 All right, all right. So I need to share my speech. Hello, welcome everybody to Sasek's Vision series. My name is Jose Moya Diaz and today I have the pleasure of introducing our next speaker, Professor Sergei Picot from the Institute of Vision in Paris. So today Sergei will be giving a real exciting talk titled, Genetic Based Brain Machine Interface for Visual Restoration. And I will give you a brief introduction about Sergei's background. So Sergei is the director of the Paris Vision Institute since January 2021. And after a PhD in Marseille and studies in Frankfurt and the University of California in Berkeley, he returned to Strasbourg and then to Paris to launch his own team on retinal information processing, enlarging then the focus to neuro-protection. His team, for instance, revealed how an antipyletic drug is leading to retinal degeneration. More recently, the team has moved to developing strategies for restoring vision in blind patients. The work involved novel material for electrodes like graphene and diamon or even based camera for visual information processing. His team has validated the photo, photovoltaic and wireless retinal implants ex vivo and in vivo on the blind primate retina paving the way for clinical trials by the company Pixium Vision. As an alternative to retinal implants of the genetic therapy was the evaluated on rodents and primates, opening the path toward the clinical trials for the company GenSight biologics. The team of Sergei is now moving toward visual restoration at the level of the visual cortex for patients with optic neuropathies. So Sergei, thank you very much for accepting our invitation and we are really pleased that you will give a talk with us today. Hello, oh, Sergei. Sorry, you are, you are, okay. So... But I hear you three times. So there's a delay when you were talking about... Yeah, there is a delay. There is a delay, exactly. There is a little delay because of the streaming connection, but you can now kind of share... Maybe he has opened the YouTube, so he should close it. All right, so, yes, it's ready. So, yeah. So, you can share your screen now and we can go. Can you see my screen? Yeah, yeah, perfect. I think you are in presentation mode and maybe, yeah, you should, you should open straight away from the PowerPoint. Is it okay now? No, you're not sharing at the moment, sorry. You need to share it. Yes, perfect. Okay, so, yeah, so thanks Sergei again for accepting our invitation and welcome to SAS exhibition series. Okay, sorry for the delay and slight problems. So I will now... Okay, so I will introduce the difference. I will start to present my presentation on all visual restoration. It's very disturbing because I have an echo of myself. He may have opened the YouTube, tell me. So, so, so, so probably you have the YouTube window open. So what I suggest you is that maybe you should close the YouTube window and just carry on with the Zoom presentation. Thanks. Okay, so let me introduce the strategy for restoring vision and especially the new genetic-based brain-machine interface. So I have some conflict of interest that are presented here and let me, so first I wanted to introduce the vision, the Paris Vision Institute and our interest in Paris is to prevent blindness and to restore vision. So we know that losing eyesight is what is the most problematic handicap because you're not only losing vision, you are also losing your autonomy and this is what when you ask patients, they are ranking blindness as the worst condition, as you can see here. And this will increase in time because as you see, I mean there are an increase in blind patients that should double in the next 20 years and the number of visually impaired patients should triple in this time. So this is due to aging, especially aging due to age-related macular degeneration and to glaucoma, I mean the increase in blind patients is due to this two disease. And so we want to try to see how we can prevent these disease. So the idea is that at the back of the eye, you either lose the photoreceptor and this is age-related macular degeneration or some rare disease, or you lose the written gain cells that are communicating the information to the brain and this is diabetic retinopathy or glaucoma. So at the Vision Institute, we're trying to develop therapeutic strategies. So either for complex disease, monogenetic disease, this was found on gene therapy for labor optic neuropathy or we are trying to impact on the neurodegenerative process like with the broad-derived conviviality factor as a gene-independent therapy or some kind of cell therapy using cells that are produced from stem cells and what I want to introduce today is what we do when patients are becoming blind and how can we restore some vision in these patients. So the idea of restoring vision is that if you lose photoreceptors, and you still have a written network that is still present and with the bipolar cell and especially also the gain cell, that are sending the information to the visual cortex. And different strategy for restoring vision have been developed from sub-retinal, epiretinal and optic nerve stimulation. And so I will come back, especially on the sub-retinal stimulation. But also, the idea is that when you lose the written gain cell, there are no connection from the eye to the brain. So if you want to restore some vision, you need to do it at the level of the visual cortex. And I will end on this, my talk. So the question of restoring vision is what should we do? And as you can see here on these first images is that if you have a matrix of 60 pixels, you cannot recognize faces. It's very complex, but you already can start to recognize faces with 625 pixels. And as soon as you go above 2000, it's clear that you can see the faces. So we need to restore at least several hundred pixels. And another question is the refreshing rate of these images. We know that when we watch the TV, it's a video rate of at least one image every 30 milliseconds. So this is what we should aim at. So when we have lost photoreceptors, this is the case of head-related macular degeneration. And in this case, the patient are losing the central vision where we have a central scosoma, but they keep peripheral vision. And then there's other disease, rare disease, like retinitis pigmentosa. And in retinitis pigmentosa, the patient are losing first peripheral vision. And unfortunately, they are also losing the central vision. So this patient with retinitis pigmentosa were the first patient to be included in clinical trials because they have no sight at all. And then the patient with head-related macular degeneration were then later involved if the device can provide better resolution than peripheral vision. So this was the case for this implant. So I will not introduce too much to the other implant, but this retinal implant called PRIMA was developed by Daniel Palenker at Stanford. So you can see this implant, it's non-wide at all. It's a fully independent implant. Each unit of this implant is produced by a central electrode here, a stimulating electrode. There are peripheral photodiode, infrared photodiode, and a ground grid here around each unit. And so the size of this unit, it are 100 microns. And as you can see here, you can have several of these units. And on the first clinical trials, there were 37, seven, three, seven, eight of these units. So what Daniel Palenker had shown is that if you put such an implant here above the retina of a blind right in such a recording units, you can record the ganglion cells, the spikes of the ganglion cells. And especially when you stimulate with infrared the implant, you can record these spikes into the retinal ganglion cells of the blind retina. So it's possible to reactivate the cells and especially the bipolar cell that then communicate to the retinal ganglion cell the information. Then he showed the in vivo that if he stimulates the implant with stripes, he can more or less measure the visual evoc potential and thus measure the visual equity of the animal showing that in fact the visual equity with the implant is fairly close to the visual equity of a normal animal in blue. So with this, I mean, or startup picture vision convinced Daniel Palenker to produce these implants in France. And so they were produced by the startup company and the question was, can we transfer these implant to patients? And so before we would transfer this patient to implant to these implant to patients, we decided to test them on non-human primates because the size of the cells in non-human primates are closer to the ones in patients. And also the types of cells are also very similar. So this is what we did. And we first tested these on isolated retina but because we have no blind retina, what we did was to take a normal retina of an animal that was atonized for other reasons than our experiments. And with this retina, we sliced it down in the thickness so that we could generate a blind retina. Then we would put the implant here above the retina and record the activity of the retina gain in cells with such an electrode array. So with this, we could stimulate with infrared the implant so that it would generate current in the tissue and then test whether we had spikes induced into the retina gain in cells. And the response was yes, we can record spikes into retina gain in cells. And what was very interesting was that if we attribute each of these spikes to a gain in cell, we can show that some gain in cell like this one here, this gain in cell is stimulated only by pixel 60 and not by the neighboring pixels indicating that we had a high spatial resolution with these implants because two neighboring pixels were not stimulating these gain in cells. After this ex vivo result, we went to in vivo experiment to test these implants. So what we did was to introduce the implant below the retina of non human primate. You see here the optical section of the retina. And what you can see is that the photoreceptor layers are disappearing above the implant. And this is clear here. If we label the photoreceptors at the level of the implant here in green, they disappear as compared to the neighboring area. So in fact, this area just above the implant is becoming a blind spot. And so if we generate a behavioral test where we asked the animal to watch a central spot and then present a peripheral spot asking the animal to generate a saccade toward the peripheral spots, he can generate saccade in all directions of sight except at the position of the implant because this is a blind spot. But if then we do infrared stimulation at that position we can see that the animal is no generating a visual saccade in this area. So this indicates that the animal is staying in infrared with the implant but not in visible light. This was done with spot of light that was sufficient to activate only one single unit of the implant indicating again that we have a high resolution with the implant. So this enabled us to go to clinical trials. And so the patients were implanted. You see here the implant in an area that was in fact the area, the scotum of a patient with age-related macular degeneration because they expected visual acuity was higher than the peripheral vision. And in fact, you see the patient here are wearing goggles allowing to keep peripheral natural vision. The rod here is taking images with a camera above the rod. And then at the back of this rod you have a video projector projecting an image in infrared on the implant here in the pink ellipse. So here you see that the patient is able to read letters on the movie. And in fact, in this case, the patient was tested with some goggles that were completely black so that he could not read with the peripheral natural vision. The visual acuity is close to one over 20. So it's 20 over 460 or 20 over 565. And what is also very interesting is that the patient can fuse the artificial infrared vision with the peripheral natural vision. So with this technology, so the patient can recover some useful vision, but we cannot reach cellular resolution. Each unit is of 100 micron. So it's much above the size of photoreceptors which are approximately a few microns in the very center of the retina. And we have only 378 units on the implant. So how could we improve and try to get to cellular resolution? So it's why we decided, so this is just to show you that we can reduce maybe the size of the implant, but at some point we cannot go further because the stimulating electrode will come close to the ground rate and we will not have any current getting into the tissue. So the technology that we developed to try to reach cellular resolution is optogenetic therapy. So the idea is to use protein opsin of algae because you have a photosensitive opsin in algae that are ionic channel. So the idea is to take the genetic code of these opcins to introduce them into an AV, inject the AV into the eye so that the AV would diffuse to the retina. And then neurons will start to express these opcins. So this was done first by Zalpan in Detroit. He was the first to introduce channel relapsing to into retina-gain cells. You can see the expression here in the retina of the one retina, the one mass with no photoreceptors here. And you see this expression here on the section into these gain cells. And when he shine light onto these retina-gain cells, you see that the gain cells are generating spikes during the stimulation, the light stimulation. So it's possible to restore some activity in the gain cell using this optogenetic top. So we decided to move this towards clinical trials. So we introduced not channel relapsing but catch into retina-gain cells. You see the expression in non-human primates in the peripheral ring, macular ring, simply because at the macular the neurons are pushed aside. And so if you take a piece of this retina and put it in an electrode array, you can record here on each of these electrodes the spikes of retina-gain cells. And even if you introduce some synaptic blockers in the chamber, you can record the spikes. So the spontaneous activity. And if you shine light, you see that you increase the spiking rate of the retina-gain cell and that this increase in spiking rate is correlated to the position of the transfected gain cells. We see that we have approximately one-third of gain cells that are transfected. In this case, we use catch, which is sensitive into the blue range and with a maximum at 460. So we decided to move to redshifted the opsin, catch the crimson. You see crimson is the most redshifted opsin. And we did this because we know that the blue range light is toxic and we need a high amount of light because these opsin are less sensitive than all photorecitors. And so we decided to test different vectors and we selected AV27M8 as the best vector. And the best opsin was crimson air fused with TD tomato. In fact, the fusion with TD tomato was increasing the expression of crimson. What you can see here is the recording of the retina of a non-human primate. You can see that the activated electrodes are really correlated, well correlated to the position of the moving bar. In this case, the image was slowed down four times so that you could see the moving bar. And so we have a fairly good spatial resolution because you see that the activation was really moving with the moving bar. And then if we look onto the next graph here, you see here the duration of the stimulation. And you see that with 20, 30 millisecond of stimulation, we reach almost the plateau here of the activity of the retina ganglion cell indicating that we can stimulate with images presented every 20 to 30 millisecond. So this enabled us to move to clinical trial. And so first year we also controlled that we had no immune reaction in this non-human primate because we have to realize that we are decorating retina ganglion cells as algae with the ionic channel that is in fact getting outside of the retina ganglion cell. So in order to move to clinical trials, we had to generate goggles so that there would be a camera to get visual information and then project at the back of the goggle, light at 600 nanometers, which is the peak of sensitivity for crimson. And in fact, we need those goggles because we need to normalize the light level in a high range to activate crimson. So the camera that were used are kind of special cameras that are not taking photos every 30 milliseconds, but instead each pixel is completely independent. And you see that each time the light intensity is increasing and passing a threshold, the camera is sending an event here. So you have positive and negative events here and you see the kind of visual information that are sent to the camera with the positive and negative events and then you can actualize the intensity level by measuring the light level. So this is very interesting because there's no saturation even if you watch the sun. And so you can, the patient can watch in all direction. So no, these are the results of the first patient we're seeing by optogenetic therapy. So it's maybe the type of camera we use give this kind of vibrating impression to the patient. And you see, you can find objects like this staple box. You can also count objects on the table. And so with these type of information, the patient can also detect some very light contrasted objects. As you can see here with this bottle of alcohol gel. So we have been working on other strategies like introducing some options into what we call dormant cones because in the retina, we've seen that during photoreceptor degeneration, we have some photoreceptors that remains despite they lost their auto-signal. And the idea was that inpatient, a blind patient could still have these photoreceptors and it's possible to reactivate these photoreceptors as well. And then the information processing is much better than if we reactivate ganglion cells. So now I would like to switch to the other types of visual restoration when patients are looting the ganglion cell. Is it possible to restore some vision by directly stimulating the visual cortex? So this was done in a long time ago by Brinley and Lewis and where you see they did have electrodes on the visual cortex, transistors to communicate the information and that it was possible to stimulate different electrodes and the patient would report phosphine in his visual field so indicating that there's really a correlation between the stimulation of electrodes on the visual cortex and the perception in the visual field. Unfortunately, these devices were not stable in time so there are no yet commercial device available but recently there were some new development especially here from the company's second site where you see that with electrodes at the surface of the visual cortex, it was possible to stimulate these electrodes and so that the patient could perceive shapes. But as you can see, it was done by some kind of sequential stimulation of the electrode and taking a fairly long time. So that it's not possible to use it in a very dynamic mode. To avoid this kind of long sequencing, sequence of stimulation, others have thought that as it's possible to activate with smaller current in the depths of the visual cortex, you see here that in order to either delay a saccade or elicit a saccade, if you stimulate in the depths of the visual cortex, you need less current to induce the saccade or to delay the saccade. So Peter or Selma, for instance, introduced such penetrating electrode in the visual cortex and so that it could record first where the stimulation of the electrode was generating a phosphine in the visual field and then the idea was to demonstrate that if you then shine a letter or what would be the form of a letter recorded by this, when you present a letter in the visual field, what are the electrodes that are activated on these arrays, you can then stimulate the same electrodes and try to see whether the animal can report the presentation of the letter. So here is the test. Normally, if you present a letter in the visual field, the animal can generate a saccade afterwards when you present two letters. So in some way, he's telling you, yes, I've seen the T letter in the previous. So when you stimulate the electrode that were activated when you present a letter, what was very interesting was that in this case, the animal can also report having seen the correct letter. So this indicates that stimulating in the depths of the visual cortex can generate perception. And so this was also achieved in-patient by the group of Fernandez in Spain, while he has also introduced such penetrating electrode in the visual cortex and the patient can then also report seeing letters like E, L, O, and indicating that stimulating in the depths of the visual cortex might be much more efficient. The difficulty with these penetrating electrodes is that they generate some kind of damage and that in fact, the visual perception is lost, is likely to be lost after some time because different studies have shown that there's a big reaction, glioses around the electrodes with such electrode rays. So at the Vision Institute, we have moved and tried the first optogenetic, but it was not very positive. So we have moved to another strategy which is called sonogenetic type. So what is sonogenetic therapy? The idea is that instead of using an op-syn, we are using a protein that is mechanosensitized. Take the genetic code of this protein, introduce it to an IV, inject the IV so that neurons will start to express this mechanosensitive protein. And then the idea is to generate an image, an ultrason image onto the brain which will represent in fact what is captured by the goggles. So first, yes, we can project that image because you can nicely make some image, ultrason images as it's done very often to see babies in models. So if it's possible to do images very deep into the body, it should be also possible to stimulate with images also in the brain. So this is also illustrated here. I mean, when you record the activity, the blood flow by ultrason imaging into the brain. So first, this is the classical images of MRI. And you see that functional ultrason imaging, you can have a much better image. What we've done with this kind of ultrason, fast ultrason imaging was to shine light into the visual field of a primate, record the visual activity. And you see that with either 500 millisecond or two seconds, we have similar kind of response. And we can show exactly where in fact, in the visual cortex, the neurons are activated when we present a visual stimulus in the visual field. So with this, we were able to generate retinotopic maps either to an eccentric stimulus or to an angular position. So this really clearly indicates that it's possible to do imaging in the brain. In this case, we had to remove the skull, but we can leave the jaw. So it's possible to do nice imaging. We also tested whether we could see the ocular dominance column, which were only reported in the depths of the visual cortex by autoradiography. And what you can see here is that with ultra-fast ultrason imaging, we can visualize these ocular dominance column with very good precision, even very deep here, as you can see in the visual cortex, like it was possible with autoradiography. So we have a good imaging solution with ultrason. Can we have also a good stimulation of the visual cortex? So what we did was to inject AV into the visual cortex of rats. We could see that we have a nice expression into the cortical neurons. So the idea was then, can we recall these neurons when we do ultrason stimulation, either with any cog array, as you can see here, or with a penetrating electrode into the brain? So here you can see the results with a penetrating electrode. You can see that we can measure, of course, visual response to light stimulation. But we can also recall some ultrason response if we have injected the AV previously. And you see that the latency here, which was at least 50 millisecond, is much shorter when we do direct stimulation of the visual cortex, of course, because the neurons are immediately activated. There's no need of transfer of information from the eye to the visual cortex. If we did not inject the AV, of course, ultrason are not generating any activity in the visual cortex. We can repeat the activity. You see that we can stimulate with 20 millisecond and we have a nice activation of the visual cortex. We can also stimulate with different frequency and up to 13 volts. Now, can we, so this gives us the high temporal precision of the stimulation. What about the spatial resolution? So to measure the spatial resolution, we use a cog arrays at the surface of the visual cortex with these cog arrays, we can measure visual response, which are quite classic. Then if we have no AV injected into the visual cortex, there's no response to ultrason. But if we did inject the AVs, we have nice response to ultrason stimulation. And you can see them here. And then you see that depending on the ultrason intensity, we have different types of response. And also if we move the stimulator, we can move the area of stimulation. So meaning that we have a high, we have a good spatial resolution also with this technology. And a resolution that would be highly compatible with visual restoration. Finally, the question was, do we have visual perception? So to test this, what we did was to train the animal to report light perception. And you can see that with time, so the animal can really respond efficiently to light perception. So in order to measure the response to light perception, what we do is we do the stimulation, we wait for some time and we provide water. And we see that during this period, the animal will start to leak, in fact, the tube. And you see that after four days, I mean the animal has learned that it will receive some water and start to leak the tube. Then we did a test by stimulating the visual cortex with ultrasound. And in this condition, what you can see is that the animal is leaking the tube exactly in the same way that it was doing with the visual stimulation, indicating or reporting that it perceived light during the ultrasound stimulation of the visual cortex. What was really interesting as well was that when we had the visual stimulation, the latency was almost 300 milliseconds. But if we perform the ultrasound stimulation of the visual cortex, the latency for the behavioral response of leaking the tube was more in the range of 200 milliseconds. So shorted because of course, we had no transfer of information from the eye to the visual cortex. So this was shortening the behavioral decision to leak the tube. So in conclusion, I've tried to show you that we can induce visual restoration in blind patients with age-related micro patient. The best visual acuity was with Prima. It's close to the threshold of legal blindness. The alternative solution is optogenetic therapy. And we measured that the theory for optogenetic therapy could be visual acuity close to 20 or 249 or even 20 over 72. So we still need to define in-patient what is the actual visual acuity that they can reach. It's still a non-going clinical trial. I forgot to mention that the patient included our patient with retinitis pigmentosa. And the patient you saw was blind for 14 years. And when you recovered this vision and our future project is really visual restoration at the cortical level by sonogenetic therapy. So with this, I would like to thank all my colleagues, especially Sarah Coedini, who did the sonogenetic work. Matthew did the optogenetic at the visual cortex. Greg did all the optogenetic part onto the isolated primate retina. Kevin Bled did all the imaging by faster tracin on the visual cortex. And Polori did all the tests on the implant on the human primates. And Fabrice was very helpful in all the primate studies. And this was done in partnership with colleagues like Denise Dalka, Yens Dubel, Riyad Benosman, and of course, Sose Sel who has really always put a lot of effort so that we could develop this project on visual restoration. And also the company Ecolies for faster tracin imaging, gene site for optogenetic therapy, Pixium for the prosthesis, PrimaPostasis, and our colleagues, Mikael Tanteur for sonogenetic therapy. Yannick Lema was the clinician for the studies on visual prosthesis. And also Daniel Paranker who generated the implants. And Smonberg was also very helpful for all the optogenetic therapy and a burden provided crimson. So thanks a lot. Okay, Fredette. Thanks a lot for this exciting talk. Really, really interesting. So we are now waiting for questions from the chat. So actually there is a question from Antonio Ninojosa Garcia which is asking that what are the advantage of retinal stimulation versus cortical stimulation and which one do you think is more promising? Well, for cortical stimulation it's quite complex because you need to make some intervention onto the visual cortex. So you need to open the skull. Whereas when you do some intervention onto the eye, I mean, you don't have to open the brain and so it's more accessible as well. So I think already from a kind of very simple issue of surgery, it's much more accessible. There are less risk because in the event of something is really going wrong. For a blind patient, you can remove maybe the eye. So I think the risks are less than what you would be having with a patient where you need to open the skull. Then in terms of resolution, I mean, this will be the report from patients. But in the eye, you have a high, I mean, you can stimulate a very large field of view. The difficulty is to target the cell in all these, this, the retina, I mean, it's not easy at the moment to have some gene therapy in a very large area of the retina. And for prosthesis at the moment, the prosthesis that we are using is only two millimeter by two millimeter. We could maybe have several as I was showing on one slide, but I mean, we will see. Patient will be reporting what to see and there will be a deciding. Okay, so we are, sorry for the noise from the background. We're still waiting for questions from the, from the people that is present on the live. I just posted the Zoom link if you want to discuss more in details, these ideas with Serje. And I have a question, Serje, it's about the first part of your talk. You talk about these implants that are non-wire, you know? So what's the duration of, or the utility life of these implants? And if you can use them again or not at all? Well, the idea is first that these implants are not powered by a battery. So it's very, it's very nice because as long as they're efficient, they can stay in place. And they are powered by the light intensity activating the photo diodes. So you don't need to have a wire to the battery or anything like that. You don't need also to fill the battery. So, so this is, this is quite, easy to manage. So I think this is one first thing. Then the stability, I mean, all the tests that were done ex vivo, like accelerating testing, did show that in fact, it was, it was well, I mean, well preserved in time that it was not degrading in time. So, I mean, again, it's only when it will be impatient for 10 years or more that we will know whether at some point it is degraded and no longer active. But it has been tested for several years already and it's functioning well. Okay, and I have another question. This is more a general question about the sonogenetic therapy you've been explaining. When you started at what are your future projections with the, you know, in the midterm and how you see, you know, this therapy could reach a real and accessible impact for the whole population. I mean, it's difficult to know. I mean, it will take, I think, longer because we are doing an intervention onto the brain. So you have to open the skull. You have to do AV injections into the brain. So you're also introducing a channel into the neurons of the brain. So there's a lot of questions there. But I think it could be useful in different neurological disease. We have to do tests. We are starting to do tests in non-human primates. And we hope to show that it's possible also maybe to, for the primate to see with this technology. And if we can show that, then why not start in clinical trials? Okay. So yes, so yes, so there is not more questions at the moment from the community. So we will be ending the stream and we can stay a bit in online, Serje, or chatting and discussing more ideas with people who wants to join us. And okay, thanks everybody who attended to this SAS exhibition meeting and see you next week. See you next time. Okay, very nice. Okay.