Scientist Stories: Abraham Beyene, Imaging Neuronal Chemical Efflux with Nanotube Hybrid Materials

This week for the Janelia Embell Optical Interest Group Bioimaging Seminar Series— its name has changed several times, so I just put it all in there— we're welcoming Abraham Veyhenne, who's a group leader in Molecular Tools and Imaging. He started just about 6 months before me, and he's doing some really exciting work. And so I'm excited to hear some updates on his work on imaging neuronal chemical efflux using DNA-carbon nanotube hybrid materials.

All right, so is it my turn now? Okay. Well, thank you for the invitation, and I'm happy to be here talking about these hybrid nanomaterials that are made from a conjugation between single-strand DNA and carbon nanotubes. They have some really interesting properties, and then I'm going to talk about some of the practical applications of these nanomaterials as well in the later part of my presentation. So I'm going to jump right in and then begin by introducing this technology that we use in the lab. We try to make biosensors out of this particular class of nanoparticles that we call single-wall carbon nanotubes. And this is depicted in this cartoon that you see here on the right-hand side, right here. And carbon nanotubes are more easily understood to be rolled sheets of graphene. So graphene itself is another really interesting nanomaterial. But I always find it easier to introduce graphene before I describe what a carbon nanotube is. Graphene itself is a two-dimensional nanomaterial. It's made entirely out of carbon, and each one of these vertices that you see here represents a carbon atom that is bonded to three other carbon atoms through this so-called sp2 hybridized carbon-carbon bonds and has, because of that bonding system, has this hexagonal planar structure to it. That is a fascinating material. It's made entirely out of carbon and is a single carbon atom thick. And if you think about rolling this graphene sheet into a very tiny hollow cylinder, you get what are known as nanotubes or single-walled carbon nanotubes, which are distinct classes of nanomaterials in their own right. The reason these two are different is because of this extreme curvature in the radial dimension, which is on the order of about a nanometer. That ends up giving these nanotubes properties that are very different and distinct from the parent graphene sheet. And there are a few properties that I want to highlight today about these nanomaterials that are going to be important for my presentation today. The first one is the fact that nanotubes are intrinsically fluorescent and they fluoresce in this range of the spectrum that we call NIR or SWIR. So that stands for near-infrared or shortwave infrared emission. And if you draw your attention to the x-axis here, you see emission happens in this 850 to 1,350 nanometer range. And confined within that NIR window, you see this collection of colors or peaks all emitting in this particular region of the electromagnetic spectrum. And the reason you see these differences in peaks and this fairly complex-looking fluorescence emission spectrum is because of the fact that photoemission itself from nanotubes is highly sensitive to the geometry of the wrapping of the graphene such that slight differences in diameter end up having a very profound effect on the electronic band gap structure that then translates into slight differences in photoemission. So takeaway from this is that this is a fluorescence emission spectrum. It's a fairly complex looking one because we have multiple collection of these nanotubes all confined within this window. And then secondly, you would also notice that the lambda excitation here is about 785, so you get this long Stokes shift which makes it very advantageous for imaging. And then secondly, the emission is also highly photostable. So, this is the red trace here shows a dispersion of a nanotube that more or less puts out the same fluorescence emission after hours-long exposure to the excitation light source. And that, of course, is in contrast to a lot of organic fluorophores and fluorescent proteins that undergo some well-known photobleaching mechanisms after some period of exposure to the excitation light source. And I'm showing ICG here, indocyanin green, which is a really bright dye, emits in this range of the spectrum, but of course undergoes photobleaching after some period of usage. So a takeaway from this slide is that these nanomaterials have some interesting photophysical properties, including emission in this unique range of the spectrum, photostability, long Stokes shift, and so on and so forth, all of which have made them intriguing for use in biology. My lab is by no means the first lab that has ever been interested in these materials. In fact, if you look at the literature of these nanomaterials in biology, it goes back for at least 2 decades. And there are a couple of reasons why researchers are interested in these materials. The first one is encapsulated in this image that I show you here, which shows the fluorescence emission from nanotubes in this blue trace overlaid with absorbance by blood that is shown here in the red trace. And absorbance by water shown here in the black trace. And you can see that this fluorescence emission straddles this boundary and targets this local minimum. That translates into some advantages for imaging. And to demonstrate that, I'm going to highlight this study. And there are many examples of usage of these materials for imaging, but I just want to highlight this particular study as just one example of these applications in fairly complex and challenging systems. This paper came from a lab at Stanford, Hongjie Dai's lab, was published in 2014. Here what they do is they inject these dispersions into the tail vein of a mouse and then they try to image the vasculature of the cortical mantle of this animal through the skull without actually having to introduce any optical window, which is, I think, quite a feat when you think about this. What is really striking for me, at least from this, study is a demonstration that the traditional NIR region of the spectrum in this 800 to 900 nanometer range really doesn't give you the type of signal-to-background ratio that proper shortwave infrared imaging does, which is around 1,300 to 1,400 nanometers. So you can see how— and trying to do this in the visible range of the spectrum is totally futile. You wouldn't— you would not be able to see anything. But you can see how this study tries to really demonstrate the utility of imaging range of spectrum and how it facilitates imaging in fairly complex and optically challenging preparations. Okay, so another one that I want to highlight today is this single particle tracking experiment. Again, nanotubes are, you know, nanoparticles in a way, so you can target them and do single particle tracking experiments with them in a number of systems. And this study that was published in 2014, for instance, targets these nanoparticles to the motor protein kinesin. And then you can watch this cargo getting carried around in the cytoplasm. You can do this extended imaging and biophysical characterizations of this particular motor protein. Another really interesting study for me is this idea that you can load these dispersions of nanotubes into the extracellular space. Where you do— you can do two things at the same time here, where you can watch these nanoparticles diffuse through the ECS or the extracellular space. And that gives you a readout for the morphology of the ECS. But you can also look at the diffusivities of these particles and get a sense for the rheological properties of the extracellular milieu through which these particles are diffusing. So, this is by no means an extensive representation of the literature. This is a very, very small subset. Of, I would say, a very robust literature of nanotubes being applied for imaging in biology and studying various biological systems. The way we want to use it in the lab, however, is not just based on the fact that they are fluorescent and therefore can be used as contrast agent. It's based on this idea that you can also make sensors out of them. And the principle behind making a sensor out of a nanotube is in a lot of ways analogous to how one goes about making a sensor out of a GFP, for example. And that idea is to conjugate that emitter, in this case the nanotube, to some kind of chemical matter that affords it to be a recognition motif, something that is able to change the local chemical environment of this fluorophore in response to some cognate ligand that this other chemical matter they have, that you have assembled on the surface, is able to recognize. And what you hope to get here is some kind of fluorescence modulation in the form of a fluorescence emission increase or decrease, or some kind of peak shift or solvatochromic shift that can be used as a readout to transduce whatever molecular recognition event may have happened in close proximity to these systems. Okay. So I know that I'm introducing a nanomaterial that I think a lot of you may have never heard of. But the takeaway from this is that these materials have high photostability. They emit in this region of the spectrum that is very unique. And very importantly, you can also make sensors out of them because the fluorescence is somehow also sensitive to the local chemical environment. So how do you square that? So how do you square the fact that these things can be both photostable in their emission, but can also be sensitized to make sensors out of. So I thought maybe I would spend this slide here to try to explain some of the similarities and differences in photoemission from these extended nanomaterials and photoemission from a small molecule fluorophore. So fluorescence can be easily encapsulated in this graphic here where you have an absorbance of an absorption of a photon that leads to some excited state of the molecule, which relaxes and then gives off a photon of a different energy. Then you detect that photon of different energy as fluorescence. In this representation, in this cartoon, this is precisely how you describe fluorescence from just about any fluorophore. However, when you take a closer look at the system, there are some really intriguing differences that are very important to understand why we can even make sensors out of these systems. And the reason is because the so-called excited state of this molecule is actually an electron-hole pair or exciton that gets generated on the surface when a photon gets absorbed, right? And the best way you can understand these electron-hole pairs is just regions of this relative positive and negative charge that are coupled electrostatically through columbic interactions and remain on the surface of these nanomaterials. And these are extended solids, meaning that you can have multiple absorptions of photons that lead to the emergence of these quasi-particles that we keep calling excitons. And excitons are not just, you know, figments of our imagination. This is not some kind of mathematical abstraction. These are real particles that generate on the surface. They occupy a physical space. They can diffuse about on the nanotube, and sometimes they even bump into each other and annihilate each other. And then every now and then, these electron-hole pairs will recombine radiatively to give off a photon of a different energy. That we detect as fluorescence. So this is how photoemission happens from these extended nanomaterials, including quantum dots and nanotubes and so on and so forth. Now, what is really striking about these materials and that something that makes them different from like a quantum dot, for example, is the fact that all of these things are all on the surface, right? So this is a quasi-one-dimensional material. This is just all shell. There is really no core-shell technology here. All of these carbon atoms, the conjugated pi-electron system, these electron-hole pairs are all on the surface exposed to the elements. And for that reason, just slight perturbations in the local chemical environment could end up having a very profound effect on the photoemission of these nanomaterials. For example, we know from an extensive set of literature that slight changes in the dielectric environment, like the relative permittivity in the immediate vicinity of the nanotube, ends up having very profound effects on these electron field lines, and that ends up having an effect on the photoemission of these materials. Subtle changes in whatever chemical matter that you have on the surface could also end up perturbing these photoemissions. And then this last point, which becomes relevant for my talk, is this activity-based modulation, where the addition of some kind of electron-active molecule could end up having a modulatory effect on this photoemission from these nanomaterials as well. So this gives me the opportunity to introduce this class of neurochemicals, class of nanosensors that we have been able to make for detecting a specific set of neurochemicals in the brain that we call catecholamines. So no surprise, so this sensor is really based on a nanotube that is conjugated to single-strand DNA. And the best way that I can describe this class of sensors is like this. Imagine a pristine nanotube that is sort of exfoliated and separated from the elements. So, think of it as something that is, like, suspended in vacuum where the local chemical environment is not deleteriously affecting the emission of these nanomaterials. So, in this case, we can imagine this intrinsic nano— this intrinsic fluorescence to have some kind of brightness. We can call that F naught that emits— that gets released from these nanomaterials. Now, um, nanotubes are really hydrophobic, right? So these are entirely made out of carbon. They don't have any kind of solubilizing side chain. And in general, if you try to take a powder of these nanotubes and try to solubilize it or dissolve it in any kind of solvent in the lab, you will be disappointed. These things are going to crash out of solution. So the only way you can actually have them remain in solution where you can take advantage of the photophysical properties is by decorating the surface, functionalizing it by different types of chemical matter. And, and one of these species that does a really good job of dispersing and creating colloidal stable suspensions out of these nanomaterials is actually single-strand DNA, right? So pristine nanotubes do not exist. They're all in kind of clumped and in powder form. But to disperse them, you have to decorate the surface. And one of those molecules happens to be, uh, dopam— uh, single-strand DNA that you can assemble on the surface. And the reason is because DNA has this amphiphilic property to has a negative charge in the form of a phosphate backbone, but it also has these hydrophobic nucleobases that can pile up onto the surface of this highly hydrophobic graphene lattice. In that way, it's able to straddle both the, you know, aqueous and non-aqueous phases and stabilize it in that manner. And what is intriguing, however, is that DNA in so doing also ends up quenching the intrinsic brightness of the flow of the, the nanotube for the reasons that I described, described in the previous slide, where you have these electron-hole pairs that get deleteriously affected as a consequence of this DNA being on the surface. So this itself is really fascinating. It is— the quenching is a function of the sequence chemistry and the sequence length, and there is a lot of information there, and some of which I'll be able to get into in the next slide. But that's this— that— but that's— this does happen. So this fluorescence quenching does happen. And then, um, uh, this quenching can be reversed for some sequences of single-strand DNA by the addition of a neurochemical that we encounter in the brain. These are called catecholamines, of which dopamine happens to be one. I'm showing that here. You see this reversal in quenching can actually be used for imaging dopamine and studying dopamine in real biological systems, which I'll get into in the later parts of my presentation. Let me show you what this actually looks like in reality, what the fluorescence emission from these dopamine sensors looks like. This is a dopamine sensor that we use quite often in the lab. This is created from a conjugation of GT6 with single-strand— with nanotubes. By GT6, what I mean is just alternating Gs and Ts, creating this 12-mer short oligo sequence that spontaneously self-assembles on the surface of the nanotube. Creating this colloid that is very stable and is also emissive. You can see this fluorescence emission from it. That is the baseline fluorescence. I want to draw your attention to the fact that this emission is highly quenched, so much so that some of the peaks are completely missing. Then the subsequent addition of dopamine rescues that fluorescence and then turns it back on. This is precisely what I mean by this modulatory effect of this particular molecule. In which this quenched fluorescence has now been recovered to a certain extent by the addition of dopamine. Let me just add a couple of points here to try to contextualize this. This is a reversible process, meaning that if I were to take the dopamine away somehow magically from this solution, this is going to go back to a quenched state. That's very important. Secondly, If you were to titrate the concentration of dopamine, creating some kind of dose-response curve, and try to compute some kind of apparent dissociation constant for this system, you'll find a value, something that is on the order of about 250 nanomolars. And then again, that gives you a sense for the relative sensitivities of these, uh, class of hybrid nanomaterials for sensing dopamine. So I know I'm throwing out a lot of information here, but I just want to like pause here and summarize what I've told you so far. I told you that there are these nano— inorganic nanoparticles that are made entirely out of carbon. They have this really interesting shape and size to them. They're about a nanometer in diameter, about hundreds of nanometers in length, quasi-one-dimensional materials that are very interesting for physics to them, emission in this unique range of the spectrum, for stability, and so on and so forth. And then I told you that that inorganic nanoparticle can be conjugated to DNA, which happens to be one of the central molecules in all of biology. To create this hybrid material that has some functionalities. This functionality in this particular case is ability to sense dopamine. This happens to be a unique ability of the hybrid material. As far as we know, the nanotube itself is not able to recognize dopamine and GT6 itself is not a known optimer for dopamine. But somehow the assembly of these two things into one hybrid nanomaterial ends up having a really beneficial functionality, in this case, the ability to detect dopamine. But why does this happen? How does this even make sense? You take Gc6, you assemble that on the surface of these nanoparticles and somehow get this really good sensitivity to dopamine from the system. So why does this happen? Why not other sequences of single-stranded DNA? Like, so DNA has these modular pieces where you can change the base or the ribose or phosphate backbone. You know, what happens, for example, if you begin to poke at all of these components? Surprisingly, the chemical space here is very, very extensive. And there is a lot that you can actually do here. And I'm not going to be able to present everything that we know about these systems. But today, I'm just going to highlight a few things, a few findings. For example, If you were to take the T and replace that with an A, adenine, and just have this short 12-mer oligo that effectively creates the same stable colloid and look at the fluorescence emission and then expose that to dopamine, what you find is something that is okay in terms of sensing dopamine, but it's not that great. It's not as good as GT6. That's really intriguing. Then if you were to take the T and then replace that with an RNAU in which the deoxyribose now just also becomes proper ribose, an RNA ribose, what you'll find is a sensitivity to dopamine that really goes through the roof. It increases several-fold. So, here we're keeping the Gs as DNA versions, but the Us are being changed here. And you can see this very profound effect. And then if you were to change one of these non-bridging oxygen atoms on this phosphate backbone to a methyl group, creating this so-called methylphosphonate backbone on the DNA, you end up having a sequence that again forms a really good colloid. It's very stable. It's very fluorescent, but it's nearly insensitive to dopamine. Why did this happen? How do you rationalize this observation? What we find is this quenching that I keep talking about in the previous slide, which is that to be able to see the fluorescence, you need to assemble something on the surface but that assembly itself ends up quenching or decreasing the intrinsic fluorescence of the nanotubes to— and that quenching is a function, a very strong function, I would say, of these different sequences where in this particular case, for example, DGRU really sort of quenches the system for you, sort of springloads it for you such that dopamine can come in and rescue that fluorescence. In fact, if you look at the baseline brightness of these suspensions for equal amounts of nanomaterial, in them, you see this opposite trend that kind of explains why we see these DFFRS trends that are depicted here. Right? But what happens on this side of the equation? So, here, I'm trying to show you what exactly might be going on on this side. But what exactly happens on this side of the equation where dopamine comes in and then rescues the fluorescence? Because dopamine, again, has to really do something. Something has to happen for for this quenched baseline to be reversed. And to explain this a bit more, I'm going to show you two molecules. Dopamine is the guy that keeps— that I keep talking about, which rescues or to a certain extent recovers the fluorescence. But let me also show you another monoamine. This is a molecule that is also a neuromodulator in the brain. It is a trace amine. So, it's an important neuromodulator. But it's distinct from dopamine in a sense that it's not a catechol. It does not have these vicinal diols on the benzene ring. This is a phenolic monoamine neuromodulator. And what we know from these systems is that this class of compounds actually does not modulate the fluorescence at all. Right? So, you see zero DFOF, if you add tyramine. You can add millimolar quantities of this, and there is really no response from it. So, why does this happen? Now, one of the ways we try to rationalize this is to— we don't— we can't really create crystal structures for these systems. We can't study them in that way, but we can study them in these modeling systems by collaboration with labs that do a very good job of this. And what we find from this particular study, for example, is that these catecholamines have this ability to intercalate into these empty spaces on this DNA that is stabilized on the surface of a nanotube. Π-star contour and then also really stabilized through these hydrogen bonding events through the phosphate and some of the hydrogen bond donors on the bases that create these stable binding events. And these stable binding events happen to be very distinct and different from the transient binding that we see with— for other types of molecules. And we believe these stable binding events are really sort of the reason why we see the type of modulation that we see here. Now, This binding then has to be followed up by some kind of electronic modulation that we still are investigating. There is more to this story than I can tell you right now, but we do feel like this is giving us that thread that we can continue to pull on in the months to come. Hopefully, I've given you a flavor of how these systems work, how they're assembled, why they even sense dopamine, and so on and so forth. But now I'm going to switch gears here and then begin to talk about the utility of this nano, hybrid nanomaterials. Like, why would you want to use this? What are some of the practical advantages when it comes to actually being able to sense dopamine? Here, my emphasis is going to be less on the photophysical benefits and more on the practical benefits. Like, what are, what are some of the use case advantages for this particular hybrid nanomaterials? And to do that, I'm going to show you a very quick study that we did early on for these systems where, because these are extremely synthetic, we can use them in combination with acute brain slices derived from wild-type animals. And we don't have to express virus here. We can just take an acute brain slice collected from a wild-type animal and then incubate that slice in a solution that contains these nanosensors. And a passive room-temperature incubation for 15 minutes is enough to give you a brain slice that is well labeled. Then you can take that slice, take it to your imaging microscope, and then begin to image. You can produce videos that look like this. This is imaging in this dorsomedial striatum of the brain. This is a region that has a very intense innervation of dopamine terminals from the midbrain. You can capture this electrically stimulated release of dopamine in this turn-on events that we see in these types of videos. If you take a closer look at these videos, they are very informative. There is a lot of information contained in them. In both the spatial and temporal domains. The first thing that is striking about this is that they have this— we see these hotspots of activity or these blobs of clustered pixels that we believe are a consequence of dopamine getting released from some chemical synapse of dopaminergic terminals in the dorsal medial striatum. Then secondly, there's also a lot of temporal information contained in them. For example, we know that when we stimulate the tissue, the turn-on is very fast, But the decay of each one of these hotspots is very different and is distinct. And there is this beautiful heterogeneity that you're able to capture. And you can do that by first fitting first-order decay curves to each one of these, extracting this so-called tau-off values, and then looking at the distribution of these tau values to appreciate that there is quite a bit of diversity that is not necessarily captured by simultaneous measurements with FSCV, which is another technique of measuring dopamine. So this type of, I think, heterogeneous information that can be extracted by imaging with these nanosensors becomes sort of a repeating theme for me. And I'm going to get back to it a couple of times in my presentation. These nanotubes are synthetic, meaning that you can also use them in some really unconventional and weird and useful ways. And one of these is this publication that came from my lab a couple of months ago. Where we've been able to create these composite nanofilms out of these nanomaterials on which you can culture primary neurons. And the best way I can describe this is like this. So, if you take a glass coverslip, you can use very basic silent chemistry to create a positive charge on the surface of the glass. And then because the sensor itself has a negative charge on it, you can electrostatically pin a very thin layer of this nanosensor on the glass. So you can think about this two-dimensional dopamine-sensitive layer on the surface, and then you can passivate that layer with polylysine and then grow or culture primary neurons on top. In this case, we culture primary dopaminergic neurons because we have a dopamine sensor at the bottom here, and then take that entire ensemble to imaging microscope and then begin to image. So mind you, what you have at this point is a two-dimensional surface that is more or less uniformly fluorescent on which these neurons are growing, and these neurons are capable of releasing dopamine. The hope is if you can capture some of these releases at subcellular level, you might be able to localize it to specific boutons or specific chemical synapses of dopamine. What do I mean by chemical synapses of dopamine? Here is a scanning electron micrograph of a single process of a dopaminergic axon. These processes have these so-called axonal swellings on them. These beads on a string type of structure, these are called boutons or varicosities. I'm going to use those two terms interchangeably. These are the chemical synapses of dopamine. This is precisely where we expect to see releases of dopamine from these neurons. The hope is that if some of these are release capable and they can release dopamine onto the substrate on which they're growing, you might be able to capture some flash of activity because that substrate is fluorescent and also the fluorescence itself is sensitive to dopamine. After some period of optimization, we've been able to generate these videos that precisely look like what we had in mind initially. This is a single process of a dopamine neuron. It is connected to the cell body, but I'm not showing the cell body here for simplicity. But we're stimulating the neuron, we're generating action potential, and then we're subsequently causing these releases of dopamine that we can detect. You can see these transient activations of the surface on which this process is growing, and you can, you can produce videos that look like this. Of course, that single process is just a demonstration, but we can also choose fields of view that are a lot more elaborate and extensive. Here is an axonal arbor of a dopamine neuron where we have hundreds of release sites of dopamine or boutons of dopamine creating this synchronously released dopamine that we're able to capture with this substrate. So what exactly is the utility of imaging in the manner that I just did? So we're imaging at a subcellular level, at the level of a chemical synapse, but what kind of utility or value would that modality of imaging provide? And to try and answer that, I'm going to take you through a series of slides, and to do that, I'm going to I'll show you a series of frames from the video that I just shared earlier in the previous slide. I'm going to begin by showing you the frame before we stimulate the field of view. Now, if you recall, the substrate itself is fluorescent, but I've normalized the fluorescence here so you don't see anything. Then right at the moment of stimulation, you see the emergence of these so-called hotspots of activity that are localized in this two-dimensional plane. A second later, these hotspots have grown bigger and they blend into one another to lose their identity. To create this semi-homogeneously activated surface. And then after some period of time has elapsed, the substrate goes back to baseline, right? So this is for one stimulation, right? But we can stay in this field of view and stimulate several times. And what we find when we do that is the emergence of these same clusters of pixels that I call hotspots, uh, at the same locations, having exactly the same shape for every stimulation. In fact, when you average these activities across several stimuli and look at the temporal property, it looks like this. You see this mean trace in black with a standard deviation band around the mean that is indicative of two things. You can see this very stable baseline, very high signal-to-noise ratio for imaging. Then secondly, you can also see a very tight standard deviation at the amplitude, suggesting that the substrate really has the ability to go back to baseline and then respond equally vigorously to subsequent stimulations. That's very important because that tells you that in regions where you actually don't see any kind of activity, that's likely because there is really no release coming from those regions, and it is not because we are somehow unable to detect release. Now, if you take a closer look at these early phases of dopamine efflux at that moment of stimulation and try to examine how well that compares with the underlying structural image, it looks like this. There is a pretty good overlay between these activities and the underlying structural image from which we expect to see these releases. So, just as a matter of housekeeping, what I mean by TH-GFP is GFP that is expressed under the control of the tyrosine hydroxylase promoter, which is a rate-limiting enzyme for dopamine biosynthesis and is often used as a marker for dopamine neurons. In a typical imaging experiment, what we do is go into a GFP channel, take a structural image, and then we can switch to the NIR channel to begin to take activity images. Because all of these things are being captured from the same microscope, same camera, we can overlay these two things to examine how well these clusters of pixels of activity that we call hotspots colocalize with these potential release sites of dopamine. You can see that to at least to a first-order approximation, there is a pretty good agreement between where we see release and where we expected those releases to occur. Now, biosensors in neuroscience are nothing new. We have some of the most advanced and beautiful genetically encoded biosensors that are now extensively used in a lot of neuroscience labs. These are now the sorts of the bread and butter for as far as tools are involved in neuroscience. You can capture calcium activity or membrane voltage changes and even neurotransmitter sensors using these genetically encoded probes. But one thing that most of these things share in common is the modality in which they are used. You have this two-photon or wide-field imaging microscopes or some miniscope or maybe even fiber photometry, where the emphasis for these biosensors is almost always on getting information from as large a number of neurons as possible. The emphasis is all about getting— and sometimes you don't even care about spatial information. You might do fiber photometry just to look at temporal kinetics in some region of the brain. And that is often usually enough for most of these studies. But the type of imaging we're advocating here really emphasizes looking at things at a subcellular level. And I would argue that it sort of adds a value. It is orthogonal to these other systems of imaging. And I hope that going about imaging in that manner contributes some really new knowledge in neuronal function and how neurons actually contribute to brain function. To make that point clear, I'm going to go back to this NIRS image that I brought up a couple of slides ago, and then try to make this really important point, which is, what exactly is the utility of imaging at a subcellular level? What is some new biology that you might be able to find in going about imaging in this manner. So here is the NIR DFI image. Here is a corresponding THJFV image. These THJFV images have this so-called varicosities. So these are the chemical synapses of dopamine. We can find where they are by just doing a very simple thresholding. You can determine exactly where these potential release sites are. And then you can use this overlay image as a guide to assign specific varicosities to each one of these clusters of pixel correlated pixels. Or hotspots. For example, for these 5 hotspots, I know that these emanate from these 5 varicosities of dopamine neurons. But if you take a closer look, something really intriguing happens. These hotspots— excuse me, these varicosities that I have in red actually do not contribute to any kind of release here. So, they don't appear to be participating in release. Intriguingly, if you stay in this field of view and stimulate several times, which we do, what we find is activity almost always comes from the same set of varicosities, whereas some other subset of varicosities remains completely release incapable or remains silent. The question is, why did this happen? This is very intriguing for me. Why exactly does this happen? Are there some molecular-level differences between these release capable and release incapable chemical synapses of dopamine? Are there protein level or transcript level differences? And even more importantly, what are some of the regulatory mechanisms that go into delineating these two populations of varicosities? I think for me, that is very intriguing. And this tool will now begin to help you get at these questions in a way that has not really been possible before. And I don't have an answer for you regarding these questions that I'm bringing up, but I'm trying to demonstrate the potential future applications for this particular technology that we've developed in the lab. Something that has been a bit more productive and more fruitful for this particular two-dimensional imaging, subcellular imaging, has been the ability to visualize dopamine release from dendrites of dopaminergic neurons. Dendrites of dopamine neurons in a very weird and unique way are also capable of releasing dopamine. In addition to release from axons, you also see release from dendrites.. But that particular release modality has not really been fully explored for a number of reasons. And that reason is primarily because we really did not have the tools to be able to visualize these in a way they should have been really well studied. And we have now been deploying these technologies in dendritic regions to study potential release sites on dendrites of dopamine neurons, which I think is really intriguing and something that we hope to continue to do in the years to come. In this particular case, this putative dendrite of a dopamine neuron gives rise to these three clusters of activities that are very precisely co-localized with the process itself. If you stay in this field of view, stimulate several times, you find a consistent set of responses. This is not just a one-off event. This is a consistent set of responses to repeated stimulations. Now, imaging in these systems can also be coupled with downstream existing techniques where we can elucidate things a bit more, a bit more after we've done imaging. In this case, for example, where I'm trying to demonstrate release from dendrites of dopamine neurons, we've been able to couple that with downstream postdoc immunostaining and then super-resolution imaging to demonstrate that some of these releases that we claim to be dendritic are really truly dendritic in nature or are coming from dendrites. And here I'm showing you a small window of a bigger field of view where a really nice hotspot can be detected and localized to a specific dendritic swelling of this process. And then we can image across multiple stimulations. And then once that imaging is done, we can fix this a particular specimen and then stain it for different types of proteins. And in this case, we looked for a protein called MAP2, which is microtubule-associated protein 2, that is enriched in dendrites, but it is completely absent in axonal processes. So this tells you that this process that is giving rise to this activity is probably truly dendritic in nature. You can also amplify the GFP signal. Here the GFP is virally expressed. It is kind of dim. It is okay. It has Good signal-to-background, but it's not the best looking. But you can also amplify the GFP signal, right? And you can reimage it just to prove that there are really no other processes that are lurking in the background here contributing to release and that everything that we see here just comes from the single lone process that we think is contributing to this release. Here is a bigger field of view with a more robust release being colocalized with a a dendritic arbor, and again, we can reimage that area and convince ourselves that these are truly processes that are dendritic in nature. There are times when we also see spontaneous activities without actually having to stimulate the neuron. Here is a dopamine neuron, and I'm showing you a smaller window of these putative dendrites. You can see the stochastic release events that are totally independent of action potential firing. We're not stimulating this neuron, You see this temporally uncorrelated but spatially specific release events that we can detect from this process. We can examine this spatial and temporal kinetics from dendrites in a way that has really never been done before, where we can cluster specific pixels based on their correlated activity and then assign them as putative release sites from a specific dendrite, we can count how many of them are present on a segment. We can characterize the full width at half maximum, the spatial extent of diffusion, and so on and so forth. I'm coming to the last couple of slides in my presentation, but I want to make a couple of more points before I conclude. The first one is that these are highly sensitive towards their analytes. Here we're imaging in our normal imaging buffer that we call ACSF and imaging in an axonal arbor. Which is fairly extensive and very dense. You see this synchronous release events in response to externally applied stimuli which generate these action potentials. We can also block these action potentials with a drug called TTX. TTX stands for tetrodotoxin, which blockades action potential firing but leaves behind these stochastic release events. These are called TTX minis. These are really, at least in principle, the smallest bits of chemical release that can be detected at a chemical synapse. You can see these tiny deflections that we believe are single vesicle fusion events that we're able to capture. You can also see these peaks that have higher amplitude, and we believe those are multiplicative events of single quantum release events. This slide here really demonstrates the fact that release from these cultured systems can be detected with a high degree of sensitivity. And all the points that I'm making in terms of release-capable varicosities or release-incapable varicosities is backed by the fact that we know that if there is any kind of release coming from some of these chemical synapses, we are more likely to be able to detect it. And this TTX imaging field of view. I want to show you a smaller field of view from the bigger field of view where we have these clusters of 4 release sites. We believe these are correlated sets of pixels emanating from 4 release sites. We're imaging here in our normal imaging buffer, and then we can blockade those action potentials with CTX and then re-image that same region with TTX. What we find in this case is only one of these four survives. Only one of these four survives. And if you're wondering which one of these four survives in TTX, it turns out to be number 1. Number 1 is a guy that survives in post-TTX imaging. And how do I know it is number 1? Is because we can determine the centroids of each one of these clustered pixels. And by centroid, what I mean is just a pixel that corresponds with the local maxima, sort of like a local peak. And what we find is that when you compare the local peaks there is a near-perfect match between this guy in the post-TTX case and number 1 in the pre-TTX case. When you look at the temporal property of that, it is highly stochastic, which is precisely what you expect to see. But what was striking for me is that number 2 and 3 and 4 do not actually contribute to release in this post-TTX imaging. The point that I'm trying to make here is that there is a temporal stochasticity in this post-TTX imaging frames, but there is a special specificity in terms of where the release is actually coming from. And for me, that was— that's not what I expected, right? This is not necessarily what I had in mind when I was thinking about this stochastic release events from TTX fields of view. And even in the bigger field of view that I was showing in the previous slide, what we find is TTX minis appear to emanate from a specific, especially specific set of varicosities, but have a temporal stochasticity as you expect them, but not necessarily on the spatial side. Anyway, very intriguing for me. And I'm highlighting this again because this is the type of information that I think would be completely lost or ignored if you're doing some kind of ensemble average measurements. But when you begin to look at things at a highly granular level, at a subcellular level, and begin to obsess over some of these nuances that are beginning to emerge, uh, when you image in this, uh, uh, subcellular level, then things that you don't necessarily expect begin to bubble to the top, and hopefully that will bring up subsequent sets of questions that we can pursue in the years to come. Here's a forward-looking slide where we hope to do patch clamp on these systems where we can very precisely depolarize these neurons. All of the simulation so far I showed you is based on optogenetic stimulation where we just flash blue lights at this cultured system, so we don't really have a very good control over how many action potentials these neurons are firing. But if we can have a very precise control of our action potential firing on these neurons, then we can carry out this very extensive biophysical characterization of individual synapses, which would be like a really dream experiment where you can look at probability of release, examine if there is synaptic variance across a subset of chemical synapses, and investigate spatial extent of release, and even couple this with subcellular multiomics where you can look at the subcellular protein level or transcript level differences in these systems. All right. So, those are all the applications of these hybrid nanomaterials. But this being an OIG, optical interest group, I thought I would be stupid not to bring this up, which is there is a bit of a challenge in terms of using these materials, right? I talked about how advantageous it is to have fluorescence in this region of the spectrum, but there is also a huge disadvantage. And this is the fact that emission in this range of the spectrum is completely undetectable with silicon-based technology. So you can see that the QE, quantum yield or quantum efficiency of these detectors falls off in this region of the spectrum that we actually are interested in. Here is an image that I just copy-pasted from Thorlabs where you have a bunch of PMTs and You don't have to worry about what these designations mean, but all that— all I want you to take away from this is that this region of the spectrum that we're interested in, all of these state-of-the-art PMTs that rely on these different photocathode materials, bi-alkali, multi-alkali, GASP, they are insensitive. They're not going to be able to help you image. So if you want to use these hybrid materials, you have to have your own custom microscopy solution, which I realized early on in my PhD. So I've been using these materials for 6 years and It was a rude awakening. If you take this into like a standard imaging microscope, you're not going to be able to use them. You're not going to be able to see any kind of fluorescence. So we spent like a year and a half really developing this microscopy solution, which is a combined widefield and laser scanning confocal that does a very good job imaging in this broad range of the spectrum. So we can image from a GFP and from a nanotube and so on and so forth. With that, I think I'm going to acknowledge some critical people that are responsible for the success of some of the stuff that I presented today. Chandima is a postdoc in my lab who began exactly 2 years ago today. I think today is his 2-year anniversary. Deepika from cell culture for culturing and creating all of these really beautiful primary dissociated cultures from rats and I want to mention her name. And then Dmitry Shbulsky, who is no longer at Genelia, but he was truly instrumental in helping me come up with this solution. So there was, uh, there is a lot of investment on in this, uh, project from him. So I want to, uh, mention his name. Uh, I want to mention Kasper Podgorski, which a lot of you know. Uh, he gave me this MATLAB program that has been very good at helping me analyze some of these videos that we've been able to produce. So I want to mention his name. And of course Anastasia, who's my lab admin and has been a critical component of the success of the lab as well. So with that, I'm going to stop here and open it up for questions. Great.

Thank you for the talk, Abraham. That was really cool. So we have, we still have some time for questions. I guess I get first dibs because I'm the person sort of moderating this, which is the best part of that. So I like, I find this super curious that you see certain varicosities which are release competent and some which are not. And so I'm curious if you've started to look in a more complex preparation like brain slice to see if you see the same kind of phenomenon, or if that's— or if it's, you know, uh, either difficult to visualize. I mean, it seems like you have really good signal-to-noise, so you might be able to see it without, you know, having lots of scattering and stuff.

Um, yeah, in brain slices, um, direct imaging would be difficult because, um, I think one of the reasons we are able to do the type of imaging we do is because we have— we impose a very high degree of sparsity on the system by controlling the density of neurons that we culture, right? So you can force these things to be further apart. And even in culture, there are times when, uh, these varicosities are just so dense that you're not gonna— you just see like a blob of activity, uh, that is just difficult to really assign to specific, uh, boutons. So doing this type of live imaging in a brain slice would be, I think, challenging. I don't think that would be successful. But I think there are— the way I was hoping to go about this is to try to prove it in a roundabout way. And here's what I had in mind. So imagine that after doing a bunch of experiments over the next couple of years, we find that these release-capable varicosities have specific clusters of proteins, like a sort of a fingerprint of constellation of proteins that come together that enables that to be the case, whereas those same proteins are absent in some other, the release incapable ones. So the question that I have is, you know, once we establish that in a culture system, can we go find those in brain slices, right? I think it's possible to do that. We can do it, you know, you can do like histochemistry, you can stain them for the same set of proteins, or you can expand the tissue if you have to, right? You can, or you can do super-resolution imaging with state microscopy. And then you can go look for those things. That is the hope. And if we find those, right, if you feel like there are like those, you know, fingerprints, even in a brain slice, then you can engage in like these more extensive like knockdown experiments where you can knock down specific proteins and image from brain slices where you can try to map some of these illuminated proteins and map that onto some kind of activity. So, yeah, so that's what I have in mind. I don't know how easy it would be to do this in a brain slice, honestly. Like, yeah.

I have like another naive question. I'm like a protein person. So material stuff is sort of But can you use strategies like spatial lithography type of things to pattern your nanotubes in a way that you could enforce sparsity in a brain slice by patterning your sensor?

Lithography on an acute brain slice.

I don't really know. I'm wondering if you can, Instead of having some homogeneous layer, if you can pattern that in a way that could be useful.

Yeah. I think, I would say, I think a more, I don't know if lithography would be successful, but something that could be successful is to do, you can do acute brain slices. You can do electrophysiology in layers and slices. You can see specific processes. Even in slices. And then, I mean, this is going to be very tedious, but you could, you know, nanoject like small, like nano volumes of this synthetic dyes into like single processes maybe, or something that looks like a bouton in a brain slice, and then image it there. I think we do have the signal-to-noise to be able to detect activity even from single boutons if you could do it that way. It would be— the throughput would be very low, but if we have to, I think we can do it that way. For me, I think the lithography approach, even if you can impose a sparsity on the sensor itself, I think my fear is that there is just so much overlapping layers of glutons or varicosities and the brain slice that I think would still be difficult to to say like this release is coming from this, this specific chemical synapse, right? Yeah, I don't know. Did I answer your question, or is that— Yeah, yeah, I'm just curious.

So does anybody else have questions? I think you can just kind of raise your hand and jump in.

Daniel Milky raised his hand.

Yeah, hi. I would like to know if the fluorescence spectrum changes or the absorption changes when you have these sensors you're showing, how the nanodroplets have all these interesting peaks. And I guess you have multiple peaks, so you have a heterogeneous ensemble of different types, different species. And would it be advantageous to use like a single species?

Yeah, that's a really good question, Daniel. So yes, it is. I think it would be really cool to try to use single purified colors of these sensors. And if you recall from the spectra that I showed, there are even differences in the vigor with which these things rise. So there are some more sensitive chiralities of species of these nanotubes relative to others. I think it would be really cool to try to purify these things and then do imaging in the single-color regimes. And I have to admit, so like that, as soon as I got here, the first few weeks was really spent on like trying to purify these things. And then, and I signed up for kind of like to spend a couple of weeks at a lab elsewhere that where they do a really good job of like purifying these things. And then the, the, my visitor status did not go through because COVID came and then everything stopped. And I ended up really switching what I was doing in the lab and just doing, uh, other things. And I never really got back to that question, uh, mostly for just bandwidth reasons. I just don't— did not have the, the time and the manpower to be able to go after those. But I think, yeah, in principle, on paper, would be feasible. The purification, I have to say, is very challenging. The throughput is still fairly low, but even if you adopt like the best technique that is published and available, you will still be able to get like tiny bits of these purified samples, but you can't get enough to be able to do imaging with them. And yeah, so intriguing and potentially possible thing we can do in if we're able to have the manpower to do it.

Mark, do you have a question?

Yes.

I was really curious about the dopamine unquenching phenomenon there. I'm wondering, is it very specific to dopamine? Does it apply to other catecholamines or?

Yes, it does apply to other catecholamines. This is a catecholamine sensor. So norepinephrine is the other catecholamine in the brain and also does modulate the fluorescence. If you look at, I guess, the relative sensitivity, so there is something of like a threefold more smaller Kd for DA than NE. But I would say that that's not probably not enough to say that this is a sensor for DA over NE. This is I call it just a catecholamine sensor that we're applying to studying dopamine. There is a sensor for serotonin that is a lot more selective than these two. In principle, that's also a monoamine that can also be studied with this system. In terms of other classes of neurochemicals like glutamate, GABA, and so on and so forth, we don't really have good sensors for those. In fact, this is an active area of investigation where we're trying to understand exactly how these sensors even work at like a really mechanistic level. Like, what are the molecular processes that exactly underpin this function? Which I tried to get at in a few slides today. I tried to like give you some of the— I was dropping hints here and there in terms of what we know and what we don't know. And, uh, yeah, hopefully maybe in the years to come, as we study this system more and more and we have a very good understanding of what the exact mechanisms are for fluorescence modulation, then we may be able to make more sensors like this. But, uh, yeah, TBD, I would say.

Yeah, I think it's really intriguing. Thanks.

Uh, yeah.

Hi, Abraham. Uh, thank you for a very wonderful talk. I really enjoyed it. I had a question about the morphology of the single-walled carbon nanotubes, and I was kind of wondering, the conjugate that you made is utilizing the outer surface, right? The fact that it's hollow, I was just curious if there's any way to decorate the interior space. That's one, or does it have strong, say, capillary actions for uptake of inorganic dyes, for instance, that you could utilize?

Yeah, no, that's a really good question, Raghavan. That's a There is a literature out there that they call it endohedral filling of nanotubes with different types of small organic compounds like hexane and, and like different, you know, different small molecules have been filled. And, uh, you know, I have to, I have to be honest, I have to read it. I have to read this to actually see what, uh, how they do it and what the effect is of filling these nanotubes in that manner and how that actually ends up in any way affecting the sensitivity, if they make some better or poorer. I, I don't know, but, um, yeah, it's, uh, I, I wouldn't be able to give you an answer, but it does happen. So I, I think, I think they just do it through capillary action or whatever, but they do fill it with different types of, uh, just passive, uh, filling of this, uh, nanotubes with different types of molecules.

Yeah, thank you.

Thank you for your honest feedback. Yeah, no problem. Brian.

Wow, Abraham, that was really fantastic. Thank you so much. I have a question about the detectors, IR detectors, and how cheap they could be, because, you know, I would envision for your 2D detection, I would envision growing you're making that layer right on the detector if they're cheap enough, right? And then, you know, increasing your resolution, right? Forget all about— I'm sorry to say this in the optical interest group, but forget about all the relays and far field, just growing it right on the detector. But I guess that would rely on really cheap IR.

Yeah, those things are really not that cheap, surprisingly. Yeah, at the beginning, like 6 years ago, they're getting cheaper, I would say. Like 6 years ago, 6 or 7 years ago when I was beginning my PhD, the first camera that we purchased in the lab of my PI, she was still a postdoc when she got it. I think she paid like $800,000 for it, for this camera that is not even, wasn't even that great.

I retract that statement then.

$800,000 or $100,000?

Yeah, sorry, I didn't say $800,000. I said $100,000. Okay. Sorry, yeah, that's my accent getting in the way. My apologies. Um, uh, $100,000. One, one, one hundred thousand.

I thought I had a stroke or something, but it has, it has become cheaper.

Uh, the, the, uh, when, when I bought these cameras for my lab, I spent about $40,000 the first time around, and then the second time around, uh, I just, I just got another camera and it was like more like $35,000. So yeah, they're getting cheaper, but not cheap enough for you to do that kind of experimentation, I would say. And they're getting better too, by the way. They're getting better. Their read noise is really improving. QE has always been good, but there's always been the electronics that has been troublesome with these cameras. QE is always in the 80-90 range, which is enough, but they have really bad read noise. I showed some of the specs to Raghav and he just couldn't believe how bad it was relative to like your, you know, fancy cameras that you can get from Hamamatsu, which is, which have like really, really good electronics in them. So, um, so yeah, electronics are catching up.

How fast can you run them?

These guys have— sorry, I have absolutely no idea about any of this, but, uh, so yeah, so all of the stuff that I presented today, uh, was mostly in the 10 to 15 frames per second range. We can do, we can do like 25, 25, 30 frames per second. I think we do have enough brightness and signal-to-noise to, to speed it up and go out to that range of speed, faster imaging.

Yeah, maybe one more question from myself. So coming from the And thanks for the talk, I really liked it. Coming from the photoacoustic field where the contrast is based on the absorption and not fluorescence, this carbon nanotubes, is it really the fluorescence emission that is changing, like quantum yield, or the absorption cross, like absorption coefficient?

Yeah, that's a good point. I think we believe this, uh, if you look at the absorbance spectra, it doesn't really change Just looking at the— if we do measurements in dopamine and without dopamine, we don't see any kind of increase in the, in the absorbance region. It's just the emission that is increasing, and we believe it's just a quantum efficiency, I guess, quantum yield that is improving.

This absorption is more or less flat in the visible region?

Um, it's not getting modulated. It is not flat. You can see this, the, this, uh, nanoparticles have, um, there are like this resonant absorbance peaks at different regions in the visible range of the spectrum. And we try to image with one of those regions. We target like one of that areas. So it is not just a flat thing. It is It has peaks and valleys, but that peak and valley is not really changing in response to dopamine, I would say. Thanks.

Great. Well, I don't think there's any more questions, but we had a nice discussion. Thanks again, Abraham, for the wonderful talk. That was really cool.

Thank you for having me.

Everybody, for, for a good discussion afterwards.