 that number of contribution in the function. And today, he's been visiting in San Francisco. The city is long and dense. You can find the line. Today, we'll discuss on perception. Thank you, Massimo. Thank you, Antonio. Thank you, Venki. And thank you, Anna, for putting together these wonderful programs in such a wonderful place. So I will try now to keep you awake. It's very hot, and probably you are feeling asleep. But I would like to share with you some recent data we got regarding what I will call the re-entry loop. So I'm going to tell you why. But so first of all, as I've been told to present in a very easy way for students, I don't know if the students are still there. Yes, OK. I would just like to make a very general presentation on the olfactory bulb, receiving olfactory inputs, and the olfactory cortex. The idea is that we should give up with this kind of cerebrocentric view about the brain receiving information from outside and processing this information and triggering output behaviors. It's more than that. You know that even some specialists are talking about the dark energy, which means that there are regions of the brain that are working together. And if you have to take into account one important thing in the olfactory bulb is that the major activity within the olfactory bulb is either intrinsic or is modulated by top-down fibers. And this is what I would like to share with you today. This vision of this olfactory cortex sending back information to the olfactory bulb and that the reason why I call this a re-entry loop, because we are still in the olfactory system. Everyone is looking for something. I'm very granular-centric, and I'm going to present all the data we have got recently on the granular cells activity, how they are firing activity, how they can be modulated in very precise context. So we're going to stick to these cell types. And we, of course, will have to integrate the other neurogenesis, as my lab has been heavily involved in this activity. So when I'm talking about granular cells, as you can see here, this was a recent work done by Kurt, if I can start the movie. All right, so no movie today. It was working initially, but here it goes. So here we are looking on a chronic imaging of granular cells. And this has been done in awake animals, behaving animals. And here you have the days after injections. And as you have seen, we have this possibility of tracking cell movements, spine movements, spine genesis, spine retraction. And interestingly, when Kurt came to some counting, we were surprised to see that about 40% of these spines are dynamic within two days. And if you do, and we have been doing this in a control cases for the cortex, if you do this in the same time window, you will see barely 1% of dynamic spines and motility. So one thing is important to take into account that these cells are here to look for partners. And what I'm going to show you that one of the major factors that will control this cell motility and connectivity would be the top-down fibers. So we have these granular cells being activated locally by the downright of mitral cells. We know that these granular cells also receive very proximal local inputs from the collateral axons. And this excitation is balanced by inhibition provided, as we heard today, as well as previous days, by the deep short axon cells that control the overall excitability of the granular cells. Now what we're going to see is that those granular cells, as well as short axon cells, as well as the mitral cells, will receive excitation from this pulmonary nerve cell from the olfactory cortex. And one of the major consequences of getting those proximal excitatory inputs will be to trigger action potential, therefore releasing GABA through a kind of synchronized way. So keep in mind that you have asynchronous way, like, for instance, lateral excitation coming from one granules to from one gemules, from one spine to another one. You have the local reciprocal synapses, but you have also another way to trigger GABA release by this excitation coming to the proximal site of the granular cells. And if you are able to look what are the partners of the output neurons using different way of labeling, what we heard already during those days that the mitral cells will reach several territory and in blue are these axonal projections from the mitral cells. But interestingly, in green are the downstream structure from the olfactory bulb. And you can see a very nice mirror images. Therefore, this is why I'm talking about re-entry loop, because most of the region of the brain receiving input from the olfactory bulb will, again, release information to the olfactory bulb, mostly to the granular cells, as we heard already, and as well as for the mitral cells. And we heard from Alexanders and others were using retrograde labeling, for instance, that the layer 2 of the piriform cortex is massively sending inputs to the piriform cortex, sending massive inputs to the olfactory bulb and mostly the granular cells. Here we are using adenoviruses that we are injecting in both the piriform cortex as well of the olfactory nucleus. And as you can see here, the terminals are mostly, not exclusively, but mostly, reaching the deeper part of the olfactory bulb, the granular cells layer. So we are reaching these images where you can segregate the local excitatory inputs, as shown here with the Tibet cray mice, where you have this major excitation impinging into the granular cells at the apical site of the granular cells, while the most proximal sites are receiving the top-down input, or the re-entry loop. And as a way to see this with the Tibet mice, as you can see here, the outside of the olfactory bulb, the inside is quite clean, while if you inject and you superimpose those two images, if you inject the adenoviruses into the olfactory cortex, both piriform cortex and olfactory AON, you can see that the major target of these top-down fibers is inside of the olfactory bulb, where, in fact, you find the other bone neurons shown here. So we are reaching this idea that the granular cells are receiving different excitatory inputs, and the question is what these inputs are doing for the function of the granular cells. Many people have been proposing that these top-down fibers control the dynamic gain, play an important role in synchronizing territory, distant territory, and the importance of beta versus gamma erasings to control the olfactory information processing into the olfactory bulb. Encoding nonsensory information, we heard from Alexander's and others, value, for instance, are important for the functioning of the olfactory bulb, and they can be carried by these top-down cortical inputs. From previous work, we have seen, as well, from Noam, actually that these top-down fibers could play a very important role with attention for finding a pheasant for a dog, or when he is training his students to find the chocolate place, where you can see exactly the same way the dog is finding his targets, his students here showing the same similar behavior to find the chocolate places. And, of course, the top-down cortical inputs are very important for Venki when he's doing wine testing. And ordering the identification is probably one important function, as well, that is under control by these top-down cortical inputs. So this is a very recent picture. This was at Leitre, Venki, and yes, yes, this is yours. It's very embarrassing, right? Sorry, it was Leitre, yes, correct. But Venki had a wonderful wine tasting, and he's going to tell us a lot later on. Right, OK, so let's see now how the granule cells integrates both sensory inputs as well as the top-down informations. One way to do this is using adenoviruses, or lentiviruses, to label granule cells. And with DT tomatoes, you can study, for instance, the cell movement of spine genesis, as well as you can feel the newborn cells with JCAMP. And as shown here, what you can observe with these chronicle recordings, he goes, that the granule cells, newborn granule cells, for instance, when you applied all different orders, they barely respond in these experiments, and this is now working the progress, while after a training and a go-no-go task, what you can observe that those newborn neurons will highly respond to the S-plus orders, the orders associated with the reward, while the S-minus orders is barely responding. At least you have this kind of pattern separation within the bulb, and we believe that the granule cell activity is a major role in this function. So let's see now, by using these adenoviruses that we could inject into the peripheral cortex to label and make these inputs light sensitive, and use these lentiviruses with GFP to look where the inputs are coming, and as you can see here for the PSD, for the newborn neurons, that these granule cells are heavily receiving gabbiologic inputs, in mostly into this proximal region, as we said before. We have been counting the number of contact that a given granule cells will receive from the peripheral cortex, and we reach this number, one to four synoptic contact per cells, which might be spared, whether this is relevant for the functioning of the cells, this is of course work and a progress, but you have to take into account that this major reentry loop is impinging into this area, and as we're going to see now, each of these domains, the distal proximal somatic as well as the basal inputs, will be challenged in a different way when animals are learning. And so we're going to see now animals associating orders with rewards. We're going to see, so this would be AE, and we're going to see animals with just order exposure without any reward, and just animals exposure to clean air. And as you can see here, none of the local inputs have been changed after this training for several days, but as you can see here, for the distal proximal and basal, just only learning increase the number of synoptic contact. By looking on slice physiology, we can shed light into these fibers and trigger glutamate input into the granule cells, and as you can see here, after learning, only the excitatory inputs were increased following this one week of order training. Tell me, I didn't really switch anything from here, but is there any way to make bigger? I can switch actually the computer if you want or make it bigger. This is embarrassing if you don't see anything. Already granule cells are tiny, so. That's, shall we, shall we keep on? Okay, so if top-down fibers arriving to the bulb are so important, one should find a way to regulate those inputs, and we have been digging for many modulators express, receptors express on these terminals and see whether once you block these terminals, whether you have different functioning to the olfactory bulb. One of these study came when we've been looking for cannabinoid receptors. Cannabinoid receptors mostly express within the olfactory bulb inside of the granule cell layer, where you find these cortical fibers. And so we decided first to record with field EPSP. These are the evoked responses when you shed light into the fibers and you trigger cell depolarization of the granule cells. And if you induce, if you infuse within the olfactory bulb, a CV1 agonist, then you switch off this excitation driven by the fibers. So, CV1 seems to be a very important or potent regulator that knocked down the excitation driven by these cortical fibers. Interestingly, with fasting, you can measure the increase of cannabinoid levers in animals after only 24 hours of fasting. You can inject CV1 antagonist into the bulb and show that you are reducing the food intake in animals induced by CV1 antagonist. And odor exploration was shown to be increased by activating CV1 receptors. So this is one indirect way to demonstrate that these fibers reaching the olfactory bulb might have something to do with food intake and odor exploration. Another way to look for the regulation of these fibers reaching the olfactory bulb was to look for GABAB receptors. And we have been looking carefully for these GABA receptors. And if you here make slice of the olfactory bulb and shed light for these fibers reaching the olfactory bulb when you inject the adenoviruses within the curriculum cortex, what you will observe by shedding light is an evoked responses within the granule cells. And as you can see here, by activating GABAB receptors, you can knock down this excitation driven by these fibers. Interestingly, if you record a granule cells, you can also observe a desynaptic inhibition. When you shed light, you don't get, oops, sorry, you don't get, okay, okay. So I will repeat again, shedding light in these fibers trigger something wrong. I don't think so. Yeah, yeah, okay, let's, I'm not going to touch it anymore. Maybe it's going to skip. So when we shed light to stimulate these fibers, we get excitation onto those granule cells, it's in an EPSP, and by adding baclofen into the olfactory bulb or in the slice here, you blow these excitations. And what I wanted to show you, I hope we're going to see it now, with, by recording a granule cell, what you can observe is also, you could observe an inhibitory event. This is a pure GABAGIC events. And if you apply NBQX, you blow this event, which means that this is a desynaptic inhibition. You are stimulating a short accent cell that will release GABAG onto these granule cells. And again, these events will be blocked by baclofen. So now let's see what's happened to the mitral cells when they are firing. So this is in vivo recordings. And you can see here the spontaneous activity of the mitral cells. And again, if you shed light to stimulate these top-down inputs, what you observed is that the pure feed-forward inhibition, glutamaturgic triggering granule cell activation will knock down this mitral cell activity. And if you applied baclofen, you totally block this desynaptic inhibition. You get a desynaptic inhibition. You block the feed-forward inhibition. And you can use GABAGIC, Cremice, just to check that what you're doing is correct, meaning that the shedding light into these fibers trigger glutamate release and GABAG activation of the presynaptic terminals will block the glutamate release. Okay, but as all of you know, things are never simple. And as we said before, if you shed light, you stimulate these fibers and you inhibit mitral cells, so pure feed-forward inhibition. But as you can see here, by shedding light, you also increase, in some cases, you also increase the firing activity of the mitral cells followed by inhibition, as you can see here. And by playing with baclofen, what you could observe is that whether the still, the inhibitory effect of the mitral cell activity is gone, the excitatory inputs to the mitral cells is still present. So GABAGIC receptors seems to be present in terminals impinging into the granite cells, but not onto the mitral cells. Okay, now let's see what's happened to the order evoke responses within the bulb using optical fibers and recording the mitral cells activity. And as this is the optical fibers here located on the top of the mitral cells and shedding lights to stimulate these top-down fibers. Here we should see an order evoke responses and the calcium responses from a population of mitral cells. And now let's see what's happened if we shed light into the pyriform cortex and see how much we're going to reduce activity of the mitral cells. And here by using different frequency, as you can see, this feed-forward inhibition, increasing at 33 hertz, and by applying baclofen, as you can see, this feed-forward inhibition is dramatically reduced. And a summary of these results is shown here with different frequency, as you can see. We reduce the firing of the mitral cell activity and baclofen reduce this feed-forward inhibition. So we have this pathway that trigger glutamate onto the granule cells, as well as onto the deep cortex and cells. And finally, the mitral cells activity, the spontaneous as well as the order evoke responses, are under control of these top-down inputs. So GABAVI receptors was described already by Jeff Isaacson and others on the sensory inputs. GABAVI receptors was known to be present on the terminal of the granule cell, but now we are adding another place where GABAVI receptors are located nearby this glutamaturgic release site. But as I said before, and as you all know, these stories are never as simple as we would like. When we have been using vigate prey animals and injecting into the puriform cortex, we have been able to look and found internal runs into the puriform cortex. That's nothing surprising, but when we look into the granule cells, we have been able to see some fibers reaching the olfactory bulb. And we're talking here about vigate prey animals. So we've been looking carefully on what kind of inhibition could reach the olfactory bulb. And here you have the YFP into the olfactory bulb with GAT67. And you can see that some of these dots coming from the olfactory cortex are indeed genus GABAVI terminals. And so when we've been able to use herpes varices for retrograde labeling, injecting the varices into the bulb and looking at different places into the forebrain, again, we found different places where we could find nice inhibitory internal runs and we're going to see them in a better shape here with this kind of spiny and as well as non-spiny internal runs. Mostly we observe those inhibitory internal runs into the puriform cortex, the anterior parts, as well as in the HDB. Those GABAVI inputs reaching the olfactory bulb seems to follow the lot, as you can see here. We could try to identify those populations of GABAVI internal runs, reaching terminals to the bulb by using different CRE mice. So the VIP, where you can see some of these neurons into the puriform cortex. But most of these GABAVI internal runs are stained with somatostatin and we haven't been seeing any parvalbumin internal runs. So this population seems to be further characterized to identify what kind of GABAVI internal runs are reaching the olfactory bulb. As we heard this morning, are we talking about different subpopulation of inhibitory internal runs, those who are dedicated for local inhibition into the puriform cortex or the GABAVI internal runs sending information to the bulb, completely different. These are completely open questions now. When we did some slides physiology, just to look what kind of targets we could find in the olfactory bulb, what we have been able to see mostly all internal runs within the bulb. So this is really puzzling because we are talking about inhibition reaching the olfactory bulb and reaching inhibitory internal runs. We have to carry more recordings to really discard or not output neurons from these GABAVI inputs. And what you will be expecting if these inhibitions coming from the cortex to the granule cells or the short accent cells is by activating using optogenity by activating these fibers one will expect this disinhibition. And as you can see here, but now we are looking in which scenario, in which conditions those GABAVI inputs are speaking to the olfactory bulb. Finally, we decided to look in which occasions, in which scenario these GABAVI terminals could be activating. And again, we are using the same task as I showed before with the excitatory inputs from the cortex to the bulb. We are using the same protocols here and using the optical fibers as you can see here. It seems that these GABAVI inputs to the bulb really bring some information about the values because you have here the S plus and the S minus which are completely different. And if you do the same kind of experiment but without any differences, reward or whatever, you don't see any differences. So there is a value information that seems to be brought by these inhibitory inputs to the bulb. Finally, we've been looking and about, sorry, about, disappointed about the function these GABAVI inputs could do regarding olfactions. We didn't find any real effect on odor discrimination by trying chemogenetics, by inhibiting or activating those fibers. Of course, maybe as we heard from Dimitri, this could be a problem of timing, but by making those kind of experiments, we couldn't find any obvious effect on odor discrimination. And when we stimulate these inhibitory inputs during odor detections, we barely see effects only at the very diluted concentration, as you can see here, by inhibiting these inhibitory inputs to the bulb. We could observe that these animals were better performer and we don't see too much when we are driving, when we are exciting those fibers. So all in all, we have this scheme where we know that these sensory inputs trigger excitation to these mitral cells. Mitral cells will send information to the olfactory cortex. We knew already in this scheme that the olfactory cortex sent back excitation to the mitral cells as well as other inhibitory internal organs. But to this scheme now, what we have to add, whether this is the same population or not, we don't know yet, but we have these GABAGIC inputs reaching mostly the local inhibitory internal organs within the bulb. And the reason why we have this loop, we don't know yet. And I hope this will feed some models and some thoughts and maybe some debates right now. So with this slide, I'd like to finish by thanking the people who have been doing the job. Gabrielle Le Poussé, who has been mostly driving this experiments with a PhD student, Camille, who has been really patient to inject. I must say that Serendipity was really the center of this experiment where we have been using adenoviruses with camcane estu and channel rhodopsin, thinking that we will be targeting pyramidal cells in the puriform. And guess what? I mean, mostly these granules cells mean inhibitory internal organs within the puriform cortex. We are stained with this kind of viruses. So you have to strongly believe in what you're doing before thinking that you have to throw all the data. But Camille was really patient enough and he did his PhD on that. And we had the great honor of having Andreas as a member of this studies. Antoine Nissant was the one doing the slice physiology and those were rotating students and Julien Grimaud initiated this work before joining Venki's lab. Right, thank you very much and I will be glad to take a question.