Listen to Our Podcast
Dear Aventine Readers,
Welcome to 2024!
Late last year both the U.S. and the U.K. granted approval for the first-ever CRISPR-based therapy. So this month we are looking at what that means for the future of such therapies given that this first treatment — for sickle cell disease — was in many ways unique in its almost frictionless path to approval.
Also in this issue: a look at the role hydrogen needs to play in the green energy transition and the potential costs and risks involved; an antibiotic that has effectively combatted — in mice — one of the most pernicious drug-resistant bacterias plaguing hospitals today; and how generative AI systems are contributing to a water crisis in Africa.
Thanks for reading,
Executive Director, Aventine
Subscribe to our newsletter and be kept up to date on upcoming Aventine projects
CRISPR Can Solve Sickle Cell Disease. What’s Next?
As 2023 drew to a close, Vertex Pharmaceuticals made history. The company succeeded, after just eight years of development, in gaining regulatory approval for the first-ever CRISPR-based therapy. The one-time treatment, called Casgevy, alleviates the symptoms of sickle cell disease, allowing sufferers to lead a normal life.
The relatively breakneck development of the therapy — CRISPR itself has existed as a tool for only twelve years — might suggest that Casgevy heralds an imminent explosion in CRISPR-enabled treatments, but the reality is more complex. Aventine spoke with several genetics experts who agreed that while the new therapy is a critical milestone, not too much should be extrapolated from Casgevy’s journey; sickle cell disease was a relatively straightforward test case that belies the complexity of developing many other potential therapies.
“It’s obviously a landmark,” said Stuart Orkin, a professor of pediatrics at Harvard Medical School, whose fundamental genetics research underpins the therapy developed by Vertex. “Still, each treatment is going to have to be taken one at a time.”
Treating sickle cell disease
Sickle cell disease is caused by a genetic mutation that makes the body produce faulty hemoglobin molecules, which give rise to unusually shaped red blood cells. These unusually shaped cells can cause periods of intense pain as well as increasing the risk of infections and anemia. The disease affects close to eight million people worldwide and is most common among people with African ancestry.
Researchers have known for decades that the symptoms of sickle cell disease do not present until a few months after birth because fetal hemoglobin, which persists in the body for a short period after a child is born, does not give rise to the unusually shaped blood cells. It’s when the body begins to produce the adult version of the molecule that the disease kicks in. Research, much of it carried out in Orkin’s lab, revealed that a gene called BCL11A prompts that switchover from fetal to adult hemoglobin by essentially turning off the production of fetal hemoglobin. By editing that gene, Orkin’s research showed, it should be possible to reactivate the production of fetal hemoglobin, enabling the body to produce regularly shaped blood cells.
And this is exactly what Casgevy does.
The result has been a “massive” step forward for sufferers of sickle cell disease, said Simon Waddington, a professor of gene therapy at University College London. Early recipients of the therapy during clinical trials claim that it has been life-changing: one, writing in MIT Technology Review, reported that it enabled him to “experience things I had only dreamed of,” including “boundless energy” and “a sense of control over my own destiny.”
The story of Casgevy is also unique in its seamless path to market. This was the first therapy that a company called CRISPR Therapeutics, one of many startups inspired by the development of CRISPR, sought to build. After partnering with Vertex Pharmaceuticals (which provided financial backing and expertise in navigating trials) the company spent the years between 2015 and 2023 working to turn fundamental genetic research into a viable treatment. In November 2023, after extensive clinical trials, the resulting therapy, Casgevy, received regulatory approval in the U.K; approval by the FDA in the U.S. followed quickly in December 2023. “It went precisely the way it should go: fundamental discoveries done in academic research labs, translated by people who know how to do that,” said Orkin.
More broadly, the therapy’s success suggests that you can safely use CRISPR to correct a defect — a proposition that was largely theoretical until this point. So far the therapy has not shown any “off-target effects,” which is an ongoing safety concern with all CRISPR-based therapies. A treatment’s off-target effect would make unintended edits and affect genes in undesirable ways. Perhaps more important, the development demonstrates institutional acceptance: “It shows the willingness, at least in this country, to approve gene editing therapies,” said Orkin.
(Vertex declined an interview on the technology behind Casgevy, referring Aventine to CRISPR Therapeutics. CRISPR Therapeutics didn’t respond to a request for an interview.)
Casgevy’s lucky break
Yet for all the excitement, experts also point to a need for restraint. Sickle cell disease has several traits that made it particularly attractive as a target for CRISPR-based treatment, so that it is a relatively straightforward test case for development and commercialization.
One important feature is that it’s easy to access the cells involved in the disease. The treatment targets blood stem cells that are produced in bone marrow: Stem cells are harvested from a patient, edited using CRISPR outside the body, then infused intravenously back into the body so that they can resettle in the bone marrow. (Before the final step, patients undergo an onerous course of chemotherapy to remove unedited stem cells still in the bone marrow — a process called “conditioning” — so that the new ones can proliferate.) Being able to do the work outside the body is a “big advantage” for the treatment of sickle cell disease, said Orkin. For genetic diseases affecting cells in muscle, the brain, or certain organs such as kidneys, CRISPR therapies would need to be delivered by binding the CRISPR editing machinery to something, such as an antibody or lipid nanoparticle, that would seek out specific cells within the body and make the edits there, he explained.
Then there’s the complexity of the editing itself. Casgevy makes just a single edit to create the intended effect in the body. (While there are actually two genes associated with suppression of fetal hemoglobin, editing just one of them is sufficient to alleviate sickle cell disease.) Many genetic diseases would require more edits, making the development of therapy far more complex. With Casgevy, It’s also relatively easy to tell if the edit has worked because patients can report changes in their symptoms within weeks.
But perhaps the most significant factor easing Casgevy’s path is economics. For a pharmaceutical company needing to recoup its R&D costs for a new treatment, finding a big enough market is vital. Crucially for Casgevy, as many as 100,000 people suffer from sickle cell disease in the U.S. alone; for many other genetic diseases, far fewer people are affected and the economics just don’t work out.
A trickle or a flood?
While many diseases check one or more of these boxes, sickle cell disease benefited from checking all of them. “The stars aligned,” said Orkin. “I call it either dumb luck or good fortune.”
For that reason, he said, there “won't be a flood” of new CRISPR therapies suddenly being offered to patients in the next year or two, because there simply aren’t enough instances of this kind of low-hanging fruit. Indeed, Fyodor Urnov, a director at the Innovative Genomics Institute, told Stat that the most advanced CRISPR companies in the world are currently developing therapies for fewer than 10 different genetic conditions — a tiny percentage of the thousands of known genetic disorders.
One of the main reasons for that is the cost of developing these therapies. While Casgevy was considered sufficiently economically viable for development, it will still cost $2.2 million per patient for a single treatment. For context, gene therapy treatments often cost upward of $1 million, reflecting the huge investment and ongoing cost of developing them for the pharmaceutical companies. With such price tags, “there's not much tolerance for failure these days” when it comes to translating fundamental research into usable treatments, said Orkin.
Driving down costs over time will be an important part of the success of CRISPR-based therapies — both for their proliferation and their reach. Currently, the prices of the resulting therapies make them inaccessible to huge swaths of the population. And in the case of sickle cell disease, the largest number of sufferers reside in African countries, most of which have neither the finances nor the infrastructure required to administer the treatment. This effectively means that much of the continent has no access to it. “I continue to be horrified by the prices that have been put on gene therapy,” said Waddington.
Still, there is hope that this will change. Monoclonal antibodies are an example of how a medical technology can move from being prohibitively expensive to widely used, and something similar could happen for CRISPR-based therapies as they’re increasingly rolled out and adopted. ” “If you give people really good therapy, I think people will accept it, and people will pay for it [and] it will be supported by pharma, by insurance,” said Orkin. “But, you know, I wouldn't expect it overnight.”
Listen To Our Podcast
Learn about the past, present and future of artificial intelligence on our latest podcast, Humans vs Machines with Gary Marcus.
Advances That Matter
An AI-designed Battery, Light on the Lithium. As we grow more dependent on batteries, finding alternatives to lithium-heavy models has become an urgent challenge: Demand for the metal will soon outstrip supply. Researchers from Microsoft recently deployed AI to tackle this huge and complex physics problem using a series of machine learning models to sift through 32 million materials that could be used as electrolytes — the materials that facilitate the movement of charged particles in batteries, allowing them to charge and discharge. Gradually, the software weeded out materials that weren’t stable or lacked certain physical or electrical properties. With a final list of over a hundred, the team collaborated with material scientists to further whittle down the list, settling on one candidate material — which contained about 70 percent less lithium than some rival designs, with an unusual and interesting amount of sodium in its place. But the big news here, really, is less about the specific material — though that could go on to be important — and more about the pace of the work. Not only did the AI zip through its analysis in just a few months, it took only nine months to develop a working prototype that could be used to power a light bulb. That speaks to the efficiency gains that researchers are likely to capture from AI in coming years.
A Potent New Antibiotic. The bacterium known as Carbapenem-resistant Acinetobacter baumannii (or Crab) is a tough cookie: it quickly adapts genetically to resist the effects of antibiotics and has a protective double cell wall that makes it both hard to penetrate with drugs and allows it to survive in inhospitable, dry conditions like those found in ventilators. That, combined with its ubiquity in hospitals and its ability to cause severe infections, has made it one of the three highest priority pathogens to treat around the world, according to the World Health Organization. Now, a new class of antibiotic called Zosurabalpin has been shown to effectively defeat Crab, interfering with the production of its protective outer shell by blocking its accumulation of long-chain molecules. The findings, described in Nature, show that Zosurabalpin was able to reduce levels of bacteria in mice that had Crab-induced pneumonia and prevent death in mice with Crab-related sepsis. If such a treatment turns out to be effective for humans it could make a substantial dent in the one million-plus global deaths caused each year by antibiotic-resistant bacterial infections, with 35,000 of those in the U.S.. Still, there’s a long road of clinical trials ahead before the Roche-manufactured antibiotic can make it into clinical use. And while experts have speculated that similar approaches could be used to tackle other bacteria, such as E coli, an editorial in Nature describes how government intervention will likely be required to ensure funding for the creation of a new breed of antibiotics capable of combating antibiotic-resistant infections.
Fresh Hopes for Supersonic Air Travel. The ability to get from New York to London in under three hours disappeared when Concorde, the first and only supersonic passenger jet, was retired from service in 2003. The plane was hugely expensive to operate and so noisy that it was banned from many flight paths; any viable alternative will have to contend with those issues. Enter the rotating detonation engine (RDE), reports The Economist. The RDE design has long existed in labs but is now ready for prime time thanks to projects at GE Aerospace and Raytheon, which are both developing such engines for use in missile propulsion systems. For years, the ability to build those engines — which operate through controlled detonation of fuel — was out of reach, but new manufacturing techniques and advances in material science have now made it possible. The engines are more energy efficient than other supersonic engines and could be made to operate more quietly — so the hope is that they may soon also find application in passenger aircraft. So far, nobody has solved the physics of silencing that pesky sonic boom, though a new experimental airplane built by NASA and Lockheed Martin is set to test whether features such as a smooth underside, elongated nose and alternative engine positioning can help quiet supersonic flight to tolerable levels.
In every plan for the green energy transition, hydrogen plays an important role. Hydrogen is an energy carrier, not an energy source; making hydrogen is akin to filling a battery with power. And while renewable electricity can keep the lights on, fuel our cars and power our cities, it cannot easily replace the fossil fuels we currently use in heavy industrial plants, shipping and airplanes. Hydrogen fuel could — but first, we need to make it.
The cleanest way to create hydrogen is a process called electrolysis, wherein a device known as an electrolyzer splits water into oxygen and hydrogen. If the electricity needed to power the electrolyzer comes from renewable sources, the hydrogen is called “green.” Otherwise it takes a different name from the so-called hydrogen rainbow, which describes different ways of producing hydrogen — pink if it’s from nuclear power, blue from fossil fuels with carbon capture, gray from fossil fuels without carbon capture, and so on.
But green hydrogen remains mostly theoretical. Producing catalyzers — materials that speed up the electrolysis reaction to make it efficient — is so costly that the resulting hydrogen is significantly more expensive than fossil fuels. Instead, almost all hydrogen produced today is gray, with only a few production facilities beginning to incorporate technology for carbon capture to make it blue. And once the hydrogen is produced, it’s still extremely difficult to store and transport in a cost-effective way that also conserves energy.
In October 2023, the Biden administration announced an investment of $7 billion into seven hydrogen hubs across the country to accelerate the creation of practical hydrogen applications. But only two of those seven hubs — in California and the Pacific Northwest — will focus entirely or primarily on green hydrogen technology. Aventine spoke with experts across the field, from engineers to environmental scientists to policy coordinators, to explore the challenges still facing green hydrogen and what it will take to overcome them.
Wind and solar electricity are so inexpensive that you have these stranded [spare] resources now. Nobody is using solar and wind electricity when you have these surges [in supply]. We want to have a way of storing that, and that’s where the idea of a reversible hydrogen cell that can operate like a battery makes sense. And I think that there’s excellent potential to create the catalyst that will make hydrogen production from green electricity much more efficient and therefore less costly.”
— Sossina Haile, professor of engineering and applied physics at Northwestern University and a researcher of solid acid fuel cells
I am really color agnostic when it comes to hydrogen, and I think the color wheel actually is pretty harmful in keeping our eye on the ball. We should be thinking about it in terms of carbon intensity; I like any form of hydrogen that has a low carbon intensity and is better than the options that we are currently using in the form of fossil fuels. I think that often, particularly in energy, we let perfect be the enemy of the good, and that certainly seems to be what’s happening with green hydrogen. It looks like the guidance for green hydrogen is going to lean toward some very strict rules and constraints in terms of ensuring that every molecule is completely free of carbon by requiring 100% fully renewable energy, and our grid is just not in a position to really do that right now. I'm of the opinion that we should work really hard to establish the hydrogen markets, recognizing that if we’re using the grid, it’s not going to be 100% carbon free, but it’s certainly a lot better than what we’re doing right now.”
— Naomi Boness, managing director of the Natural Gas Initiative at Stanford University
The biggest challenge is cost. If I can replace a diesel [engine in a] truck with a hydrogen fuel cell and it costs the same, the technology becomes a no-brainer. But if it could be done, it would have been done years ago, so practically the big challenge is: How can we increase the efficiency and the durability of the electrolyzers? Or decrease the amount of precious or expensive materials we use [as catalyzers] and still maintain the same efficiency or performance in the technology? If we can use fewer materials and have long durability, such that we only need (install and replace) one electrolyzer every 40 years, that’s going to really decrease the cost of the hydrogen being produced, such that the cost will be dictated by the inlet feeds, like how much the electricity is. Similarly, if we can increase the durability of a hydrogen fuel cell so that we can go a million miles (without maintenance), then that starts to look advantageous, assuming that the cost of hydrogen is somewhat similar to diesel.”
— Adam Weber, chief technology officer for ARCHES, California’s green hydrogen hub
In my opinion, the biggest challenge is hydrogen transportation. There was a huge debate and discussion about the hydrogen economy in the early 2000s — it seemed like hydrogen was the answer to everything — and then very little really materialized. I think now we are in another cycle in which we can see hydrogen as important for a green future, but I do believe we’re still in the same place when it comes to taking hydrogen from Point A to Point B. How do we transport hydrogen? We’re doing quite a bit of work in advancing other areas of hydrogen when transportation is lagging behind, and that may kill the whole hydrogen economy.”
— Matteo Cargnello, director of a research group focused on carbon capture, emissions control, electrocatalysis and other green energy technologies at Stanford University
What do you do with the hydrogen you’ve created? It’s very hard to put [hydrogen] in pipelines and put it in tanks and carry it around, [because] it's a very light gas and it’s dangerous. So the industry is thinking of transforming hydrogen into something else, and one of the most prominent options that we as a world are considering is ammonia. Ammonia is toxic, it’s corrosive, it’s poisonous, it’s not something we should do lightly. You could reconvert it back to hydrogen and then burn hydrogen directly, but that is expensive. Or you could directly burn ammonia, but if the combustion is not perfect, if the air-fuel ratio is not ideal, you produce other nasty stuff like nitrous oxide, which is 300 times more powerful than C02 as a greenhouse gas. If we’ve now done all of these things to avoid the problem of C02, it’s a huge problem. But as long as we do things well, I’m generally optimistic. I think people have the right mindset to pay attention to this problem and fix it.”
— Amilcare Porporato, professor of civil and environmental engineering at Princeton University
Technology’s Impact Around the Globe
1. Kapolei, Hawaii. Overlooking Barbers Point Harbor on the southwestern tip of the island of Oahu is a patch of land that’s home to four neat rows of nondescript metal boxes, each about the size of a small outbuilding. It might not look like much, but it's one of the world’s most advanced grid batteries, which went online just before Christmas, reports Canary Media. Designed and built to help replace Hawaii’s final coal power plant, which shut down just over a year ago, the grid battery — made up of 158 huge storage devices supplied by Tesla — smooths out power supply from renewable sources so that renewable sources can be used 24/7. It’s also built and wired up in such a way that it can kick-start the entire grid back into life if, say, a natural disaster knocks the entire thing offline. Over time, you can expect to see more of this kind of installation, but probably only as the cost gradually falls over time — the Kapolei installation ran a cool $220 million. For now, Hawaii is at the cutting edge.
2. Lagos, Nigeria. A thirty-question back-and-forth with GPT-3 reportedly uses around seventeen ounces of water. This fact is an urgent concern in certain sub-Saharan African communities, according to SciDevNet, because the water demands of data centers are straining local supplies. Data centers — integral parts of modern technology that underpin almost every aspect of our digital lives — foster technological development and boost local economies. But they also demand a significant amount of power and water — demands that are skyrocketing with the advent of generative AI systems like GPT-3. Already short of potable water, cities such as Lagos now find supplies dangerously strained. Water and security experts told SciDevNet that stronger regulation, monitoring, reporting and innovation are all required to ease the potential impact of a world bent on using generative AI.
3. Everywhere. Higher temperatures, more rainfall, greater numbers of storms: These are some of the ramifications of climate change. They may also, it turns out, be fostering a rise in antibiotic-resistant microbes. This story in Nature describes research that’s beginning to piece together how the two phenomena may be linked. Some early observations: Wetter, warmer conditions increase the growth rate of bacteria; greater fluctuations in temperature appear to induce genetic changes in bacteria that lead to greater antibiotic resistance; extreme weather events cause injury and infection that promote the use of antibiotics and in turn increase the likelihood of antibiotic resistance. The situation is more complex than that, of course, and there are not yet any provable causal links between the two phenomena, but health policy experts told Nature that global collaboration may be required to battle this possible new side effect of a warming planet.
Magazine and Journal Articles Worth Your Time
A Tunnel to a Magma Chamber Could Unleash Unlimited Energy, from New Scientist
2,400 words, 10 minutes
Drilling a hole into a magma chamber beneath a volcano sounds, frankly, like a stupid idea. But it’s what the Icelandic Deep Drilling Project accidentally did all the way back in 2000 while prospecting around the Krafla volcano for super-hot water to generate geothermal energy. It turns out that the team had stuck a magma chamber just over a mile beneath the Earth’s surface, destroying the drilling equipment in the process — but, crucially, not causing an eruption, as you might have expected. Fast forward to the present day, and the Krafla Magma Testbed (KMT) project, established in 2014, is gearing up to drill two more holes in the same area starting in 2026. One will drill right back into the chamber and serve as what KMT calls an observatory, allowing it to study magma conditions below the Earth’s surface for the first time. A second will venture just shy of the chamber itself, with the intention of tapping superheated steam with which to drive turbines — potentially creating a kind of geothermal energy on steroids that could someday be used in locations around the world.
All the Carcinogens We Cannot See, from The New Yorker
6,200 words, or 26 minutes
The causes of cancer are many, varied and complex — and the mechanisms through which they can occur are even less well known. That’s particularly true of carcinogens, the substances that appear to promote the development of cancers. This wonderfully written story in The New Yorker traces the history of our understanding of carcinogens right through to a landmark research paper published in 2023, exploring the shift in thinking from a belief that carcinogens directly caused cancer through cellular mutation to a more complex view that they rouse already mutated cells from stasis. Carcinogens, it seems, have to be present in the wrong place, at the wrong time and for long enough to have an impact, and even then there’s still plenty we don’t understand about genetic differences that could cause them to have a greater or lesser impact on individuals. The story doesn’t leave you with a nice, neat punchline, but it’s a fascinating and informative look at how biological science is more complicated, interdisciplinary and effective than ever.
Thousands of AI Authors on the Future of AI, from AI Impacts
9,500 words, or 40 minutes
An international team of researchers recently surveyed 2,800 experts who had presented work at six of the world’s top AI conferences on the future of their field, and this research paper lays out what they found. Perhaps the biggest message: Things are accelerating. The survey asked the researchers to predict when AI might outperform humans at a series of thirty-nine different tasks — from building a website from scratch and writing a Top 40 pop song, to physically wiring an entire house and proving unsolved math theorems. The experts gave even odds of AI outperforming humans at all those tasks by 2047 — a full thirteen years sooner than predicted by a similar, smaller survey just a year earlier. There are obvious caveats: Just because AI researchers spend all their time thinking about AI doesn’t necessarily mean they’re particularly good at predicting the future, and asking only that group obviously introduces its own biases. But the paper is still an interesting look at what some of the best qualified AI experts think about the future of the technology.