gut.
Limitations of this study include a limited sample capacity and a lack of attention to strain-level identi cation. Additionally, gut bacteria-speci c media were used to activate sIgA-coated bacteria, which may have led to a preference for speci c genera. Whether GMM and De Man Rogosa Sharpe medium could restore the microbial structure of the intestine should be further con rmed.

Conclusions
Breastmilk and infant faeces samples showed unique microbial composition and diversity in the three lactation stages. Fractions of sIgA-coated microbiota of maternal faeces, breast milk, and infant gut showed a similar bacterial pattern, shared with genera including Bi dobacterium, Enterococcus, Streptococcus, Lactobacillus, Klebsiella, Escherichia-Shigella and an unclassi ed genus of Enterobacteriaceae. According to shotgun sequencing, the co-occurrence species of Bi dobacterium and Lactobacillus were dominated by B. longum, L. reuteri, and L. gasseri, sharing functional genes to survive in an ever-changing hostile environment.

Milk sample collection
Thirty-six mothers were recruited after delivery at The A liated Wuxi Maternity and Child Health Care Hospital of Nanjing Medical University from March 2018 to June 2020 (Fig. 8). The recruitment criteria for mother-infant dyads were as follows: (i) vaginal delivery at term (≥37 weeks gestation), (ii) exclusive breastfeeding during the sampling period, (iii) no antibiotic/probiotic exposure of either the mother or infant during pregnancy, intrapartum, or postnatal. Clinical data including maternal age and weight, delivery mode, neonate weight, lactation stage, gestational time, blood biochemistry, and routine blood tests were collected (Table 1). Ethical approval for this study was provided by The A liated Wuxi Maternity and Child Health Care Hospital of Nanjing Medical University, and all participants provided written informed consent.
Pregnancy duration was >38 weeks for all mothers, and samples were collected within 1 month after delivery. Breast milk was divided into three stages according to the time after delivery: colostrum (days 1-5), transitional (days 6-15), and mature (day 16 onwards) [45]. We collected breastmilk at three stages, and the corresponding infant and maternal faecal samples. Faecal and breastmilk sample collection was standardised for all subjects. Prior to milk sample collection, the maternal gland was sterilised with 75% alcohol and the rst drops were discarded to minimise contamination [46]. Spot infant faecal samples were collected from paper diapers on a super clean bench. Maternal faecal samples were stored in a plastic box. Maternal and infant faecal and breastmilk specimens were stored in plastic containers at 4 °C in home refrigerators until they were brought to the study clinic no more than 24 h after collection. At the study clinic, samples were frozen at -80 °C until analysis.

Enrichment of sIgA-coated bacteria through MACS
Milk samples (10 mL) were divided into two parts, one (5 mL) for direct DNA extraction, and the other for identifying sIgA-coated bacteria. Nanoscale aminated magnetic beads and streptomycin a nity nanomagnetic beads were prepared using methods previously developed in our lab [18]. Thereafter, 5 mL of maternal breastmilk was centrifuged at 10,000 ×g for 10 min [47], and the precipitate was resuspended in phosphate-buffered saline (PBS). Either 5% goat serum (40 μL), biotin-labelled rabbit anti-human IgA (20 μL), or streptavidin-labelled nanomagnetic beads (250 μL) were added to the bacterial suspension with PBS, which was subsequently placed in an ice bath for 20 min. Nanomagnetic beads were adsorbed with a magnet, and the supernatant was discarded. Finally, the magnetic bead and bacterial cell combination were washed several times with PBS until the supernatant was clean and the enriched bacteria were resuspended in PBS. Aliquots (100 μL) were added to De Man Rogosa Sharpe medium and GMM [26], and anaerobically cultured for 48 h at 37 °C.
Infant and maternal faecal samples corresponding to maternal breastmilk samples were weighed and dissolved in peptone buffer to prepare a 20% faecal bacterial suspension. Half of the sample was stored with 30% glycerine for DNA extraction, and half was supplemented with 0.5% Tween 20, vortexed, and centrifuged at 1,000 ×g for 10 min. Thereafter, the supernatant was removed by washing the precipitate with peptone buffer three or four times. Breastmilk samples were prepared in the same way; however, peptone was used instead of PBS.
All steps were carried out in an anaerobic glove cabinet (DROID Instruments and Equipment Co., Ltd., Shanghai, China).

DNA extraction
Bacterial genomic DNA was extracted using a reaped bead-beating method using an Ezup Column Bacteria Genomic DNA Puri cation Kit (Sangon Biotech, Shanghai, China) with a minor modi cation as previously reported [48]. Brie y, cell lysis was achieved by bead beating with zirconium beads (0.1 g, 0.7 mm: 0.3 g, 0.15 mm) on an oscillator at 6,000 rpm with two circulations (30 s per circulation; Precellys 24, Bertin Technologies, Montigny-le-Bretonneux, France) in the presence of 4% (w/v) sodium dodecyl sulphate, 500 mM NaCl, and 50 mM ethylenediaminetetraacetic acid. Ammonium acetate was used to precipitate and remove impurities, and sodium dodecyl sulphate in addition to isopropanol precipitation was used for nucleic acid recovery. RNA and proteins were removed or degraded using RNase and Proteinase K, respectively, followed using an Ezup Column Bacteria Genomic DNA Puri cation Kit. Genomic DNA concentration was determined using a Nanodrop 1000 spectrophotometer (Thermo Fisher Scienti c, Wilmington, DE, USA), and DNA integrity was determined by electrophoresis on 1% agarose gels. DNA concentration for all samples was standardised to 10 ng/μL.

16S rRNA gene sequencing
Microbial pro les were analysed by 16S rDNA sequencing at GENEWIZ, Inc. (Suzhou, China). To maximise the effective length of the MiSeq 250PE and 300PE sequencing reads, a region of approximately 469 bp encompassing the V3 and V4 hypervariable regions of the 16S rRNA gene was targeted for sequencing. The PCR primers used to amplify V3 and V4 hypervariable regions were as follows: forward 5'-CCT ACG GRR BGC ASC AGK VRV GAA T -3' and reverse 5'-GGA CTA CNV GGG TWT CTA ATC C -3'. In addition, an indexed linker was added to the end of the 16S rDNA PCR product for nextgeneration sequencing (Illumina, San Diego, CA, USA). First-round PCR products were used as templates for a second round of PCR amplicon enrichment (94 °C for 3 min, followed by 24 cycles at 94 °C for 5 s, 57 °C for 90 s, and 72 °C for 10 s, and a nal extension at 72 °C for 5 min). PCR reactions were performed in triplicate using a 25 μL mixture containing 2.5 μL of TransStart Buffer, 2 μL of dNTPs, 1 μL of each primer, and 20 ng of template DNA. DNA library concentration was validated using a Qubit 3.0 Fluorometer. The libraries were quanti ed to 10 nM, and subsequently multiplexed and loaded on an Illumina MiSeq instrument according to the manufacturer's instructions (Illumina, San Diego, CA, USA). Sequencing was performed using PE250/300 paired-end; image analysis and base calling were performed using the MiSeq Control Software embedded in the MiSeq instrument.
Metagenomic sequencing DNA was sequenced using an Illumina HiSeq 3000 at GENEWIZ Co. (Suzhou, China). Cluster generation, template hybridisation, isothermal ampli cation, linearisation, and blocking denaturing and hybridisation of sequencing primers were performed according to the work ow speci ed by the service provider.
Libraries were constructed with an insert size of approximately 500 bp, followed by high-throughput sequencing to obtain paired-end reads with 150 bp in the forward and reverse directions.
For data quality control, Prinseq [49] was employed to: 1) trim reads from the 3′ end until reaching the rst nucleotide with a quality threshold of 20; 2) remove read pairs when either read was <60 bp or contained "N" bases; and 3) de-duplicate reads. Reads that could be aligned to the human genome (Homo sapien, UCSC hg19) were removed (aligned with Bowtie2 [50] using --reorder --no-hd --no-contain -dovetail).
High-quality paired-end reads from each sample were used for de novo assembly with IDBA_UD [51] into contigs of at least 500 bp. Genes were predicted using MetaGeneMark [52]. A non-redundant gene catalogue of 4,893,833 microbial genes was constructed with CD-HIT using the parameters "-c 0.95 -aS 0.9". High-quality reads were mapped onto the gene catalogue using SOAPaligner [53]. Aligned results were sampled and downsized to 21,399 per sample. The soap.coverage.script was used to calculate the gene-length normalised base counts in each downsizing step. The sampling procedure was repeated 30 times, and the mean abundance value was used in further analyses. Based on the orthologous genes from the KEGG database, gene function was annotated and quanti ed.

Sequence data analysis and statistical analysis
The QIIME data analysis package was used for 16S rRNA data analysis. The forward and reverse reads were joined and assigned to samples based on the barcode and truncated by cutting off the barcode and primer sequence. Quality ltering on joined sequences was performed, and sequences that did not ful l the following criteria were discarded: sequence length <200 bp, no ambiguous bases, mean quality score ≥20. Thereafter, sequences were compared with those in the reference database (RDP Gold database) using the UCHIME algorithm to detect chimeric sequences; all chimeric sequences were removed. The remaining sequences were used in the nal analysis. Sequences were grouped into OTUs using the VSEARCH clustering program (1.9.6) against the Silva 132 database, pre-clustered at 97% sequence identity. The Ribosomal Database Program classi er used the Silva 132 database, which contains taxonomic categories predicted to the species level, for assignment to all OTUs at a con dence threshold of 0.8.
Sequences were rare ed to the minimum library size and total sum scaling was applied. Alpha diversity indices were calculated in QIIME from rare ed samples using the Shannon index for diversity and Chao1 indices for richness. Beta diversity was calculated using PCoA based on the distance between the matrix Brary-Curtis and visualised through PCoA plots, and similarity was measured using the Anosim test (9,999 permutations) [54]. An UPGMA clustering tree was constructed using the weighted clustering hierarchy and the group average method.
The relative contribution of the mother's gut and breast milk to the assembly of bacteria in the infant gut was analysed with feast (V.1.0) [55]. OTUs present in <1% of samples were rst ltered, and the resultant OTU table was imputed with default parameters, with infant faeces as the "sink" and the samples from the mother (faeces, milk) identi ed as the "source". The results were aggregated into three categories: mother gut, milk, and other (unknown).
The core community was de ned as the selection of information on genus taxonomy with a relative abundance of at least 0.01% and present in more than 90% of tested samples.

Consent for publication
Not applicable.

Availability of data and material
The data sets supporting the results of this article are available in the NCBI Sequence Read Archive, BioProject is PRJNA542027 (accession # SUB5578272) (https://submit.ncbi.nlm.nih.gov/subs/sra/SUB5578272/overview).

Competing interests
The authors declare that they have no competing interests.  Beta diversity (Bray-Curtis distance of the operational taxonomic units) of the total microbiota of breast milk (A) and infant gut (B) from different lactation periods and sIgA-coated microbiota from mother gut, breast milk, and infant gut (C). To examine the effect of lactation periods or source of sIgA-coated microbiota on β-diversity, the Anosim test was conducted for pairwise comparison.  indicates the common genus of bacteria among samples. C, Core genus present in at least 90% of individuals with a minimum of 0.01% mean relative abundance in speci c types of samples. C, T, and M stands for sample from colostrum, transitional and mature milk stage, respectively. IF, MF, and BM stand for infant faeces, maternal faeces, and breast milk, respectively.

Figure 5
Signi cantly different family between different sample types. Different families between samples type were identi ed in edgeR and expressed as the log2 fold-change; the threshold was set to FDR < 0.5, |log2 fold-change | > 1.5, and log CPM > 10.0. C, T, and M stands for sample from colostrum, transitional, and mature milk stage, respectively. IF, MF and BM stands for infant faeces, maternal faeces, and breast milk, respectively.  Heatmap of Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway level 4 categories. K07497, putative transposase; K02529, LacI family transcriptional regulator; K02004, putative ABC transport system permease protein; K02003, putative ABC transport system ATP-binding protein; K01990, ABC-2 type transport system ATP-binding protein; K06147, ATP-binding cassette, subfamily B, bacterial; K07024, sucrose-6-phosphatase; K01992, ABC-2 type transport system permease protein.