New worries about Earth’s asteroid risk, and harnessing plants’ chemical factories

Itturalde crater in Boliva with podcast overlay


First up this week, have we been measuring asteroid impact craters wrong? Staff Writer Paul Voosen talks with host Sarah Crespi about new approaches to measuring the diameter of impact craters. They discuss the new measurements which, if confirmed, might require us to rethink just how often Earth gets hit with large asteroids. Paul also shares more news from the recent Lunar and Planetary Science Conference in Texas.

Next up, pulling together all the enzymes used by a plant to make a vaccine adjuvant—a compound used to boost the efficacy of vaccines—in the lab. Anne Osbourn, a group leader and professor of biology at the John Innes Centre in Norwich, England, talks about why plants are so much better at making complex molecules, and an approach that allows scientists to copy their methods.

This week’s episode was produced with help from Podigy.

About the Science Podcast





0:00:05.9 Sarah Crespi: This is the Science Podcast for March 24th, 2023. I'm Sarah Crespi. First up this week, a look at large recent impact craters. Staff writer Paul Voosen joins us to talk about a preliminary finding that suggests we've been underestimating the size and frequency of asteroids hitting Earth. We discussed the evidence needed to shore up this claim. Next up, pulling together all the enzymes used by a plant to make a vaccine adjuvant. This is a compound used to boost the efficacy of vaccines. I talked with researcher Anne Osbourn about why plants are just so much better at making complex molecules than chemists and an approach that does allow scientists to copy these plant methods.

0:00:52.9 SC: Last week, staff news writer Paul Voosen spent some time in Texas at the Lunar and Planetary Science Conference, LPSC. And one of the sessions he attended on the size of relatively recent asteroid impacts on Earth, some researchers are basically saying the big ones are more frequent than previously thought. This is great news. Thank you for bringing this, Paul. Let's talk about it.


0:01:16.3 Paul Voosen: This is a hypothesis. That's the first thing to lead with. This is something that is not proven. If it was proven, it would be something to scream from the rooftops. It would be a big, big news, but right now it's an initial reconsidering of high-resolution satellite imagery is making people think, oh, we need to go out and figure this out.

0:01:40.8 SC: Let's start from how recent are these impacts and then we'll talk about how we measured them in the past and how they're being measured with this new approach.

0:01:50.1 PV: These researchers are looking at large impacts from the last million years. Earth really likes destroying impact craters. We have erosion, plate tectonics, everything, just landforms, just go away.

0:02:03.5 SC: Rain.

0:02:04.0 PV: Rain.

0:02:06.4 SC: Yeah.

0:02:07.0 PV: Or they're hidden under jungle canopies. It's just not easy to find impact craters. So they're looking at three known impact craters and one that is highly likely one, but not technically proven yet.

0:02:19.1 SC: So is it young ones because we care about what happened more in the last million years or is it because they're the only ones we're gonna be able to really see?

0:02:29.4 PV: They're the best preserved. So they're testing out this new kind of data set that they have access to. So they figured, let's look first at these most recent ones that we can best analyze and they're really at first just testing software that had previously been used on Mars because Mars does preserve its impact craters and it's just easier to study for some topics like that.

0:02:48.9 SC: We tried it out on Mars. Now we're pointing it down at Earth. What's different about this technology than what's been used before to get a size on these impact craters?

0:02:56.7 PV: Before, it was really kind of just what looks like a crater rim, that's the impact crater.

0:03:02.3 SC: The circle.

0:03:02.3 PV: Yeah, the circle. All these have a very obvious kind of circle that has been the traditional measurement of this. These researchers from NASA are using this new data set from Planet that is super high resolution imagery and putting all these images together, like 7000 for a crater to create 3D elevation models at super high resolution and then you can further constrain that with space lasers that measure height, and that gives you this very detailed map of the landscape. And then with this new detailed map, you can run this software that had previously been developed for Mars to detect what is the most circular looking structure starting from a center point. For these craters doing this, it kept coming up with these larger circles that are these subtle features that maybe are at diameters two or three times larger.

0:03:56.6 SC: So how big are the starting crater values and what are these new estimate values?

0:04:02.7 PV: The accepted diameters now are around 10 kilometers. Each is a little different and this increased them by two to three, so up to 30 or 35 kilometers wide. This, again, is not proven, but a lot of them seem to increase in size by two to three times diameter which is the power of the impact, that would be a 10 times larger impact.

0:04:26.9 SC: And how big were the big, big ones like the dinosaur killers?

0:04:29.3 PV: Dinosaur killers that's up to 200 kilometers wide.

0:04:33.4 SC: Obviously, this is hypothetical. This is only four craters that are being remeasured with this technique and not everybody's on board. But before we get to the objections, what does it say about the way we thought about this in the past? How does it update how we should think about asteroid impacts, their likelihood, and their history?

0:04:50.9 PV: We shouldn't change our thinking at all yet. But what we should do is be considerate. These kind of impacts that, let's say like a kilometer-wide asteroid or comet hits the Earth maybe every 600,000 to 700,000 years, which is a long time. These are not dinosaur killing type impacts, but they are like and seriously mess up kind of the area around where they hit could cause local extinctions for animals that are continent restricted. They could, with the big, these, talking like the 30 kilometer kind of crater types, not these smaller ones that they've been accepted to be. You don't want these hitting more frequently. And seeing that these four could be larger and they're now applying this to other craters as well, it would imply that perhaps this impact rate is higher than we would have thought. But there are other ways that scientists already think, we know this rate that conflict with this, so it would be a big rethink. It's a big kind of claim. So there's a lot to kind of go into that.

0:05:56.4 SC: Right. So this was presented at the meeting that you're attending and not everybody who was in the session or who heard about this research was on board. What were some of the objections or big questions that came out in talking to people about this research?

0:06:09.4 PV: Everyone's skeptical in a kind of healthy skepticism. The researchers themselves say we have to go into the field and show that this is true. This is not something you just make drawing a circle on a map. We know about the cratering rate from lunar impact craters.

0:06:26.9 SC: Right. When I hear about craters, I'm always thinking... I immediately picture what the surface of the moon looks like and that's pretty well-preserved, right?

0:06:33.9 PV: Yeah. And fairly well-dated, but we assume that they have the same gravity well, so the impact should be the same on both. But no, this is like perfectly understood.

0:06:45.4 SC: Right.

0:06:46.2 PV: And also you have... We study asteroids around earth and just knowing the size distribution of those is another important data point.

0:06:53.2 SC: Right. Do we have a population in this time span of big impactors that would be hitting us?

0:06:58.7 PV: Yeah. That's how you get kind of that 600,000, 700,000. Right. So to have this many, these are on land, right? So the oceans are two-thirds of the planet. It would imply like a lot hitting and maybe a couple of them are, maybe all of them are. The other thing is these are so big, they should be seen in climate records, which the researchers haven't looked for that yet. No one's saying it's not there, but that's something to look for.

0:07:23.7 SC: What can they do on the ground to corroborate this? If they physically go to these sites, what will researchers be looking for?

0:07:30.4 PV: This is hardcore field geology. So, you're looking for distinctive... I'm not going to perfectly answer this, to be honest, but the kind of bends in the landscape, the deformation that you would expect at the rim of a crater. There are other things. Most of these are known asteroid sites. You know that there's already like shock quarts or whatever else, but there are ways you can define what's a rim. There are also geophysical measures of gravity and all this stuff. These are not in the easiest places to work. They're in Nicaragua, in Kazakhstan, Bolivia. So this would be a big project and quite a lot of time to figure out.

0:08:10.3 SC: Yeah. I mean, I guess my big question is how do we disprove these monsters, Paul? No, I'm just kidding.


0:08:17.3 PV: It's a testable thing. They've made this prediction. So they're going to put a paper out about this later this year and it'll get reaction. And hopefully we are being the moderate early voice in this, but this will be something that should be checked out, but should be also, would be highly, highly worrying if it was at all true and it's not known to be true yet, but could be. Science in a nutshell.

0:08:43.1 SC: Exactly. Exactly. All right. So before I let you go, you've been at this meeting, you brought a story back for us. Is there anything else you want to mention that caught your attention while you're there?

0:08:52.6 PV: Yeah. So this is always a nice meeting to check in on with various NASA missions that has its roots out of Apollo. There are some nice updates about the Perseverance Rover, which is now stowed its first cache of samples and is moved up on top of the fossilized delta that it's been studying.

0:09:11.4 SC: This is Mars, by the way.

0:09:13.6 PV: This is Mars. Yes, Mars. Sorry.

0:09:16.5 SC: It's okay.

0:09:17.1 PV: They're kind of looking at some of the samples they collected because these are gonna be picked up and returned to Earth and trying to figure out what do we have here. You can see kind of fine mudstones, which they were searched for and they targeted that seem to have like little sulfate crystals that we know on Earth can preserve organics, like signs of life on Earth. They're getting what they're hoping to get is kind of the general vibe. Speaking of Mars, there was also, earlier this year, we covered some work suggesting that Earth got most of its water and stuff from where it formed and not far out in the solar system. People had thought that, oh, it must have come out from the outer solar system because water has zero time existing in the early solar system. This kind of defied that. This new work on Mars using Mars meteorites suggests Mars got everything from around it.

0:10:07.1 SC: Oh, interesting.

0:10:07.2 PV: Not even the small amount that Earth got and that is weird because you expect them to be similar and no one has a good explanation right now for why that could be, so that's a new mystery.

0:10:19.5 SC: That's very cool.

0:10:20.8 PV: Yeah. And there's the water theme a little bit. There's also a lot of excitement about the new Artemis missions from NASA, especially the first human return to the moon mission, which is Artemis 3 on the South Pole of the moon, which NASA still has scheduled for 2025. And scientists are now getting excited about what astronauts will be able to do, what rocks they can bring back, where they're going to go in the South Pole. So there are a lot of sessions thinking about that, looking at the candidate regions.

0:10:49.1 SC: Cool. I'm sure we're going to end up talking about that again in the next year or so.

0:10:51.9 PV: Yes.

0:10:53.5 SC: All right. Well, thank you so much, Paul. It sounds like a great meeting.

0:10:55.0 PV: Right. Thank you.

0:10:56.8 SC: Paul Voosen is a staff news writer for Science. You can find a link to the story we discussed at

0:11:02.4 SC: Up next, Anne Osbourn talks about harnessing the chemical making ability of plants for exploring new useful compounds and for pharmaceutical production.


0:11:21.4 SC: For a vaccine, sometimes it takes more than just giving our bodies dead bacterial cells, inactive virus, or mRNA to get our immune systems to respond, to get that maximal stimulation, and to make sure the immune system pays attention and stays impressed. In order to do that, drug makers add adjuvants, these are compounds that basically shout, "Look what we have here. It's an invader. Definitely pay attention to this." And while many vaccines have adjuvants in them, there's still a lot to learn about how they work, why they work, and how to make them better. This week in Science, Anne Osbourn and colleagues write about taking the very complex chemical synthesis of an important adjuvant compound out of a rare plant and making it happen in the lab without the plant. This allows them to create more than you would get from the plant and skip extracting it from tree bark. Hi, Anne. Welcome to the Science Podcast.

0:12:17.6 Anne Osbourn: Hi, Sarah.

0:12:18.6 SC: Tell me more about this adjuvant. It comes from a rare Chilean soap bark tree. How did this end up in so many vaccines?

0:12:28.1 AO: Yes, so that's an interesting story. I mean, historically, the adjuvants have been things like aluminium salts or emulsions containing squalene, which have relatively low potency and substantial side effects, respectively. It turns out that extracts from the bark of this tree, the soap bark tree, which grows in Chile, happen to be particularly good at promoting, stimulating the immune response and this was first discovered in animals. And then in 2017, a fraction from this bark extract called QS-21 was approved for use in human vaccines and it's now in vaccines for shingles, for malaria, for COVID, and is in a series of vaccines currently under development for other things.

0:13:16.6 SC: Taking it out of the tree, it's a rare tree. And also, you're not getting a high yield. And it's become incredibly expensive to actually get this compound out, right?

0:13:27.4 AO: So the tree grows in Chile, and it's quite common. But that's the only place in the world where it grows on a scale where it's used for extraction from the bark. And the tree actually makes over 100 different structurally related molecules, some of those are abundant in bark extract and some of them are not. The one that's mainly used is this fraction, QS-21, which has good adjuvant activity and low toxicity and is quite abundant in the bark extract. But there are other molecules, such as QS-7, which features in our paper, which are also very good immunostimulants and have low toxicity, but are present only in trace amounts.

0:14:14.8 SC: Why is it so hard to just make this in the lab without the plant being involved? Why are plants so much better at making compounds like this than chemists?

0:14:23.3 AO: Yeah. So plants are outstanding chemical engineers, and they make all kinds of exotic molecules including molecules such as morphine. Morphine is still too structurally complex to be made completely using synthetic chemistry. One of the most complicated adjuvant molecules produced by the soap bark tree, this QS-21 molecule, is very, very structurally complex. Although chemists have made it in the lab, this has involved multiple steps with low yields, and chemistry is not believed to be a commercially viable route for large-scale production of molecule. So by harnessing the instruction manual from the tree, we can learn how the tree makes not only this molecule, but also the hundred or so other structurally related molecules that it makes, and then we can put those into a system where we can actually plug and play and access the whole repertoire from the tree, not just the most abundant ones that are present in the bark extract.

0:15:23.2 SC: So what does it mean to figure out how the tree does this? You looked at the genome and what enzymes were important?

0:15:31.2 AO: This was a big job, and at the time we started this, I think most people thought it was science fiction because these molecules are so structurally complicated, and that means that their biosynthetic pathways are multiple steps, 10, 20 steps. And so trying to find the genes that encode the enzymes that make these molecules isn't trivial. And so we sequenced the genome of the tree, and in parallel we looked at the genes that are expressed in different tissues of the tree, and when we did that we learned that there are quite high levels of QS-7 and QS-21 in young tissue and primordia, for example, and lower levels in the old leaves, and that started to give us some clues that we were looking for genes that are expressed at higher levels in the young tissue, and so we used a combination of looking at transcript levels.

0:16:24.1 AO: We also mined the genome for all family members of large gene families that might possibly encode the types of enzymes that we were looking for. And in addition we have found that in some cases the genes for plant natural product pathways are actually organized in genomes like beads on a string right next to each other. And so we did some mining to look for those sorts of beads on a string things that would be guilty by definition. And so using a combination of those approaches we were able to go from the total number of genes in the tree, which is over 30,000, down to the 20 or so that make the compounds that we're interested in.

0:17:07.0 SC: And so then you had to figure out what the steps were, how the different enzymes were involved?

0:17:11.4 AO: Yes, so we had to start building up the pathway from scratch. To do this we used a really neat and quick and powerful transient expression system that we've developed here, and what this involves is you have your candidate genes of interest and introduce them into Agrobacterium, which is the bug that is used to transform plants. In this case we're not making stable transformed lines. What we're doing is taking these Agrobacterium strains and putting them, infiltrating them into the leaves of a surrogate host, which happens to be a wild relative of tobacco called Nicotiana benthamiana. And for various reasons this wild relative of tobacco is really good at just taking the sequences that have been introduced using Agrobacterium and expressing them. And this is done in a transient way so that it's five days from infiltrating Agrobacterium into the leaves to extracting the leaves five days later. And during that time the genes have been put to work and they're making enzymes and they're hopefully making the chemistry that we want to find.

0:18:21.3 SC: So you put like a subset of the enzymes into these leaves to see if they could perform the steps that you were looking for?

0:18:27.0 AO: We started by looking for the gene encoding the enzyme that makes the scaffold, the first building block. And so we knew that the starting scaffold, which is a complex chicken wire-like structure, was beta-amyrin. We found a number of candidates for the gene that might encode the scaffold generating enzyme and we looked at expression patterns and we cloned those genes and then we used our transient tobacco expression system to see which one made beta-amyrin and we found that one. So we then used that as our bait. That was our starting point. So we could then co-express additional pathway genes or candidate genes with that and look for enzymes that successively modify that scaffold and make it more and more complicated.

0:19:18.1 SC: You were able to kind of put together the sequence and figure out which enzymes were involved and create QS-7 from scratch, if you will, in your setup?

0:19:30.1 AO: Using a total of 16 enzymes, we built an advanced bridgehead, a sort of a core that is common to QS-21 but also simpler molecules such as QS-7, which we elucidate the pathway to in this particular paper. So that's an advanced bridgehead that can be used to make not only QS-7 and QS-21 but potentially other structurally similar molecules that the soap bark tree produces.

0:19:58.0 SC: And how do the yields compare doing this in the lab? Are you going to be able to get much more? I don't know how to exactly express the comparison here.

0:20:06.7 AO: That's a good question. So one of the nice things that we can do with this tobacco system is we can scale it up. We still have challenges to overcome to make these molecules in quantity. What we've shown for QS-7 is that we can make the molecules in tobacco at around about the same levels that they're found in the leaves of soap bark, but in the bark it's three times higher. Nevertheless, this is the first demonstration of the elucidation of the gene set for this pathway and so it's the starting point for learning how to make it at scale in heterologous expression systems.

0:20:43.5 SC: You can see as you talk about understanding the different intermediates and understanding the different enzymes involved that you can start to vary what you produce in a systematic approach.

0:20:53.3 AO: Yeah. And it's interesting that because these molecules have been difficult if not impossible to access systematically. Now that we have the instruction manual of the tree and we have a platform where we can engineer, we can recreate these pathways and also mix and match and make variance molecules. It now makes it possible to take a more systematic approach to trying to understand what makes a really good adjuvant, to start to look at the structure activity relationships and you know what are the best molecules with the best immunostimulant activity and the lowest toxicity to human cells.

0:21:29.1 SC: So interesting. Okay. Total sidebar. Do people make anything this complicated? Do mammals make anything this complicated with their bodies?

0:21:37.3 AO: So plants are very good at making exotic chemicals and I think that's because they can't run away so they have to defend themselves. We don't know what these molecules do in the soap bark tree but they may well be protecting the tree against attack by pathogens especially in the bark. As far as animals go, most animals don't make specialized metabolites with a few exceptions and we got very interested recently in marine animals that do make defense-related molecules and that includes the sea star and the sea cucumber.

0:22:12.3 SC: They're slightly mobile, they kind of move around but it's hard to defend yourself.

0:22:17.2 AO: They're slow, they're really slow. And unlike other situations, so quite often when you find animals that contain specialized metabolites, those are actually being made by endosymbionts and bugs that are helping them. But in the case of the sea star and the sea cucumber they're doing it themselves.

0:22:34.3 SC: Super interesting. So back to the plant world, what do you see as further rich avenues to explore with this setup where you're figuring out the enzymes and the pathways to make complex compounds that normally only plants can make?

0:22:50.1 AO: The platform that we've established is really powerful because we can very quickly mix and match and express genes and we can detect the chemicals, we can purify them, we can do structural analysis, but importantly we can make suites of molecules and analogues that can go into bioassays for different kinds of activities. So in this case we're talking about adjuvant activity but my lab has been working with the instruction manual not just from the soap bark tree but from many many different plant species. And we've been building a toolkit of genes and enzymes that ultimately we want to be able to make designer molecules on demand for different applications, for anti-cancer, for antimicrobial. And we want to be able to learn about the relationship between structure and activity of molecules so that we can build an engine that will get more and more powerful where we do go through the design, build, test, learn cycle and we can use machine learning as well to help us with this. Ultimately we'd like to build a platform to make first in class therapeutics.

0:23:57.1 SC: Thank you so much, Anne.

0:23:58.2 AO: Thank you very much. I really enjoyed talking with you.

0:24:00.0 SC: Anne Osbourn is a group leader and professor of biology at the John Innes Centre in Norwich. You can find a link to the paper we discussed and a related commentary at

0:24:10.2 SC: And that concludes this edition of the Science Podcast. If you have any comments or suggestions write to us at [email protected] You can listen to the show on the Science website, or search for Science Magazine on any podcasting app. This show was edited by me, Sarah Crespi and Kevin McLean with production help from Podigy. Jeffrey Cook composed the music. On behalf of Science and its publisher AAAS, thanks for joining us.