Read the paper here: https://rdcu.be/ckjSG
Our gut microbes affect our health and wellbeing throughout our lives. Recent data suggest that our relationship with our intestinal passengers is greatest in early life, during critical stages of immunological and physiological development. How our gut microbes participate in shaping our earliest development is still unclear, however. If we want to understand how microbes influence our early-life development, first we must know how and when we are colonized.
Recent studies have sparked controversy by claiming that we are colonized by bacteria before birth. However, studies such as these are fraught with confounders, and methods for controlling for contamination are often not as robust as needed. We believe that many (if not all) of these studies failed to control or account for contamination. Therefore, we set about to develop stringent methods to control for contamination, to test whether the fetal gut is actually colonized before birth. In our study, we found no evidence for prenatal colonization when we controlled for potential sources of contamination and bias, suggesting that our microbiome is indeed born, at birth.
How we did it
- Reduce contamination.
Our first step was to develop a sample collection protocol that minimized contamination. To avoid potential contamination from the mother’s vaginal microbiota, we collected fetal meconium (prenatal stool) samples from elective, breech cesarean sections. To minimize contamination from the mother’s skin microbiota during delivery, we also developed specialized surgical draping and expanded the area of disinfection. Collection from breech deliveries allowed us to sample the meconium before the babies were fully delivered, which minimized contamination from the surgeon and the environment.
- Control for contamination.
Contaminating bacteria and bacterial DNA are everywhere, so it is not possible to completely avoid contamination of your samples. The weaker your “true” bacterial signal is, the greater the impact of contamination “noise”; so, you can’t rely on aseptic technique or “post-hoc decontamination.” Instead, samples must be compared directly to negative controls collected at each contamination opportunity. We collected negative controls at sampling, DNA extraction, and marker gene amplification.
- Control for sequencing noise
Low biomass samples are particularly susceptible to sequencing “noise,” or errors. This is particularly true for mixed-biomass sequencing as abundant microbial signals from high-biomass samples can incorrectly appear in low-biomass samples (such as meconium). To control for this, we compared 2 independent sequencing runs for all our meconium samples. If a microbe shows up in a sample for only 1 of these 2 sequencing runs, it is likely due to sequencing noise and is not a “true” bacterial signal. In our study, we found only one single microbial genus (Bacteroides) in one single meconium sample across both sequencing runs and not in negative controls. As this bacterial genus is very abundant in our high-biomass infant stool samples, we hypothesize that this one signal in one low biomass sample of meconium was due to physical or computational cross-contamination.
Science is thought to be self-correcting, with incorrect findings deemed “irreproducible.” However, if the incorrect finding is due to systemic experimental errors, findings will seem reproducible when the systemic errors are repeated/replicated. This has been the case in studies of in utero colonization, as inadequate use of proper controls has been the rule rather than the exception. After all; “Science advances faster when people waste less time pursuing false leads.” (Nature, Challenges in Irreproducible Research Special).
Our study serves two purposes: it shows 1) stringent experimental methods with robust controls are critical in producing highly reliable, reproducible robust data; and 2) we are not normally colonized before birth. Rather, our relationship with our symbionts emerges during and after birth, which makes it not only vulnerable to early environmental influences, but also offers a window of potential intervention.