C-section delivery is a common surgical procedure intended to increase the chances of successful delivery and to protect the health of the mother. Yet this highly overused intervention [1
] has been associated with higher offspring risk of immune and metabolic disorders [5
] and shown to cause excess weight gain in a murine model [9
]. It is hypothesized that these associations are due to C-section-delivered neonates not receiving the full inoculum of maternal microbes at birth [7
Neonates delivered vaginally are exposed to maternal vaginal microbiota in labor, and additionally to maternal intestinal microbiota at birth [10
]. C-section and the prophylactic antibiotics administered during this surgery [14
] may cause newborns to miss this natural microbial inoculum [14
]. C-section-delivered neonates are first exposed to microbes from the delivery room [17
] and maternal skin (e.g., Staphylococcus
]. Recent longitudinal studies suggest the postnatal impact of delivery mode on the infant microbiome may extend for up to two years [16
]. Yet, since the intestine of the newborn at birth is filled with in-utero contents, which are passed postnatally as meconium, it is unclear when the differences in the gut microbiota of C-section versus vaginally-delivered neonates begin to emerge.
Findings of bacterial DNA in the placenta, amniotic fluid, and newborn meconium have given rise to the hypothesis that the human fetus may be exposed to microbiota in utero through ingestion of amniotic fluid [16
]. Yet no evidence exists for fetal exposure to live bacteria in healthy pregnancies. If microbiota colonize the fetal intestinal tract before labor and birth, this microbial colony should be reflected in the meconium, but not be impacted, or be impacted to a lesser extent, by delivery mode. Thus, we hypothesized that a higher delivery-mode signal should be found in the microbiota of the feces passed after meconium (i.e., transitional stool). To test this hypothesis, we compared the composition, structure and predicted metabolic function of microbiota in meconium and the subsequent transitional stool from neonates born by scheduled elective C-section vs. vaginal delivery.
2. Materials and Methods
2.1. Design and Subjects
Trained staff invited women to participate in the study if they were scheduled for a vaginal delivery or an elective C-section delivery between 38 and 42 weeks of gestation (confirmed by an ultrasound taken before the 20th week of pregnancy) in the participating private hospital in Porto Alegre, Brazil. Women were not eligible for the study if they were administered oral antibiotics in the third trimester of pregnancy; if they had diabetes, hypertensive disorders, autoimmune disorders, or HIV/AIDS during pregnancy; if they had smoked during pregnancy; or if they were on a restrictive diet during pregnancy. For vaginal births, we further only included women whose water broke less than 12 h before delivery. All vaginal deliveries were by spontaneous vaginal birth and no instruments were used. Women also had to consent to allowing study staff collect meconium or stool from their neonate, and to completing a postnatal questionnaire. The Research and Ethics Committee of the Hospital de Clinicas (protocol No. 11/0388) and the Hospital Mae de Deus (protocol No. 524/11) in Porto Alegre, Brazil, approved the study protocols and consent.
From medical records, we ascertained information on mode of delivery, gravidity, parity, history of urinary tract infection during pregnancy, antibiotic use during pregnancy, birth weight, labor length, head circumference, placenta weight, sex and race. Other clinical information was derived from a questionnaire administered after birth before discharge.
2.2. Sample Collection and 16S Ribosomal RNA Sequencing
Of the 89 mother-child pairs who consented and enrolled in the original study, we were able to collect 78 meconium samples (collected within 48 h of birth; all but 2 were collected <24 h). From the 78 meconium samples, 59 (75.6%) had detectable bacterial DNA. Transitional stool samples (collected 2–3 days after birth) with detectable bacterial DNA for analysis were available for 50 of 59 of these neonates. All specimens were collected from diapers with sterile spatulas, transferred to sterile tubes, which were kept at 4 °C for <6 h, and then frozen at −80 °C until DNA extraction.
Total DNA was extracted using QIAamp®
DNA Stool mini kit (QIAGEN) according to the manufacturer’s instructions. After DNA extraction (200 µL of final volume), we concentrated DNA samples to 20 µL using 3 M Na Acetate precipitation and kept it stored at −20 °C. We determined the DNA concentration of prepared samples using the Quant-iT PicoGreen dsDNA reagent and kit (Invitrogen, Grand Island, NY, USA) based on the manufacturer’s instructions. We amplified the V4 region of the 16S ribosomal RNA (rRNA) gene by polymerase chain reaction (PCR) using the Illumina-adapted universal primers 515F/806R [22
]. PCR was done in triplicate using the Bio-Rad CFX 96 thermal cycler (Bio-Rad, Hercules, CA, USA). We then pooled the amplicons in equimolar ratios, purified them using QIA quick PCR purification kit (Qiagen Inc., Chatsworth, CA, USA), and sequenced them on the Illlumina MiSeq platform (Genome Technology Center of NYU Medical Center, NY, USA) using a paired-end technique. Reagents for DNA extraction and for PCR amplification were sequenced as negative controls.
2.3. 16S rRNA Gene Sequence Analysis
We analyzed 16S rRNA gene sequence data using the Quantitative Insights Into Microbial Ecology (QIIME) software package (v1.8) [23
]. Using qualified sequences (Phred ≥ Q20), we identified and quantified Operational Taxonomic Units (OTUs) using an open-reference method that maps sequences with 97% identity to known sequences in the Greengenes database (v13_8) [24
] using UCLUST [25
] and PyNAST [26
] alignment algorithms. We used ChimeraSlayer to identify chimeric sequences. Negative-control derived OTUs were discarded from the OTU table using a filtration script (filter_otus_from_otu_table.py) in QIIME.
We rarefied to 1634 reads per sample to calculate microbial diversity. We calculated alpha diversity using the phylogenetic distance and the detected number of species metrics. Principle coordinate analysis (PCoA) with weighted Unifrac distance metric were used to evaluate beta diversity (i.e., the variation in microbial community composition) between C-section and vaginally-delivered neonatal fecal samples [27
]. We performed permutational multivariate analysis of variance (PERMANOVA) to determine the significance of differences in beta diversity by delivery mode. PERMANOVA, which partitions inter-group and intra-group distances, is a permutation-based extension of multivariate analysis of variance to a matrix of pairwise distances [28
]. We also used non-parametric t
tests with 10,000 Monte Carlo permutations for significant test of microbial diversity.
We used linear discriminant analysis (LDA) effect size (LEfSe) [29
] to identify biologically and statistically significant differences OTU relative abundance. We then used Phylogenetics Investigation of Communities by Reconstruction of Unobserved States (PICRUSt) to predict the metabolic function of the metagenomes from the 16S rRNA
gene dataset [30
], with Kyoto Encyclopedia of Genes and Genomes (KEGG) Orthologs classification [31
]. An LDA score of >3 (among OTUs with at least >1% relative abundance in any group) was used to determine significant differences in abundance of OTUs and metabolic pathways.
In our cohort from Southern Brazil, we found that the transition from meconium to stool involves altered bacterial structure, composition and decreasing alpha diversity, and that delivery mode-associated differences in bacterial diversity, OTU abundances and predicted metagenomic functions arise not in the meconium, but in the transitional stool passed after meconium.
Our findings on the transition in microbiome composition and structure from meconium to transitional stool are consistent with previous work reporting loss of alpha diversity postnatally in mice [32
] and human neonates [33
]. Selective environmental pressures, such as human milk oligosaccharides found in breast milk, may drive the reduction in microbial diversity from meconium to transitional stool. Our finding (Figure S2A
) that initial blooms of Streptococous
and several Proteobacteria phylotypes in meconium were replaced by bacteria in the Clostridiaceae
family (e.g., Lactobacillus
spp.) in transitional stool of vaginally born neonates, is also consistent with findings in mice [32
are facultative anaerobes that ferment milk lactose and casein and produce lactate [34
]. Lactic acid lowers the pH of the intestinal contents, inhibiting growth of obligate anaerobes [35
] commonly found in meconium. The role of this initial reduction in diversity on education of host immune development and metabolic health remains to be elucidated.
Previous investigations on the impact of delivery mode on meconium bacterial composition have been mixed, with some studies having found differences [15
], and others [39
] not. Rocio et al. reported differences in bacterial counts in the meconium by mode of delivery among 108 neonates (80 vaginally-delivered and 28 C-section) [16
]. They also reported differences in microbiota composition over the first 3 months, analyzing 33 different bacterial taxa by quantitative reverse transcription PCR in meconium, 2nd-day stool (transitional stool), and 7-, 30-, and 90-day stools; it is unclear from their study when compositional differences emerged [16
]. Our statistical power to detect large differences may have been limited by the relatively small number of vaginally-delivered neonates in our sample. However, we were able to discern microbiota differences in transitional stools from 10 vaginally-delivered and 40 C-section neonates, and these findings are consistent with a larger sample of transitional stools [40
]. Discrepancies between our study and other studies on this topic might reflect collection, storage processing or analytic platforms [41
]. Differences may also be due to inter-hospital differences in delivery-mode practices, including intrapartum antibiotic administration. While we excluded 3rd-trimester antibiotics, the standard protocol for operative C-section in our hospital in Porto Alegre is intravenous intrapartum administration of cefazolin (half-life ~120 min), 30 to 60 min before surgery [42
]. A recent study found intrapartum antibiotics were associated with an impacted infant microbiome [14
]. Thus, it is possible the impact of C-section observed in our study includes the effect of intrapartum antibiotics.
Our amplicon-sequencing results compliment previous studies [10
] that demonstrated mother-to-newborn transmission at the strain level. Makino et al. have shown that stains found in the mother before delivery, were also found in the meconium [10
]. They have also reported the differences in transmission Bifidobacterium
strains by delivery mode in 12 vaginally-delivered neonates and 5 C-section neonates, but only at 3 days after birth [11
]. In the light of studies showing evidence of bacterial DNA in amniotic fluid [44
], chorioamnion tissue [46
], cord blood [47
], fetal membranes [48
], and the placenta [19
], some investigators have taken evidence of bacterial DNA in meconium to suggest potential translocation of microbiota from mother to fetus before birth. This has been called the ‘in-utero colonization hypothesis’ [41
]. However, no study, to our knowledge, has shown evidence of live bacteria in utero in healthy pregnancies. Regardless of viability, intrauterine maternal bacterial DNA might be important in intestinal transcriptional profiles programming of the offspring conferring to them protection against pathogen colonization [49
The detection of bacterial DNA in the meconium samples in our study does not clearly support nor refute the in-utero colonization hypothesis, as meconium may reflect not only prenatal but also peri- and postnatal exposures [41
]. About 75% of meconium samples in our study contained detectable microbial DNA; other studies have found similar [16
] or lower prevalence of microbes in meconium [50
]. Presence of microbes in meconium has been shown to increase with time [51
], which one might take to support the hypothesis that intestinal colonization starts postnatally [41
]. Yet, if the bacterial DNA present in the meconium of our study was acquired postnatally, one would expect the composition and structure to be altered by delivery mode, as was observed in the microbiota of the subsequent transitional stools in our study.