University of Bristol spin-out Ziylo made headlines recently after the company was sold to Novo Nordisk in a staged acquisition with a potential value of up to US$800M. Novo Nordisk plans to use the glucose binding molecules developed by Ziylo to enable glucose responsive insulins (GRIs).
The former Ziylo team have now formed a new company, Carbometrics which has two roles; to deliver the chemistry for Novo’s GRI program and to use its exclusive licence for diagnostics to deliver a glucose sensing platform that can be used to create market-leading continuous glucose sensors.
Dr Andy Chapman is the former Chief Scientific Officer at Ziylo, now CSO and Co-founder at Carbometrics. He sat down with Anna Fleming to discuss glucose sensing, synthetic chemistry and what makes the molecules so uniquely desirable.
Can you give me some background on Carbometrics? What’s the key technology?
The core patent that we filed (and principally the one that Novo Nordisk bought Ziylo for) is for a synthetic glucose binding molecule. It’s designed to specifically recognise the shape and chemical surface of glucose. It’s so well optimised that even in a complex mixture like blood – which has thousands of other dissolved small molecules which it could potentially bind, only glucose will bind in the cavity. What’s more, the amount of glucose that’s bound in that cavity depends on the concentration of glucose in the medium. Right there you have the makings of a glucose sensor that can work in blood.
You can also use glucose binding – and some very clever molecular scale chemical engineering – to play with the properties of insulin and moderate its potency depending on the glucose concentration. That’s what interests Novo Nordisk – creating a glucose responsive insulin, which harnesses the properties of Ziylo’s molecule to act as a glucose-based switch.
This molecule is completely synthetic?
Yes. Before this, the only synthetic molecules that stood a chance of recognising glucose in these complex solutions were boronic acids. They’re simple chemical entities that recognise a specific chemical feature of glucose. The issue is that that feature is very common on other molecules, so they can’t recognise glucose selectively enough. To add to that, boronic acids aren’t very stable, so the chance that they will degrade inside a living system is very high. Insulin is life threatening if the concentration is too high, so if you’re considering moderating its potency, your off-switch has to be based on a very robust molecule.
Nature’s glucose sensors are highly complex – massive protein architectures with huge 3D structures just to form pockets around small ligands like glucose. Because they’re not from us we have immune reactions against them, which makes them difficult to deploy, and they’re difficult to use in a glucose responsive insulin.
So are those natural molecules and the boronic acids the main competing products in the field at the moment?
Yes, they’re competitors. Principally, glucose sensing is done using glucose oxidase – an enzyme that binds glucose with high specificity. But it’s not an ideal sensor because it has interference problems – traditionally, taking paracetamol, for example, would create a false reading. Those problems have largely been engineered out now, but the system’s not infallible.
Using boronic acids in glucose sensing is also very successful. The FDA has recently approved a long term implantable device using boronic acids, which is (supposedly) very resistant to interference. But its lifetime is limited by the chemical entity which is doing the sensing. For patients who have this implanted, the longer you can go without having the surgery to replace it, the better, and so the more patient uptake we’ll get. The chemistry of our system is essentially indestructible, so it has the potential to be very long term indeed.
How did you find this synthetic molecule?
It came out of Prof. Tony Davis’ lab at The University of Bristol. He’s a supramolecular chemist – he’s trying to design the host, containing a ‘cage’ area to which small molecules (the guest) will bind. Tony decided to target glucose, not just because it’s a key macronutrient, but also because its chemical properties make it ideal to build a cage around: it’s sort of symmetric – a flat disk with an equivalent top and bottom, and a ring around the outside. For synthetic organic chemistry the cage you would need isn’t such a weird shape. It’s hard to build weird shapes, but this is pretty regular.
A big issue was working in water. That’s really difficult, because you’ve got to get your organic building blocks to dissolve in water, and like oil in water, they tend not to mix. Tony developed these tricks to design and build an organic-soluble version of the Glucose Binding Molecule and then right at the end convert it into a water-soluble one once all the chemistry to make it is done.
He started with this concept that he called the temple, which is a ‘roof’ and ‘floor’ which are very hydrophobic; very well matched to the top and bottom of glucose, and then ‘pillars’ around the outside which are polar, matched to the polar rim of glucose. They modelled and built various iterations of this design, and early on they got very weak binding of glucose in water, which – to bind it at all, in water – was absolutely unheard-of. After 20 years or so of designing and modelling these molecules he found one which was fantastic.
So how did that become Ziylo?
Harry (Destecroix) spun out Ziylo with a molecule that he’d worked on during his PhD. It was quite easy to make and had the right affinity for glucose – that is, the cage can be completely unoccupied, partially occupied or fully occupied over a range of physiological concentrations. But what this molecule lacked was the selectivity for glucose over all the other molecules in blood. Millions of things bind to this molecule, some much more strongly than glucose – caffeine for example, will bind really strongly.
During this time, Rob Tromans (a recent PhD graduate from the Davis lab and now Senior Scientist at Carbometrics) was working on another design from Tony. He realised the synthesis and tested the binding of this new design, and it was fantastic. Ziylo completely pivoted, dropped everything on the old molecule and focused completely on this new one. At first it was much harder to make, but over time it’s become easier. Now Carbometrics is working to optimise it as a sensor, and for the applications Novo Nordisk are planning.
Where do you plan to go next?
Right now we’re focused on delivering the molecules and research input into this glucose responsive insulin project with Novo. It’s only been a couple of months but it’s going very well. We’re also going to do our own proof of concept glucose sensing systems with a view to partnering again. The key is developing a binding platform that’s robust, synthetically accessible and selective enough for glucose. Once you’ve got that, you’re building from the bottom up. We have this open platform which can bind glucose in solution. Now, what can we do with it? Can we make it tell us that glucose is bound using light? Using sound? Using electricity? Whatever you want. In principle, the world’s your oyster.