Welcome to the DeciBio Precision Medicine podcast where we speak to key opinion leaders on the cutting edge of biotechnology, pharmaceuticals, and research that are driving the scientific breakthroughs of tomorrow.
Here at DeciBio Consulting, we're joined by our special guest, Jacob (Jake) Becraft, the CEO and co-founder of Strand Therapeutics — a company leveraging synthetic biology to create the next-generation of messenger RNA (mRNA) therapeutics using a platform of programmable, precise, and multi-functional mRNAs through the implementation of genetic logic circuits and other genomic features.
To give a little background, Jake co-founded the company in 2017 with Tasuku Kitada, Strand’s current President and Head of R&D, during their work at the Massachusetts Institute of Technology (MIT) Synthetic Biology Center and the labs of their other scientific co-founders Ron Weiss and Darrel Irvine. Strand Therapeutics has gone on to earn various research and commercialization endorsements including the Bristol Myers Squibb 2018 Golden Ticket Award, two Phase I National Institute of Health Small Business Innovation Research Grants (NIH SBIR), and an international commercialization licensing agreement with BeiGene. It's a very exciting time with a big Series A ($52M) and Strand potentially starting their first in-human clinical trials in 2023. That was a mouthful, but a lot of great accomplishments and exciting news on the way.
To get started, it'd be great to hear about the key differentiating aspects of your mRNA technology, what really makes this platform programmable, and why it's different from conventional mRNA therapies popularized during the COVID-19 Pandemic?
Thanks for the background. To start, we need to talk about why messenger RNA is so powerful even beyond vaccines. What has always excited me about the potential of messenger RNA technology is pretty much any protein that you can imagine, synthetic protein sequence or natural protein sequence, can be encoded on nucleic acids — on DNA or messenger RNA.
What messenger RNA allows us to do is essentially use a single unified platform and single unified manufacturing infrastructure and design tools to create a messenger RNA transcript, which when added into any cell within the body can then cause that cell, basically inducing that cell, to produce that protein sequence. Going further than what's currently possible with biologics, mRNAs have the potential to do all sorts of interesting things with proteins.
Current biologics today are pretty much limited to extracellular protein structures. You have antibodies and other sorts of secreted proteins that need to interact with surface receptors, but with messenger RNA, since we're transcribing or translating the protein directly from inside of the cell, we can do intracellular based proteins, we can do transmembrane-based proteins, so the sky is really the limit in terms of the types of various proteins.
The big challenge to messenger RNA therapeutics going beyond vaccines is how do we get mRNA into different tissues throughout the body? Everyone who has been injected with messenger RNA, which is a wild thing — now, that wasn't the case when I started in mRNA or even when I started this company — but everyone who has been injected with an mRNA vaccine knows that it's injected right into your arm, then goes into those muscle cells in your arm, and sort of is active right there.
However, if you want to be able to deliver messenger RNA throughout the body in order to intervene either in a tumor, or to intervene in some sort of genetic-based disease - some lack of a correct protein that someone needs in maybe the kidney — what you're going to need is both delivery and specificity. Because delivery to the correct tissue is going to get you the efficacy that you need and the specificity is going to have the lower toxicity. You don't want to be expressing random things in random organs with all sorts of problems there.
This goes beyond just even messenger RNA, really the broad field of gene therapy or gene delivery technologies has pretty much been trying to wrestle this problem for the past 20 years or so by engineering different sorts of nanoparticles. This has led to a number of successes, but not enough specificity to really make a rigorous drug platform. One of the things that lipid nanoparticles do is they deliver well to the liver and you can actually get them to go other places within the body, but they're very promiscuous. They still end up in the liver, they still end up in other organs, and so this idea of specificity has been a huge issue for the field for decades.
Utilizing the synthetic biology tools that we had at our disposal at MIT we were able to build and take a different approach. The idea was that I don't have to only deliver messenger RNA to my target tissue, but I can deliver messenger RNA to a number of tissues and so long as I hit my target in a sufficient dose, then I can build a messenger RNA molecule itself that doesn't express in these other tissues. It's a different way of thinking and it's actually the way that viruses and other sorts of biological agents work where you have sort of a two-step process to achieve ultimate specificity. It leads us to really be able to develop drugs that no one else within the field is going to be able to.
That's fascinating. You touched on a number of topics there: The ability to create a wider array of proteins with actual intracellular delivery, the importance of transfection delivery vehicles or nanoparticles in order to get into different in vivo tissue or organ systems, you also talked about how when you program these mRNAs, you can create conditional expression of actual proteins and translation.
It seems from our reading of the different publications you have that there are two major components that you briefly touched on: These externally controlled genetic logic circuits, which have different genomic aspects such as subgenomic promoters, destabilizing binding domains, and then the second part, is the transfection delivery vehicle, which may be ionizable lipid nanoparticles (LNPs).
It'd be interesting to hear a little bit more in depth about what the components are in the genetic logic circuit that make it autonomous with conditional activation and what sets apart the transfection delivery vehicle?
Yeah, absolutely. The way that we think about it is sort of a two-step process. You have a nanoparticle that's going to get you the sufficient bio distribution that you need. You need the nanoparticle to get enough dose to where you want it to be. You need that nanoparticle also to have other sorts of characteristics such as low toxicity and biodegradability that are important to not have issues with the off-target delivery that invariably is going to happen with any sort of therapeutic modality of this type.
Second, then you have the logic circuits and that's really what's driving your specificity. The way that we're able to build that is we start with looking at the intracellular environment of different tissues. You look at different elements such as micro RNA (miRNA) levels in these different tissues, then you can identify ones that are differentially expressed.
Then the second piece of technology that you need to lay over the entire area in order to get specificity are the logic circuits. These logic circuits are essentially autonomous sequences of messenger RNA, which can sense the environment that they're transduced into and actually identify the environment for what type of cell they're likely inside of, and then use that information to determine whether or not to express the ultimate protein that we've encoded. The major way that we control these in a cell type specific manner is to leverage micro RNAs (miRNA) and other biomarkers that are specific to and differentially expressed in various cell types. Micro RNA are essentially the small silencing RNAs that are differentially expressed within cells; they're used in gene regulation.
What we can do then is use next-generation sequencing (NGS) and other tools to pull out various micro RNAs that are differentially expressed between different tissues, then computationally design circuits that respond to those certain miRNAs, model their performance, and then we can build messenger RNA so that it will essentially only express their cargo when there is a certain number of different miRNAs that are either over or under expressed and that we use as a marker of cell-type.
I love hearing things like that where you get to use the body and nature's tools, exploiting them with these programmable mRNAs, which then have specific performance that’s tuneable. That's a fascinating concept.
This is for our listeners, but how is this significantly different from the mRNA vaccines we've seen commercialized during the pandemic? What would the major differentiators be?
The first generation of messenger RNA therapeutics, being the vaccine technology, they're not really controlled. There's no logic gated circuits or anything. They're not really controlled whatsoever. They're locally administered into the muscle. You want the muscle to express the viral protein that's encoded on the messenger, so you inject it into the muscle, it goes into the muscle,and it creates the protein.
Why the first generation of messenger RNA was so well suited for vaccines is because the limitations of things like specificity are not really necessary to create a vaccine. You don't have to figure out how to express mRNAs just in a tumor, just in a T-cell or just in some cells throughout the body. You can simply inject it into the muscle and have it go to work.
Antigen design is of course a huge piece of the success there as well. That is something that we look at not in antigen design, but in cargo design. You're always designing the cargo and there's a lot of results that show the type of translation that you get from inside of a cell is going to vary based on a number of different variables some of which we don't even fully understand. This will impact the protein that eventually gets produced.
Our next topic is more on the clinical applications of your technology; what you're aiming for as a company. We know you potentially have in-human clinical trials coming in 2023, which we're excited to hear about and see come into fruition. We got a glimpse of maybe your initial disease areas of interest with a focus in immuno-oncology through the Phase I NIH SBIR grants approved for two indications: Melanoma and triple negative breast cancer (TNBC). Are these in fact the indications that you're initially pursuing and that you will be pursuing in clinical trials or are there other indications, generally, that are best suited for your technology?
We're designing drugs that are broadly applicable in the solid tumor category. Melanoma and triple negative breast cancer we think both represent great places to start the trials, and also especially with triple negative breast cancer, high unmet medical need that we can really create a lot of value for patients within this space.
Not necessarily as a replacement to pembrolizumab or other sorts of anti-PD-1 checkpoint inhibitor therapies. We're looking at seeing how these tumors will respond to our drug, but we do know that our drugs synergize rather than replace.
If you think about it, checkpoint inhibitors sort of take the foot off the brake. They press the gas of the immune system. Taking the foot off the brakes is probably the best analogy, you empower the immune system to just be able to really supercharge and go very fast in a number of cancers. For the immune system to then find tumors and actually attack and eradicate them, however, in the ever growing numbers of non-responders that we know is one of the giant problems in immunotherapy, that's not always enough. What our drugs really seek to do is give the extra boost. If the immune system is a car that we're taking the brakes off of and hoping that it will find its target and it's not being successful, then our drugs will provide the actual directions. Really lighting up the tumors and giving great signals for the immune system to really come in and eradicate them.
Outside of cancer, are there any diseases or other therapeutic areas that you're interested in that you could see your technology applied to in the future?
Publicly we've discussed one other area that we're very interested in, which is broadly in in vivo cell therapy generation.
If you think about the base case technology that we're developing, we're developing mRNAs that can be expressed for a longer amount of time in a cell type-specific manner. That's important for solid tumors and having solid tumors that can create different sorts of proteins.
But the same approach can be applied to say T cells or NK cells where you're delivering a messenger RNA and arming that cell type with any number of different effectors. And so from that perspective, we're very interested in building in vivo cell therapy capabilities.
Innovating on the many multitude of limitations that current cell therapies have in clinical and in the commercial market, and then beyond that we are in conversations with a number of other potential partners to really expand the outlook of this. For oncology, we know where we are and we were developing pretty much alone and through our partnership with Beijing.
With cell therapy we will likely find we have a number of interested partners, once those projects go down the line. But of course, as we want to look at developing therapeutics in a number of other indications where we don't have in-house expertise, partnerships are incredibly powerful and provide us near-term capital to continue to expand the platform across a number of indications.
That's an exciting development for sure. The combination of your technology with these different cell therapies could be a very powerful tool for the immune system.
The next question is about what you've seen in your studies and your ongoing validation studies. Are there any key findings you'd like to share or any sort of differentiators you've seen in the actual responses you've gotten from your technology?
Yes, we are seeing incredible responses pre-clinically across a number of different syngenetic models and robust expressions in PDX models as well. I would say that, what mRNA really offers us is this ability to leverage even known effector proteins, previous biologics, which have been complicated for a number of reasons, possibly from, the therapeutic index or systemic toxicity of things like IL-12 for instance, or just the inability to really have an on target efficacy at the safe doses. We see that as a huge opportunity since we're creating an environment where the tumor itself is going to be producing the proteins directly.
And we can of course leverage this for any protein that we want or any protein that a partner may want. We can encode those proteins on there and we see incredibly high tumor microenvironment concentrations of the proteins with very low levels of systemic exposure. And so that really drives the ability to both create very high efficacy, but also incredibly high tolerability.
So in some of our IND-enabling studies, now we're seeing just an incredible amount of tolerance for these drugs which gives a therapeutic index that we're frankly, given our efficacy studies as well, it's a therapeutic index that we're just incredibly excited about the potential of.
That is a great accomplishment, needing to manage not only the immunogenicity, but also have a very precise response. You brought up both hematological and solid tumors, for your technologies, how do you see the applicability or even efficacy in your preclinical studies in both types of malignancies?
In solid tumors, we have a platform that's essentially built to be able to express robustly inside of those solid tumors and the fact that we are very passionate about building in-vivo cell therapies or the ability to create essentially targeted delivery to various immune cells. That then opens up hematological cancers and even broader than that, in our mind, our platform doesn't have to be even limited to tumors. There's a number of other indications that we think could be interesting within the CAR-T space or other sorts of immune cell therapies.
Once you can master the ability to just deliver or create whatever protein you want inside of the immune cells, then your possibilities sort of really skyrocket. And so there's a number of immunology, auto-immune and even some other diseases that we think are very interesting once you can get into the T-cell arena.
We’re getting a sense that for solid tumors, we're talking about going with a combination therapy approach to boost let's say inhibitors and/or blockade therapies versus in more liquid tumors, we're going after delivering more material to the CAR-T cell itself for example. Is that the approach Strand Therapeutics is going after initially?
Yes right, more material for the CAR-T itself. I mean, the idea is that if you think about sort CAR-T cell therapies, there is a revolution in the way that we treat cancer for a lot of people. But if you think about the actual infrastructure of it, it's sort of ridiculous, right?
It is very crude for as advanced of a therapeutic as it is. It's like all you want is a CAR-T, a chimeric antigen receptor protein expressed inside of a T cell. But in order to do that, you pull the blood out of the patient, then you have to purify the T-cells, then take them to a lab. You then have to genetically engineer them with a lentivirus or another vector to get a chimeric antigen receptor into their genome. Then, you grow those out and then you inject them back into the patient, right? At 750,000 plus dollars later in three weeks, hopefully your patient is still alive or doing alright because it's like fourth and fifth line therapy. Obviously both financial and infrastructural strains and the demand versus supply is completely out of whack. It is a very simple thing that you're doing at the end of the day, right? It's just putting a gene in a cell, it's just that we lack the tools to do that in a way that isn't completely crude.
And so when I think about allogeneic cell therapies, it is great and it would be a huge advancement and definitely very important for the future of cell therapy as well. However, I don't know if I dared to dream about the future of cell therapy, it would just simply be an infusion like anything else, and possibly if you could drive the same levels of efficacy but transiently. You need a long lasting expression, but maybe not a lifetime expression.
Lentiviruses are sort of messy in their incorporation in the genetic material. So you just think about all of these sorts of limitations. I think messenger RNA is sort of a perfect opportunity to access that particular treatment.
There is also the entire scrutiny around AAV-based (adeno-associated virus) therapies and the regulatory requirements of them in terms of manufacturing and chain of custody among others, you would be basically avoiding all of that potentially.
There is an incredible amount of problems that we put up with, because our therapy cures people of cancer. When I have conversations about the challenges remaining to do when we talk about in-vivo mRNA delivery, I just kind of say: ``Do you have any idea how much we put up with current CAR-T therapies, right?” I mean, it's amazing what the early pioneers were able to accomplish and hats off to every single person involved in that. Some of them are friends of mine, but of course, if we think about what's an optimal future solution, I think our current standard of care of CAR-T cell therapies and current infrastructure of CAR-T cells is just simply un-american.
To follow-up on the supply chain part that you brought up, we've all been witnessing the widespread supply chain bottlenecks affecting biotech and biopharma. Has Strand Therapeutics been affected by the mRNA-related supply constraints, such as lipid nanoparticles shortages for example, and what are your plans going to be to mitigate future supply chain uncertainties?
Fantastic question. Given that we're targeting a clinical entry next year and we've started, or have almost pretty much finished our acquisition of different components for our GMP manufacturing runs ahead of those clinical trials, we've been pretty lucky to be able to manage them.
There are challenging problems, but we have one of the most phenomenal process development and manufacturing teams in the entire industry. Not even “one of the,” I think actually THE best, especially for a company of our scale and because of that, we've been able to weather it. It’s definitely been for some components — especially mRNA raw materials and some commonly used lipid nanoparticles raw materials — they have been probably some of the biggest challenges.
I think those are starting to subside, the supply side is meeting the demand side and the demand side is subsiding. As the demand for new vaccines is reducing over time we're seeing that demand is definitely sliding. But in general we've been pretty fortunate.
I'm not completely hammered by anything, but like any good company, you need to constantly be vigilant and adapt. So that's one of the things we pride ourselves on.
Fascinating and amazing to hear that. As you know, companies are accelerating their mRNA work and if they haven’t already started, are jumping on the mRNA train especially after the success of COVID vaccines. We've all witnessed the exciting regulatory and commercial environments of mRNA-based therapies during the COVID pandemic. How do you see both the regulatory and commercial landscapes evolving and changing going into a post-pandemic era and what are some of the drivers and barriers that you foresee for the adoption of mRNA therapies moving forward?
I am hopeful that the regulatory landscape will continue to evolve. I think that we will still end up regulating these therapies as if they are small molecules, because we started building our regulatory framework around small molecule therapies.
One of the greatest examples of where regulatory, best practices and sort of common sense run into each other, has been like the ability to pivot the mRNA vaccines to more relevant variant subtypes. When mRNA vaccines first came on the scene and even during the early parts of the pandemic, we saw Moderna and BioNTech speaking regularly about mRNA versatility with their ability to sort of respond to emerging variants with new vaccines very rapidly or potentially create multivalent vaccines. But we have seen an inactivity from companies to develop and really push that, in favor of developing and continuing to push the product that they're of course getting paid for.
The second thing we have been witnessing is the lack of appetite from the FDA to adapt their regulatory structures. You're seeing people, going all the way back to phase one or phase two for changing a few base pairs within the spike protein sequence in the mRNAs. While I think that we need to understand better what changing the antigen does to therapeutics, we need different sorts of regulatory patterns that don't start us back at square zero or square one and that is going to be a changing paradigm, right? We're going to see that with neoantigen vaccines, with custom-based neoantigens, like personalized vaccines that BioNTech has already started working on, we're going to see that. I think that adapting that regulatory framework will also simplify and empower the drug development process.
Imagine a world where you're developing like our therapeutic for solid tumors where we're expressing an IL-12 cytokine, perhaps we get into the clinic and we see some behavior among that IL-12 that we think we can improve with a small alteration based on some preclinical data, the ability to make that change and roll that out without starting back as if it's a brand new drug if it's a small change.
I don't have all the answers and I'm far from a regulatory scientist or a regulatory expert. However, I think more advanced ways of thinking about regulatory science need to be incorporated into the new world that we live in. We proved that we could both test and approve a vaccine in nine months of a brand new modality that had never been approved before when people actually wanted to get it done. So I guess the question is just a matter of resources and drive and in my head based on that experience is interesting.
Very interesting and definitely thought provoking statements. What makes Jacob very optimistic and super excited about the future and what are your thoughts about the future of the mRNA landscape moving forward?
I've been in the mRNA space for almost ten years now. It’s gone through every layer of skepticism and hype and excitement that you could think of across the entire spectrum from being told when I first started my PhD to not work on this garbage technology. That it will never be translatable to humans. This is by MIT scientists and professors, friends of mine still to this day. No, I will, but that's what they told me with their best advice in 2013, all the way to where we're at now, where I think there's probably a number of technologies that don't really deserve to be companies that really shouldn't be and that maybe aren't sufficiently differentiated, or for example, which happens whenever there is hype applied to a field. Ask any machine learning scientist as well, they'll have a similar thought process for you.
But, what I'm most excited about is, over the past two years, we've pretty much de-risked whether or not messenger RNA is capable of, both being active and being therapeutically relevant in humans at scale. I mean, we did it across the world, across every demographic we showed that this modality works in these people well, it works in people and can create therapeutic effects, at least from a vaccine context. That's a huge de-risking factor in my mind for a brand new modality. And now it really opens up.
For myself when I started the company, I used to have to tell this story that was two parts and more complicated, which was that mRNA will be a future cornerstone of biomedicine and that we're the future technology that enables the future of messenger RNA. And that's obviously a more complicated story than the one that we can tell today, which is that the mRNA that you've been injected with, we’re the future of that.
But I think that the way that the story has evolved is in part sort of indicative of how the science has evolved, right? Now, we know that messenger RNA does work, we can actually sit, and we've seen this with other companies that have emerged within the space as well to sort of bring new approaches to developing messenger RNA drugs and to sort of expand the purview of what we can do with it.
The idea is very exciting to me because it really proves that now we're really able to have entire companies like my own, which can sit around and think solely around what is the future of this messenger RNA technology and how do we adapt it? In my mind, it can replace so many different types of biologics and it can — much like any good enabling technology that I think is sufficiently powerful — in the future, enable entirely new marketplaces and entirely new drug design. Up until now, we've only been able to make biologics that are extracellularly active, bind receptors, bind cells, and block things. What happened now with messenger RNA, we can actually start thinking about things that we haven't thought about with protein therapeutics before.
With small molecules we think about drugging intracellular proteins and domains and transmembrane domains, but with protein-based therapeutics, we haven't. And so, it's something that they talk about in the tech industry a lot, which is like, what, why does your technology not attack a market, but in fact actually enable a completely new market, an entirely new class. And for me, messenger RNA being so adept at getting into cells and expressing protein that it opens up a completely new thought process of how we'll develop drugs and really positions my company, Strand, and even others that are trying to solve this problem of not only being drug development companies into the future, but really being a hub and wheel of innovation centers where we can develop, tens to hundreds of different types of drugs going into the future with with all sorts of partners across.
mRNA is a major scientific moment, it's had a major moment throughout the pandemic, and now, you made great points about regulatory bodies needing to maybe reconsider what is a substantial change, especially in mRNA platforms. Do you really need to go all the way back to phase one for a couple of base pairs?
We are coming to the end here, but one last topic we want to talk about is something you're very familiar with. It's our Dexter expert service platform here at DeciBio. As you know, here at DeciBio, Dexter is a major contributor to the quality consulting work we do. For those unfamiliar with Dexter, Dexter is our expert service platform that acts as an exclusive channel for companies to access a diverse array of life science professionals and key stakeholders to inform wide-ranging projects with their insights, ultimately driving innovation within the precision medicine industry. From your experience, what are the major benefits and uses of the Dexter platform by life science companies? How do you think Dexter works to foster community within the life sciences industry?
Absolutely, especially as messenger RNA got more and more popular and then into the pandemic of course became the subject of national news, I'd say the amount of expert network and consulting requests that I've gotten has gone up pretty dramatically. And so I've been exposed to a few different platforms that sort of are used to recruit people across different areas of technical sciences and interface them with different groups.
I would say Dexter is by far the smoothest platform and sort of leads to me working with DeciBio. Does the job more than other groups because of the smoothness and the more immediate feedback areas. I mean for me, it's sort of a go-to, especially when more opportunities come up, then I really have time to get to.
It really helps me to have concrete interactions and sort of see like, okay, here are all the different opportunities of what might be helpful to people and, you know, what are you interested in? What can you actually do? So I really enjoy that aspect of sort of the very clear cut system, easy to work with, and I know what I'm getting into whenever I get into it and it makes it easier.
I think it made it pretty easy for me to upload different credentials so that folks throughout the industry can sort of find me or that you guys, the consulting teams, can sort of find me if that's the expertise that’s needed, and as Dexter continues to grow, I think it will be very powerful. It's just really a well put together platform.
We love to hear that, and maybe from a different perspective, as a biotech company in space, when you turn to get primary research and comments from key stakeholders, how do you see Dexter playing into that and standing apart?
Well I think for the same reason, it becomes incredibly easy to utilize, it makes it easier on the consultants and different key opinion leaders to really enter their expertise and then makes it easily searchable by people looking for different kinds of networks that they can then engage with and access.
One of the things that becomes difficult as you sort of want to boil the landscape, for instance, like with messenger RNA, we want to look across ten different types of modalities and really get different sorts of answers to what we and how we could actually build in different disease areas, searching through that or trying to go through our own networks even though Strand Therapeutics, myself, and other people who work at the company are highly networked individuals, it’s heavy lifting sometimes to find all sorts of different folks. And so it's easier if you can have a system like Dexter that has people united together, it's easily searchable, and then sort of feeds right into the interaction as well.
That's great to hear. I think that wraps everything up. We love working with you, Jake, it's always a pleasure. You know, you have a lot of exciting developments happening there at Strand Therapeutics and we really can't wait to see what the future holds for your company and the mRNA therapy field as a whole. So it's going to be an exciting time.
We're wildly excited at Strand. I worked on this technology now, like I said, for almost a decade and I've quite literally never been as excited as I am today. The progress that the team has made - I don't work in the lab anymore, so I can't even take credit for it - but the progress that the team has made is nothing short of extraordinary, and I think that we'll continue to see that acceleration of advancement happening over the coming year. So I wake up every day pretty excited.
- Mc Cafferty, S., De Temmerman, J., Kitada, T., Becraft, J. R., Weiss, R., Irvine, D. J., Devreese, M., De Baere, S., Combes, F., & Sanders, N. N. (2021). In vivo validation of a reversible small molecule-based switch for synthetic self-amplifying mrna regulation. Molecular Therapy, 29(3), 1164–1173. https://doi.org/10.1016/j.ymthe.2020.11.010
- Wagner, T. E., Becraft, J. R., Bodner, K., Teague, B., Zhang, X., Woo, A., Porter, E., Alburquerque, B., Dobosh, B., Andries, O., Sanders, N. N., Beal, J., Densmore, D., Kitada, T., & Weiss, R. (2018). Small-molecule-based regulation of RNA-delivered circuits in mammalian cells. Nature Chemical Biology, 14(11), 1043–1050. https://doi.org/10.1038/s41589-018-0146-9