Scientist Stories: Andrew Smith, Nanoparticles for Molecular Imaging

So first of all, I wanted to acknowledge my lab. So this is what we used to do back before COVID was meet all the time, have great outings, and I miss that. And so if anybody here is from my lab, hopefully we'll return back to these types of fun events and going to immerse ourselves in science in person sometime in the near future. And this is a kind of an overview scheme of my lab. So what I'm going to talk about today is what's on the left side here. So there's kind of 4 main thrust areas. One is in imaging agents, and these are mostly optical imaging agents. And then on the bottom left, we have single-cell analysis. This is one of the favorite parts of my lab. It's really focused on the life sciences, fundamental discoveries in the life sciences, and developing tools for that, especially at the single-cell level, and then what's happening at the single-molecule level within individual cells. On the right side are more applied areas of our lab that have only really spun out in the last like 5 or 6 years or so, and these are in clinical molecular diagnostics for making diagnostic assays using blood or biopsies, as well as targeted therapies. And the targeted therapies section of our lab has really expanded a lot, going into disease states such as metabolic disorders, metabolic syndrome, wound healing in diabetes, as well as in cancer. And so I'm going to start talking on the left side here. I'm going to kind of go through just some of our most recent papers, a little bit of background about how we approach these fields, and also where we're going. If anybody has any questions at any time, feel free to stop me. Hopefully I can see in the chat here if anybody has anything in real time, and also at the end I'll hopefully have time for, for questions as well. So a little bit of context here is that we make imaging agents, and we do single molecule imaging in individual cells. So how do we do that? We have imaging involved. And so imaging, depending on what you're trying to image, you have a different instrument and a different set of physical principles that allow you to resolve certain objects and to track them and to evaluate something about them. And from our perspective in the life sciences, many of us constantly use optical imaging in some way. And optical imaging provides a lot of very valuable insights because of primarily its wide range of resolution. So our eyes, of course, are optical imaging tools, and we can resolve things that are in our surroundings in very wide ranges of scale very quickly, high sensitivity, and also what we call multiplexing. So we can see, we can see at least 3 distinct color channels with our eyes, and we can resolve different objects because of that. And so in the last really 15 to 20 years, there's been a big revolution in the optical imaging field as well. That's allowed us to now go from what we could previously resolve, which is around the size of a bacterial cell, about 200 nanometers or so. We can now connect this down to the single molecule level and see individual molecules and resolve objects that are in the range of individual molecules, like macromolecules such as antibodies, proteins, like 10 nanometers in size. Some of that has been led by people at Illinois, so like Paul Selvin, who's one of my best collaborators here. He's really one of the godfathers of this field. At the same time, optical imaging can't provide everything, so there's many contexts in which we need to resolve objects in thick tissues, in living subjects, and without any sort of interferences due to potential toxic effects. And of course, we know that X-ray, ultrasound, MRI, PET, SPECT. These are all great tools, especially for clinical imaging and imaging in animals. And we also know that if you look at biological textbooks, most often many objects that are shown in biological textbooks and many objects that are resolved in many types of fundamental studies are by electron microscopy. So electron microscopy provides this really great ability to see things that are sub-nanometer in the range of nanometers in size. And that's something that has not been provided by any other great tools so far. So optical imaging is where we're focusing. So our lab focuses on probes. And so if you've done a lot of studies in molecular biology and in cell biology, you very likely use probes very frequently. And we're particularly interested in ones that allow you to resolve individual macromolecules. And these are mostly protein targets and in DNA and RNA. And so these give us a lot of biological insights into a cell's phenotype, into molecular processes. And so to do that, usually you have something that resembles one of these two probes that's shown at the top here, an antibody conjugate or a nucleic acid conjugate. And this antibody conjugate allows you to bind something to a protein, which is usually invisible, or it's something that does not scatter or emit light in a way that we can track it and resolve it. And that is usually a dye of some sort. It's usually like a fluorescent dye or something that emits light. It could also be something that absorbs a lot of light. And so depending on how you're trying to image an object, particularly in tissues, you might want to have something that absorbs or emits. So this is generally the context of what we're working with. We're trying to make these better. And trying to make these tools that we can use to resolve individual objects in cells, individual molecules in cells, and then track them for long periods of time. And that leads me to nanoparticle labels. And so what is this thing that's attached to this antibody or this nucleic acid that absorbs or emits light? It's usually an organic molecule. Historically, it's been in most contexts that are actually used widely in the biological sciences, it's usually an organic dye or sometimes an organic fluorescent protein. In the last 20 years, there's been a really big revolution in the ability to make these out of a wide range of new types of materials, and in particular, nanomaterials have become a very big focus of making the next generation of molecular labels and molecular probes for the biological sciences. So, they provide a lot of unique values that I'll go into in a moment, but also there's a very big wide variety of these. So, these can be made out of macromolecular organic structures like liposomes and dendrimers, and they could be made out of inorganic materials that are made out of metals, semiconductors, as well as magnetic materials. So, there's a big space now available for making labels and probes that can facilitate new types of contrast in imaging. A primary focus of my lab that I've been working on since graduate school really, has actually, um, now about 15 years or so, are quantum dots. These are semiconductor nanocrystals. And so, some of the preeminent features of these are that they have tunable light emission. If you look at the spectra on the right, these are fluorescence light emission spectra. And you can see that the emission spans from the 400 nanometer wavelength up to about 700 nanometer wavelength. That's the visible spectrum. That's what our eye can see. And if you look at vials of these materials above them, with one material you can span a wide range of emission colors. So one material here, which is cadmium selenide, if you tune the size of that nanomaterial from about 2 nanometers to 7 nanometers, you could span the majority of the visible spectrum. If you look at the TEM image on the left, this is what they look like at high resolution. You can see individual atoms or atomic lattices lined up, and so these are crystalline materials. But what you can't see in this image is that this is surrounded by organic materials, so it's a composite structure. It has a semiconductor crystalline material that gives the ability to absorb and emit light, and then surrounding that you have a a wide range of potential coatings for it that can make it biocompatible, make it interact with biology in a useful way, and also allow it to bind to something with molecular specificity. So that's what's depicted in this yellow to brown object that's surrounding, or that's attached to this quantum dot that's depicted here in blue, is a molecular recognition capability. So in the last 15 years, these have really become useful tools for one particular major application, and that's single molecule imaging and tracking. And so these are one of the most foremost used materials for this field now. So what's shown on the left is an individual cell, and what's shown on the right is that cell in a different type of contrast. So on the left, you have a brightfield image where you're just looking at the transmission of light through this one cell. You can see the blue hue, which is fluorescence from the nucleus. And on the right, you can see individual molecules here that are diffusing on the cell because they're bound— that you can now resolve. So we can now see this one individual protein and what it's doing. And we can get a wealth of information out of stochastic processes in biology that otherwise you can only get out an average value. So now we can see the distribution of individual molecules and how that contributes to an average behavior that we've resolved previously. So this is, uh, single-cell imaging together with single-cell, single-molecule tracking. So this has been very transformative for this application. Um, so why are they so transformative for this application? It's because of their photophysical properties. So these are widely used in in a variety of contexts because of these unique collection of properties for quantum dots compared with other materials. For instance, they have size-tunable color. And so if you have size-tunable color, you can potentially span any wavelength range that you want. In particular, the infrared is a very useful range for biology because a lot of biology is fairly transparent and background-free in that range. To do that with a fluorescent dye or fluorescent protein is much more challenging. Instead of just changing the size which you grow the nanocrystal to, you have to develop a whole new molecule and a whole new synthetic routine for these proteins and organic dyes. They also have extremely bright emission, about 10 to 100 times that of an organic dye or fluorescent protein, and they're much more photostable. And so this shows a comparison of the signal you get out of an individual molecule in an individual cell for a dye versus a quantum dot. So quantum dots, potentially you can image them for hours to sometimes longer than hours continually, and dyes you can't do that. So dyes typically will degrade. They'll, um, it's called photobleaching, and the actual ration or reason for that is because they they chemically change because, usually because of oxidation or hydrolysis. So they'll actually degrade. So quantum dots don't have a tendency to do that under, under excitation. So there's several different designs for these materials. So these are sold commercially, but also in this, um, for many applications they have to be specifically designed for your goals. And so some of those design goals are in the photophysics of these materials. One major goal is to design these to have longer emission wavelength. And that's because of what's depicted on this top chart here, which is a spectrum of quantum dot fluorescence that's in the colors, and as well as autofluorescence from, in this case, this is fixed cells. Most autofluorescence, when you're exciting at something like 400 nanometers, 500 nanometers, you have a lot of fluorescence that's at the spectral range close to that excitation band. And so that excitation band, um, you want it to be far away from the emission band. This is something that is great for quantum dots. You can have what's called a long Stoke shift of excitation, so you can excite at like 400 nanometers and you can emit at like 700 nanometers. For dyes and fluorescent proteins, they're usually close together and usually have a strong autofluorescence signal. So we can make these emit in the infrared, and in there, in the infrared spectrum, you have very low autofluorescence compared to other spectral ranges. So that's very advantageous. At the same time, you can get many colors too. So a design goal for these is often to create many distinct colors that can be resolved and distinguished based on their emission spectrum, and you can get anywhere from currently around 500 nanometers to about 1,500 nanometers. Another goal is to create ones with controlled brightness. So currently a big problem with these materials that you don't see in other materials, like organic dyes, you can use 2, 3 colors, and you look at them at a single molecule level, and they're all kind of the same brightness. That's not the case with commercial quantum dots. That's what's shown on the top here for this vial that's shown with a laser excitation. If you use a blue and a red one together, The ones that are used commonly right now, they are much, much brighter in the red spectrum. That's very problematic if you're trying to image things with multiple colors at the same time. You have to adjust your cameras and adjust the ability to excite these in a useful way. And it's complicated, but it can be solved. But what can't be solved is the spectral crosstalk that occurs because of this. So a very small amount of crosstalk between the red channel and the blue channel will completely wash out the blue channel if you're trying to image it simultaneously. So we're trying to create these to be equivalent brightness, also very high photon emission rates, so we can detect individual molecules with statistical significance, long-term photostability, narrow bandwidths, minimal blinking, which is basically these flash on and flash on repeatedly at the single molecule level. And an ultimate goal is to make these photo-switchable. Photo-switchable is a, is a big need right now for their use in super-resolution imaging. In super-resolution imaging with fluorescence, you have to have the capacity to turn these on and turn them off with light. So these are the photophysical design goals. On top of that is also photo— physical chemical and biomolecular design goals. And so, um, it's usually the goal is to be able to use these and to, um, make use of their unique optical properties. But it's hard to do that because you also have to think about their physical principles. And so everybody wants to be able to use these off the shelf and apply them and evaluate something in biology. These materials are still not there. They still have to be designed with specific physical principles in order to reach the desired optical principles. And we still have to consider what their impact is on the imaging goals. So right now, most nanomaterials that are used in biology look like something that's shown in the bottom left here, which is a fairly large crystal surrounded by bioaffinity molecules like antibodies. These are usually in the range of 20 to 50 nanometers in size, depending on what they're made out of. But if you put that into something that resembles what biology actually looks like on the top here, which is a schematic diagram just kind of depicting the crowded space of the cytoplasm of a cell. This is not that different than what it looks like on the outside of a cell too within a solid tissue. It's very crowded and densely packed with solid materials. And this is a porous hydrogel. And depending on what the tissue or the cell type is, it might have a pore size that is much, much smaller on average than the nanomaterial that you're trying to use in it. And if you're trying to use this to evaluate a natural biological process, it's probably going to be interrupted and interfered with by the probe itself. And so our ultimate goal is to get to what looks like in the bottom right here, which is a very small probe. Very small quantum dot attached very precisely to a bioaffinity molecule, and to have that have as minimal interruption or interference with the biological process you're evaluating as possible. So, steric hindrance is a big thing. That's the take-home of this slide. We want these to be as small as possible, and we want them to have as few nonspecific processes occurring as possible. So I'm going to take you through a few papers that we published in the last few years that relate to the photophysical design principles and physicochemical design principles and how they interrelate with each other. So this first one was published earlier this year in the Journal of the American Chemical Society. This is on making quantum dots that absorb and emit light in the short-wave infrared. So the short-wave infrared, like I inferred before, infrared is great because you have low background signals. You also can see deeply into the body relatively to the visible spectrum. And so Suresh, who is a postdoc in our lab, he made a series of materials now that can span this full range. With one material, you can go from the visible spectrum to the near-infrared, which is in the range of 700 nanometers to 900 nanometers, and then also throughout the short-wave infrared, which is just beyond that. Sometimes that's called the short-wave infrared, sometimes it's called the near-infrared too. So to do this, we redesigned these nanocrystals. Actually, this is done originally by Sungjun Lim in our lab about 5 years ago, which is to take these materials which I inferred previously have been problematic because of how different they are in brightness. So we re-engineered these to have a very similar brightness by replacing the materials. So these materials, we now make them out of what is shown in the bottom here. These are quaternary alloy materials that are called mercury cadmium selenide sulfide. We use these to replace these prototypical materials that are shown at the top called cadmium selenide. And in doing so, we can change the color without changing the size. That alone gives us a pretty good ability to control brightness, because one of the reasons these are so different in brightness is because of how different they are in size. Brightness is approximately proportional to the volume, and the volume is proportional to radius cubed. And so if you change the radius a very small amount, you change the brightness by a large amount. So what we did was we made all these the same size change the composition of the alloy core, and then we finely tune the brightness by creating two material composites on this. One is a cadmium sulfide shell, a CDS here that's shown here, that allows us to finely tune the excitation spectrum. That's this epsilon. And then the quantum yield, which is the efficiency of light emission after absorption, is controlled by a separate shell, a zinc sulfide. So we can now tune these all three independently from one another. And previously, these were all intertwined together. So excitation and efficiency and wavelength were not independent previously, and now they are independent. And now we've taken this and Suresh made these now emit into this shortwave infrared or near-infrared 2 spectrum. And we see really great results here. The big problem is that these detectors you need to use this are very expensive. Depending on the detector you buy, it might be somewhere in the range of $90,000 to $200,000. So this field is still not widely used yet, but as use grows, these instruments become cheaper and cheaper. And so like CMOS cameras are in the range of just $20,000, $25,000. So ultimately these will likely catch up in the next decade. But we see great biological results here. An example of that is on the bottom. Where we have, we're imaging nucleic acids here, mRNA nucleic acids in tissue. And one of these is labeled with quantum dots and one is not. The top row is labeled with quantum dots, the bottom row is not. So what you can see is that if these are emitting at 630 nanometers, which is a visible spectrum, the autofluorescence that comes naturally from this tissue without a label looks very similar in intensity to what you see with the labeled tissue. This is not the case when you go out to the shortwave infrared. You essentially see no background fluorescence. And so this is a really great outcome, because if you're trying to detect something, you're always making a decision about what your detection threshold is. Here, that detection threshold is very easy to choose. You choose it to be very low. And then almost anything above the threshold is what you're actually your molecule is that you're trying to image and detect. And so you can see this is actually a progressive increase in improvement going from the visible spectrum on the top right that's shown here in this bar graph, from 630 nanometers, which is visible in the red, 770 and 950, which is near-infrared 1, and this really, really big boost when you get out to the shortwave infrared 2, where you've increased your signal-to-noise ratio by more than 2 orders of magnitude and sometimes up to 3 orders of magnitude. So it's a really big improvement for trying to detect rare events, individual molecules in a very noisy, complicated environment, particularly in tissues where you have a lot of fluorescence coming from lots of cells and lots of molecules, such as cofactors and blood vessels, which are very autofluorescent. Another application for this is trying to, again, detect RNA in cells. But this time do it at the single molecule level. This work was led by Yang Lu, who now has a faculty position in China. And the goal of this is to use quantum dots in the context of FISH. FISH is a field that's been around for more than 50 years. If you've taken a class in the biological sciences where you talked about biological analysis, you probably talked about FISH. It's fluorescence in situ hybridization. There's several ways to do this, but in general what you do is you have a nucleic acid target that's in a cell, either RNA or DNA, and what you do is attach fluorescent molecules to it through hybridization. So you take a short oligonucleotide that's usually in the range of like 20 to 25 bases, single-stranded, and you chemically attach a dye to it. And if you chemically attach a dye to it, then they can hybridize with this target in a cell, and that target is then fluorescent. The very big important thing here is if you want to do this on the single molecule level, these individual fluorophores are usually not bright enough to do that. And also, they might bind non-specifically to a cell. So you usually need to have multiple nucleic acids bound per target in order to resolve them. That number is usually in the range of like 10 to 50. So it's actually a pretty large number. And so what you get out of a cell if you're trying to image mRNA, which is mostly in the cytoplasm, is something that looks like this. These bright punctate spots, and some experimental data might look like this, where these are mostly outside the nucleus, which is labeled in blue with like a HOX dye. And these bright green spots are individual mRNA targets. And they're also surrounded by this greenish hue that's kind of hazy and faded out compared to the brighter spots. That's a combination of autofluorescence and nonspecific binding. And so this combination though of multiple targets binding to the same, or multiple labels binding to the same target, that gives you the ability to resolve these individual molecules. This has become a very big field in the biological sciences too, with the capacity to now buy these commercially fluorescent dyes attached to short oligonucleotides And there's several companies based off of this too. So doing this with quantum dots has more challenges. Here's an example of 3D images of individual HeLa cells, where we're labeling GAPDH, which is usually a control nucleic acid that's used, that's expressed at high numbers, and it's usually used as a reference. When you do this organic dyes, this is what the cell looks like. Somewhere in the range of about 400 copies in the cytoplasm. You could do this with quantum dots too, and it works, but only in certain size ranges. Very large quantum dots, what you get out is you get specific labeling, but you don't get comprehensive labeling. The counts that you get out when you try to count these molecules are lower than what you'd expect based on controls. And so we evaluated this pretty rigorously in terms of trying to determine what the size threshold is that gives you accurate labeling. So on the left here, what you see are TEM images on the top, as well as FPLC, which is basically size-specific chromatography that gives us the size, that's how much these behave in terms of what they look like in water. So these are always bigger in water than what the TEM images make them appear to be. That's because they're coated with a polymer, they're attached to streptavidin, which allows them to bind to nucleic acid, and they also have a hydrating shell of water molecules and salts surrounding them. So these sizes, we engineer them to be pretty uniform in the range of about 9 nanometers up to about 25 nanometers, and this commercial one is commercially, it's about 25 nanometers. And so what we see is that there's a very distinct size threshold. So on the— what's shown here on this scatter diagram is what we can expect based on our control experiment of a dye that's shown in orange. So a dye gives us a count about 200 counts per cell for this one particular target. We get very similar counts if we use quantum dots, quantum dots labeling these nucleic acids in the cytoplasm when these quantum dots are in the range of 9 to 13 nanometers. But beyond that, 17 nanometers and above, we get much lower counts out. So these are specific counts, meaning that we're getting accurate readings out in that we don't have background counts when the target is not there, but we simply are not getting all the targets labeled. Note that this has to be done under very specific conditions. With dyes, even, I alluded to the fact that there's background binding, but you have to get a certain number labeled per cell in order to detect those brighter spots over this hazy background. The same thing is true with quantum dots, but you need different blocking cocktails. These blocking cocktails we found to be very specifically needing a small amount of, these are, and a large amount of these colloidal block agents, like human. So that's one particular application. I wanted to go back to this real quick, that this 15-nanometer size threshold is something that we've seen repeatedly now. So I'll talk about this again. This seems to be something that's kind of fundamental for fixed cells in biology. These cells are dead, these cells are permeabilized, and that membrane is broken down., and we're trying to label the targets that are in there, and you would expect that you would have a lot of free access of a nanomaterial to bind to something in a cell if it is permeabilized, but it does not seem to be equal access depending on the size. It seems that there are very specific size thresholds that remain even when you have the cell broken down to some extent. The next paper I wanted to talk about is related to this, trying to also find the optimal size, but instead of trying to label intracellular nucleic acids, this is about labeling receptors on the surfaces of cells. This is led by Fong Lee, who is now a postdoc at UCSD, and also this is a collaboration between our lab, Paul Selvin's group in physics, Pablo Perez-Pinera, who's also in bioengineering, and Hee Jung Chung, who is in MCB. And so what this image depicts here is of a neuronal synapse. And so this is a big focus of Paul Suvins' group, is to evaluate receptors in the neural synapse, which is a very important region for using nanomaterials, because we want to see individual receptors, individual molecular events occurring. But it's also a very crowded space. It's about 20 nanometers separating two interconnected neurons. So, you'd have to have a small probe in order to use these. So, we went through a similar type of production of multiple sizes of quantum dots, and this is what's shown here. I'm not going to go through this again. It's very similar quantum dots to what were used for the evaluation of intracellular nucleic acid labeling. And in this case, instead of labeling cells that are fixed, we both labeled cells that are fixed as well as ones that are living. And to do so, we use an assay in which we have both labeled cells available to compare directly in the same field of view with ones that are not labeled. And we can determine what is specific and nonspecific binding based on whether the quantum dots are bound to the labeled target versus ones that are unlabeled. And what we see here on the top right is we can see labeling per cell with 4 different size quantum dots. The quantum dot size is shown on the x-axis as QD 9.2 up to QD 17.4. And we see this increase in both specific binding, which is shown in blue, that occurs pretty dramatically increasing at above 13 nanometers, as well as an increase in the nonspecific binding, which is shown in gray., and so this specific binding— specific binding, if you divide the two, you get the signal-to-noise ratio, or the signal-to-background. And that's what's shown on this bar graph below, that you have very low signal-to-background ratio for the large sizes, 15 nanometers, 17 nanometers, very high for 13 nanometers. And so this is very similar to the same size range that we observed as allowing specific binding in the context of intracellular mRNA, but what's unique here is that we see a very low signal-to-background ratio for QD9.2. And so it turns out that this QD9.2 is simply too— and not engineered well enough yet. So the reason for this, in this paper you can read through this, there's a rationale for this, is simply because this quantum dot that's 9.2 nanometers is the smallest we can make it right now. We're trying to make them smaller than this and to make them even better, but you lose a lot of the great properties when you do that. You lose— many of the quantum dots are not emitting anymore. Many of them are off, meaning that they, they don't emit light. And also, they seem to be more susceptible to their surroundings, like acid. In some, some contexts, these are— if they're exposed to acid, they'll degrade. So, we're still trying to improve these, and we still think we can make these even better than what we have right now. However, they're very satisfactory for labeling living cells. So, fixed cells can be very challenging sometimes, but in the context of evaluating living cell binding to receptors, this data was acquired in Paul Selvin's group, and the goal is to evaluate what percent of these quantum dots bound to receptors are within synapses or extra— synaptic. And so you can see with progressively decreasing the size from the bottom here, 17 nanometers up to 9.2 nanometers, more of these become synaptic, meaning that they're actually bound in their native locations within the synapse, as opposed to only labeling ones that cannot actually enter the synapse. So again, we see this really great improvement going to smaller sizes. For certain applications like rigorous molecular counting, in fixed cells, there's still room to improve these for the very small ones. The next context I wanted to talk about was not actually size related, but what happens when you have different types of physical chemical properties with the same size. This is a project led by Ziyuan Han in our lab. She's a material science and engineering student. She's finishing up her PhD now, probably going to graduate in the late, fall semester, early spring. And so she created a whole series of quantum dots that have the same size, same coating, but the coating has different chemistries. And so these chemistries are depicted here on the right. Some of these are neutral in charge, such as this one that's called POEG, that's oligoethylene glycol. Other ones have charges, so this one that's PNH3+, that's a, that's a an amine polymer. The one that's shown below that is a carboxylic acid polymer. So one is positively charged, one is negatively charged. Then we have these mixtures at the bottom. This PC4ZW, this is a zwitterion, has a balance of both negative and positive charges. And then the one below that is a combination of this uncharged OEG together with the zwitterion. And so what we see is that for cultured membrane receptors, —so cultured cells trying to evaluate membrane surfaces—that they all kind of behave the same. There's no nonspecific binding, except for this one that's cationic, this one that's positively charged. This is a well-known phenomenon. Positive charged molecules bind very strongly to cells in general. In general, cell surfaces are anionic, negatively charged, and in general, most macromolecules in biology, such as globular proteins, are negatively charged too. So this is not surprising. What is surprising is that when you do this in fixed cells, again, when you have access to the intracellular cytoplasm, the membrane is broken down, and you can get nanomaterials into cells, you get very different outcomes when you change the physical properties of these without changing the size. So again, the positively charged one binds very strongly to cells, but what's surprising is that this negatively charged one binds just as effectively, which is in this context a negative consequence of nonspecific binding, because these are not actually labeling a specific molecule. But what's great is that we can reduce this with these two neutral coatings, the one that is uncharged, OEG, and the one that has a net zero charge, the ZW. What's really surprising is that their combination works even better than either one alone. So this is what we call synergy. So combining these two effects together of having reduced nonspecific binding from OEG and azithromycin is, to me, surprising. That's very interesting. It also leads us to new design principles for how to make these even better than they are now. So we believe this is the case because these are blocking two different independent mechanisms of nonspecific binding. OEG strongly adsorbs to water, but it itself is fairly hydrophobic, as opposed to these charged ones which combine very strongly to anions. So the capacity to bind to water and anions too, and cations, gives us a unique capacity to further resist non-specific binding. So one application we've been using these for, particularly for zwitterions, is for intracellular imaging. So usually when you want to image something that's inside of a cell, you have to lyse the cell. You have to break down its membrane in some way in order to get something into the cell. There's some instances where dyes will spontaneously diffuse across membranes, and that's been very, very transformative uses for those types of biological applications where you can image them in living cells. But for nanomaterials, this does not happen. You have to break down the membrane. There's many ways people have explored to try to get nanomaterials into the cytoplasm in order to use them. This is still an ongoing field that's not actually being used to do anything particularly useful yet, but the goal is to eventually have some sort of membrane transport mechanism for a nanomaterial to have these enter cells and to function as a reporter or some sort of a probe for living cell. One example of that is this delivery mechanism that's called osmotic pinosomolysis. It's a 3-step process where you initially have a cell as shown on the left here under normal osmotic conditions, and then if you add high salt buffers to this cell, what happens is that the water leaves the cell to balance its osmotic strength between the inside and outside of the cell,, and when it does that, its surface has to reduce, and so it blebs off endosomes into the cell, and those endosomes actually capture a lot of what's outside the cell. And so if you add a probe during this step, which is only about 5 to 10 minutes, then potentially that enters the cell but is in an endosome, and so it's not actually in the cytoplasm. But in a second step, if you add a really brief exposure to a hypotonic solution, which is much higher concentration of water or lower osmotic strength, then what you get is these endosomes that formed under conditions of high salt rupture, water enters them rapidly and they break apart, and your products end up in the cytoplasm. So this has been used actually for more than 50 years. This was discovered a long time ago. And it's been used originally to deliver macromolecules like proteins into cells. And so we studied this in the context of quantum dot delivery into cells, and you can track the process with both optical techniques, and you can also do paired studies where you evaluate these with TEM. And TEM gives you the ability to see these individual molecules in really high resolution, although you cannot determine if they directly correlate back with the optical images. Because you can't see the fluorescence under TEM. And so we saw is that when you take the same materials again, same sizes, but you just change the physical properties, changing from properties that are either neutral or ones that are neutral with a net zero charge because they're zwitterionic, you get a very different outcome. What's shown on the left here is a weakly bound ligand that's neutral, And on the right here is a strongly bound ligand that's also neutral, it's also zwitterionic. You can see a very big difference in how they behave inside of a cell. One moves very rapidly, one is mostly stationary and kind of bright spots. So we developed image analysis tools for this to determine both how these are moving and what are their photophysical properties in the context of what they're doing inside of a cell. And so this is kind of a product of what you see, that the ones on the left here that were mostly immobile have this scatter plot showing a, um, what we extract out from diffusion measurements of these individual molecules as being a low diffusion constant, low diffusion coefficient, and low alpha value, which is a kind of how much these can move spatially. This is very different than what you see here in this zwitterionic case on the right, that these are high mobility, highly diffusive, and they can move long distances, indicated by this alpha value. Also, we've been evaluating how these are clustered together. So the ones that are very diffusive and can move long distances are also not bright, and that's what you would expect because these are not individual molecules anymore. Some of them have become aggregated and trapped inside of an endosome, and there's many of them in an endosome, and so they behave as one particle that emits many photons simultaneously. So we can independently analyze both of these effects, um, independently analyze are they clustered together as well as are they moving a lot. And so what we're trying to extract out here is this one particular type of intracellular nanocrystal, one that's cytoplasmic, moving a lot, freely diffusive, and also single and individual. This is the one that we hope in the future will allow us to probe events in a cell without any sort of interference. A free single probe inside of a cell is our ideal. And so we see in order to achieve that, it depends strongly on the physical properties of a quantum dot. In particular, this PZW1 is a strongly bound polymer that's zwitterionic. That one outperformed all the others, even ones that were also neutral, like this PEG one. And so this kind of goes back to the same type of phenomenon we saw in fixed cells, in that once you have the membrane permeabilized, the zwitterion has a lot of benefit for use inside of a cell. It seems to resist these nonspecific binding effects that inhibit the use of ones that are charged, and also inhibit the use of ones that just have polyethylene glycol by themselves. So, free cytoplasmic quantum dots are strongly facilitated by stable neutral surfaces. The last paper I wanted to talk about came out last year, also again by Fong, and so her previous paper was on evaluating receptors and making the optimal quantum dots to to prepare them or to evaluate them with. Our main project has been to use quantum dots in order to evaluate discrete events inside cells. In particular, the main goal here is that our current understanding of a lot of cell physiological processes, as well as phenotype changes, relates to how a cell interprets an input. And then decides to do something. So that decision could be that it's going to proliferate, invade into its surroundings, or change phenotype. And it typically won't do that under random scenarios or without conditions of some sort of biochemical event that tells it to do so. These biochemical events are often growth factors, cytokines, hormones, and these input events are usually at low concentration. We have a lot of amazing tools now, particularly ones that have come out in the last 10 years, that gives us this single-cell distribution of phenotype or behavior. So we have tons and tons of tools, a single-cell analysis, particularly single-cell sequencing, and just in the last 5 years or so has been really transformative for evaluating these distribution of cell properties. But what we don't have is one that tells us what is the distribution of input stimuli that led to this particular output distribution. So that's what this technology is developed toward, is toward connecting these dots between a single-cell stimulation distribution leading to a single-cell output distribution of a phenotype. The context of which we're studying this is, um, binding factors. As this is being— which leads to some transcription factor translocation to the nucleus like this, what's shown here in the MAPK process, and that leading to change in transcription, which then leads to something that's a biological change in a cell, often a phenotype change or migration or something like this. This is what it might look like at the single molecule level dynamically on a cell. This is EGF bound to a quantum dot moving on a single cell, and you can see events that take place. This is a big field of evaluating growth factor signaling right now. This is different than that though, in that we're trying to count them as opposed to evaluate dynamics. So we want to know the number of binding events. It turns out we can't do this with most types of fluorescent molecules. Fluorescent proteins, fluorescent dyes, you get something that looks like what's shown here in the bottom. But we're trying to evaluate these individual spots of an individual growth factor binding to a cell, but that's really occluded by this really big background haze of autofluorescence. We found that this disappears when you use quantum dots, when you get out to the range of about 720 nanometers, 740 nanometers, and then the signal-to-noise ratio is so great that you can really confidently say this is an individual molecular binding event. So this is an ROC curve depicting that for these different colors of quantum dots. What you can see is we get really great confidence and accuracy of determining whether this event is actually an individual molecule or some sort of background event once we get out to the 700 nanometer range. So the AUC, or area under the curve, for the ROC is about 99%. When you try to do this with the dye, it's extremely low. Very low confidence that you're actually detecting all the events and very low confidence that you're actually detecting an event that's not autofluorescence. And so we can now quantify these molecular binding events across diverse types of cells. That's what's shown here with 3 different types of cells exposed to the same number of growth factors. Counting these growth factors on these cells, it should be proportional to the number of receptors that are there. And that is what we see. These 3 different cell types that are all human all expressing EGF receptor at different levels. We get relatively a proportional number of binding events to these cells, depending on the number of receptors that they have, under this one condition of nanomolar concentration of this growth factor. And you see a great, great improvement in the ability to resolve these and count them when you use quantum dots as compared with dyes. That's what's shown here in a comparison. The same cells here are bound by quantum dot EGF conjugates, as well as dye EGF conjugates. And in order to image these in 3 dimensions, you have to progressively scan through Z-stacks of acquisition. And what you can see here is if we're imaging from the bottom to the top, quantum dots look like they're bound uniformly throughout the cell, as opposed to dyes, which are shown here summarized on the right, that you get a really bright signal at the bottom of the cell. By the time you get to the top, you've photobleached all of them. The inverse occurs when you image from the top to the bottom. So most of the dyes look like they're at the top, and then you get to the bottom and they're all disappeared. So effectively, you can't do this experiment and quantify and count all these molecules that are there without a quantum dot. So the outcomes that we get are here, which is what I'm most excited about, is that we can now evaluate stimuli together with an output. So here we're evaluating this output as being internalization of EGF. That's the first step in the signaling cascade, is that EGF binds to EGF receptor. That receptor complex then usually dimerizes, heteromerizes, gets internalized, and that leads to a cascade of intracellular signaling events that are like through the MAPK pathway. And so we can correlate this, how much of it enters versus the number that is there. And what's great is that we can get out these dependencies based on a drug. So what's shown here is that going from the initial events of binding to about 30 minutes later, all these EGF molecules are internalized. That tells us that the signaling cascade has begun. If we add a receptor blocker to this, in particular this one is a kinase inhibitor called gefitinib, If we add that in low and high concentrations, we see that we inhibit this effect from occurring. This is a well-known phenomenon. This inhibitor is a small molecule kinase inhibitor. It blocks the receptor. It also blocks all the downstream processes as well. So if we track these events, we can now get a population distribution out of which cells are responding to the drug. And then we can backtrack if that cell responded or not, is because it had a low or high level of stimulation. So that's what's shown at the bottom here. Under no stimulation, effectively all the receptors are internalized. With minimal— with small levels of increase of gofitinib, the blocker, to high levels, which are well above the IC50 of this drug, what you see is that fewer of the receptors are internalized, which you would expect. But also there's a dependence on the number of stimulation events. And so this means that basically you could overcome this drug by simply stimulating more. And this is what you should expect to see too. So basically, if you have very low stimulation events, which is kind of what's shown on the, on the bottom left here, fewer numbers of EGF per cell, a larger fraction of the blocked events are effectively blocked. High stimulation events, you can't effectively block it. So this gives us an indication of one of the sources of drug resistance, being you can oversaturate the biological process that you are targeting with your drug. And in local microenvironments where you might have very high concentrations of EGF, even if you have an EGF blocker, the same signaling cascade can proceed simply because you've highly saturated the process beyond what the drug can inhibit. So this is one potential thing that you can get out of these types of experiments with quantum dots if they're designed correctly. So that's the conclusion of what I wanted to go through today. These are the last, really, 2 years of research products from our lab in the context of quantum dot engineering for single molecule imaging, and I've been very fortunate to work with some fantastic scientists and engineers in in this project and since I've been at Illinois since 2012. In particular, Yang, Suresh, Sungjun, Ziyen, Fong, and Mohammad really made these projects their own and they really took them to the next level. And I worked very closely together with many collaborators both on campus and outside of campus. In particular, Paul Selvin, Pablo, Fres Piñera, Heejung Chung are really key for these projects to proceed. And everything to do with quantum.fish really took place because of some great collaborations with the Mayo Clinic that originated about 5 years ago. And we're grateful that these projects were supported by both the Cancer Center at Illinois as well as NIH through the NCI and NIDDK— and actually NINDS, sorry. So thanks for your attention. I'd be happy to take any questions.

Thank you, Andrew. You think the cytoplasmic release of nanoparticle probes in living cell, has that problem been solved?

No, I don't think so. I think there's a lot of groups that have explored this. At a pretty high level of depth, and the outcome is that there's something that prevents these from binding to the targets, and it's not clear what it is. So there's been many publications now on the use of antibody-based probes of these orthogonal ligation chemistries like HaloTags, and it seems that a large fraction of the molecules don't bind to their targets when you have a nanoparticle. There's been a good number of papers where people have used dyes and kind of more simplistic types of probes in cells, and they seem to be much less obstructed. Particular things like a phalloidin dye conjugate, if you can get into a cell, it can bind to actin pretty sensitively, probably not all 100% is bound. This does not seem to take place with nanomaterials, and I project that there is a very much more stringent size threshold for this than what you have in a fixed and permeabilized cell. It seems that there are microscopic microenvironments that these cannot access, So no, that has not been solved with nanomaterials.

So these nanoparticle probes, including quantum dot, still work much better outside the cell, either for receptor binding, imaging, in vitro diagnostics, than inside the cell, right?

So the applications still are mainly outside cell. For living systems, yeah, I would definitely say so. Fixed cells, yeah, they do. I think they do work very well for fixed cells. Yeah, the dead cell. That's right.

Once they're permeabilized. Living cell, intracellular, looking at signal transduction, there's still some challenges.

And definitely, that's even beyond nanomaterials, too. I think that's still the case for organic dyes. Definitely, fluorescent proteins are the way to go for anything living that can be expressed. Transgenically. Thank you.

Thanks, Zhiming. Hi, Andrew. This is Fan. Thanks for a great talk. So I'm curious, has anyone looked into using quantum dot beyond as a sensing mechanism and using it as some kind of stimulation vehicle? For example, is anyone exploring using quantum dot for optogenetic applications?

Yeah, it's a great question. That their use for what I presented here is all as a label and a probe, meaning that you basically— they don't turn on and turn off with their surroundings, or at least you don't want them to. You want them to just label and then wash everything else away. But the ability to have them immersed in an environment and responding to its environment is something very different. That has been a very big field. There's a lot of labs that work on that. In terms of making something that's a product that's widely used, there's really none. Almost everything is based on FRET, fluorescence resonance energy transfer, and it's very hard to make these functional for FRET in comparison with dyes, which are much more molecularly precise, and it's easier to interpret changes in fluorescence. In the context of optogenetics, yeah, people have— there are several labs that are focused on trying to make these tools change in emission in response to a voltage change on a cell. That's very hard to do unless you can get it into the membrane in some way. So people have made FRET reporters where the original source of the signal comes from a fluorescent protein, and then that transfers to a quantum dot, which gives you some value because you can make them emit at longer wavelengths, or potentially track them for longer periods of time. The value for that, I don't know, is that substantial yet. In terms of actually using them to modulate something biologically, that's also not quite there yet. The problem is that these materials usually want to have an insulating coating around them, and that insulating coating functions like an electrical insulator. It provides the ability to prevent the optical properties from degrading, because the nanocrystal itself is a semiconductor, and if it's surrounded by a cladding layer like a laser, it has better optical properties. It absorbs and emits light better, and it doesn't degrade and oxidize. And so if you try to make it a balance between that and being able to respond to an electrical change in its surroundings, it's complicated. So potentially you could do that by integrating into a membrane in some way. That's still very challenging. So those, I think, are very exciting fields, and a lot of what's missing is simply the biophysics of how to integrate it. We know under very controlled conditions they can stimulate a local voltage change, and they can respond to them, but in a cell, we simply don't understand that microscopic interface. That needs to be solved, I think, in order to really make them integrate into cells. I'm not sure that they'll catch up though. I think the value of optogenetic fluorescent proteins is so high, and they keep getting better and better. I'm not sure they're going to provide a value in the next 5, 10 years that's going to be worth the development costs.

Cool. Thanks.

Good question.

Since your emission is so bright, so I was wondering whether it could be used as a more efficient modulation or stimulation device instead of just a probe and sensing.

Yeah, they should be able to. In that, if you're trying to modulate something with them, It's kind of the opposite goal for most of quantum engineering, which is trying to maximize their fluorescence. So you want to have a zero quantum yield, then you want to not convert any of the absorbed light into fluorescence, and you want to convert that into a local electrical change, like a current. And, um, that's, that's the goal of photovoltaics. And so photovoltaics, you want to have this really nice interface between the semiconductor absorbing material and the ability to have a current or an electron released from the material. And in biology, everything's oxidizing. Typically, you have oxidizing conditions, and that dominates when you have absorbed light occur. Instead of actually releasing the electron, you actually get an oxidizing event that occurs on the surface and it breaks down. That's for the most materials that are used right now. We have binary materials that have an anion and a cation in them. Theoretically, if you make these out of, um, completely covalent materials like silicon, carbon, those potentially would not be prone to any of the chemical degradation processes that inhibit the ability to use these for modulation. So materials that are being used right now are not ideal. So there is, I think, a lot of potential for that, but somebody needs to refocus the use of these for the correct compositions that will allow it to succeed. Great, thank you. If you want to do that, let me know. Thanks. Lots of projects. There's a, there's a question in the chat. Why dots around 9 nanometers can't function well? That's a great question. So it's not a fundamental reason why they can't in the long term. It's that the best ones we have so far are optically not as great as the others. So there's a big challenge to making these really tiny. And the big challenge is that you have a core and a shell, and this core gives you the color and it gives you the absorbing efficiency, and the shell dictates the stability and the quantum yield, basically how efficient they are at emitting light after they've absorbed light. And we have to minimize the size somehow to get it really tiny like this, like the 9.2 nanometer one. Its core size is 3.3 nanometers. So the coating is actually what dictates the ultimate size. And so to make it this small, we really had to reduce the shell as well as the core. And it's a balance. So basically, if we reduce the core too much, it won't be bright enough to see very well. If we reduce the shell too much, the quantum yield drops. So we kind of have to balance them together, and that 3.3 nanometers is that balance. But it's still not perfected. We can make these out of slightly different materials. We can anneal them in ways that we get better coating consistency. And so ultimately, we don't think this will be a limitation. It's just this is the current state of the art. So the current state of the art is that we now realize that these smallest materials need to be more perfected in the manufacturing, processing, and probably the treatment of how we make them emit light in this spectral range when they're this small. So it's not an ultimate failure of them, it's that we need to spend more time developing them. There's many, many ways to make these same materials. So that's one of the unique features of these is that you have the whole landscape of semiconductors to work with, and you also have heterostructures where they're made out of core and shell materials with different compositions together in different ways. And so finding the right ultimate design is something that's, um, we do these experiments to inform our designs. And so now we have a better understanding of how to redesign them, which is what we're doing now.