Humans and bacteria are in a constant war.
For most of history, bacteria won. Before 1928, you could die from an infection simply by scraping your knee, getting a cut while cooking dinner, or giving birth.
The balance changed with the discovery of penicillin, a molecule secreted by fungi. For the first time, humans had a way to fight back. Since then, several generations of antibiotics have targeted different stages of bacterial growth, effectively eliminating them before they can spread through the body and infect others.
But bacteria have an evolutionary advantage. Their DNA readily adapts to evolutionary pressures, including antibiotics, allowing them to mutate over generations to evade drugs. They also have a sort of “telephone wire” that transmits their adapted DNA to other nearby bacteria, giving them the power to resist even antibiotics. Rinse and repeat. Soon the entire bacterial population will have the ability to fight back.
We may be slowly losing the war. Antibiotic resistance is now a public health threat, causing an estimated 1.27 million deaths worldwide in 2019. The World Health Organization (WHO) and other organizations say that without the next generation of antibiotics, surgery, cancer chemotherapy and other life-saving treatments will become increasingly common. Risk of death from infection.
Traditionally, it takes about 10 years to develop a new antibiotic, test it, and finally reach patients.
“New methods for antibiotic discovery are urgently needed,” Luis Pedro Coelho, Ph.D., a computational biologist and author of a new study on the topic, said in a press release.
Coelho and the team leveraged AI to speed up the entire process. They analyzed a huge database of genetic material from the environment and discovered nearly a million potential antibiotics.
The research team synthesized 100 types of antibiotics discovered by AI in the laboratory. Tested against bacteria known to be resistant to current drugs, 63 were found to easily fight infections inside a test tube. One was particularly effective in a mouse model of skin disease, destroying bacterial infection and allowing the skin to heal.
“AI in antibiotic discovery is now a reality and has significantly accelerated our ability to discover new drug candidates. What once took years can now be done in hours using computers,” study co-author César de la Fuente, MD, of Penn Medicine, said in another press release.
the enemy of antibiotics
It’s easy to take antibiotics for granted. Let’s say you develop an ear infection from always wearing wireless earbuds. After receiving the prescription and applying the medicine, everything goes smoothly.
Or is it? Over time, the drops may have a harder time preventing infection. This “antibiotic resistance” is central to the evolutionary battle between bacteria and humans.
Antibiotics generally work by preventing bacteria from replicating in several ways. Like human cells, bacterial cells have a cell wall, a wrapping paper that keeps DNA and other biological components inside. One type of antibiotic destroys the walls, preventing pathogens from spreading. Others target genetic material or inhibit metabolic pathways that bacteria need to survive.
All of these strategies required decades of research to discover and advance medicine. However, microorganisms mutate rapidly. For example, some bacteria create “pumps” on their surface that literally expel the drug. Others evolve enzymes that block antibiotics by neutralizing their effects by slightly altering protein target sites through DNA mutations.
Each strategy on its own is difficult to develop. But bacteria have another trick up their sleeve. This is horizontal movement. Here, antibiotic resistance genes are encoded as small circular pieces of DNA that can travel through biological “highways” (physical tubes) to neighboring cells, giving recipients antibiotic-like abilities.
Finding ways to kill invading bacteria is difficult. As bacteria evolve to evade their targets, other antibiotics that are chemically similar to the antibiotic quickly lose their effectiveness. So is there a way to find antibiotics that neither bacteria nor nature itself has ever seen before?
AI solutions
AI is beginning to revolutionize biology. From protein structure prediction to antibody design, these algorithms are addressing some of humanity’s most serious health disorders.
Traditionally, finding antibiotics has been mostly trial and error, with scientists often scraping samples from exotic moss or other sources that could potentially fight infections.
In the new study, the team aimed to find new versions of antibiotics based on antimicrobial peptides (AMPs). Similar to proteins, they are made up of strings of relatively short molecules called amino acids. These peptides are found throughout the living world and can disrupt microbial growth by destroying cell walls and “exploding” bacteria. They have already been used clinically as antibacterial agents and are currently in clinical trials for yeast infections. However, like any other antibacterial agent, there is a risk of resistance.
As the discovery of penicillin suggested nearly 100 years ago, the natural world is a rich source of potential antibiotics. In this study, the team used machine learning to search for antimicrobial peptides with possible antibiotic properties in more than 63,000 publicly available metagenomes (genetic information isolated from multiple organisms in the environment) and nearly 88,000 high-quality microbial genomes. Sources come from all over the world, in the seas and on land, and also include the gut microbiota of humans and animals. These data have been merged into a public AMPSphere database for anyone to explore.
de la Fuente said this resource will allow scientists to mine “the entire microbial diversity we have on Earth, or a huge representation of it,” and find nearly a million new molecules encoded or hidden within all microbial dark matter. tutelar.
To test their findings, the team selected 100 candidates and synthesized them in the lab. In test tubes, 79 destroyed cell membranes, and 63 completely killed at least one of the dangerous bugs.
“In some cases, these molecules were effective against bacteria even at very low doses,” de la Fuente said.
Next, the team developed an antibiotic peptide from their database to address the dangerous bugs that cause skin lesions in mice. After just one injection, the drug discovered by AI inhibited bacterial growth, and body weight measurements showed that the mice suffered no side effects.
“We were able to accelerate antibiotic discovery.” de la Fuente said. tutelar. “So instead of having to wait five or six years to find one candidate, you can now find hundreds of thousands of candidates in just a few hours on your computer.”
Image courtesy of antibiotic-resistant Staphylococcus aureus (yellow) and dead white blood cells (red). National Institute of Allergy and Infectious Diseases (NIAID)/NIH
