The connection between ERVs and placentas is already well established.

Without a shadow of a doubt, the ability of mammals to generate placentas was, in part, accomplished by domesticating an endogenous retroviral env protein.  In humans, we call this protein syncytin, or ERVWE1.

This new paper in Nature Genetics takes the ERV-Placenta connection a step further:

Endogenous retroviruses function as species-specific enhancer elements in the placenta.

Even though all mammals have placentas, mammalian placentas are different, even between species we would think should be very similar, like the subjects of this paper, mice and rats.  Between mice and rats, the building blocks of a placenta, the genes, are related, and yet the mature placentas are distinct.


You might have run into this situation before-- You know how humans and chimpanzees and bonobos share most of the same genes... and yet humans and chimpanzees and bonobos are all distinct species?  The 'trick' is the regulation of the 'same' genes.  More of Gene A. Less of Gene B. Gene C doesnt turn on at all.  Gene D is expressed at the same rate, but earlier.  Gene E later.  Different species are different recipes using the same ingredients.

And it appears as if ERVs are helping write the recipes in mouse and rat placentas.

These researchers started by mapping the epigenetic landscape of the mouse and rat embryonic cells that start making mouse and rat placentas. And by 'mapped the epigenetic landscape' I dont mean they measured how happy the mouse stem cells were, or fed the rat cells all organic non-GMO media. I mean they mapped the location of epigenetic markers that indicate actively transcribed DNA, and silenced DNA.  Please note the lack of magic:

...promoters (trimethylation of histone H3 at lysine 4 (H3K4me3)), enhancers (H3K4me1 and H3K27ac) and repressed regions (trimethylation at histone H3 lysine 27 (H3K27me3) and trimethylation at histone H3 lysine 9 (H3K9me3))...

Even though the genes to build placentas in mice and rats are relatively similar, the epigenetic profile of the embryonic cells that turn into the placenta were totally different between mice and rats:

We predicted 9,460 mouse and 7,932 rat TSC promoters on the basis of H3K4me3 enrichment over gene transcriptional start sites (TSSs), which were associated with expressed genes (Fig. 1a,b). We predicted 52,476 mouse and 41,142 rat TSC enhancers on the basis of distal enrichment of H3K4me1 (>5 kb from a gene TSS) and 25,736 mouse and 4,471 rat regions of distal H3K27ac enrichment. These predicted enhancers were significantly enriched near genes with annotated placental function (P = 1 × 10−51, binomial test; Supplementary Fig. 1).


We found that, although the majority of promoter regions were conserved between mouse and rat, both enhancer and repressed regions were predominantly species specific.

Whats more, is that of the 80% of enhancers that were mouse-specific or rat-specific were next to transposable elements.  What kind of TE?  ERVs. What kind of ERVs? Species specific ERVs.

There are many, many ERVs in mouse/rat genomes.  The only ones they would be interested would be the ones near the 52,476 mouse and 41,142 rat TSC enhancers... but... um... that doesnt entirely help narrow things down.

So they focused on a mouse-specific ERV family they found hanging out with their epigenetic enhancer regions, RLTR13D5.  They poked around, trying to find potential enhancer sequences of DNA hidden in the carcass of dead ERVs.  And they found them. In the ERVs hanging out with transcription factors that have previously been associated with placental development-- Eomes, Cdx2 and Elf5.

But that doesnt mean that the ERV carcasses are actually used as enhancers. Just because those three proteins can bind to the ERV carcass, it doesnt mean they do.

Sometimes, they do:

Overall, we found that 35% (336) of all genomic regions triply bound by Eomes, Cdx2 and Elf5 were derived directly from the mouse-specific RLTR13 ERV superfamily, demonstrating their central role in substantially reshaping the TSC core regulatory network.

Okay, well, just because they do bind to ERV enhancers, that doesnt mean they actually do anything in vivo.

Sometimes, they do. They found some genes downstream of putative species specific ERV enhancer regions, and used a reporter system to determine if that really resulted in species specific expression of their reporter (could they make rat cells, which dont have RLTR13D5, use RLTR13D5 in an artificial expression system).  It worked.

Overall, these results show that RLTR13D5 is capable of driving gene expression in placental cells and provide strong evidence that RLTR13-derived enhancers have facilitated the evolution of mouse-specific gene expression patterns in TSCs.

As if this wasnt enough, these folks looked in non-placental tissue for similar patterns... and didnt really see it.  The putative ERV carcasses involved were old (read: not species specific).

We have ERVs to thank for our ability to make placentas. Go figure :-/

We suggest that species-specific ERVs contribute to the rapid divergence of the placental gene regulatory network.

More like this

Hi Abbie! I'm the first author on the paper. As a long time fan of the blog I'm flattered that you featured the work. I'd be happy to answer any questions, and I also would like to draw attention to another hot-off-the-press paper (published by the same group who found previous syncytins) that discovered yet another syncytin, in ruminants, that plays a slightly different role. Other syncytins appear to mediate cell fusion between fetal trophoblast, but this version seems to promote "heterologous" fusion between fetal and maternal cells. Together our papers show that one simply cannot think about placentas without also thinking about ERVs.


Obviously a ton of work in that paper, Edward. A ton of work and not many authors (HIV Nature papers frequently have upwards of hundreds of authors...)-- And very cool findings to boot-- Congratulations!

And I might write about that PNAS paper for tomorrows ervalanch post :-D

Hey Edward (if youre still around)--

Do you happen to know if immune responses to ERV proteins are selected against (central tolerance)? Or are they not expressed, ever, so you can still make antibodies to them? Females dont make antibodies to ERVWE1 when they are pregnant (right?), but what about other ERVs?

Good question, admittedly I don't know much about immunology. But, a recent paper in Nature ( showed that antibody-deficient mice exhibit greater ERV activity, so host antibody production does seem to be involved in ERV suppression in general.

The placenta may be a different story. Placental ERVs have been detected traveling through the umbilical cord, but it's unknown whether this triggers any immune response in the fetus. So, there is a possibility that the first time a female immune system is subject to these ERVs is when she is pregnant. But, whether pregnant females produce antibodies against those viral particles is still unknown.. though there is clearly a potential for conflict, and other viral proteins could conceivably serve to downregulate the local maternal immune response.

One, perhaps off-the-wall possibility, is that the placenta may act as a "testing ground" for ERV expression where consequences are minimal, and if mom produces antibodies, these antibodies may travel back to the fetus to help control potential ERV activity there. There isn't any evidence at all for this in mammals, but something similar basically happens in plants (with siRNAs) and their endosperm, which is like a plant placenta in many ways.


Great paper Edward. And thanks for talking about it here.
I have a few questions regarding the RLTR13D5 and I am hoping one of you could answer them.

Can any of you explain whether the 608 copies of RLTR13D5 are derivatives of the same ancestral copy or they may have integrated multiple times in the mouse genome? For instance, in humans, there are over a dozen human specific HERVs that are ~90% identical but are considered different members of the same HERV-K family.
Also, how it was determined that RLTR13D5 "no longer replicates"? I would be highly grateful for your comments.