Discovering new drugs remains an inefficient, trial and error-based process. The founding team behind the Italian startup Sibylla Biotech explain how their interdisciplinary approach can speed up drug discovery in the protein degradation space.\n\n\n\nProtein degradation is a hot field in the biotech industry, with many pharmaceutical companies forging lucrative deals with protein degradation players. \n\n\n\nInstead of blocking a target protein, as most pharmaceutical compounds are designed to do, protein degraders can instead trigger the destruction of the proteins in the cell. This allows them to hit proteins that would otherwise be considered undruggable.\n\n\n\nThe startup Sibylla Biotech was launched in 2017 to open a new frontier in the development of small molecule protein degrader drugs. Its targets are proteins that haven’t yet folded into their final state after being produced in the cell -- a state that lasts only a short time in the protein’s lifespan.\n\n\n\nTo identify these elusive targets, Sibylla’s founding team pooled together expertise from a wide range of disciplines including biotechnology, biology and quantum physics. Being just a few buildings away from each other at the University of Trento, Italy, members of the Sibylla team met thanks to links formed by a master's program in Quantitative and Computational Biology. \n\n\n\nThey devised computational drug discovery technology, dubbed the pharmacological protein inactivation by folding intermediates targeting (PPI-FIT) protocol, to screen drugs that can degrade proteins that haven’t yet completed the folding process. \n\n\n\nPPI-FIT was originally derived from research into protein folding disorders such as Creutzfeldt-Jakob disease. Now, Sibylla’s pipeline is focused on oncology targets such as the protein KRAS, whose mutations are linked to a range of cancers including pancreatic cancers, colorectal cancers and adenocarcinoma. \n\n\n\nTo finance its research, Sibylla bagged €23 million ($25 million) in a Series A round in October 2022 -- one of Italy’s biggest biotech financing rounds to date.\n\n\n\nIn an interview, the founding team behind Sibylla, led by CEO and co-founder Lidia Pieri, discussed how their PPI-FIT approach is tackling common challenges in drug discovery. They also explained how academic institutions can break barriers between departments and create cross-disciplinary innovations in drug discovery.\n\n\n\nWhat methods are used in drug discovery, and why do they need improvement?\n\n\n\nThroughout the history of medicine, drugs were discovered by isolating the active substances of a natural product or by sheer luck as exemplified by Alexander Fleming's discovery of penicillin. Through the development of modern scientific methods of molecular and cellular biology, biochemistry, medicinal chemistry and pharmacology, combined with the ever-growing number of computational technologies, discovering new drugs has gradually moved toward conceptual and rational methods. \n\n\n\nCurrent drug discovery approaches span from fast phenotypic screenings of tens or hundreds of thousands of molecules, enabled by the introduction of lab automation systems, to computer-based ("virtual") screenings of ultra-large libraries of millions or even billions of different molecules. The latter approach has experienced the fastest growth in recent years with the introduction of artificial intelligence aimed at boosting screening efficiency and reliability. \n\n\n\nRegardless of the approach, and despite critical advances in our understanding of many biological systems, drug discovery is still an expensive, lengthy and largely inefficient process. This is demonstrated by the low number of new drugs that reach the market every year, especially in critical diseases such as pathologies affecting the nervous system. \n\n\n\nImproving drug discovery requires innovation at many levels, from solving the nature of incurable disorders to inventing new technologies, to easing regulatory obstacles and defining new methods for efficient clinical trial evaluation. \n\n\n\nUltimately, the entire process should be made sustainable from an economic standpoint to allow the improved search and identification of therapies for rare indications that lack commercial or public health interest.\n\n\n\nWhy is it hard to find the structures of proteins that are still folding, and how can Sibylla’s algorithm tackle this challenge?\n\n\n\nFrom the experimental standpoint, the characterization of partially folded conformations is challenging, because these structures are not stable, at least not in their monomeric phase. \n\n\n\nIn order to study transient structures, one needs to resort to time-resolved methods, with a time-resolution of at least a millisecond. It can be very challenging to apply nuclear magnetic resonance (NMR) and X-ray diffractions to this goal. We are getting to achieving such a result, but it is a target-dependent approach, and we are developing and testing other viable strategies that may hold for all targets. \n\n\n\nTo study transient structures, advanced time-resolved biophysical methods have been developed, including optical techniques, chemical experiments and mechanical experiments. Some of these methods enable probing of the folding at the single molecule level.\n\n\n\nHowever, all these methods provide only partial information and cannot yield the full three-dimensional structure with atomic resolution, like in an X-ray or NMR experiment.\n\n\n\nFrom a simulation standpoint, the problem of observing partially folded structures in plain molecular dynamics simulations is due to the fact that folding is a very rare event. A typical protein can spend from a millisecond to several seconds, up to even hours “attempting” to find a way to fold, before a productive folding event takes place and the chain reaches the native state. \n\n\n\nEven relying on the largest special purpose supercomputer (developed and privately owned by DES Research), it is only possible to simulate molecular motion for up to a few milliseconds, using a simulation box that can only contain a relatively small protein (up to about 80-100 amino acids) in its folded state. Biologically relevant proteins fold over longer time intervals and require a much larger box. \n\n\n\nFurthermore, to characterize the folding intermediates, one needs a statistically relevant sample of conformation, i.e. to reproduce many independent folding events. With conventional molecular dynamics, this simply cannot be done on any existing computers or even those in development, for large and biologically relevant proteins. \n\n\n\nOur methods overcome this problem by relying on a completely different theoretical set up. In a nutshell, our PPI-FIT technology has two main advantages with respect to plain molecular dynamics: \n\n\n\nIt enables us to focus directly on the productive folding events without spending time in simulating the failed attempts. It can be shown that this leads to a doubly exponential gain in computational efficiency.\n\n\n\nIt capitalizes on the available information about the structure of the native state to further exponentially lower the computational cost of the simulation. \n\n\n\nWhy did Sibylla Biotech choose to target cancer when one of the founders investigated protein folding disorders such as Creutzfeldt-Jakob disease?\n\n\n\nOur technology is agnostic with regards to the therapeutic area, however our choice was driven by three criteria. First, the simulation feasibility. Then, the targets covering unmet medical needs. Among those passing the first two criteria, we selected targets that were more straightforward at the wet lab level which were more operationally feasible with the resources we received from our seed investment. Cyclin D1 and KRAS were the targets that emerged based on these factors. \n\n\n\nHow does Sibylla have an advantage over other companies working in protein degradation?\n\n\n\nSibylla is strategically well-positioned in the targeted protein degradation platform space.\n\n\n\nSmall-molecule drug discovery has traditionally focused on the protein function, and this approach typically precludes targeting proteins that lack binding sites in the native state. \n\n\n\nSibylla’s simulations can rescue those undruggable targets by identifying druggable pockets on the surface of the thus-far unexplored folding intermediate states. \n\n\n\nThe PPI-FIT-based folding interfering degrader (FID) technology combines the benefits of various approaches. It is applicable to intracellular and extracellular proteins and is degradation pathway agnostic, as the degradation follows the physiological pathway.\n\n\n\nBeing small molecules, the chemical structure of FIDs is relatively drug-like and amenable to iterative medicinal chemistry optimization. It is found by rational drug discovery as opposed to the serendipitous finding of non-chimeric small molecules that promote degradation. \n\n\n\nThe FIDs concept combines the simplicity and drug-likeliness of monomeric small molecule degraders with a rational hit identification approach, and in doing so occupies a unique and differentiated position. \n\n\n\nWhat are some of the biggest challenges for academic researchers when creating a spinout company, and how did you solve them?\n\n\n\nOne of the biggest challenges for academics who want to launch a startup is changing their mindset from that of an academic researcher to that of an entrepreneur. It is not so easy to make an immediate shift from being a scientist in the academic research lab to focusing on downstream technological development. Furthermore, most academic researchers do not necessarily have prior business and finance experience and this gap can make the transition more challenging. \n\n\n\nWe at Sibylla also faced this transition, and the initial efforts were not trivial. However, we have always firmly believed that our innovative approach could have a significant impact in the drug discovery industry, especially for currently incurable diseases. This certainty, combined with a well-knit team where passion, dedication, collaboration and excellence were paramount, ensured that our spinout company overcame the initial hurdles to launch successfully.\n\n\n\nSibylla Biotech formed when a physicist joined forces with a biologist. Can you explain how we could encourage more researchers to meet from different disciplines and form new ventures?\n\n\n\nWhat we experienced in the cross-disciplinary research effort that led to the inception of the PPI-FIT paradigm and later to Sibylla Biotech was a true exchange of knowledge, experience and expertise. This provided unexpected opportunities to tackle the complexity of the drug discovery process. \n\n\n\nResearchers need to possess specific qualities for effective interaction in cross-disciplinary environments. First, they should recognize its value. Second, they need proper motivation and courage to explore new and innovative ideas. They should be inclined to take risks and be comfortable outside of their comfort zones. Finally, they should be able to explain complex concepts in simple words. Pushing disciplinary boundaries can be difficult. In some way, it requires the attitude of an explorer.\n\n\n\nThere are several actions that research institutions can implement to promote a thriving cross-disciplinary research culture. These may range from monetary incentives (e.g., intramural grants dedicated to cross-disciplinary research programs) to establishing inter-departmental research centers -- where researchers from different disciplines share spaces -- to creating educational programs explicitly dedicated to the future leaders in cross-disciplinary science. \n\n\n\nIt was thanks to such a program at the University of Trento that the founders of Sibylla Biotech started to interact, and several current employees of the company graduated from there as well.