Gut microbiota plays a fundamental role in various aspects of health and numerous diseases, but the multifarious dynamics of early-life intestinal microbiota colonization are only now beginning to come to light. For decades, it had been believed that the fetus as well as the intrauterine environment is universally sterile, and the earliest microbial exposure and colonization of infant gut begins during delivery and promptly thereafter by the acquisition of maternal (vaginal, anal, and skin microbiota) and environmental bacteria. However, some recent studies have elegantly reported the presence of bacteria in placenta (Aagaard et al., 2014; Collado et al., 2016), amniotic fluid (Oh et al., 2010; Collado et al., 2016), and umbilical cord blood (Jimenez et al., 2005), thereby manifesting that the fetus is already exposed to a variety of bacteria in utero. Several studies have also detected various bacteria in the meconium of healthy babies (Gosalbes et al., 2013; Hu et al., 2013; Del Chierico et al., 2015; Hansen et al., 2015; Collado et al., 2016), further evidencing that the gut microbial colonization may already begin before birth. However, quantitative data on the types and particularly the population levels (i.e., the absolute count) of bacteria that may be present in the meconium is still sparse. Also, the information on how the composition of ‘1st-day flora’ differs between infants born vaginally or by cesarean section is limited. Given that the early-life microbiota plays important role in infant’s physiological, metabolic and immunological development (Matamoros et al., 2013; Arrieta et al., 2015; Kerr et al., 2015) and is strongly impacted by birth mode (Penders et al., 2006; Tsuji et al., 2012; Backhed et al., 2015; Nagpal and Yamashiro, 2015; Bokulich et al., 2016; Nuriel-Ohayon et al., 2016), the numeric data about the population levels of bacteria present in the first intestinal discharge of vaginally-and cesarean-born newborns could be very informative and relevant. In this milieu, we herein aimed to quantify the bacteria present in the meconium of healthy term infants born vaginally or by C-section. Bacteria were quantified by using a highly sensitive culture-independent reverse-transcription-quantitative-PCR (RT-qPCR) based analytical method that targets the bacterial 16S rRNA molecules (Matsuda et al., 2009). We applied this method because it provides quantitative data; whereas 16S rRNA gene-based sequencing methods are largely qualitative and do not provide numerical data of bacterial population levels. Moreover, it is highly sensitive, whereas DNA-targeted PCR methods have relatively lower sensitivity and hence may overlook several subdominant but important bacteria, especially in neonatal stages when the microbiota is underdeveloped. Another advantage is that it provides information on viable cells, whereas the results of DNA-based analyses might also include background DNA (e.g., free bacterial DNA, or DNA of dead bacteria) that may be present in prenatal niches such as amniotic fluid and placenta and also in meconium. We have previously validated that the counts obtained by this RT-qPCR method are equivalent to the bacterial counts enumerated by culturing and fluorescent in situ hybridization (FISH) methods and that the detection sensitivity of RT-qPCR is approximately 100- to 1000-fold higher than that of other molecular methods such as qPCR and terminal restriction fragment length polymorphism (t-RFLP) (Matsuda et al., 2007, 2009; Kubota et al., 2010; Kurakawa et al., 2015b). Therefore, owing to these advantages, we specifically implemented RT-qPCR for the present study. Herein we demonstrate the quantification of bacteria that are present in the first meconium of healthy term Japanese infants born vaginally or by C-section.

We detected several bacterial groups in the first meconium samples of both VG and CS infants. There was no significant difference in the birth weight, height or other generic parameters according to the birth mode (Table 1). The count and detection rate of intestinal bacteria are presented in Table 2. Overall, the meconium of 95% infants were found to harbor one or more types of bacteria; whereas the meconium of remaining 5% infants appeared to be sterile, i.e., no bacteria were detected in these samples. B. fragilis group, Enterobacteriaceae, Enterococcus, Streptococcus, and Staphylococcus were most prevalent members (detected in more than 50% infants), with counts ranging from 10⁵ to 10⁸ cells/g sample, followed by bifidobacteria, lactobacilli, C. coccoides group, C. leptum subgroup, C. perfringens, Atopobium cluster, and Prevotella (Table 2). Of total 151 infants, C. difficile was detected only in one baby (in VG group).

Compared to VG infants, the meconium of CS infants were significantly less often colonized with Lactobacillus genus (6% vs. 37%; P = 0.01) (Table 2). Further examination of eight Lactobacillus subgroups and species revealed that L. gasseri subgroup was the most prevalent subgroup in VG infants (detected in 31% babies) whereas rest of the subgroups and species were detected only in less than 6% babies (Table 3). The detection rate of L. gasseri subgroup was significantly lower in CS babies compared to those born vaginally (6% vs. 31%; P = 0.04). Overall, total seven Lactobacillus subgroups/species, i.e., L. gasseri subgroup, L. ruminis subgroup, L. casei subgroup, L. reuteri subgroup, L. sakei subgroup, L. plantarum subgroup, and L. brevis were detected in the meconium samples of VG group, whereas only two members, i.e., L. gasseri subgroup and L. brevis were detected in CS group (P = 0.04). Other than lactobacilli, no significant difference was observed in the carriage of any other bacterial group in the meconium samples. However, the count of Enterobacteriacea was insignificantly lower in CS babies compared to VG babies (P = 0.058) (Table 2). Also, the overall ratio of facultative vs. obligate anaerobes (in proportion to total bacterial count) was significantly lower in CS babies than those delivered vaginally (P = 0.005) (Figure 1A); and accordingly, the overall proportional bacterial composition appeared to be slightly different in VG and CS babies (Figure 1B). Follow-up analysis of lactobacilli revealed that, compared to VG infants, the detection rate of Lactobacillus genus and several subgroups and species remained significantly or insignificantly lower in CS infants at different time-points during the first 6 months of life (Figure 2); but these differences tended to disappear by age 3 years.

To our knowledge, this is the largest birth cohort studied demonstrating a quantitative profile of the meconium bacterial microbiota in babies delivered vaginally or via C-section. Several remarkable studies have recently reported the presence of bacteria not only in meconium but also in placenta, amnion, and umbilical cord blood (Gosalbes et al., 2013; Hu et al., 2013; Del Chierico et al., 2015; Collado et al., 2016); but these were primarily based on DNA-based sequencing methods which, while being elegantly elaborative, may overlook several subdominant bacteria (e.g., ≤10⁴ cells/g feces), particularly during the earliest phase of life when the microbiota is immature. In this context, our data should fortify these evidences with important information of population levels of not only the predominant but also the subdominant bacteria that may inhabit the infant gut during critical developmental stages. We detected bacteria in the meconium of 95% babies while rest 5% samples appeared to be sterile (i.e., no bacteria detected). However, previous studies on term infants reported the presence of bacteria in 66% (using 16S rRNA gene-targeted FISH method) (Hansen et al., 2015) and 77% (using 16S/23S rRNA gene-targeted qPCR) (Martin et al., 2016) of total babies studied. This difference can most possibly be ascribed to differences in the detection sensitivities of the methods used, and hence could corroborate the advantage of high sensitivity of our RT-qPCR assays. We detected several bacterial clades, albeit with varying detection rates and numbers, in the meconium of both VG and CS infants (Tables 2 and 3), which clearly hint that the intrauterine environment harbors several bacterial communities which may provide the early inoculum for intestinal microbiota colonization. Further, these data show that the facultative anaerobes thrive–and hence are the first settlers–in the primitive gut, thereby pointing toward a relatively aerobic intestinal environment (Backhed et al., 2015; Del Chierico et al., 2015; Bokulich et al., 2016; Collado et al., 2016; Martin et al., 2016). This also suggests that the pioneer microbiota core is comprised predominantly of facultative anaerobes such as Proteobacteria, Staphylococci, Streptococci, Enterococci etc.; whereas obligates belonging to Firmicutes, Bacteroidetes, and Actinobacteria take over the predominance later on during infancy and early childhood, as also evident from the follow-up data of this cohort (Supplementary Table S1). This also concurs well with a recent study by our colleagues who reported that the 1st-day microbiota of healthy infants is by and large predominated by Enterobacteria or Staphylococci, but this predominance is mostly, rapidly transitioned to Bifidobacterium-dominated microbiota within a few days after birth (Matsuki et al., 2016).

Early-life microbiota colonization is strongly impacted by birth mode (Penders et al., 2006). During vaginal delivery, the infant is seeded with certain bacteria from the birth canal; but C-section bypasses this seeding thereby resulting in a different spectrum of pioneering microbiota (Dominguez-Bello et al., 2010; Backhed et al., 2015; Bokulich et al., 2016; Nuriel-Ohayon et al., 2016). Therefore, to determine whether–and how prominently–these differences exist in the first-pass samples, we compared the meconium microbiota of vaginally- vs. cesarean-born infants. The lower detection rate of Lactobacillus genus and the L. gasseri subgroup in the meconium of CS babies was intriguing. To know how long this difference persisted, we followed up the Lactobacillus carriage, and interestingly–albeit not surprisingly–VG and CS infants were found to exhibit dissimilar degrees of Lactobacillus detection rate at several time-points during the first 6 months of life, with detection rate being relatively lower in CS infants (Figure 2). It might also be possible that some vaginal materials are also transmitted from mother to baby during vaginal delivery that may have promoted the growth and activity of lactobacilli in vaginally-born babies; however, we were not able to verify this issue because of the unavailability of the maternal samples in this study. Interestingly, by using same RT-qPCR assays, we recently reported that the vaginal microbiota of healthy Japanese women of reproductive age is predominated numerically by Lactobacillus communities among which L. gasseri subgroup is a major constituent (Kurakawa et al., 2015a). Also, lactobacilli are rarely detected in amniotic fluid and placenta (DiGiulio et al., 2010; Aagaard et al., 2014). Collectively, these studies suggest a prominent mother-to-baby transmission of vaginal Lactobacillus communities during vaginal delivery, while indicating a delayed sampling of lactobacilli in CS babies’ gut. However, given the limitation of large difference between the numbers of vaginally- and cesarean-born infants, this finding needs to be corroborated in studies with comparable number of subjects in the two groups. Also, given that L. gasseri subgroup comprises several beneficial commensal species that can elicit probiotic health benefits such as anti-pathogenic activity, bacteriocin production, immune-modulation, and maintenance of gut homeostasis (Selle and Klaenhammer, 2013); studies are needed to investigate the impact, if any, of its lower/delayed colonization in CS babies. As for the clinical milieu, this delayed sampling does not seem to directly indicate any disease risk, because all these infants were apparently healthy (at least until age 3 years). Nevertheless, differences in Lactobacillus carriage in the ‘first flora’ justify further research and should be an exciting theme for future studies about viable therapeutic targets (probiotics, prebiotics, etc.) in context to the gut dysbiosis, particularly in cesarean-born and/ or preterm babies.

While the birth mode did not have a significant effect on the carriage of other bacteria (apart from lactobacilli), we noticed differences in the ratio of facultative vs. obligatory anaerobes and also in the relative proportions of different bacteria between VG and CS infants (Figure 1). Assuming VG babies as healthy controls, this might contemplate some sort of imbalanced microbiota in CS babies (Arboleya et al., 2012) and might also hint that the trajectory of C-section-associated gut dysbiosis may already instigate as early as the 1st day of life. Slightly higher count of facultative anaerobe community belonging to Enterobacteriaceae in VG babies might indicate the transmission of maternal perianal bacteria during birth, because enterobacteria are typical fecal bacteria and defecation during parturition is not uncommon (de Muinck et al., 2011). We have previously reported that the carriage of B. fragilis group, bifidobacteria and several other bacteria remain inconsistently lower in CS babies until age 6 months, but these differences begin to appear after 3–7 days of age (Tsuji et al., 2012). Together, these reports suggest that while the infant microbiota is definitely, strongly influenced by delivery mode, these differences (with the exception of lactobacilli) are not noticeable at meconium stage and become prominent thereafter (Backhed et al., 2015; Bokulich et al., 2016). This is perplexing and calls for further research, because if the initial dose and pool of microbes in VG and CS babies is similar, then what is it that drives the salient differences in the carriage of certain bacteria (such as B. fragilis group, bifidobacteria etc.) during subsequent stages. Nevertheless, our data show that irrespective of birth mode, the prenatal microbiome is characterized by specific founding bacterial clades which subsequently become more diverse with age as exposure to various environmental and dietary factors increases (Figure 2; Supplementary Table S1). While the source(s) of these bacteria in the meconium remain largely unclear, they are speculated to be of intrauterine origin and suggestive of swallowed amniotic fluid (Ardissone et al., 2014; Collado et al., 2016). Given that the meconium bacterial composition could reveal several features of maternal health (Hu et al., 2013) and might also be associated with several infantile diseases (Gosalbes et al., 2013), further research on the characteristics of these bacteria and their interaction with the host may lead to novel interventions and therapies for newborn’s health.

Our study did have some limitations most important of which was our inability to evaluate the influence of prenatal factors such as maternal diet, antibiotics, stress etc. because of the paucity of information. Also, the influence of maternal microbiota (e.g., fecal, vaginal, and skin microbiota) was not studied because maternal samples were not collected in this open observational study. Even though all samples were from first stools, the time of sampling may have varied in each case and hence might have some influence on the bacterial composition in babies who passed their first stool at an earlier vs. later time. However, because 148 out of total 151 samples were collected within the first 24 h of life, it is highly implausible that sampling time could have any significant bias on the results. While the possibility of environmental contamination during sample collection cannot be completely ruled out; we followed all standard and potential laboratory procedures to avoid as much as possible the chances, if any, of contamination. Moreover, the finding that 5% samples were found to be sterile (no detectable bacteria) also corroborates the sterility during sample collection. While no baby received any kind of formula-feed up to the point of first defecation, some of the babies might have been breast-fed during this short period. However, because of lack of precise information, we could not distinguish such infants to exclude the possibility of contamination with breast-milk’s bacteria. Nevertheless, given that all the mothers and babies were apparently healthy, the data could be extrapolated to the general population.

In summary, our data demonstrate the quantitative picture of bacterial pool dwelling in neonate’s first meconium, and corroborate the evidence that the colonization of gut microbiota may begin already before birth, i.e., in utero. Furthermore, delayed/lower colonization of lactobacilli and relatively different proportions of facultative vs. obligatory anaerobes in the meconium of CS babies indicate that the elements of gut dysbiosis associated with birth mode may start building up as early as the 1st day of life. Given that early-life microbiota acquisition is crucial for newborn’s growth and development, our data underscore the need for broader studies to elucidate the sources, routes and significance of these microbial clades in the prenatal niches, as well as to investigate the impact of these primitive microbial exposures on newborn’s long-term health.

RN, HT, KN, and YY: conception and design of study. RN and HT: performed experiments. HT and KK: coordinated sample collection. RN and HT: analyzed data. RN, HT, KN, and YY: interpreted data. RN: drafted manuscript. RN, HT, TT, KN, KK, SN, and YY: edited and revised manuscript. RN, HT, TT, KN, KK, SN, and YY: approved final version of manuscript.

RN is supported by postdoctoral research fellowship from Juntendo University. HT, TT, and KN are employees of Yakult Honsha Co. Ltd, Japan. SN and YY have received unrestricted research grants from Yakult Honsha Co. Ltd. The author KK declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.