petri dish with bacteriaShare on Pinterest
Researchers are testing new, AI-led ways of finding more potent antibiotics. Image credit: Gennaro Leonardi/EyeEm/Getty Images.
  • Pathogens are getting better and better at resisting current antibiotics, a medical crisis in the making.
  • There is reason to believe that many bacteria contain natural, powerful antibiotics yet to be found.
  • A paper describes the discovery of one such antibiotic, found through the use of bioinformatic algorithms that can predict the products of silent biosynthetic gene clusters.

Dr. Cesar de la Fuente-Nunez, of the Perelman School of Medicine at the University of Pennsylvania in Philadelphia, described the problem to Medical News Today.

“Many antibiotics no longer work. The current prediction is that, by 2050, 10 million people will die annually in the world from untreatable infections. This corresponds to one death every three seconds,” he noted.

“In other words,” said Dr. de la Fuente-Nunez, “we really need novel approaches to treat drug-resistant bacteria.”

Researchers from The Rockefeller Institute in Philadelphia, PA, have just published a new paper that presents one such approach.

It describes the use of bioinformatic algorithms to discover existing natural antibiotic agents “hiding” within bacteria that can overcome drug resistance.

The paper introduces cilagicin, a new anti-drug-resistant antibiotic, discovered using the new process.

Cilagicin protected mice threatened by acute infection, and exhibited broad, potent, antimicrobial activity against several drug-resistant pathogens.

The study, whose lead author is Dr. Zongqiang Wang, appears in Science.

Dr. de la Fuente-Nunez was not involved in this study.

Corresponding author of the study, Dr. Sean F. Brady, told MNT that “[m]any of our most useful therapeutics come from bacteria.”

“Traditional methods for identifying antibiotics — and other natural therapeutics — rely on biological processes, i.e, fermentation, to convert the genetic instructions contained in bacterial genomes into antibiotics,” said Dr. Brady.

“Unfortunately, it is often difficult to coax laboratory-grown bacteria into producing all of the different antibiotics they are capable of making,” he pointed out.

Dr. Brady noted: “Historically, about 10% of bacterial fermentation broth extracts showed antibacterial activity. It is now clear that even very well-studied bacteria can contain a large number of silent biosynthetic gene clusters (BCGs).”

There is no way of knowing, admitted Dr. Brady, whether the products of these BCGs will turn out to be as useful as those that have been easily expressed and identified.

Still, said Dr. de la Fuente-Nunez, “[o]ne way of thinking about this is by teaching computers to design and discover novel antibiotics, which is the underlying concept of the beautiful Wang et al. paper.”

Dr. Brady explained: “We therefore developed a ‘biology free’ discovery approach where, instead of decoding genetic instructions using natural biological processes, bioinformatic algorithms are used to predict the chemical structures produced by bacteria, and then chemical synthesis is used to build these potential antibiotics.”

The molecules from which these antibiotics are derived are called “synthetic bioinformatic natural products (syn-BNPs).

“We are just scratching the surface, but there are exciting biosynthetic gene clusters in numerous bacteria that can potentially encode for novel drugs,” believes Dr. de la Fuente-Nunez. “Outside-the-box approaches are urgently needed, and this work and this field of research is a great example of how to think about the antibiotic discovery problem differently.”

Researchers from Dr. Brady’s lab searched a database of about 10,000 BCGs for genes that might encode nonribosomal peptidesynthetase-encoded lipopeptide antibiotics. These lipopeptides have a history of inhibiting bacterial growth through a variety of modes of action.

Many BCGs have not been previously explored. One, which the researchers named the “cil” cluster, caught their attention due to the close common ancestors it shares with other genes associated with antibiotics.

The researchers fed it into an algorithm that predicted the BCG would produce several compounds, including one, cilagicin, that was an active antibiotic.

Cilagicin was pitted against, and proved potent against, several known drug-resistant bacteria, including those grown specifically to resist cilagicin.

They discovered that cilagicin also caused no harm to human cells, and once converted into a bioavailable drug form, fought off infections in mice.

Cilagicin is so effective at defeating drug-resistant bacteria, say the researchers, because of two molecules upon which bacteria depend for the maintenance of their cell walls.

The molecules are known as C55-P and C55-PP, and most antibiotics bind with one or the other, rendering them prone to developing resistance. Drug-resistant bacteria can make do with their one remaining molecule. Since cilagicin binds with both, bacteria have no workaround and are defeated.

The researchers hope the process put forward in the paper can provide one way out of our drug-resistance crisis. Dr. Brady said:

“The remaining useful time of our current antibiotic arsenal will completely depend on how carefully we use it. With good stewardship, I am very hopeful that our current antibiotics can last long enough to allow for the development of the next generation of antibiotics that scientists are working on today.”

The paper’s approach is welcomed by Dr. de la Fuente-Nunez, who said: “I believe in the potential of AI and computers to help us design and discover novel antibiotics. I think we will need to merge machine intelligence with human intelligence to make this happen.”