Penn Engineering and Stanford researchers leveraged AI to discover dozens of potential new antibiotics in the human gut microbiome. (ChrisChrisW via Getty Images)
The average human gut contains roughly 100 trillion microbes, many of which are constantly competing for limited resources. “It’s such a harsh environment,” says César de la Fuente, Presidential Assistant Professor in Bioengineering and in Chemical and Biomolecular Engineering within the School of Engineering and Applied Science, in Psychiatry and Microbiology within the Perelman School of Medicine, and in Chemistry within the School of Arts & Sciences. “You have all these bacteria coexisting, but also fighting each other. Such an environment may foster innovation.”
In that conflict, de la Fuente’s lab sees potential for new antibiotics, which may one day contribute to humanity’s own defensive stockpile against drug-resistant bacteria. After all, if the bacteria in the human gut have to develop new tools in the fight against one another to survive, why not use their own weapons against them?
In a new paper in Cell, the labs of de la Fuente and Ami S. Bhatt, Professor in Medicine (Hematology) and Genetics at Stanford, surveyed the gut microbiomes of nearly 2,000 people, discovering dozens of potential new antibiotics. “We think of biology as an information source,” says de la Fuente. “Everything is just code. And if we can come up with algorithms that can sort through that code, we can dramatically accelerate antibiotic discovery.”
Cesar de la Fuente (left), Fangping Wan (center), and Marcelo der Torossian Torres (right). Fangping holds a 3D model of a unique ATP synthase fragment, identified by their lab’s deep learning model, APEX, as having potent antibiotic properties.
“Make sure you finish your antibiotics course, even if you start feeling better’ is a medical mantra many hear but ignore,” says Cesar de la Fuente of the University of Pennsylvania.
He explains that this phrase is, however, crucial as noncompliance could hamper the efficacy of a key 20th century discovery, antibiotics. “And in recent decades, this has led to the rise of drug-resistant bacteria, a growing global health crisis causing approximately 4.95 million deaths per year and threatens to make even common infections deadly,” he says.
De la Fuente, a Presidential Assistant Professor, and a team of interdisciplinary researchers have been working on biomedical innovations tackling this looming threat. In a new study, published in Nature Biomedical Engineering, they developed an artificial intelligence tool to mine the vast and largely unexplored biological data—more than 10 million molecules of both modern and extinct organisms— to discover new candidates for antibiotics.
“With traditional methods, it takes around six years to develop new preclinical drug candidates to treat infections and the process is incredibly painstaking and expensive,” de la Fuente says. “Our deep learning approach can dramatically reduce that time, driving down costs as we identified thousands of candidates in just a few hours, and many of them have preclinical potential, as tested in our animal models, signaling a new era in antibiotic discovery.” César de la Fuente holds a 3D model of a unique ATP synthase fragment, identified by his lab’s deep learning model, APEX, as having potent antibiotic properties. This molecular structure, resurrected from ancient genetic data, represents a promising lead in the fight against antibiotic-resistant bacteria.
These latest findings build on methods de la Fuente has been working on since his arrival at Penn in 2019. The team asked a fundamental question: Can machines be used to accelerate antibiotic discovery by mining the world’s biological information? He explains that this idea is based on the notion that biology, at its most basic level, is an information source, which could theoretically be explored with AI to find new useful molecules.
Almost a century ago, the discovery of antibiotics like penicillin revolutionized medicine by harnessing the natural bacteria-killing abilities of microbes. Today, a new study co-led by researchers at the Perelman School of Medicine at the University of Pennsylvania suggests that natural-product antibiotic discovery is about to accelerate into a new era, powered by artificial intelligence (AI).
The study, published in Cell, the researchers used a form of AI called machine learning to search for antibiotics in a vast dataset containing the recorded genomes of tens of thousands of bacteria and other primitive organisms. This unprecedented effort yielded nearly one million potential antibiotic compounds, with dozens showing promising activity in initial tests against disease-causing bacteria.
“AI in antibiotic discovery is now a reality and has significantly accelerated our ability to discover new candidate drugs. What once took years can now be achieved in hours using computers” said study co-senior author César de la Fuente, PhD, a Presidential Assistant Professor in Psychiatry, Microbiology, Chemistry, Chemical and Biomolecular Engineering, and Bioengineering.
Nature has always been a good place to look for new medicines, especially antibiotics. Bacteria, ubiquitous on our planet, have evolved numerous antibacterial defenses, often in the form of short proteins (“peptides”) that can disrupt bacterial cell membranes and other critical structures. While the discovery of penicillin and other natural-product-derived antibiotics revolutionized medicine, the growing threat of antibiotic resistance has underscored the urgent need for new antimicrobial compounds.
In recent years, de la Fuente and colleagues have pioneered AI-powered searches for antimicrobials. They have identified preclinical candidates in the genomes of contemporary humans, extinct Neanderthals and Denisovans, woolly mammoths, and hundreds of other organisms. One of the lab’s primary goals is to mine the world’s biological information for useful molecules, including antibiotics.
The growing threat of antimicrobial resistance demands innovative solutions in drug discovery. Scientists are turning to artificial intelligence (AI) and machine learning (ML) to accelerate the discovery and development of antimicrobial peptides (AMPs). These short strings of amino acids are promising for combating bacterial infections, yet transitioning them into clinical use has been challenging. Leveraging novel AI-driven models, researchers aim to overcome these obstacles, heralding a new era in antimicrobial therapy.
A new article in Nature Reviews Bioengineering illuminates the promises and challenges of using AI for antibiotic discovery. Cesar de la Fuente, Presidential Assistant Professor in Microbiology and Psychiatry in the Perelman School of Medicine, in Bioengineering and Chemical and Biomolecular Engineering in the School of Engineering and Applied Science, and Adjunct Assistant Professor in Chemistry in the School of Arts and Sciences, collaborated with James J. Collins, Termeer Professor of Medical Engineering and Science at MIT, to provide an introduction to this emerging field, outlining both its current limitations and its massive potential.
In the past five years, groundbreaking work in the de la Fuente Lab has dramatically accelerated the discovery of new antibiotics, reducing the timeline from years to mere hours. AI-driven approaches employed in his laboratory have already yielded numerous preclinical candidates, showcasing the transformative potential of AI in antimicrobial research and offering new potential solutions against currently untreatable infections.
Recent advancements in AI and ML are revolutionizing drug discovery by enabling the precise prediction of biomolecular properties and structures. By training ML models on high-quality datasets, researchers can accurately forecast the efficacy, toxicity and other crucial attributes of novel peptides. This predictive power expedites the screening process, identifying promising candidates for further evaluation in a fraction of the time required by conventional methods.
Traditional approaches to AMP development have encountered hurdles such as toxicity and poor stability. AI models help overcome these challenges by designing peptides with enhanced properties, improving stability, efficacy and safety profiles, and fast-tracking the peptides’ clinical application.
While AI-driven drug discovery has made significant strides, challenges remain. The availability of high-quality data is a critical bottleneck, necessitating collaborative efforts to curate comprehensive datasets to train ML models. Furthermore, ensuring the interpretability and transparency of AI-generated results is essential for fostering trust and wider adoption in clinical settings. However, the future is promising, with AI set to revolutionize antimicrobial therapy development and address drug resistance.
Integrating AI and ML into antimicrobial peptide development marks a paradigm shift in drug discovery. By harnessing these cutting-edge technologies, researchers can address longstanding challenges and accelerate the discovery of novel antimicrobial therapies. Continuous innovation in AI-driven approaches is likely to spearhead a new era of precision medicine, augmenting our arsenal against infectious diseases.
The de la Fuente Lab uses use the power of machines to accelerate discoveries in biology and medicine. The lab’s current projects include using AI for antibiotic discovery, molecular de-extinction, reprogramming venom-derived peptides to discover new antibiotics, and developing low-cost diagnostics for bacterial and viral infections. Read more posts featuring de la Fuente’s work in the BE Blog.
Susan Davidson, Cesar de la Fuente, Surbhi Goel and Chris Callison-Burch speak on AI in Engineering in episode 4 of the Innovation & Impact podcast.
With AI technologies finding their way into every industry, important questions must be considered by the research community: How can deep learning help identify new drugs? How can large language models disseminate information? Where and how are researchers using AI in their own work? And, how are humans anticipating and defending against potential harmful consequences of this powerful technology?
In this episode of Innovation & Impact, host Susan Davidson, Weiss Professor in Computer and Information Science (CIS), speaks with three Penn Engineering experts about leveraging AI to advance scientific discovery and methods to protect its users. Panelists include:
Chris Callison-Burch, Associate Professor in CIS, who researches the applications of large language models and AI tools in current and future real-world problems with a keen eye towards safety and ethical use of AI;
Surbhi Goel, Magerman Term Assistant Professor in CIS, who works at the intersection of theoretical computer science and machine learning. Her focus on developing theoretical foundations for modern machine learning paradigms expands the possibilities of deep learning; and
Cesar de la Fuente, Presidential Assistant Professor in Bioengineering, Psychiatry and Microbiology with a secondary appointment in Chemical and Biomolecular Engineering, who leads research on technology in the medical field, using computers to find antibiotics in extinct organisms and identify pre-clinical candidates to advance drug discovery.
Each episode of Penn Engineering’s Innovation & Impact podcast shares insight from leading experts at Penn and Penn Engineering on science, technology and medicine.
NAM, founded in 1970, is an independent organization of professionals that advises the entire scientific community on critical health care issues. Each year, NAM chooses up to 10 new ELHM Scholars who are early-to-mid-career professionals from a wide range of health-related fields, including biomedical engineering, internal medicine, psychiatry, radiology and journalism to serve a three-year term.
“We are delighted that Dr. de la Fuente is receiving recognition from the National Academy of Medicine for his breakthrough contributions and exceptional leadership in the life sciences,” says Vijay Kumar, Nemirovsky Family Dean of Penn Engineering. “His pioneering work using computers to accelerate antibiotic discovery is extraordinary. We proudly celebrate his selection as part of this outstanding group of scholars.”
In an era peppered by breathless discussions about artificial intelligence—pro and con—it makes sense to feel uncertain, or at least want to slow down and get a better grasp of where this is all headed. Trusting machines to do things typically reserved for humans is a little fantastical, historically reserved for science fiction rather than science.
Not so much for César de la Fuente, PhD, the Presidential Assistant Professor in Psychiatry, Microbiology, Chemical and Biomolecular Engineering, and Bioengineering in Penn’s Perelman School of Medicine and School of Engineering and Applied Science. Driven by his transdisciplinary background, de la Fuente leads the Machine Biology Group at Penn: aimed at harnessing machines to drive biological and medical advances.
“Biology is complexity, right? You need chemistry, you need mathematics, physics and computer science, and principles and concepts from all these different areas, to try to begin to understand the complexity of biology,” he said. “That’s how I became a scientist.”
Perelman School of Medicine’s Maayan Levy, and Christoph Thaiss. (Image: Courtesy of Penn Medicine News)
When we hear about gut bacteria, we may think about probiotics and supplements marketed to help with digestion, about how taking antibiotics might affect our intestinal tract, or perhaps about trendy diets that aim to improve gut health.
But two researchers at Penn Medicine think that understanding the microbiome, the entirety of microbial organisms associated with the human body, might be the key to deciphering the fundamental mechanisms that make our bodies work. They think these microbes may work like a call center switchboard, making connections to help different organs, biological systems, and the brain communicate. Maayan Levy, and Christoph Thaiss, both assistant professors of microbiology at the Perelman School of Medicine, argue that the microbiome is instrumental to revealing how signals from the gastrointestinal tract are received by the rest of the body—which may hold the key to understanding inter-organ communication in general. Perelman School of Medicine’s Maayan Levy, and Christoph Thaiss. (Image: Courtesy of Penn Medicine News)
While the gut sends signals to all parts of the body to initiate various biological processes, the mechanisms underlying this communication—and communication between different organs involved in these processes—is relatively unknown.
“The more we learn about the role the microbiome plays in a wide range of diseases— from cancer to neurodegenerative diseases to inflammatory diseases—the more important it becomes to understand what exactly its role is,” says Thaiss. “And hopefully once we understand how it works, we can use the microbiome to treat these diseases.”
Levy and Thaiss joined the faculty at Penn Medicine after completing their graduate studies in 2018. Here, they continue to investigate the role of the microbiome in various biological processes.
In his lab, Thaiss focuses on the impact of the microbiome on the brain. He recently identified species of gut-dwelling bacteria that activate nerves in the gut to promote the desire to exercise. Most recently, Thaiss published a study that identified the cells that communicate psychological stress signals from the brain to the gastrointestinal tract, and cause symptoms of inflammatory bowel disease.
Meanwhile, in her lab, Levy examines how the microbiome influences the development of diseases, like cancer, and other conditions throughout the body.
A recent publication authored by Levy suggested that the ketogenic diet (high fat, low carbohydrate) causes the production of a metabolite called beta-hydroxybutyrate (BHB), that suppresses colorectal cancer in small animal models.
Now, Levy is collaborating with Bryson Katona, an assistant professor of Medicine in the division of gastroenterology who specializes in gastrointestinal cancers, to investigate whether BHB has the same effect in patients with Lynch syndrome, which causes individuals to have a genetic predisposition to many different kinds of cancer, including colon cancer. These efforts are part of a growing emphasis at Penn on finding methods to intercept cancer in its earliest stages.
“It’s remarkable that we were able to quickly take the findings from our animal models and rapidly design a clinical trial,” Levy says. “One of the most exciting aspects of our work is not only making discoveries about how our bodies work on a biological level, but then being able to work with the world’s leading clinical experts to translate these discoveries into therapies for patients.”
Further, studies led by Levy and Thaiss often utilize human samples and data from the Penn Medicine BioBank, to validate animal model findings in the tissue of human patients suffering from the diseases which they are investigating.
While Levy and Thaiss pursue different research interests with their labs, they also collaborate often, building on their previous research into what the microbiome does, and its role in the biological processes that keep us healthy. Their long-term goal is to learn about the mechanisms by which the gastrointestinal tract influences disease processes in other organs to treat various diseases of the body using the gastrointestinal tract as a noninvasive entry point to the body.
“Some of the most common and devastating diseases in humans—like cancer or neurodegeneration—are difficult to treat because they are no existing therapies that can reach the brain,” says Thaiss. “If we can understand how the gastrointestinal tract interacts with other organs in the body, including the brain, we might be able to develop treatments that ‘send messages’ to these organs through the body’s natural communication pathways.”
“Obviously there is a lot more basic biology to be uncovered before we get there,” adds Levy. “Most importantly, we want to map all the different routes by which the gastrointestinal tract interacts with the body, and how that communication happens.”
Christopher Thaiss is Assistant Professor in Microbiology in the Perelman School of Medicine. He is a member of the Penn Bioengineering Graduate Group.
Fabrication steps of the biodegradable BC substrate and the electrochemical devices. (1) Incubation of the bacterium Gluconacetobacter hansenii. (2) BC substrate collected and treated, resulting in a clear sheet. (3) The biodegradable BC sheet is screen-printed, (4) resulting in a device with 3 electrodes, (4) which are cut out using a scissor, (5) resulting in a portable, biodegradable, and inexpensive electrochemical sensor.
The availability of rapid, accessible testing was integral to overcoming the worst surges of the COVID-19 pandemic, and will be necessary to keep up with emerging variants. However, these tests come with unfortunate costs.
Polymerase chain reaction (PCR) tests, the “gold standard” for diagnostic testing, are hampered by waste. They require significant time (results can take up to a day or more) as well as specialized equipment and labor, all of which increase costs. The sophistication of PCR tests makes them harder to tweak, and therefore slower to respond to new variants. They also carry environmental impacts. For example, most biosensor tests developed to date use printed circuit boards, or PCBs, the same materials used in computers. PCBs are difficult to recycle and slow to biodegrade, using large amounts of metal, plastic and non-eco-friendly materials.
In addition, most PCR tests end up in landfills, resulting in material waste and secondary contamination. An analysis by the World Health Organization (WHO) estimated that, as of February 2022, “over 140 million test kits, with a potential to generate 2,600 tonnes of non-infectious waste (mainly plastic) and 731,000 litres of chemical waste (equivalent to one-third of an Olympic-size swimming pool) have been shipped.”
In order to balance the need for fast, affordable and accurate testing while addressing these environmental concerns, César de la Fuente, Presidential Assistant Professor in Bioengineering and Chemical and Biomolecular Engineering in the School of Engineering and Applied Science, with additional primary appointments in Psychiatry and Microbiology within the Perelman School of Medicine, has turned his attention to the urgent need for “green” testing materials.
The de la Fuente lab has been working on creative ways to create faster and cheaper testing for COVID-19 since the outbreak of the pandemic. Utilizing his lab’s focus on machine biology and the treatment of infectious disease, they created RAPID, an aptly named test that generates results in minutes with a high degree of accuracy. An even more cost-effective version, called LEAD, was created using electrodes made from graphite. A third test, called COLOR, was a low-cost optodiagnostic test printed on cotton swabs.
The team’s latest innovation incorporates the speed and cost-effectiveness of previous tests with eco-friendly materials. In a paper published in Cell Reports Physical Science, the group introduces a new test made from Bacterial Cellulose (BC), an organic compound synthesized from several strains of bacteria, as a substitute for PCBs.
Marrying artificial intelligence with advanced experimental methods, the Machine Biology Group has mined the ancient past for future medical breakthroughs, bringing extinct molecules back to life. (Image credit: Ella Marushchenko)
“We need to think big in antibiotics research,” says Cesar de la Fuente. “Over one million people die every year from drug-resistant infections, and this is predicted to reach 10 million by 2050. There hasn’t been a truly new class of antibiotics in decades, and there are so few of us tackling this issue that we need to be thinking about more than just new drugs. We need new frameworks.”
Marrying artificial intelligence with advanced experimental methods, the group has mined the ancient past for future medical breakthroughs. In a recent study published in CellHost and Microbe, the team has launched the field of “molecular de-extinction.”
Our genomes – our genetic material – and the genomes of our ancient ancestors, express proteins with natural antimicrobial properties. “Molecular de-extinction” hypothesizes that these molecules could be prime candidates for safe new drugs. Naturally produced and selected through evolution, these molecules offer promising advantages over molecular discovery using AI alone.
In this paper, the team explored the proteomic expressions of two extinct organisms –Neanderthals and Denisovans, archaic precursors to the human species – and found dozens of small protein sequences with antibiotic qualities. Their lab then worked to synthesize these molecules, bringing these long-since-vanished chemistries back to life.
“The computer gives us a sequence of amino acids,” says de la Fuente. “These are the building blocks of a peptide, a small protein. Then we can make these molecules using a method called ‘solid-phase chemical synthesis.’ We translate the recipe of amino acids into an actual molecule and then build it.”
The team next applied these molecules to pathogens in a dish and in mice to test the veracity and efficacy of their computational predictions.
“The ones that worked, worked quite well,” continues de la Fuente. “In two cases, the peptides were comparable – if not better – than the standard of care. The ones that didn’t work helped us learn what needed to be improved in our AI tools. We think this research opens the door to new ways of thinking about antibiotics and drug discovery, and this first step will allow scientists to explore it with increasing creativity and precision.”
