Dulce Lima Cunha is a postdoctoral researcher from UCL Institute of Ophthalmology, currently working at Radboud University in the Netherlands.

Last year she published the results of a study demonstrating that Amlexanox rescues PAX6 levels in aniridia stem cell-derived models, in comparison with Ataluren and other translational readthrough inducing drugs.

Other researchers involved in the study were Hajrah Sarkar, Jonathan Eintracht, Philippa Harding, Jo Huiqing Zhou and Mariya Moosajee.

Dulce spoke about the outcomes of the study and answered questions at our 2023 Conference in Birmingham last September. You can watch her presentation in the following video, beneath which you can also read the transcript. The full research paper is also available online if you want to delve into the details further.

Transcript

[Dulce] I’m Dulce, I was a postdoc researcher at Mariya Moosajee’s lab around the same time as Viv was working there as well, so we worked quite a lot together.

And I’ll be showing today part of the results that we had from my main project at the time, which we now published as well end of June, I think it came out in the journal, so it’s also open access if you’re interested.

And I’ll be talking about Amlexanox as a possible new therapy for aniridia and how it rescues PAX6 levels in stem cell models of this disease.

So, just a quick introduction. We all know about this and Viv also introduced it before. We know that aniridia or congenital aniridia is rare and inherited in a dominant way.

It’s mostly characterised by the underdevelopment of the iris. That’s what really gives away the initial symptom.

We do know that there’s also macular foveal hypoplasia, nystagmus, and then this progressive appearance with cataract, glaucoma and corneal keratopathy.

We do know that it’s mostly related to mutations or changes in one of the copies of PAX6.

So we all get one copy of each gene from the mum, or the other from the dad, one problem in the PAX6 gene is enough to cause aniridia.

And the general mechanism that we think is behind it is called PAX6 haploinsufficiency, which means that because one copy is affected we have less amounts of PAX6 in the cells and that’s what really causes the problems.

So we have only one copy of functional protein as I’ve written there.

We also know that the majority of mutations affecting the gene are nonsense, like Vivienne mentioned.

And these nonsense mutations, what they do is there’s a little letter change in the DNA code, and that is enough to introduce a stop signal. So then, when the machinery is reading the DNA, the code, there’s a little stop signal there and the machinery stops, and that protein is affected.

So there’s a type of therapy that addresses this kind of problem called nonsense readthrough therapy.

So there’s two options when it comes to nonsense mutations.

They introduce the stop signal, so then one option is the machinery continues and it produces this shorter, we call it truncated protein.

The other option is the machinery doesn’t even read it at all. And that mutated code is degraded by a process called nonsense mediated decay.

I was trying to find a simple scheme, but I don’t think we can see it here very well, but that’s pretty much what I’m explaining now.

So there’s these compounds called translational readthrough inducing drugs that act on this machinery, and they sort of confound it so that when it gets to that early stop signal, instead of stopping the machinery, they confound it and allow the process to continue.

So there’s an amino acid introduced there, and then a full length normal protein comes again. So that’s what we’re trying to do here in this case.

Ataluren, as I think some of you might have heard, also kind of works in the same sort of way.

So we work with stem cells and a specific type of stem cell, which is human induced pluripotent stem cells. So these are stem cells that we generate in the lab from a sample of skin or blood or urine even.

We can get those cells, we give them specific compounds that force the cell to rewind and go back to a stem cell way. And then we get those stem cells and we force them into different cell types.

So from that stem cell we can grow, for example, neuronal cells, we can grow pancreatic cells. We can grow, in our case, eye, cornea or retinal cells.

And that’s what we do and this is a great way to study these kind of genetic diseases, because from the same sample of the patient we end up being able to grow cells from all different parts of the body. So this was the concept behind it.

We had two patients that we got samples from, they both have nonsense mutations. That was the location of each of the mutation in the PAX6 gene, but they are both causing this stop signal.

So the strategy for this research, we do have the two aniridia patients. We did use two controls, so patients who did not have a PAX6 mutation. We generated these stem cells, and then we developed two models.

So we wanted two models where we could test these drugs on. One which we called early retinal organoid, or an optic cup, and then limbal epithelial stem cells, which are the cells that maintain the corneal epithelium and we know the cornea is quite affected in aniridia patients.

So we tried to model a bit of the back of the eye and a bit of the front of the eye to test the drugs. And then we use them for the drug screening.

So we tested four different compounds.

G418, which is one of the most famous of this class. It’s very, very effective but quite toxic.

And then we did Amlexanox, which was new, recently reported. Also FDA approved, which was quite important for us, because if this goes further it actually saves a lot of time in clinical trials.

We tested Ataluren, which we knew already was quite good for PAX6 and it did even go in clinical trial.

And then we tested a very recently identified one called DAP.

So what we did first, the first model, we grew these cells for 35 days. So we induce the stem cells into these kind of 3D shape organoids, we call it. They’re very distinctive.

Here you cannot see, but there’s like this golden kind of layer outside, which is the neural retina, and then there’s always a black bit in the middle, which will then become the retinal pigmented epithelium, but that’s why it’s dark.

So this takes approximately 35 days to get to that stage. And then we dose them from day 15 onwards.

So we tracked this process and we saw when PAX6 was being turned on, and then we started working with the drugs. And the first thing we saw, even before we test the drugs, was that by measuring the levels of PAX6 in the aniridia versus the control.

So in here the controls are in blue here and throughout, they will always be the blue bars or the blue lines, and the aniridia will be in red.

So what we see is, throughout, there’s a trend of a reduction of PAX6 levels, compared to the controls, and it’s only really significant. So it’s like the results are consistent enough for us to say, okay, it is indeed less than in the controls around day 35 so that was our end point.

And what we also saw is that we actually don’t see just half the levels of PAX6. We see less than half, we see sometimes three, even four times less.

But anyway, what we saw, and this was actually consistent with what we know from mouse, is that even though there is this less PAX6 it is enough to form the structure.

So this structure kind of mimics a very early developmental eye, and we know if we have no PAX6 at all, the eye is not formed, but people have one copy, one good copy of PAX6, and that is enough to have the eye formed, at least on these very early stages.

And we see this in the mice as well and we see this here in these cells, that even though we have less PAX6, they do form the same kind of structure. And they do have these very two key layers that I mentioned – the golden ring, which is in the neural cells that will become the retina, and the dark bit, which is this mid F marker, which will become the pigmented part of the eye.

When we dosed it, so with the four compounds, G418 and actually surprisingly DAP were cytotoxic, so they killed all the cells. G418 we sort of expected, because we knew there were a lot of toxicity reports. Amlexanox, Ataluren were quite alright. And that’s just a figure of how the structures look.

And then here, when we quantified PAX6 levels, we see that… so the blue is again the control, the untreated, so UT is the not treated aniridia cells, so you see there’s a reduction, right?

And then when we dose them with Amlexanox and Ataluren, we see that the amount of PAX6 increases. And Amlexanox is actually significantly increased, so we are confident of this. Ataluren showed a bit more variable results.

We also tested if then, when we increase PAX6 levels, if that PAX6 is actually functional, if it goes and does what it should do. So we checked a few markers that we saw were affected in the untreated cells.

For example, this VSX2 which is, I won’t go into detail, but we saw that there was more VSX2 in the patient cells and in the controls.

And then when we dose them, this is just a figure of the fluorescence, but focusing on that graph there, you see the dosed one, so the two last bars, they go lower, to similar to the blue dots, so to the controls. This was quite encouraging for us. And we did test other markers as well in the paper, but I won’t show here.

So this shows that we increase PAX6 levels when we dose them, and that it seems to have a good effect, because it lowers certain markers to levels similar of the controls.

The second model, so the cornea model, was about 15 days of cell growth. So we see these cells growing on a dish which have this kind of cobblestone appearance.

Again we tracked for the days when PAX6 was on. And then we saw that it was on already at day 15. We dosed them from day 13 to 15 with the drugs.

Again, first of all, we checked the PAX6 levels. We do see, again, a lower expression of PAX6 compared to the controls.

At this point, for the cornea model, we tried both aniridia patients, so you will see aniridia one and aniridia two are the two different patients. For the first model we could only grow the first one patient’s cells.

So when we checked PAX6 levels, we do see again a reduction, and we do see again less than half reduction.

And then when we dosed them, we see first G418 cytotoxic again, so this really confirmed that we could not use this compound. And then we actually see a good increase in the PAX6 in Amlexanox, Ataluren and DAP. So all three of them seem to have increased the levels of PAX6.

When we look at the functional, so when we saw again if PAX6 was functional after we dose, we saw that DAP did not really show. So we are wondering why, but there’s also discussion about that I can mention in the end.

But we do see that Amlexanox again, this was a marker that we saw was increased in the aniridia cells compared to the controls. And these two concentrations of Amlexanox lowered it again to very similar levels of the controls. DAP, like I said, did not cause a change. Ataluren was again a bit complicated.

And this actually we could replicate on the second patient as well, so it showed that it works for more than one mutation.

So to conclude, we grew these models. This technology was not done to aniridia patients yet. And we think, well I personally think, it’s a very good model to use, not just to understand how the disease works, how PAX6 acts in these cells, but also to use these cells as a drug screening.

And then we tested these four drugs. To recap, G418 was toxic in both.

DAP was toxic in the organoids in the optic cups. It did seem to induce PAX6 readthrough in the cornea cells, but we did not see a clear functional rescue, so we are thinking it’s probably because of the amino acids that the drug allows to be introduced.

Ataluren, again supporting the clinical trials and the nice pre-clinical data before. It seems to induce PAX6 readthrough. We did have some variability, but I do think it was more technical than biological.

And Amlexanox we propose as a promising candidate, because it did increase PAX6 levels in both models, and we do see a functional effect, we do see that it’s changing something in the cells for the better.

So just to wrap up, so I’ve since left the UK actually, I’m now in the Netherlands, but I will be continuing working in aniridia.

I actually got a grant approved to grow corneal organoids. So cornea 3D structures that will have not just one cell type of the cornea, but actually we can model all three main cornea layers. And we will be using, for me, some very cool lab techniques to really understand how PAX6 affects the cornea development, but also the cornea maintenance. And then hopefully also use these for drugs and for drug screening. So I will be staying around for two years in the field at least.

And yeah, these are everyone that helped me in this project, Mariya Moosajee and Vivienne as well from UCL and other colleagues. Now at Radboud University in the Netherlands, Jo Zhou is my current supervisor. And then people in Finland as well who helped with the cornea cells.

I think Aniridia Network also funded our projects and were also very much involved. I think a lot of you I ended up meeting throughout. And thank you for listening.

[Applause]

[James] Thank you. Any quick questions? If I can ask you to keep it just one question each please. Any queries?

[Man 1] Yeah, just a quick one scientifically. Has any science been looked on around using AI and would it?

[Dulce] Using, sorry?

[James] AI.

[Man 1] Artificial intelligence.

[Dulce] For aniridia?

[Man 1] Yeah.

[Dulce] I know there’s a very… at Moorfields and UCL there’s Pearse Keane, he does a lot of very cool AI studies for the eye. I’m not sure it’s aniridia specific, but definitely he’s the top. I think he’s at the forefront in AI applied to eye diseases and eye research. But I don’t know if anything’s been applied to aniridia yet.

[Man 2] Would the treatment work at any age? If you’ve given the treatment to an older patient, would it have any effect or would it just affect the sort of…

[Dulce] Well, I’m not really a clinician, so I’m really very careful in mentioning.

[James] Maybe you’re talking about the previous study in Canada?

[Dulce] In theory, PAX6 has kind of two main roles, which is one is in the development, so while the baby is being formed, and that part is very difficult to treat, right? But then it has this progressive part, which for example affects the cataracts, the cornea mainly. And that’s why we also wanted to grow cornea models, because we know as a therapy it’s easier and actually more quite relevant for the patients, right?

So, in theory, you could treat at any age the cornea part at least, but of course if you already have it and it’s too damaged and you have all these other complications, I think that would be very patient-specific, and talking with the clinicians and everyone to really decide when would be the best. Because the developmental part would be very difficult to treat in any way, right?

So of course then these progressive parts, the earlier you would start. This is something also I want to do in this new project that I’m starting, is really finding out the very early changes that happen in the cornea, so that we can hopefully find a way to clinically treat them as soon as they appear, or even before they appear to prevent them from appearing. But in that sense the cornea has more probability of success, so targeting the cornea would be better in that sense, or easier.

[Man 3] I think it’s slightly outside of the box, but a significant percentage of us will probably at some point require stem cell transplants. Now, would there be a viable scenario where you could have… you’re growing and forcing cells, the stem cells, to use your own cells to be made into new stem cells, so you don’t have the rejection and being able to correct them?

[Dulce] That has been explored for other diseases. Of course if you use your own cells, it has the advantage of not rejecting them, but also they carry the same mutation, right? So in theory they would also keep failing at… the problem could be appearing.

But there are strategies about that for other genetic diseases, where you could for example correct. Of course that’s very personalised, right? That’s very specific for each person, which is for example less attractive for companies and pharma.

[Man 3] I was just wondering, using what you’d said, and trying to think outside of the box, and how that could correlate to people and their own personal treatments which they have?

[Dulce] In an ideal world, yes, everyone would have, okay, I have this change that is causing my problem, we can correct it, we can grow your own cells in the dish, give them back to you, error free. But, yeah, in real life it’s a bit more complicated. But it is being done for other diseases, yeah.

[James] Any other questions? No. Alright, thank you very much Dulce.

[Applause]

---

Thank you to Glen for the video editing and write-up.