SuperMUC-NG Supports Search of RNA-Based Therapies

The field of RNA-based medicine has grown since the 1990s, with researchers making strides in fighting cancer using RNA for over a decade. The technology came into popular consciousness during the COVID-19 pandemic, where the first vaccines created to combat the virus used messenger RNA (mRNA). As researchers continue to better understand how to use RNA to fight disease, they’ve explored how to use it to treat new conditions as well as how to improve its ability to arrive where it is intended in the body. “You cannot just put RNA in your body through a capsule or a vaccine, because it will not end up in your cells – you need to design ways to get it there,” said Dr. Benjamin Winkeljann, a group leader within the Drug Delivery Chair led by Prof. Olivia Merkel at Ludwig-Maximillians-Universität München (LMU). “We use nanocarriers that can essentially bring the RNA into a cell. More of the therapeutic entering the cell makes a medicine more effective and getting it into the correct, diseased cells specifically lowers the risk or severity of side effects for patients.” Winkeljann has spearheaded efforts to incorporate high-performance computing (HPC) simulations into the Chair’s larger efforts to improve drug delivery in the body. Recently, Winkeljann and his collaborators used SuperMUC-NG at the Leibniz Supercomputing Centre (LRZ) to run molecular dynamics simulations capable of guiding the team’s efforts in better understanding how certain classes of nanoparticles can help safely guide RNA therapies to their target. Working together with LRZ, the team recently developed computational methods to more accurately simulate these interactions between RNA and nanoparticles. 

Simulations chart the course to improved drug delivery

Unlike DNA—a double-stranded molecule that imparts important genetic information to human cells about their role and function in the body—RNA is a class of molecules that either carry out specific cell functions or create proteins to do it for them. Researchers in the late 90s discovered a specific class of RNA, called short-interfering RNA (siRNA), which limits or blocks gene expression—essentially, the process by which the constituent building blocks of DNA and RNA, called nucleotides, produce proteins that give instructions to certain cells. When it comes to genes, there can be too much of a good thing. “Certain pathological conditions are caused when genes get overexpressed, which leads to the symptoms we associate with a specific disease,” Winkeljann said. “This can happen in severe asthmatic disease, COPD, or certain types of cancers.” Winkeljann continued, noting that one of the primary benefits of designing siRNA therapies is its ability to attack illness at the root cause. “You can use these kinds of therapies very early in the onset of a disease rather than trying to interfere and correct for damage already done,” he said. 

While siRNA is well-suited to limit or silence excessive gene expression, it is not adept at finding its way to the right place to do it. RNA in medications can degrade before it ever reaches the correct place in the body, so those developing RNA-based therapies must combine the RNA with other carriers that can serve as delivery vehicles. Thus far, the several approved RNA medications on the market rely on lipids, or fat molecules, as their delivery agents, but Winkeljann and his collaborators wanted to investigate whether they could get better results using cationic polymers. Polymers are large molecules with consistent, repeating structures—DNA and RNA are both subcategories of polymers. Positively charged cationic polymers are attracted to negatively charged RNA molecules and form so-called “polyplexes” that can safely guide the vulnerable RNA to its intended location. 

When Winkeljann joined the LMU Drug Delivery Chair in 2021, the team already had ample experience working on developing RNA therapies, but he sought a way to augment the iterative trial-and-error experimental process. He focused his postdoctoral work on integrating simulation into the team’s workflow. “I thought it would be interesting to involve computational methods in our research, and it also felt like I was able to challenge myself with a new skillset, as I had not worked with molecular dynamics simulations before.” To study how drugs interact in the human body, researchers use molecular dynamics (MD) simulations. MD allows scientists to observe the movements of various molecules in a system and chart how that system evolves and changes over time. While this method is accurate in recreating molecular interactions, it is also computationally expensive when trying to simulate more than a few molecules interacting for a short period of time. 

Because he knew the team needed to run larger simulations to accurately model polyplex formation and interactions between multiple polyplexes, Winkeljann worked closely with LRZ user support staff to develop a computational workflow using “coarse-grained” molecular dynamics (CGMD). Instead of simulating every atom from first principles, CGMD allows researchers to group certain atoms and model them as a single “interaction site.” While this approach makes some assumptions about parts of the molecular system, it also allows researchers to run simulations efficiently while maintaining good accuracy. For Winkeljann and his collaborators, having the ability to run many good simulations of a larger system is more valuable than running one or two simulations with fewer particles from first principles. “In order to understand patterns and concepts present in these complex systems, we need to be able to change parameters and run suites of simulations instead of simulating one gigantic system,” he said. Using SuperMUC-NG, the researchers created simulations modeling the molecular organization and dynamics of siRNA polyplexes and compared those results with experimental findings. Its simulations showed good agreement with their experiments, providing a new research and validation tool for use in the development of new siRNA therapies. The team published its results in Nano Letters. 

Fndamental research underpinning the drug discovery 

With a trusted CGMD method now in its toolbox, the team can complement its experimental work with insights from simulations. Winkeljann credited the close collaboration with LRZ staff in helping the team scale up its application to take full advantage of SuperMUC-NG. “I did not have extensive computational experience before joining this team, and it was a big jump to take our application from our Chair’s cluster to a large system like SuperMUC-NG,” he said. “LRZ helped us go through a scaling up process when we started to think about applying for time, so we had a test project on the system, then we had a regular project, and now we’ve got a large-scale allocation.” The team was paired with Dr. Helmut Brüchle at LRZ, whose experience with MD simulations allowed the team to refine its application. 

Further, Winkeljann appreciated working at a public HPC center that gives his team the freedom to do fundamental research. “Having public support versus commercially interested support allows us to maintain the independence of our research,” he said. “We are doing work that is more fundamental to the field than the usual pharmaceutical formulation developer because we are digging into the underlying mechanisms of these systems. If this was privately or commercially funded research, there would undoubtedly be other goals in mind that would make it difficult to pursue this kind of research.”  Eric Gedenk | GCS