Targeted mRNA therapy tackles deadly pregnancy condition in mice
28:32 Share

Hey guys!

Have you heard of Goldbelly? It's this amazing site where they ship the most iconic, famous foods from restaurants across the country, anywhere, nationwide. I've never found a more perfect gift than food. They ship Chicago deep dish pizza, New York bagels, Maine lobster rolls, and even Ina Garten's famous cakes. So if you're looking for a gift for the food lover in your life, head to goldbelly.com and get 20% off your first order with promo code GIFT.

This is an ad by BetterHelp. What's your perfect night? Is it curling up on the couch for a cozy, peaceful night in? Therapy can feel a bit like that. Your comfort place where you replenish your energy. With BetterHelp, get matched with a therapist based on your needs, entirely online. It's convenient and suited to your schedule. Find comfort this season with BetterHelp. Visit BetterHelp.com to learn more and save 10% off your first month. That's BetterHelp, H-E-L-P dot com.

Welcome back to The Nature Podcast. This time, a potential treatment for preeclampsia and engineering a custom cellular switch. I'm Emily Bates and I'm Nick Pettichow.

Researchers have developed a way to treat preeclampsia, a deadly complication of pregnancy, and they're doing it by delivering mRNA to the placenta. Preeclampsia accounts for about 15% of maternal deaths and 25% of fetal deaths worldwide.

High blood pressure or hypertension and reduced blood flow in the placenta are both telltale signs, but current interventions only really manage the symptoms, rather than treating the underlying causes. Now, the team behind this new work hopes to change that. I caught up with two of the authors, Kelsey Swingle and Mike Mitchell, to discuss how their mRNA delivery technique works.

Kelsey started by explaining the current options available to people when they have preeclampsia. So I think that's one of the biggest challenges right now with preeclampsia and a lot of other pregnancy disorders. There's not a lot of or really there's no therapeutics in the clinic that address the underlying problem which is in the placenta. When you're diagnosed with preeclampsia in the clinic you're

Really, clinicians are just trying to manage your symptoms. So they're managing your high blood pressure. They're trying to prevent seizures from developing. They're trying to promote gestation for as long as possible to ensure viability of your baby. So really, it's symptom management and not addressing the underlying disease pathology. And unfortunately, with preeclampsia, most of the time the mother has to induce early because there's no curative option.

And the challenge becomes if preeclampsia is very severe and you have to induce early on in the pregnancy,

then it might not even be an option, which becomes a very big challenge for treating preeclampsia. And so clearly preeclampsia needs quite urgent interventions. And in your case, you were looking at an mRNA molecule inside a lipid nanoparticle, LNP, which would end up in the cells and get them to produce certain proteins. Right.

And so in your case, what was the idea here? How might this mRNA technique work for preeclampsia? So how this project actually started in my lab was we were getting questions from pregnant mothers over email.

As the COVID-19 vaccines were coming out, they were asking us if the vaccines were safe to take during pregnancy. And of course, I'm not a clinician. I can't advise. But we did know that these vaccines were safe. But given that my lab's a drug delivery lab, we did ask the question, when these are injected, where do the lipid nanoparticles go and where does the mRNA go?

And Kelsey led these fundamental studies where she started injecting these lipid nanoparticles into the bloodstream of pregnant mice to see where they go. And what we found was none of them went to the fetus, but in the actual placenta tissue.

A lot of these particles collected and delivered mRNA to those cells in the placenta, the trophoblasts, the endothelial cells that make up the blood cells, and also some immune cells. And I'll let Kelsey take it from here, but that led us to ask the question, well, can we design a lipid nanoparticle that specifically goes to the placenta? And if it does, can we leverage it therapeutically?

So like Mike said, once we identified that particles are accumulating in the placenta, they're delivering the mRNA, we had to ask, okay, how can we tailor this technology that's been used in the COVID vaccines for preeclampsia? And that's kind of when picking the right mRNA or the right therapeutic cargo comes in. So in this case, we're delivering VEGF mRNA or vascular endothelial growth factor mRNA to try to induce

reduce vasodilation or promote a healthy blood flow environment in the placenta. So you knew that some of these lipid nanoparticles were going to the placenta, but most actually go to the liver. So how did you go about tweaking them to make sure that most of them are actually going to the placenta and delivering the mRNA there?

You're right. Absolutely. Most LMPs, when you administer them, IV tend to accumulate in the liver. And that's why a lot of applications of LMPs to date have been for, you know, liver based or liver centric disorders. Here, we actually used a high throughput screening approach to screen a large library of LMPs because we,

really didn't know what was going to work or what wasn't going to work. So we kind of just made a large library of about 100 LMPs and were able to test those in vivo and identify which ones enabled more delivery to the placenta and less delivery to the liver. And when you're saying in vivo, these were experiments in mice, right? Yes, absolutely. And so by looking through this high throughput screening, you found better lipid nanoparticles for the placenta.

how well were they able to get there? So we definitely found a particle that enabled really great mRNA delivery to the placenta. So we benchmark kind of our particle against, you know, some that are used in the field pretty typically to deliver mRNA. And our particle enabled more than an order of magnitude greater mRNA delivery to the placenta than some of these, you know, industry standard particles, which is really exciting.

And one thing I'll add to what Kelsey said is when we think about these lipid nanoparticles, they are not one size fit all, where you could take one type of LNP and use it for a range of applications. You really need to tailor it to the biological microenvironment that the nanoparticle sees.

And what we noticed was interesting about these particles is they have what is known as a protein corona that forms around them that is very unique compared to other types of lipid nanoparticles. And it's really the chemistry that dictates it. The chemistry of the particle will dictate what types of proteins ultimately interact with it. And Kelsey shows in her paper that this seems to play a key role in delivery to the placenta.

So are there particular proteins in the placenta that I guess are more attractive to the lipid nanoparticle? It's actually the proteins that are in the bloodstream, right? When we administer these lipid nanoparticles, pretty much instantaneously proteins from your blood kind of are attracted to the nanoparticle. And which exact protein is kind of attracted to your nanoparticle kind of dictates where the particles go. So here, the protein that we identified that

is really promoting this delivery to the placenta. It's like a glycoprotein, it's called beta-2-glycoprotein-1. So it enables some future applications potentially of this work to try to really rationally design further nanoparticle formulations that are even better at delivering mRNA to the placenta based on this idea of protein corona formation. So the lipid nanoparticles were getting to

the placenta. I guess the key question is, how much did this help with preeclampsia? So we did two different mouse models of preeclampsia. And in both of these, you're kind of inducing preeclampsia pretty early in mouse gestation at about day seven, because mice are pregnant for about 19, 20 days or so. So we're inducing preeclampsia pretty early in gestation. And then we administered our therapeutic at about mid gestation at about day 11. And

And with the placenta particle that delivers VEGF mRNA to the placenta, we saw that we were able to reduce hypertension pretty much instantaneously and blood pressure stayed reduced through the end of gestation. So it was a really exciting outcome. And I think it suggests that

perhaps we're remodeling the vasculature that's in the placenta to kind of see a really sustained therapeutic effect. And what I'll add is in the paper, you could really see the differences. If you look at the healthy mice,

The fetus from the mother, it's large, it's healthy. If you look at the fetuses from the mother with preeclampsia, those fetuses are much smaller in size. They almost look shriveled up because of poor nutrient and oxygen transport. But with just one dose of that mRNA lipid nanoparticle to basically normalize that vasculature,

the fetus from the mother looks almost exactly the same.

as the normal mouse without preeclampsia. So the result is quite striking. Now, as I understand it, the placentas of mice and humans are quite different. So do you anticipate that there'll be further tweaks to make if this were to become a human therapy? It's certainly true that mouse placentas and, you know, human placentas are quite different. The differences lie a lot of times in the exact cellular composition of the placenta across species. I think something that Mike and I have talked a lot about is

what animals have placentas closest to humans and guinea pigs actually have placentas that are quite close to humans, which is quite funny. So I think that that's something that we're interested in exploring in the future. And then other things that I think we have to think about when we start to think about translating this to humans is

You know, how many doses would we have to give a pregnant human patient, right? In mice, they're only pregnant for about 20 days. So our single injection of our VEGF LMP therapeutic worked really well. In humans, we don't know. So we'd only have to test larger animal models that have longer gestational periods to really answer that question. That was Kelsey Swingle and Mike Mitchell, both from the University of Pennsylvania in the U.S.,

For more on that story, check out the shoutouts for some links. Coming up, researchers have modified key molecular switches to drive custom cellular behaviours. Right now though, it's the Research Highlights with Dan Fox. Excavations in north-eastern Iraq have unveiled evidence of organised societies, stacks of mass-produced bowls. But their abandonment points to an eventual rejection of the institutions that made them.

Mesopotamia was home to the world's most ancient cities and states, such as the Copper Age Uruk civilization. Now a team of researchers excavating a site that shared close cultural ties with Uruk have found mass-produced bowls with beveled rims that indicate the existence of institutions that fed large numbers of people, possibly in exchange for labor.

The team found evidence of multiple periods of occupation but no sign of violent attack or natural disaster that might have led to the site's abandonment. Instead, it seems that the region's population deliberately dispersed and urbanism didn't appear in the region again for another 1500 years. The authors say the evidence suggests that the formation of state-level institutions is not an inevitable trend.

Read that research in full in Antiquity. The planet Venus has probably been dry inside and out for its entire history.

Venus has no water now, but it's at a distance from the Sun that could allow liquid water to exist on its surface. One theory was that Venus once had watery oceans, but that these desiccated early in the planet's history, leaving a dry, uninhabitable world. To test this scenario, a team of researchers looked to the composition of gases in Venus' atmosphere. From that, they could determine what the interior of the planet was like.

Their results suggest that Venus' interior contains relatively little hydrogen and therefore little water. The team also found that the molten rock that erupts from Venus is much drier than the lava from similar eruptions on Earth. They conclude that if early Venus once had water, it was probably in the form of steam floating above a fiery surface, not life-friendly oceans. Find out more in Nature Astronomy.

Next up, a team of scientists writing in Nature have found a new way to modify a large family of receptors, allowing them to drive a huge range of cell behaviours using a single platform. There are around 30 trillion cells in the human body. And for those cells to do their jobs properly, whether that's detecting light, fighting off pathogens or secreting hormones, they need to be able to sense what's going on around them so they can respond in the appropriate way.

Crucial to this are a vast group of proteins called cell surface receptors, which allow things in the environment outside the cell to drive specific processes inside the cell. Synthetic biologists have long been interested in producing their own cell surface receptors as a way to reprogram cell behaviour to their own ends, including new therapeutics. And now the team believes that they have done it for a huge family of these receptors –

G-protein coupled receptors, or GPCRs. Reporter Anand Jagatia spoke to co-first author Nick Kalagropoulos about the work.

and started by asking him about the kinds of things you can do with an engineered cell surface receptor. The major example would be synthetic receptors based off T cell receptors. So they're expressed in T cells, and T cells, they're a major immune cell in the body. And by creating synthetic receptors, we're able to reprogram those receptors to recognize tumor antigens expressed on tumor cells that they otherwise wouldn't have been able to recognize. And so there's been a lot of progress in the sort of the cancer space for therapeutics.

So why did you decide to engineer these receptors in particular, these GPCRs? Actually, the largest class are these G protein coupled receptors. Pretty much any aspect of biology and physiology likely has some G protein coupled signaling in it. They actually engage entirely different pathways inside the cell that no other synthetic receptor can access. And so by having a synthetic receptor that can tap into GPCRs,

these pathways, it allows us to really open up the types of outputs we can do and the different types of cell behaviors we can control. What are some of the challenges then of harnessing these GPCRs and how did you manage to overcome those challenges? Yeah, the major challenge is synthetic receptors in these other classes, they're actually a lot more structurally modular protein. So you can kind of cut off a piece and replace it and it still generally functions the same.

And that's not true with GPCRs. So structurally, they're a lot more complex. So the challenge is how do we really have this sort of programmability, this modular gating on them? And the way we overcame it, we didn't actually add modularity specifically to the receptor. We added an additional component that gave us this modularity. And so if we can think about receptors as sort of a lock and key mechanism where you need to sense...

different antigen or different signal out there as the key, and it fits into this lock, this pocket on the receptor. What we did really is we fused a component that blocks that pocket, so it can't be activated. And we can then control the activation by then removing that blockage, that inhibitor, when it senses something of interest. Lots of people will have heard this lock and key idea of how proteins and how cell receptors work. But rather than changing the lock,

you've basically like added a bouncer, like somebody who's going to be in front of the blocking the doorway and then allowing things in or out. Yeah, exactly. A hundred percent. People have taken other approaches to try to engineer the lock itself. And that's a major undertaking, many years and resources. And it's kind of a brute force approach to then just re-engineer it to recognize one new thing. But to get this really modular programmable aspect to it has really just not been possible before.

And so that really allowed us to have this sort of new mechanism of gating that hasn't been used previously. So that's how the platform works. How well did it work? I mean, what kinds of stuff were you able to demonstrate in the paper when you gave this platform a run?

It turned out to be quite robust, and we were pleasantly surprised with how well it worked. We were able to have different types of outputs, including transcriptional expression, so expression of a gene and a protein. We were able to drive, for example, the secretion of effector molecules like cytokines or therapeutic antibodies. We were able to drive, when we put it in T-cells, drive T-cell migration towards an antigen-retraced

When we put it in neurons, we're able to inhibit neuronal activity. So again, because GPCRs are expressed in so many different cell types and do so many different types of things, we really have this wide array of different things in cell behaviors that we were able to control, which I think is really exciting. Aside from all of these new pathways now that you can access in the cell, are there any other pathways that you can access in the cell

Are there any other advantages of this platform? Yeah, so the major limitation of a lot of the other synthetic receptors is that what's required is actually a surface antigen. So they can respond to things that are expressed on other cells. But as we know,

a lot of biology, a lot of this intercellular communication is driven by soluble factors. And those pretty much are inaccessible to sense and respond to from most of the other synthetic receptor platforms, but not in this case. So we're now able to access those, which again, opens up, I think, a lot more biological context in which we can apply this platform. Okay, so it's kind of like the sky's the limit in terms of how people want to apply it to drive cell behavior. Yes, exactly. And I'm really excited for people to take it and apply it because

we don't have the expertise and can't put it everywhere and do everything. So people who study these different areas of biology know more of what they want to sense and what they want to drive, can really take it and put it to use. Where are you hoping to sort of take this in the future? And what things are you hoping to improve? This is really a proof of concept that this works and we can do it. And so there is definitely room to improve, kind of reducing the complexity, making it even easier for people to use

and then really kind of broadening the scope of using more GPCRs to be able to apply to more places and drive more behaviours. That was Anand Jagatia talking to Nick Kalagopoulos from Stanford University in the US. For more on that story, check out the show notes for some links. Finally on the show, there's been a new advance in the world of quantum computing.

Researchers at Google have reached a key threshold for error correction. So joining me to discuss it is Davide Castelvecchi, physics reporter here at Nature, who's been reporting on this story. Davide, how's it going? Excellent. Thanks for having me, Nick. Well, it's always nice to have you on the show, Davide. And now we're talking about quantum computers. Maybe it's worth doing just a little bit of a refresher. How exactly is it that these quantum computers work? How does it differ from more classical computers people might be familiar with?

A quantum computer encodes information in quantum bits.

which are different from the traditional bits of classical computers, which can be a zero or a one, because they can also be a quantum superposition of a zero and a one at the same time. And they can also be quantum superpositions of multiple zeros and one in a collective state that physicists call entangled. Now, these machines...

Because they have so many possible entangled states, a certain number of qubits can encode a lot more information than the same number of ordinary bits. And not only that, it's often described as

the superposition of multiple states means that you can run calculations in parallel. Now, the purists say it's not quite that way, but there's something to it, which means that certain computations end up being exponentially faster

in principle, on a quantum computer than on a classical computer. So they could potentially open up a lot of applications. But one thing that's been a real difficulty for quantum computers is errors. And so there's been work to try and do error correction. How, in principle, would this work, this error correction? So the idea is in a classical computer,

you can just copy information or you can do calculations based on the bits you have in the memory to figure out if an error has developed or appeared. In a quantum computer, it's much more difficult because you can't simply read out the qubits or you will spoil the quantum state.

And you also cannot copy the qubit or the quantum state without spoiling it. In quantum physics, there's no measurement you can do without actually changing the state of the thing that you're measuring. And so people have had to come up with much more ingenious ways to do error correction, which amount to basically spreading the quantum information of one bit across multiple qubits in a redundant way, but also...

a way that is not simply copying and the way also that allows you to detect if an error has occurred and correct it. And so what was the approach specifically then of the team at Google? Well, a lot of it relied on just improving the quality of their qubits. And by doing so, you get to a point where the error correction kicks in and actually starts to make a difference.

And it's not just one thing. So this is something that the Google researchers emphasized in interviews. You have to make everything better. Basically, every single step, every single component has to improve in quality. And if just one of the things, one of the many little components doesn't improve in quality, then the whole thing

will not work. And so this has been a painstaking process of daily improvements. And the other thing they've emphasized is that their quantum computing lab now has its own chip fabrication facility where they are able to, you know, go back and forth between manufacturing a new prototype and testing it and then going back and manufacturing a new one.

And that has really made a difference. This has been just in the last couple of years. So it's like a set of quality improvements that's just allowed this idea of using extra qubits to do the error correction to work. Correct. And in your story, the headline is that this is a milestone for quantum computing. Why?

Why is that and what does it mean to the community? On paper, when you encode a qubit into, say, a network of little superconducting circuits, as in this case, so a network of physical qubits represents one piece of quantum information, the error rate goes down, but then it doesn't go down by itself enough.

So the idea is you want to make it even bigger and spread it over a larger number of qubits. And on paper, at each step that you increase the size of this kind of network of qubits, the error rate should go down. But that's on paper.

That had never been shown before. So Google and other labs had been able to show before is this first step of showing one little improvement in one step. But they hadn't shown that at each successive increase in the size of the qubit, the error rate goes down. And in this case, they've shown that in two successive steps, they cut the error rate by half or better than half twice. So it results in an exponential improvement

Because in principle, by extrapolating, you could just keep increasing it. And once you get to a network of around 1,000 qubits, then you have an extremely error-resilient system that has errors only like once per 10 million qubits.

calculation steps. Sounds like very few, but as you say there, it's a network of a thousand qubits to essentially represent a single qubit's worth of information. So does this mean that these kind of quantum computers will have to be relatively large? Yes. So the current kind of conservative estimate is that you'll need 1000, although with the newer

error correction schemes, the overhead might go down to perhaps a few hundred. But yes, to have a useful computer that does the wonders that people have been promising, you will probably need a thousand qubits per unit of information, and then you'll need to have a thousand units of information together. So overall, you need a million.

And so that means you also need to find a way to connect them, to make them exchange information or to make them share quantum states.

And that's going to be, that's all in the future for now. As you said, this is something that people have been trying to do for a long time. What's been the reaction of people in the field? A lot of people were very excited because the preprint of this Nature paper was already online for a few months and physicists in the field had familiarized themselves with it. And this certainly contributed to a feeling that quantum computers are

about to get real. And yeah, in that case, what's next then for quantum computing? Where do they get real? Well, according to Google, they'll be able to reach the size...

where they can beat classical computers at useful tasks, such as scientific experiments, by the end of the decade. But there's also really interesting applications that are starting to appear now, where, like I said, it's not something that you couldn't do on a supercomputer, but maybe it's cheaper or easier to do it on a quantum computer already. In the short term, I think we'll see more and more sophisticated kind of proofs of principle of calculations

done with multiple qubits with very low error rate. What Google has done now, this demonstration was lowering their error rate just for a single unit of quantum information basically stored in there as a memory. So what they showed was that the memory was resilient, could hold the information for a longer time. The next step will be to have this quantum error correction

do wonders for actual quantum calculations where you have multiple qubits interacting and there will be

Probably soon. And I'm sure you'll be back again to talk about it when it happens. But I think that's all we've got time for this time. Thank you so much for joining me, Davide. Thank you very much for having me. And listeners, for more on that story, check out the show notes for a link to Davide's news article. And that's all we've got time for this week. As always, you can keep in touch with us on X, we're at Nature Podcast, or you can send an email to podcast at nature.com. I'm Emily Bates. And I'm Nick Petrichow. Thanks for listening. Hey, guys.

Have you heard of Goldbelly? It's this amazing site where they ship the most iconic, famous foods from restaurants across the country, anywhere, nationwide. I've never found a more perfect gift than food. They ship Chicago deep dish pizza, New York bagels, Maine lobster rolls, and even Ina Garten's famous cakes. So if you're looking for a gift for the food lover in your life, head to goldbelly.com and get 20% off your first order with promo code GIFT.

Enjoying your podcast? We'll be brief. If you're looking for the perfect holiday gift, gift scratchers from the California Lottery. With so many to choose from, you're sure to find the right gift for anyone on your list. Now that your holiday shopping list is figured out, enjoy this bird singing jingle bells.

Give the gift of Scratchers from the California Lottery. A little play can make your day. Please play responsibly. Must be 18 years or older to purchase, play, or claim.