AI uncovers new antibiotics in ancient microbes

(Press-News.org) They’ve survived for billions of years in boiling acid, deep-sea vents and salt flats. Now, some of Earth’s oldest life forms — microbes called Archaea — are offering a new weapon in the fight against one of today’s most urgent health threats: antibiotic resistance. 

In a new study published in Nature Microbiology, researchers at the University of Pennsylvania used artificial intelligence to identify previously unknown compounds in Archaea that could fuel the development of next-generation antibiotics.

“Previous efforts to find new antibiotics have looked mostly at fungi, bacteria and animals,” says César de la Fuente, Presidential Associate Professor in Bioengineering and in Chemical and Biomolecular Engineering in the University of Pennsylvania School of Engineering and Applied Science (Penn Engineering), in Psychiatry and Microbiology in the Perelman School of Medicine and in Chemistry in the School of Arts & Sciences, and the paper’s senior author. 

In the past, de la Fuente’s lab has used AI models to identify antibiotic candidates in a range of unlikely sources, from the DNA of extinct organisms to the chemicals in animal venom. Now, they’re applying those tools to a new set of data: the proteins of hundreds of ancient microbes. “There’s a whole other domain of life waiting to be explored,” says de la Fuente. 

Exploring a Microbial Frontier 

Distinct from both bacteria and from eukaryotes (which include plants, animals and fungi), Archaea occupy their very own branch on the tree of life. 

Though they resemble bacteria under a microscope, Archaea fundamentally differ in their genetics, cell membranes and biochemistry. These differences allow them to survive in some of Earth’s most extreme environments, from superheated undersea vents to blistering hot springs like those in Yellowstone National Park.

Because Archaea often thrive where few other organisms can — enduring crushing pressures, toxic chemicals and extreme temperatures — their biology has evolved in unusual ways. That makes them a promising but largely untapped source of new molecular tools, including compounds that may function like antibiotics but operate differently from those currently in use.

“We were drawn to Archaea because they’ve had to evolve biochemical defenses in unusual environments,” says Marcelo Torres, a research associate in de la Fuente’s lab and the paper’s co-first author. “We thought, if they’ve survived for billions of years under those conditions, maybe they’ve developed unique ways to fight off microbial competitors, and maybe we could learn from that.”

Hunting for Antibiotics with AI

To uncover potential antibiotic compounds hidden in Archaea, the researchers turned to artificial intelligence. The team leveraged an updated version of APEX, an AI tool that de la Fuente’s lab originally developed to identify antibiotic candidates in ancient biology, including in the proteins of extinct animals like the woolly mammoth.

Having seen thousands of peptides — short chains of amino acids — with known antimicrobial properties, APEX can predict the likelihood that a given sequence of amino acids will have similar effects. 

By retraining APEX 1.1 on thousands of additional peptides and information about bacteria that cause diseases in humans, the researchers prepared the tool to predict which peptides in Archaea might inhibit bacterial growth. 

Scanning 233 species of Archaea yielded more than 12,000 antibiotic candidates. The researchers dubbed these molecules “archaeasins,” which chemical analysis revealed differ from known antimicrobial peptides (AMPs), in particular in their distribution of electric charge. 

The researchers then selected 80 archaeasins to test against actual bacteria. “Trying to find new antibiotics one molecule at a time is like looking for needles in a haystack,” says Fangping Wan, a postdoctoral fellow in de la Fuente’s lab and the paper’s other co-first author. “AI speeds up the process by identifying where the needles are likely to be.” 

On Par with Existing Antibiotics

Antibiotics work in a number of ways. Some punch holes in bacterial membranes, while others shut down the organisms’ ability to make proteins. The researchers found that, unlike most known AMPs, which attack a bacterium’s outer defenses, archaeasins seem to pull the plug from the inside, scrambling the electrical signals that keep the cell alive.

In tests against a range of disease-causing, drug-resistant bacteria, 93% of the 80 archaeasins surveyed demonstrated antimicrobial activity against at least one bacterium. The researchers then selected three archaeasins to test in animal models. 

Four days after a single dose, the archaeasins all arrested the spread of a drug-resistant bacterium often acquired in hospitals. One of the three compounds demonstrated activity comparable to polymyxin B, an antibiotic commonly used as a last-line of defense against drug-resistant infections. 

“This research shows that there are potentially many antibiotics waiting to be discovered in Archaea,” says de La Fuente. “With more and more bacteria developing resistance to existing antibiotics, it’s critical to find new antibiotics in unconventional places to replace them.” 

Future Studies

Next, the researchers plan to further enhance APEX so that it can predict antibiotic candidates based on their structure, potentially improving the tool’s accuracy. The researchers also hope to better understand the long-term efficacy and safety of archaeasins, with the goal of one day bringing them to human clinical trials. 

“This is only the beginning,” says de la Fuente. “Archaea is one of the oldest forms of life, and clearly has much to teach us about how to outsmart the pathogens we face today.”  

César de la Fuente-Nunez holds a Presidential Professorship at the University of Pennsylvania and acknowledges funding from the Procter & Gamble Company, United Therapeutics, a BBRF Young Investigator Grant, the Nemirovsky Prize, Penn Health-Tech Accelerator Award, and the Dean’s Innovation Fund from the Perelman School of Medicine at the University of Pennsylvania. 

Research reported in this press release was supported by the Langer Prize (AIChE Foundation), the National Institute of General Medical Sciences of the National Institutes of Health (R35GM138201), and the Defense Threat Reduction Agency (DTRA; HDTRA11810041, HDTRA1-21-1-0014, and HDTRA1-23-1-0001). 

END