clock menu more-arrow no yes mobile

Filed under:

Using AI, scientists bring Neanderthal antibiotics back from extinction

Neanderthals are extinct. But their molecules are back and they just might save our lives.

A Homo Neandertalensis skull, originally from the Near East, at the Marseille Prehistory Lab.
Marc Charuel/Sygma via Getty Images
Sigal Samuel is a senior reporter for Vox’s Future Perfect and co-host of the Future Perfect podcast. She writes primarily about the future of consciousness, tracking advances in artificial intelligence and neuroscience and their staggering ethical implications. Before joining Vox, Sigal was the religion editor at the Atlantic.

Maybe you remember the movie Jurassic Park, where scientists bring dinosaurs back from extinction. Or maybe you’ve heard about the real-world scientific quest to de-extinct the dodo and the woolly mammoth.

Whether we’re talking dinos or dodos, de-extinction is risky business. It’s problematic on a pragmatic level (that T. rex you resurrect just might gobble you up while you’re sitting on the toilet) and on an ethical level (can you imagine how lonely the first resurrected dodo or woolly mammoth would be?).

But what if instead of bringing back a whole species, we bring back just one tiny part — like, say, a molecule?

That’s what scientists have just achieved at the University of Pennsylvania’s Machine Biology Group. They’ve resurrected molecules with antibiotic properties found in extinct organisms — specifically, our close relatives, the Neanderthals and Denisovans. (Neanderthals went extinct 40,000 years ago, while Denisovans might have survived until 15,000-30,000 years ago.) The breakthrough throws open the doors to a brave new world of “molecular de-extinction,” which holds promise for drug discovery.

The researchers started by gathering the sequenced genome data of Neanderthals and Denisovans, which is publicly available thanks to paleontologists who have painstakingly collected and analyzed ancient DNA from bones and artifacts.

Then they trained an AI model to make predictions about which molecules might make effective antibiotics for modern humans. After the algorithm identified the strongest candidates, the researchers created those molecules in the lab and tested them in infected mice. Some of the molecules effectively fought off bacterial infections, according to a new study published in the journal Cell Host & Microbe.

“This is completely new. We came up with the term ‘molecular de-extinction’ and this is the first peer-reviewed paper that describes it,” César de la Fuente, who co-authored the study, told me. “So it’s quite exciting for us.”

If the burgeoning field of molecular de-extinction turns out to yield clinically successful results in humans, it could be exciting for the world, too, because we urgently need good ways to create new antibiotics.

We’re entering a post-antibiotic era. Here’s how AI is helping.

The CDC has warned that we’re entering a post-antibiotic era — a time when our antibiotics are becoming increasingly useless. We’ve created this crisis by overusing antibiotics in the treatment of humans, animals, and crops. The bacteria have adapted to our drugs, morphing into superbugs that can all too easily decimate human health.

In the time it takes you to read this article, one person in the US will die from an infection that antibiotics can no longer treat effectively. The annual global death toll from drug-resistant infections could rise to 10 million by 2050, a major UN report warned in 2019, if we don’t make a radical change.

Big Pharma and biotech companies haven’t been creating the new antibiotics needed to address the crisis because it takes many years and lots of funding to do the research and development. Most new compounds fail. Even when they succeed, the payoff is small: An antibiotic doesn’t sell as well as a drug that needs to be taken daily. So for many pharma companies, the financial incentive just isn’t there.

But if we can use AI to ramp up the speed of antibiotic discovery, that could change the calculus. De la Fuente’s team did this in a two-step process.

The proteins in our cells are like long strings. But they often get cut up by enzymes into short fragments at specific junctures. These short fragments — small molecules known as peptides — can have antimicrobial properties.

So, de la Fuente’s team trained an AI model called panCleave, which can look at all the proteins encoded in a genome and predict where their junctures will be. That way, the researchers could identify the peptides in the Neanderthal and Denisovan genomes. They then ran other models on the peptides to predict which ones would have antimicrobial properties.

They weren’t all slam-dunks. When tested on infected mice, some of the peptides predicted to be strong candidates didn’t manage to kill the bacteria. Others were effective but required high doses to work. That could mean researchers need to either revamp the predictive algorithm or retool the most promising peptides to make them more effective.

A further limitation of this study is that the researchers didn’t test whether the infected mice developed resistance to the peptides. “It’s something to do in the future,” de la Fuente acknowledged, since there’d be little point in manufacturing a new antibiotic only for our bodies to soon become resistant to it.

But at a high level, molecular de-extinction is “a creative approach” that can help us get past the current bottlenecks in drug discovery, according to Jonathan Stokes, a biochemistry professor at McMaster University who was not involved in the study. “I think this technique will augment other antibiotic discovery efforts to help us discover structurally and functionally novel antibacterial therapies that overcome existing resistance mechanisms,” he told me.

Molecular de-extinction could end up working alongside other efforts in the use of AI for antibiotic discovery. In 2020, for example, MIT researchers discovered a new type of antibiotic by training their AI on molecules that we know have antimicrobial properties. And in 2021, IBM researchers designed two new antimicrobial peptides in a matter of days by training their AI on a much broader database of all the known peptides that exist in nature today.

Resurrecting the peptides of yesteryear gives us more possibilities to sift through and potentially more buried treasure to discover.

“Molecular de-extinction” brings up big ethical questions. Nobody knows the answers.

What does it mean to bring molecules back from the dead?

De-extincting molecules doesn’t suffer from the same ethical concerns that plague de-extincting a whole species — like the concern that revived dodos or woolly mammoths would be miserable in today’s world, if they could even survive. Still, on a philosophical level, it’s not obvious how we should think about efforts to revive molecules that currently exist nowhere in living organisms.

For instance, would de-extincted molecules be eligible for patents? Existing patent law tells us that nobody can patent molecules that occur in nature, but it doesn’t tell us whether someone can own the rights to a resurrected molecule that was once expressed in Neanderthals. Arguably, that should be considered the heritage of all humankind, and no individual or business should own it.

But if patent law ultimately comes out on that side, will it disincentivize scientists from doing research in molecular de-extinction — research that could ultimately help us with the antibiotic resistance crisis?

I asked de la Fuente if he would want to patent his work. He replied, “I don’t know.”

But, he told me, he did walk over to the patent office on his campus one day to find out if that would be possible. The legal minds at the University of Pennsylvania couldn’t tell him. This is a new legal frontier, and so far, no one knows the answers.

Sign up for the newsletter Today, Explained

Understand the world with a daily explainer plus the most compelling stories of the day.