Lab Spotlight: Theunissen

Stem cell models are a powerful window on the embryo, and the cells which support it.

We live at a pivotal time. In the span of a few decades, biologists have gained a deeper understanding of life’s earliest and most crucial moments. Now, we at e184 stand poised to take the next step, and address medical conditions that have chased humanity since we first evolved.

In this piece, we interviewed Thorold Theunissen, an Associate Professor of Developmental Biology at Washington University in St. Louis, Missouri. Theunissen’s career spans these developments, from biologists’ first appreciation of the versatile power of stem cells to the development of stem cell-based embryo models that can shed light both on the hidden first weeks of embryonic development, and the crucial role played by extraembryonic cells. We asked him about the history of key advances in the field, his own lab’s work modeling the implantation of embryos in the uterus, and what he expects in future.

What follows is an edited and curated version of our discussion.

To start, why don’t we go back to the beginning? What first motivated you to study stem cell biology?

This goes back to my freshman days at Harvard. Doug Melton was there. He was one of the early pioneers in stem cell biology, and he had just launched the Harvard Stem Cell Institute. This was the first organization of its kind, really, a dedicated institute at a major research university focused on stem cells. The Harvard Stem Cell Institute had the resources to derive embryonic stem cell lines in the Boston area from IVF clinics, that would then be distributed to researchers around the world. So they were leading this early stem cell wave, and they just had phenomenal scientists associated with that institute. I was successful as a young undergrad to get Doug Melton to be my advisor, and I asked him, “Can you recommend any good lab opportunities?”

He very kindly wrote to a scientist in the Netherlands (which is my home country), Christine Mummery. She’s also a pioneer in this area. She works on cardiomyocyte differentiation from embryonic stem cells. And I was very fortunate to spend a summer in her lab, that was such a great experience. There was just so much buzz around stem cells, being basically a method to generate any cell type of interest in the body. And the cool thing with the cardiomyocyte is you get these beating areas. So it’s something you can actually tangibly see under the microscope, there’s contractions, there’s a beating heart almost.

Then I returned to Boston for, I think it was maybe my junior year, and Doug kindly arranged for me to talk to Stuart Orkin, who is a famous blood stem cell biologist. They had this ongoing project mapping the protein interactions in embryonic stem cells, and I was actually able to contribute to that. It was a Nature paper, just really exciting for a young scientist to get an opportunity to contribute to that level of science, and they got me hooked.

Then, as a graduate student in Cambridge, I was fortunate to join the lab of Jose Silva, a pioneer in induced pluripotency. That experience solidified my interest in stem cell biology and motivated me to pursue postdoctoral training in this field.

What was the path that led to the first stem cell embryo models?

The big question at the time was, “Can we create human stem cells that are more similar to mouse ES cells [embryonic stem cells]?” This was one of the fundamental problems in the field: mouse ES cells are really good at controllably differentiating them into different fates. If you put them back into a mouse embryo they will give rise to many different tissues. Austin Smith published this key review in 2009 with Jenny Nichols, where they argued you have these different stem cell states, naive and primed, and that was a nice conceptual framework to think about these different types of stem cells that we see in mice and humans. And the obvious missing piece in all of this was the human naive state.

What ended up happening is in 2014 I was able to publish in Rudolf Jaenisch’s lab these defined conditions to culture naive cells, and Austin’s lab published right around the same time. If you look at epigenetics, a cool sign of the naive state is they undergo X chromosome reactivation. Males have a single X, females have two, so you have this process of X chromosome dosage compensation where one of the X’s is inactivated in females. And this normally happens in a random manner. That’s how you have the spotted coat color of the calico cat. But what happens in very early development in the early embryo, the pre-implantation naive epiblast, is that in females both of those X’s are reactivated. And what’s really cool is that the typical primed human embryonic stem cell lines, they tend to show one inactive X. But when we apply these naive media, they undergo X reactivation. So they end up with two active X chromosomes.

But there was still kind of this nagging doubt in the field if these are truly naive stem cells. And that brings me right back to what e184 is interested in. In the past five to ten years, a number of groups have shown that you can really use these naive stem cells to not only make embryonic cells but also extraembryonic cells, and we were one of the first to do that.

After I started my lab at WashU in 2018, my first graduate student found that when we apply media for placental cells, remarkably these naive human stem cells can actually make placental cells. This was very surprising because in the mouse system, that’s not possible.

Mind you, there are different kinds of extraembryonic cells that support the fetus. You have the placenta, you also have the yolk sac that plays a key role in blood supply, and there’s another thing called the amnion, which forms an amniotic sac around the embryo. And subsequent work, not from us but by other groups, also showed that these naive stem cells can make yolk sac precursors. They can make primitive endoderm, they can make extraembryonic mesoderm, so it was really another exciting wave of papers showing that these naive cells are a starting point for extraembryonic differentiation.

The ultimate test of these cells came with the embryo models. Full credit to the people that led the wave here. First, Nicolas Rivron showed that you can do this in the mouse system. You can combine embryonic stem cells with trophoblast stem cells and make a blastocyst-like structure, a blastoid. Magda [Magdalena Zernicka-Goetz] did really cool work, in again the mouse system, combining embryonic stem cells, trophoblast stem cells, and primitive endoderm cells to make these first synthetic embryos, what she called ETX embryos. They’re post-implantation-like. So we realized that if the human naive cells are able to make placental cells in vitro and they are themselves like the inner cells of the embryo, the ICM, then theoretically you should be able to make a whole embryo-like structure just from the naive cells. And we were playing around with this. I had quite a few people working on this in the lab. But as we were working on that, there were actually a couple of papers from other groups showing that you can make what are called human blastoids starting from naive cells. One of those first papers came from Jun Wu, Austin’s former postdoc Ge Guo had a similar paper, and then Nicolas Rivron as well later that year. So those were really the first papers showing that yes, you can take naive stem cells and make a blastoid.

One impressive recent result from your lab is the development of a stem cell embryo model into the post-implantation stage. What were some of the key insights that made that possible? What were the biggest challenges?

With the blastoids, I obtained permission at WashU to take them up to 21 days. I was a little concerned whether we would get permission, but I explained we’re not going beyond gastrulation. We’re going right up to the stage where these specialized germ layers begin to form. There’s no heartbeat yet. There’s no neural tube. And we have this committee, the ESCRO committee, that oversees these types of experiments. They have all sorts of stakeholders. They have clergy, I think there’s a rabbi and a priest, and all sorts of administrators and scientists that actually look at our proposal. And they decided 3 weeks is a fair limit.

Here I need to give credit to my former postdoc, who’s now an independent PI at the University of Colorado, Rowan Karvas. We tried a 2D protocol but it didn’t work very well at all, the blastoids were disorganized. So at that point Rowan tried these 3D thicker matrices, essentially a thicker ECM gel. The idea is, the three-dimensional surface is more mimetic of what actually happens when an embryo implants into the uterus. We tried a bunch of different substrates, and we ended up settling on a combination of Matrigel and Geltrex.

Using that matrix we were able to culture these blastoids up to gastrulation stages, and we could see very good development of the placental structure. The outside of the blastoid on that gel forms these beautiful invasive projections, and it really resembles early placental development when you get these villi that sprout out from the embryo and go into the uterus, and we see that very nicely. On the inside of the blastoid we start to see the embryonic compartments organizing, but they’re still pretty disorganized.

What is clearly still a challenge for us and others is to allow that blastoid to form structures with the right spatial organization. You’ve got the cell types, but you don’t have the structure. And we’ve been honest about that. I think that’s an issue for the field. We do see this early sign of a primitive streak forming, which is where you start to get this asymmetry within the embryo where you’ve got the specialized cells forming on one end of the epiblast, it’s where the early mesoderm, the muscle cells, arise. So, we do see that happening, but it’s clear that the blastoids don’t have the right organization yet.

Exciting. While there are still challenges ahead, it sounds like the field is making great progress. Can you tell us a bit about how these models can lead to improvements in reproductive health?

Yeah. I think it’s unlimited in terms of the potential ways in which this can develop. To give a specific example, just recently in the past few weeks there were three studies that came out using blastoids to model implantation. And I’ll give these guys full credit, they sort of scooped us on it. It’s really exciting work from Peter Rugg-Gunn’s lab in Cambridge, his former postdoc Matteo Molè’s lab at Stanford, and a group in China led by Leqian Yu. They showed that you can take blastoids, combine them with endometrial organoids and endometrial stromal cells, and essentially recapitulate early maternal-fetal interactions.

And what’s really nice, the group from Beijing showed that you can not only model interactions between trophoblast cells, blastoids and endometrial cells, but you can even use patient-derived endometrial cells to recreate the conditions that lead to implantation failure. They took endometrial cells from women who experience recurrent implantation failure and they found that the blastoids actually attach to those much less frequently. And then they went even further. It was a Cell paper, to me it was like three papers in one. They even did a whole chemical screen to identify FDA-approved compounds that can overcome this implantation barrier. A lot of couples go through IVF, and even if the embryos are genetically normal, they still don’t implant. This is really the first study to suggest that you may be able to fix that with the right drugs.

The other key point, though, is they saw a very high degree of patient variability. Some of these patient cells required certain compounds to overcome the implantation failure. Others require different compounds. It’s really a need for patient-specific tailored precision medicine.

So I think the sky is the limit.