Antibiotics of the future: an innovative approach by César de la Fuente

BIOTECHNOLOGYARTIFICIAL INTELLIGENCE

As part of the launch of the report Neurotechnology for Human Well-being , we concluded a series of webinars with the outstanding participation of Dr. César De la Fuente. This report is the result of the last meeting of our think tank Future Trends Forum, where we convened more than 40 internationally renowned experts to discuss the repair and enhancement of human capabilities through neurotechnology and other innovative areas, and to analyze the opportunities and risks involved.

Dr. César de la Fuente, who was awarded the Princess of Girona Award for Scientific Research in 2021, is an outstanding scholar and researcher, leader of the Machine Biology Group at the University of Pennsylvania, USA. This group is mainly dedicated to the development of new antibiotics through computational technologies. With a background in Biotechnology from the University of León and a PhD in Microbiology and Immunology from the University of British Columbia, Canada, Dr. de la Fuente has also extended his education and research at renowned institutions such as the Massachusetts Institute of Technology (MIT) in the U.S. Author of nearly a hundred scientific papers, his work is a benchmark in the field.

In a context where no new antibiotics have been found in recent decades and bacteria show increasing resistance to existing treatments, Dr. de la Fuente’s research focuses on using computation to find new classes of antibiotics. These efforts aim to save millions of lives and have been widely recognized, receiving awards such as the Langer Prize, the ACS Kavli Emerging Leader in Chemistry, AIChE’s 35 Under 35 Award, the ACS Infectious Diseases Young Investigator Award and the GEN Top 10 Under 40.

In this webinar, Dr. César De la Fuente tells us about his fascinating, 10-year journey of research and discovery.

Here you can watch the webinar with César de la Fuente:

Developing new antibiotics with artificial intelligence, with César de la Fuente

Below, we summarize the key insights and discoveries shared by César de la Fuente during this webinar:

Development of new antibiotics: a race against bacterial resistance

Antibiotic resistance is a global challenge, one of the most pressing threats to public health. This resistance occurs when bacteria mutate and become immune to the effects of the drugs that used to kill them. Over time, these antibiotic-resistant bacteria have become more common and dangerous, jeopardizing the medical advances of the last century.

Antibiotic resistance not only makes common infections more difficult to treat, but also increases the risk of disease spread, severe infections and more deaths. The World Health Organization estimates that by 2050 deaths from untreatable infections could exceed those caused by all current forms of cancer. This grim prognosis underscores the critical need to develop new antibiotics.

In this context, the work of César de la Fuente and his team is particularly relevant. Their innovative approach uses artificial intelligence (AI) to discover and design new antibiotics. This involves analyzing large databases of chemical and biological structures to identify potential new drug candidates. AI is particularly useful in this area because of its ability to process and analyze massive amounts of data faster and more accurately than traditional methods.

De la Fuente’s work also focuses on better understanding how existing antibiotics interact with bacteria and how they develop resistance. This knowledge is crucial for developing strategies to overcome resistance and for designing antibiotics that are more difficult for bacteria to evade.

Innovation in bioengineering: a hybrid between AI and medicine

De la Fuente began integrating computational concepts into biology and microbiology during his PhD studies at the University of British Columbia and continued this approach at MIT. This integration marked the beginning of his journey to use AI in the development and discovery of new antibiotics. Now, his lab, called the Machine Biology Group, brings together experts from multiple disciplines, including chemistry, physics, computer engineering, microbiology and synthetic biology, creating a multidisciplinary environment that transcends individual disciplines and is enabling them to accelerate the discovery of new antibiotics using AI.

A significant advance in this field is the design of antibiotic molecules that can target specific bacteria. This falls squarely into the realm of personalized medicine, enabling the treatment of infections accurately without harming the patient’s beneficial gut microbiome. The initial perception that biology was too complex for the application of AI was challenged by de la Fuente, who saw the potential of using computers to accelerate antibiotic discovery. 

AI not only speeds up the antibiotic discovery process, but can also make antibiotics more accessible and affordable, especially in developing countries. This reduction in the cost of production could translate into greater availability of these vital medicines. Compared to traditional antibiotic discovery methods, which can take years, AI has enabled the discovery of hundreds of thousands of antibiotic candidates in a matter of hours, significantly increasing the likelihood of success in clinical phases and, ultimately, in saving lives.

De la Fuente identifies three key concepts in his approach: the efficiency of computers to explore molecular sequences, the ability to generate new molecules never before seen in the biological world, and the use of computation to systematically explore biology in search of useful new molecules, including antibiotics. 

A revolutionary first discovery: Guavanine

César de la Fuente points out that, until recently, only a tiny fraction of all possible molecules in the course of evolution had been explored. De la Fuente underscores the ability of computers to expand the frontiers of molecular research, exploring sequential spaces unheard of in natural evolution. This computational approach offers new opportunities to address current challenges, including antibiotic resistance. The de la Fuente team considered how to teach a computer to innovate at the molecular level. They decided to apply Darwin’s algorithm of natural selection theory, accelerating the evolutionary process on a compressed time scale. This project, started in 2015 and published in 2018, began with an initial population of molecules encoded in binary language. Using an iterative loop, the computer was able to evolve these molecules, generating new antibiotics. One of the resulting molecules, called Guavanine-2, proved to be tremendously effective against pathogenic bacteria in the laboratory. What is remarkable about Guavanine-2 is its unique mechanism of action: unlike conventional antibiotics that depolarize the bacterial membrane, Guavanine-2 hyperpolarizes it. This discovery was an unanticipated innovation by artificial intelligence, evidencing the emerging capabilities of AI in fields such as biocomputing.

Finally, Guavanine-2 was tested in a mouse model, showing a significant reduction in infection compared to a control. This success marked a breakthrough in the use of artificial intelligence for the design of new antibiotics, a field that has grown significantly since the team published its paper on Nature Communications. De la Fuente expresses his hope that this approach will one day result in drugs that can save human lives.

A second surprising discovery: antibiotics generated from human body proteins

After his time at MIT and then at the University of Pennsylvania, de la Fuente and his team set out to answer a fundamental question: Could computers probe biology to find new antibiotics and other useful molecules? This line of inquiry broke new ground in biomedical research, driven by the urgency to speed up the slow and costly process of drug development, especially antibiotics. The inspiration for this project came from existing algorithms used in pattern recognition, such as those used in apps like Siri and Alexa. However, instead of identifying faces or sounds, the goal was to recognize molecular patterns that might indicate the presence of potential antibiotics. The team developed algorithms to scan entire proteins, identifying regions as potential antibiotics. This methodology made it possible to examine proteins and fragments of their structure, searching for new antibiotics. 

Thanks to advances in computational and algorithmic capabilities, de la Fuente’s team was able to scale up their project to analyze entire proteomes, i.e., all the proteins encoded by a genome. This approach led the team to explore the human proteome as a source of antibiotics for the first time. Using advanced algorithms, they analyzed approximately 42,000 proteins, including isoforms, to discover peptides with antibiotic properties. Surprisingly, the entire computational analysis was completed in just one hour, thanks to the simplicity and efficiency of the algorithm used.

The most significant discovery of this research was the identification of a new type of antibiotics, called “encrypted peptides”, previously unrecognized in the human proteome. These peptides showed remarkable efficacy against a variety of bacteria in laboratory tests and in preclinical mouse models, including resistant strains such as Acinetobacter Baumannii. In addition, at least four physiologically relevant antibiotics were identified, expressed in different parts of the human body and possibly playing a role in the immune system.

This work not only provided new insights into the potential of the human body as a source of antibiotics, but also suggested that these encrypted peptides could be found throughout the tree of life, thus opening new avenues in the search for solutions against antibiotic resistance. 

Third revolutionary discovery: antibiotics in the genomes of extinct species and the concept of de-extinction.

Following their first two groundbreaking milestones, de la Fuente and his team decided to explore a previously unimaginable scientific horizon: molecular de-extinction. They began to delve into the vast diversity of the tree of life, with the goal of deciphering molecular sequences that reveal hidden evolutionary histories, similar to how DNA is studied to understand evolution. This search is not only focused on the present but also on unraveling secrets of the past and foreseeing future developments.

Inspired by the idea of resurrecting, not whole organisms, but specific molecules, de la Fuente has developed the concept of “molecular de-extinction“. This innovative approach led him to examine the genomic and proteomic sequences of our closest evolutionary relatives: Neanderthals and Denisovans. Through advanced computational methods, his team has been able to identify fragments with antibiotic potential in archaic proteins from these extinct humans, including a remarkable finding called Neardentalin-1, a molecule with therapeutic properties discovered in Neanderthal proteins. The discovery process involves splitting proteins encoded in the genomes of Homo sapiens, Neanderthals and Denisovans into smaller fragments. Artificial intelligence filters are then used to predict the antibiotic capacity of these fragments. This approach has enabled the computer to identify antibiotics encoded in ancient proteins, a promising breakthrough in drug discovery.

However, de la Fuente’s project does not stop there. The chemical synthesis of these molecules has revealed fundamental differences in the mechanisms of action between modern and archaic molecules. While the modern ones tend to attack the outer membrane of bacteria, the ancient ones target the inner membrane, suggesting new avenues for antibiotic development. These discoveries have proven effective even in preclinical mouse models, especially the molecule he has named Neardentalin-1.

The concept of molecular de-extinction proposed by de la Fuente opens up a completely new field in science, allowing us to explore molecular sequences never before examined. This approach not only offers the possibility of discovering new drugs, but also expands our understanding of human biology and evolution.

The efforts of de la Fuente and his team extend beyond Neanderthals and Denisovans. Using more powerful Deep Learning algorithms, they have researched several extinct organisms, from the Holocene to the Pleistocene. The most promising molecules they have found are a molecule from the mammoth, which they have named Mamutusin-2, one molecule from an extinct elephant seal, one from the giant sloth and one from an extinct deer.

This research journey also poses significant ethical and legal challenges. De la Fuente’s team is committed to responsible innovation and has consulted with bioethicists and patent experts to navigate these uncharted waters. The question of whether extinct molecules can be patented is an unprecedented legal dilemma that is spawning a new field in patent law.

And this has only just begun: the future of the search for antibiotics in nature and in bacteria themselves.

Dr. César de la Fuente and his team are now focused on mining more than 100 million proteomes and more than 200 million proteins. Following the remarkable progress outlined above, de la Fuente has turned his attention to an even vaster territory: bacteria.

In collaboration with international experts, De la Fuente has initiated a comprehensive exploration of the global microbiome, encompassing some 90,000 microbial genomes and more than 60,000 metagenomes. This research has resulted in the discovery of nearly one million new antibiotic candidates encoded in the planet’s microbial diversity. Some of these antibiotics have shown potential in preclinical models using mice, offering new hope in the fight against diseases resistant to conventional treatments.

This progress is not limited to the mere identification of new substances. The application of artificial intelligence and the use of powerful computers have started a revolution in research. Traditionally, the discovery of preclinically-relevant antibiotic candidates could take between three and six years, as mentioned above. However, with today’s technological tools, this process has sped up dramatically. It is now possible to identify hundreds of thousands of such candidates in a matter of hours, significantly increasing the likelihood of advancing to the clinical phases and ultimately saving human lives.

Dr. De la Fuente also highlights the rapid growth and dynamism of a field that combines artificial intelligence with antibiotic discovery. Until 2018, there were hardly any publications in this area, but since then it has experienced exponential growth, becoming a promising and exciting field of study. Hopefully and optimistically, he anticipates that some of these new compounds will have a positive impact on our society and contribute significantly to global public health.

Questions and answers with César de la Fuente

To conclude the webinar, César de la Fuente answers some of the many questions posed by the audience. Below is a summary of these questions:

Antibiotic development through AI and clinical practice:

When asked about the integration of AI-developed antibiotics into clinical practice, De la Fuente expressed optimism. He stressed that AI could significantly accelerate the discovery of new antibiotics, potentially offering solutions to currently untreatable infections. In addition, he highlighted the role of computational models in predicting synergies between existing antibiotics, which could improve their efficacy by combining them in innovative ways.

Ethical and regulatory considerations:

Regarding ethical concerns and regulations in drug design, Dr. de la Fuente underscored a responsible approach. He mentioned specific precautions to avoid autonomous replication of molecules and assured that, although consultations with bioethicists minimize their concerns, his team strives to maintain the highest standards of safety and accountability.

Collaboration across disciplines:

Regarding collaboration between experts in different areas, De la Fuente highlighted his role as a “translator” between disciplines. His laboratory benefits greatly from diverse perspectives, leading to innovative projects that would be impossible in monodisciplinary environments.

AI and treatment customization:

On the personalization of treatments, the scientist expressed great enthusiasm for the potential of AI to design specific antibiotic molecules. These molecules could be effective against specific bacteria without affecting the beneficial microbiome, a significant step toward personalized medicine.

AI in the fight against cancer:

Speaking about cancer, De la Fuente mentioned that, although they specialize in infectious diseases, they have started collaborations with cancer experts who seek to develop models to discover anticancer molecules.

Mechanisms of antibiotic action and bacterial resistance:

On the variety of antibiotic mechanisms of action detected by artificial intelligence, although they initially focused on the bacterial cell membrane, his team is now also researching whether they can direct antibiotics to other targets such as bacterial DNA, RNA, or proteins within the bacteria. This represents a significant advance, as exploring multiple mechanisms of action could lead to more effective and targeted treatments.

Antibiotic resistance and bacterial adaptability:

In response to concerns about bacterial resistance to the new antibiotics, De la Fuente shares encouraging observations. So far, they have not detected significant resistance in bacteria exposed to the new antibiotics developed in their laboratory, unlike traditional antibiotics such as ciprofloxacin. However, he recognizes the adaptive capacity of bacteria and does not rule out the possibility of future resistance.

Accessibility and affordability of antibiotics:

Finally, addressing the issue of accessibility, De la Fuente highlighted the role of AI in reducing antibiotic production costs. He envisions a future where advanced systems can deliver essential medicines to remote and underserved areas, improving access to healthcare worldwide.

If you want to learn more about this field and other technologies and innovations for human welfare, be sure to check out our report.