Scientists at the University of Pennsylvania Machine Biology Group have developed what they call a groundbreaking approach to drug discovery, which uses artificial intelligence to discover antibiotics in extinct organisms. In a newly published study in Cell Host and Microbe, the team described the use of the “molecular de-extinction” technology to discover antimicrobial peptides (AMPs) in our closest hominid relatives, the Neanderthals and Denisovans. Initial tests showed that the newly discovered archaic peptides encrypted in these extinct human proteins displayed anti-infective activity against bacterial infections in different preclinical in vivo models. The achievement could mark the start of a new chapter in the search for antibiotics and other valuable biomolecules, allowing scientists to harness AI and systematically explore long extinct organisms to help us better understand life’s molecular diversity and sequence space.

Senior and corresponding author César de la Fuente-Nunez, PhD, and colleagues reported on their study in a paper titled, “Molecular de-extinction of antimicrobial peptides enabled by machine learning.” In their paper, the team concluded, “These results suggest that machine-learning-based encrypted peptide (EP) prospection can identify stable, nontoxic AMPs …  we establish molecular de-extinction through paleoproteome mining as a framework for antibacterial drug discovery.”

De-extinction refers to the process of resurrecting extinct species, and the focus has primarily been on reviving entire organisms. “The idea of reintroducing extinct organisms into extant environments has captured the public and scientific imagination,” the authors wrote, but this concept does raise “profound ethical and ecological questions.” In contrast, molecular de-extinction aims to resurrect extinct molecules—nucleic acids, proteins, and other compounds no longer encoded by living organisms—rather than the complete organisms, as a means of addressing contemporary challenges. “By synthesizing only isolated compounds, molecular de-extinction circumvents many of the ethical and technical problems posed by whole-organism de-extinction,” the investigators continued. “Such molecules could be of biomedical or economic utility by bolstering defenses against future challenges that resemble stressors from environments past, including climate change or infectious disease outbreaks.”

Technically, molecular de-extinction offers a more achievable and controllable process compared with the resurrection of entire organisms. This approach harnesses the latest capabilities in machine learning, synthetic biology, and chemistry to discover, synthesize, and test extinct molecules in a laboratory setting. Scientists then have the opportunity to tap into previously unexplored molecular sequence space, and gain insights into the evolutionary history and potential functionalities of these molecules without the need for challenging de-extinction procedures.

For their newly reported study, the researchers used the AI-driven molecular de-extinction strategy to prospect for antimicrobial peptides—encrypted peptides—hidden within extinct and extant human proteins. To achieve this, they harnessed the panCleave machine learning model, designed for proteome-wide cleavage site prediction. “This open-source machine learning (ML) tool leverages a pan-protease cleavage site classifier to perform computational proteolysis: the in silico digestion of human proteins into peptide fragments,” they explained. Remarkably, the scientists found that this machine learning model outperformed several protease-specific cleavage site classifiers for three modern human caspases, despite its pan-protease design.

In vitro experiments revealed the antimicrobial activity of both modern EPs (MEPs) and archaic protein fragments identified using the machine learning approach. The team further evaluated lead peptides to understand their mechanism of action, resistance to proteolysis, and efficacy as anti-infective agents in two preclinical mouse models. “Antimicrobial activity was observed in vitro for modern and archaic protein fragments identified with panCleave. Lead peptides showed resistance to proteolysis and exhibited variable membrane permeabilization,” they wrote.

Encouragingly, some of these peptides exhibited stability, nontoxicity, and potent antimicrobial properties, with initial in vivo tests confirming antimicrobial activity. “… representative modern and archaic protein fragments showed anti-infective efficacy against A. baumannii in both a skin abscess infection model and a preclinical murine thigh infection model,” the team stated.

Of note, their approach identified peptides encrypted within known precursor protein groups, and also rediscovered a known antimicrobial EP. “In addition to discovering antimicrobial EPs, the panCleave pipeline unintentionally uncovered an MEP containing a known bioactive subsequence,” the scientists further commented. “The unintentional rediscovery of this antimicrobial motif in lysozyme C supports the use of the present pipeline for antimicrobial EP discovery. Similarly, all antimicrobial EPs discovered in this work originate from proteins belonging to groups previously described in the EP literature.”

Today’s achievement serves as a testament to the extraordinary potential of AI in the field of molecular de-extinction and, more specifically, antibiotic discovery. “… this work lends further support for computational EP prospection in the modern human proteome, which was previously proposed as an antibiotic discovery framework,” the investigators concluded.

“This study highlights the power of machine learning to realize the concept of molecular de-extinction and to accelerate scientific discovery,” added de la Fuente-Nunez. “Our findings suggest that molecular de-extinction holds tremendous potential as a framework for antibacterial drug discovery.”

Noting limitations of their study, the authors also further suggested, “Future work might consider whether prospection within modern and archaic humans might minimize pharmacological risks such as toxicity, relative to mining evolutionarily distant or synthetic protein spaces.”

The researchers acknowledge the importance of ethical considerations and legal frameworks, and say conversations with bioethicists and patent lawyers have already begun. These will address key questions. For example, what does it mean to resurrect molecules that are no longer expressed in living organisms? Are de-extinct molecules eligible for patents in this new frontier of patent law? Natural molecules are not patentable, but the emergence of de-extinct molecules presents a unique challenge to existing patent law. The patentability of extinct molecules presents a novel and thought-provoking challenge within the fields of de-extinction, biotechnology, and medicine.

The potential of this AI-driven strategy extends also beyond antibiotic discovery. By exploring ancient organisms, it is possible to focus on previously uncharted molecular sequence space, uncovering new horizons for potential therapeutic agents that hold promise in combating diseases. The use of AI could also help to shed light on the significance of the peptide sequences uncovered, and unravel their role in immunity throughout evolutionary history. Resurrecting these ancient molecules may also address present-day problems, offering innovative solutions to the challenges faced.

Previous articleNew Tool Combines scRNA-Seq and AI to Uncover Novel Neuronal Cell Types
Next articleHow Cancer Derails a Stem Cell’s Fate