AI discovers antibiotics in archaea, the microorganisms that explain the origin of complex life

Archaea are single-celled microorganisms that make up one of the three domains of life (evolutionary categories into which all living beings are classified), along with bacteria and eukaryotes, the group that includes humans, animals, and plants. They are members of the original stages of evolution on early Earth and, in that hostile environment, learned to live in extreme temperatures (more than 80 degrees in geysers), high salinity, acidity, or alkalinity, and high pressures (they are present in the depths of the oceans). They have also managed to outlast their biological neighbors, bacteria, with whom they compete for space and resources. This led the laboratory of Spanish biotechnologist César de la Fuente at the University of Pennsylvania to believe that, under these conditions, they would need defense mechanisms that could open the door to new antibiotics in response to the resistance microorganisms develop to existing drugs. Using artificial intelligence (AI) and computational deep learning, the team found antimicrobial agents, which they called "archaeains," in 93% of the 80 compounds identified by the AI. Archeasin-73 performed in vivo on a par with polymyxin B, a last-resort antibiotic, according to a publication in Nature Microbiology on Tuesday.

Existing antibiotics arise from chemical weapons developed by microbes to defend themselves against other species and have been sought in all kinds of environments, including extinct animals and humans. But, apart from very limited research, such as that published in The Microbe on bacterial and archaeal communities in the Roman baths of the British city of Bath, no investigations had been made into the more than 20,000 species of these resistant organisms.

Exploring any of these domains is essential in the face of the rise in drug-resistant and life-threatening infections , considered by the World Health Organization to be one of humanity's greatest threats. In 2019, bacterial antibiotic resistance was associated with 4.95 million deaths worldwide, and if alternatives are not found, the number will double in the next two decades.

"Since the discovery of penicillin, the search for new antibiotics has focused almost exclusively on bacteria and fungi. With our work, this paradigm changes because we discover antibiotics in a virtually unexplored domain of life," emphasizes the University of Pennsylvania scientist.

In this way, De la Fuente's research with archaea opens up an important source of future treatments through a technique that avoids spending decades discarding and identifying compounds with anti-infective capabilities. "Artificial intelligence can reveal new antibiotics from unexpected biological sources. Combining algorithms with rapid experimental tests allows us to accelerate discovery at digital speed," explains the Galician biotechnologist.

“Our study,” insists biochemist Marcelo Torres, co-author of the study, “reveals that archaea, a domain of life yet to be explored, harbor a vast reservoir of antimicrobial molecules with the potential to combat antibiotic resistance.”

For this work, the team relied on the detailed existing information on archaea and used an artificial intelligence program (ApexOracle) , an improved model compared to its previous versions and specifically trained for this task of delving into the archaeome. "We explored a virtually unexplored domain of life and discovered a new gold mine of antibiotics. From a biological perspective, we position archaea, along with bacteria and fungi, as a rich source of useful molecules," emphasizes De la Fuente.

The system has improved since the lab began using AI and deep learning to identify compounds. In the initial work, results were obtained for just over 60% of the computer's proposals, which was considered a success. With the reprogramming and refinement of the model, this percentage has increased by 30 points. "This tells us that the more experimental data we use to train the model, the better the result," explains the researcher.

The combination of computer tools with chemistry is a growing field. A study led by Younes Smani, a researcher at the Andalusian Center for Developmental Biology and a professor in the Department of Microbiology at Pablo de Olavide University, has found the basis for developing a new series of potential antibiotics in tamoxifen (a common cancer treatment) and raloxifene, a related compound.

In addition to identifying new antibacterial molecules, research is also advancing new drug delivery formulas to increase their effectiveness. This is the proposal of a research team from the University of Huelva , the University of Seville , and the Virgen Macarena University Hospital , which has used carbon nanotubes, a million times thinner than a hair, to enable the drug to act more precisely on the focus of the infection and increase the time during which it is effective. This more efficient delivery of the active ingredient is another strategy to combat antibiotic resistance, reports the Descubre Foundation, based on research published in the Journal of Drug Delivery Science and Technology .

Antibiotic resistance (AMR) rates, according to a Médecins Sans Frontières report , “are alarmingly high in settings experiencing conflict, population displacement, climate-related disasters, or where the healthcare system is fragile.” The NGO explains that “the limited availability of essential antibiotics, their frequent shortages, and the recurrent infections that often occur in these settings, among other factors, can lead to inappropriate use of antibiotics by patients, which increases AMR.”