Tell us about the work you are leading at the Roche RNAHub.
The RNAHub is a community of 60+ scientists that drives the early discovery pipeline of the oligonucleotide modalities at Roche. This includes ASOs and siRNAs in all their forms and functions. Beyond taking care of the portfolio of molecules on their journey to help patients, we also innovate. One must keep in mind that the modality can be considered “young” compared to small molecules which have a 100-year history or antibodies, which have been around for more than 50 years.
To this end, we run platform projects dedicated to asking and attempting to solve the difficult questions in the field, from design, to safety, potency, screening and anything else that plays a crucial role in drug development.
The third important aspect is that we work as a fully matrixed organization rather than the hierarchical structures often found in pharma companies. We believe that such a flexible expert network is well suited to solve the complex issues we are tackling. But it also requires “community work” to bridge the classic gaps of matrix structures such as decision making or communication and to create an enjoyable environment that fosters creativity and team play.
What kinds of complex issues are you trying to solve?
With ASOs, siRNAs and other types of nucleic acid-based medicines, their delivery to the right tissue and cell type is a major challenge. It is difficult to access tissues beyond those that the field has already successfully targeted without pairing the oligo with an innovative delivery technology.
Also, as mentioned before, the modality is relatively young, so there are quite a few questions like “What should the screening cascade look like?” and “How can we derisk safety, efficacy, dosing, cost of goods, etc?” still to be explored. These are complex questions that other modalities have already worked out and that we need to invest time into understanding them better.
What are the delivery challenges for ASOs, siRNAs and other nucleic acid-based medicines?
Extrahepatic delivery. There are various options to achieve tissue-specific targeting outside of the liver including devices, conjugation to ligands, encapsulation or even local administration. Only if one matches the right delivery technology with the right choice of oligo to the cell type of interest, such a molecule has a realistic chance to become a drug.
A little bit more specific, at Roche we have a strong focus on eye and brain delivery. Both are tissues to which oligonucleotides have no access if merely administered peripherally, since they are protected by cellular barriers, such as the blood-brain barrier (BBB) or the blood-ocular barrier.
"Only if one matches the right delivery technology with the right choice of oligo to the cell type of interest, such a molecule has a realistic chance to become a drug."
How can we achieve tissue-specific delivery for these therapies, especially in CNS and ophthalmic?
Our team has made big strides in delivering oligonucleotides via eye drops to the front of the eye and there is also an option for intravitreal administration to reach the back of the eye. Such local approaches can achieve sufficient biodistribution while remaining limited to the tissue of interest.
For the brain, we are taking a different route. Here, we conjugate oligonucleotides to Roche’s Brainshuttle™, which will allow for peripheral administration. We are profiting a lot from the fact that the company already has two molecules in the clinical pipeline that employ this type of delivery approach, which is a great source of knowledge. Nevertheless, it’s important to keep in mind that like all conjugate-type technologies this will not be exclusively specific to the CNS which means we have to understand all relevant clinical aspects.
For both approaches it is imperative to balance the size of the design space with the capabilities of your test systems. For example, if you attach an antibody like the Brainshuttle™ to an oligonucleotide, you need to consider the biophysical properties, the target affinity, the linker, and many many more aspects. The design space is huge and at the same time, one must accept the limitations of our ability to screen in vitro and test in vivo. Matching these two aspects really requires a deep understanding of what translates, and which properties can have which impact.
Can you explain the Brainshuttle™ delivery technology?
The Brainshuttle™ is one approach to address the fact that Mother Nature has decided to protect the brain very well. Basically, all roughly 400 miles of blood vessels we find in a human brain are surrounded by a tightly connected layer of endothelial cells, the BBB, which carefully controls which molecules and substances can enter the brain. Nucleic acid-based medicines are pretty much excluded from this tissue as they are too large and too charged to cross the BBB.
Roche and others in the field have made encouraging progress in developing what we call today the Brainshuttle™. In essence this is an antibody that binds the transferrin receptor, which usually functions as a transport system for iron-bound transferrin across the BBB via a mechanism called transcytosis. Once this antibody binds the transferrin receptor, it can trigger the same transport mechanism and thus shuttle any drug connected to the Brainshuttle™ into the brain parenchyma.
Can you speak about the role of AI in drug delivery for these kinds of therapies?
AI is based on machine learning approaches and machines can only learn if fed with a decent amount of high-quality data. Unfortunately, that sort of data is still missing in many areas of our field. If we scientists are not able to offer good data to these machine learning algorithms, these tools will not deliver on the original promise. We first need to generate good data by running experiments and working on getting the design space right. Nevertheless, we have of course activities in-house that harness the power of AI, where it makes sense, and there are encouraging developments happening also in the field. One example is activities focusing on LNPs where AI is used to screen hundreds and thousands of molecules in silico.
"If we scientists are not able to offer good data to these machine learning algorithms, these tools will not deliver on the original promise. We first need to generate good data by running experiments and working on getting the design space right."
What do you think about non-AAV and non-LNP approaches to delivering these therapeutics?
It’s exciting to see these different approaches being developed. To really move the field forward we need to make sure we create a solid data foundation. For example, it is a good start to successfully deliver a fluorescent dye or reporter system but that doesn’t tell you that you have crossed the BBB. As scientists we should create data packages that address the critical questions. In the end it comes down to data quality, which ideally is being generated very early on.
What are the difficult questions we should be asking?
I consider two aspects as imperative: Number one is to understand if we’ve successfully delivered cargo to the tissue of interest. For me, an ideal proof consists of the “data trinity” of pharmacokinetics profile (PK), exposure (i.e., accumulation in the target tissue) and a meaningful pharmacodynamic effect (PD). Where does the cargo end up and how long is it in the bloodstream? Is it realistic that with a very short half-life we can deliver this cargo to the brain? How much cargo ends up physically in the target tissue? While this can be difficult to measure for some modalities, for example due to stability, it often is straightforward. And does a given system also work in a relevant disease model?
The other pillar is to ask if a molecule or formulation is something that can eventually become a drug that will help patients. For example: what about CMC properties and cost of goods? While this doesn’t need to be the first thing to figure out, a successful PoC should trigger these considerations to make smart investment decisions.
What kinds of partnerships are you looking for at the RNAHub?
Right now, the RNAHub is built around three “Grand Challenges:” to develop and deliver on our oligonucleotide portfolio, to develop a modality toolbox including new delivery approaches and to solve the remaining safety questions for this area. To support what we do internally, we have an external innovation team that keeps an eye out for new technologies emerging outside of Roche. We are always excited to learn about potential partners who could help us with any of the three “Grand Challenges.”
"The RNAHub is built around three “Grand Challenges:” to develop and deliver on our oligonucleotide portfolio, to develop a modality toolbox including new delivery approaches and to solve the remaining safety questions for this area."
Are there any specific delivery technologies that most excite you?
With regards to ligand-based delivery I am of course most excited about our Brainshuttle™, which we hope to see entering clinical development in the not-so-distant future. On the device side (combined with cleverly designed oligo conjugates), I am looking forward to hearing more about options for inhaled delivery, which some companies are working on. And of course, as mentioned our work in-house on eyedrops.
On the particle side, I would hope to hear more from the world of polymer nanoparticles - LNPs seem to have found their niche in the vaccine field. The polymer particle field has been very active for a long time and multiple challenges like CMC and stability remain. But if we can solve these challenges, these therapeutics could make a big splash in the future.
Anything else?
Let’s keep working on these challenges! Things in our industry – and in natural science generally – take time but eventually can have a huge impact on a patient's life. I am convinced that ASOs and siRNAs in a few years’ time will be looking back at a successful history just like large molecules and small molecules can do today.
