Precision Design of Antimicrobial Peptides Against Bacterial Infections - PROJECT SUMMARY Antibiotic resistance of bacterial pathogens is one of the greatest public health challenges of our time. It causes difficult-to-treat infections and jeopardizes modern healthcare advancements. As the emergence of bacterial resistance is outpacing the development of new antibiotics, we must find cost-effective, innovative approaches to discover new antibacterial therapeutics complementary to small-molecule antibiotics. Antimicrobial peptides (AMPs), as a new class of antibacterial agents, represent one of the most promising solutions to fill this void, since they generally undergo faster development, display rapid onsets of killing, and most importantly show lower risks of induced resistance, compared to small-molecule antibiotics. Yet, very few analogs or modified derivatives of natural AMPs have been approved in practice, and most of the failure is caused by systemic or local toxicity associated with broad-spectrum antibacterial activity. Toward a long-term goal to discover effective, selective AMPs as therapeutics to target a narrow spectrum of specific antibiotic-resistant pathogens, our objective is to develop the new capacity needed for such discovery, by integrating innovative approaches and applications of machine learning, multiscale modeling, peptide synthesis, and microbiology. We have developed the first generative adversarial network model (AMP-GAN) to produce AMP candidates with diverse sequences and structures, as well as accurate multiscale models and methods to study the mechanisms of AMP aggregation and target interactions. It is our central hypothesis that AMP selectivity may be achieved via controlling their sequence, structure, interaction, aggregation, and co-aggregation. In pursuit of three specific aims to establish a novel methodology toward discovery of narrow-spectrum AMPs, we will (i) generate selective AMP sequences with predictable activity and pathogen targets, (ii) identify AMPs to target characteristic biomolecules in pathogens, and (iii) modulate AMP aggregation to tune cell selectivity or to achieve synergy. We will advance our computational techniques like AMP-GAN and top-down simulations in conjugation with chemical characterizations (for structure and dynamics) and cellular assays (for activity and toxicity). We anticipate gaining a fundamental understanding of how to design narrow-spectrum AMPs, as well as how to combine new computational and experimental tools to achieve desired AMP selectivity. Overall, this contribution can be significant since it will establish new avenues for precision AMP design and bring more AMPs closer to the clinic by overcoming their known pitfalls. The resulting knowledge will be widely shared in the scientific community for AMP research and development. Our concepts and approaches are innovative, as they shift the current paradigm of broad-spectrum AMP design towards higher accuracy, diversity, and target selectivity through precision AMP design. Collectively, given the increasing need for treatment options against antibiotic-resistant infections, the methodology and tools from this proposed research will enable the discovery of new therapeutics for challenging infectious diseases.
