Healthy human placentas do not harbor a population of microorganisms, and recent reports that concluded otherwise were likely the result of contamination, according to the largest study of the issue to date (Nature 2019, DOI 10.1038/s41586-019-1451-5). Placental microbes are therefore not associated with common pregnancy complications such as low birth weight and a dangerous condition called pre-eclampsia, the study’s authors conclude.
Traditionally, most scientists have believed that the placenta is sterile. But in the past decade or so, researchers have identified a small community of nonpathogenic bacteria present there. In 2014, a landmark study of placentas from 320 women found a small population of several types of nonpathogenic bacteria in the placenta and reported that these low-abundance communities differed between women who had full-term and pre-term deliveries (Sci. Transl. Med. 2014, DOI: 10.1126/scitranslmed.3008599). Some labs, however, have failed to find placental microbes in follow-up studies.
In the current work, researchers sequenced the DNA from more than 500 placenta samples to determine whether they could detect a placental microbiome, and if so, link the microbes to pregnancy outcomes. They also wanted to see whether genetic material, either from the molecular biology kits used for sequencing or from the vagina during delivery, could be contaminating the samples.
In one series of experiments, they spiked their placental samples—20 from mothers with pre-eclampsia, 20 from mothers who had babies small for their gestational age, and 40 controls—with known quantities of Salmonella microbes that weren’t present in the samples and aren’t found in humans. They then sequenced each spiked sample using two different techniques. “We just sequenced [the samples] to death,” says Stephen Charnock-Jones, a reproductive biologist at the University of Cambridge, who is one of the study’s lead authors.
In the first technique, called metagenomic sequencing, researchers identify every base pair of DNA present, filter out the human DNA, and then determine the identity of microbial sequences. In the second method, called 16S rRNA amplicon sequencing, researchers sequence just one region of a gene that exists in all bacteria but varies widely between them. Because the researchers knew how much Salmonella they had added, they could use that quantity as a benchmark to estimate the quantities of other microbes they detected in the sample.
“If there’s a real signal with your sample, you should detect it with both methods,” Charnock-Jones says. But for the most part, they didn’t. Only one bacterium, Streptococcus agalactiae, came up with both techniques, in very small quantities and in just a few samples.
In the second series of experiments, which included 100 pre-eclampsia samples, 100 low-birth-weight samples, 100 preterm birth samples, and 198 controls, the researchers performed only 16S rRNA amplicon sequencing, but they did it twice—each time with a kit from a different manufacturer. They also paid scrupulous attention to the batches of reagents they used. They could not find bacterial species that were consistently present when sequenced with different kits and reagent batches, suggesting contamination was the source of the microbes.
“That indicates it’s not true signal,” Charnock-Jones says. “We would hope that our paper puts the question to bed.”
But not all researchers are convinced by the new data.
Most people assume a microbiome is a large thriving bacterial community, says Indira Mysorekar, a reproductive biologist at the Washington University School of Medicine in St Louis, but what her team and others have described in the placenta is instead a small collection of microbes, each at low abundance. Yet the main technique this study used, 16S sequencing, struggles to identify low-abundance microbes, she explains.
That’s true, when working with samples that have high bacterial loads overall, Charnock-Jones says. But in his team’s experiments, there weren’t many microbes in the entire sample, so 16S should have found any microbes that were there.
Kjersti Aagaard, a reproductive biologist at Baylor College of Medicine, who was the lead author of the 2014 placental microbiome study, still stands by her team’s conclusions. She argues that the fine details of how Charnock-Jones’s team analyzed their data led them to incorrectly discount signals of microbes as contamination. For example, their process classified certain microbes as contaminants from the vaginal microbiome, even in cases in which the babies were delivered by Caesarian section.
In fact, the types of bacteria the study identified were exactly those that she and her colleagues found in their study, Aagaard says, and “would indeed appear to be a low biomass, low abundant placental microbiome.”