This week in eLife, our lab published a study entitled Gut bacteria are rarely shared by co-hospitalized premature infants, regardless of necrotizing enterocolitis (NEC) development. Spearheaded by a talented Banfield Lab post-doc, Tali Raveh-Sadka, in collaboration with Michael Morowitz’s Lab, the study aimed to find the causative agent in an outbreak of NEC cases that happened last summer. NEC is a life threatening gastrointestinal disease that primarily afflicts preterm infants and the cause of disease remains cryptic. While the hypotheses and aims of the study were human focused, the results have a direct impact on how the readership of microBEnet might view built environment microbial communities. But first, the main findings from the study:
- The gut microbiome of infants that develop NEC will share a subset of microbial strains that likely cause or contribute to the onset of NEC, while co-housed healthy infants will not share these strains.
- The recovery of whole genomes necessary for comparison of microbial populations at the strain-level is achievable within a clinically relevant time frame.
- Surprisingly, there is almost no overlap between the 10 infants in the cohort. Out of 149 strains, only 4 are shared between infants! Additionally, microbes typically classified as pathogens colonized both diseased and healthy infants. Table 1 in the main text plots the strain-level distribution, where heatplots in the same row reflect the same strain found in multiple infants. Additionally, no pattern differentiating diseased and healthy infants at higher taxonomic level or through our metabolic analyses were detectable.
- From the last sample collected (55 total fecal samples) to genome assembly completion and general metabolic analysis, our approach took as little as 3 days for an infant sample set. I wont elaborate on this too much, other than to say this is quite fast and clinically relevant for some applications. As sample prep, sequencing, and informatics workflows get faster, soon whole genome analyses will be the norm. Check out the paper and visit ggKbase (one of the main tools used in the project) for more details.
At first the negative results (#1 above) were extremely unsatisfying. We wanted to save the babies! We wanted to find the smoking gun. Previous reports of the distinct colonization pattern of preterm infants and observations that NEC tends to occur in outbreak clusters indicated that microbes could be the etiological agent. Our results certainly don’t preclude microbes from playing a role, but the data suggests the cause to be more multifaceted — likely a combination of abnormalities in host development, microbial community maturation, and environmental factors.
The most interesting and unsuspected finding of the study (to me at least), was the diversity of strains found in infants housed in the same ward, at the same time. As a graduate student studying how the built environment effects building occupants, this leaves a number of important questions to be answered:
- What does the strain-topology or biogeography of the neonatal intensive care unit (NICU) look like? This has been done in a few studies doing surveys of the 16S rRNA gene (these approaches yield family to species level classifications depending on the method used and environments surveyed), but the strain-level resolution offered in this study offers orders of magnitude more information. Many of the 16S genes assembled in this study were identical, masking the true nature of the shifting strain populations in the infant gut. Whole genome recovery unmasks these ambiguities, offering the chance to study microbial populations in their natural context.
- Once the environment has been mapped, are there environmental factors that could predict how an infant is colonized while housed in the NICU? It’s important to note, most of these infants received fairly aggressive antibiotic treatments within the first weeks of life. This likely decouples infants from the typical source inoculum acquired during the birthing process.
- If there are environmental indicators or vehicles for microbial dispersal identified, what intervention methods are possible to deliver “better” microbes. The scientific community still has a long way to go to determine what constitutes a “better” microbial consortium, so perhaps blocking delivery of “bad” microbes may be a better approach (I’m thinking pathogens known to cause serious outbreaks).
The paper is definitely worth a look! There are several other interesting topics I didn’t mention, such as the use of ddPCR for biomass quantification (could be interesting for those who are attending the live/dead workshop or generally interested in techniques for bacterial counts) and the use of the fast evolving CRISPR loci to better differentiate strains in a population.