Antibiotic resistance is a public health problem and ‘resurrecting’ molecules using cutting-edge technologies has become a promising avenue. Now, a new study has shown that deep learning can be used to extract proteins from all extinct organisms, including mammoths. The results were published this Tuesday in the journal Nature Biomedical Engineering, in an article led by the Spaniard César de la Fuente, from the University of Pennsylvania, in the United States.
Molecular de-extinction aims to resurrect molecules to solve antibiotic resistance and other current biological and biomedical problems. Thanks to machine learning, De la Fuente’s laboratory has discovered the first therapeutic molecules in extinct organisms, launching – according to the group on its website – the field of molecular de-extinction.
“We believe that resurrecting molecules from the past can help solve current problems, such as antibiotic resistance. In fact, resurrection biology is an emerging field that aims to bring chains of molecules and more complex organisms back to life with the ultimate goal to benefit humanity”.
In the new article, the team from the University of Pennsylvania shows that deep learning can be used to extract the proteomes – a group of proteins made by an organism – of all extinct organisms available for the discovery of antibiotic peptides. To do this, as the scientist explains, and to accelerate the discovery, they developed a new AI model called APEX, thanks to which numerous antibiotic compounds have been “successfully” discovered in creatures of the past, such as the woolly mammoth.
This model is the culmination of several years of work, based on decades of previous research on sequencing methods for ancient genetic material. And his laboratory had previously explored the proteomes of our closest relatives, the Neanderthals and the Denisovans. “We identified antibiotic molecules in modern and ancient humans, including Neanderthals. This led us to hypothesize that similar molecules could be conserved throughout evolution,” she says.
To explore this hypothesis, the researchers had to go from extracting several proteomes to hundreds at a time, and to do so they had to develop a more powerful method. This is how APEX was born, “our new artificial intelligence model designed to analyze all extinct organisms known to science: the ‘extinctoma’,” detailed in a thread on X César de la Fuente.
The researchers trained sets of deep learning models for antimicrobial activity prediction and used it to extract 10,311,899 peptides. The models predicted 37,176 sequences with broad-spectrum antimicrobial activity, 11,035 of which were not found in existing organisms.
Experiments in mice
The team synthesized 69 peptides – a short chain of amino acids – and experimentally confirmed their activity against bacterial pathogens. Most of the peptides killed the bacteria by ‘depolarizing’ their cytoplasmic membrane, contrary to known antimicrobial peptides, which tend to target the outer membrane, the authors write in their paper.
In particular, the main compounds (including mammothusin-2 from the woolly mammoth; elephasin-2 from the straight-tusked elephant; hydrodamine-1 from the ancient sea cow; mylodonin-2 from the giant sloth; and megalocerin-1 from the extinct giant moose) showed anti-infective activity in mice with skin abscesses or thigh infections.
Many of these compounds were effective both in vitro and in two preclinical mouse models, with an activity comparable to that of the antibiotic polymyxin B,” says the researcher, who last week published in the journal Cell, together with scientists from the University Technological Institute of Queensland (Australia), research that identified, with the help of AI, almost a million potential sources of antibiotics in nature.
Molecular decay aided by deep learning can accelerate the discovery of therapeutic molecules, the authors conclude in their Nature Biomedical Engineering article. “The molecules discovered by APEX (…) are now candidates for preclinical antibiotics. Our work with AI has dramatically accelerated the discovery of antibiotics. years of work can now be done in hours!” says De la Fuente online social.