Scientist Stories: Carolyn Bertozzi, Chemistry in Living Systems

So good afternoon and welcome to this 5th edition of the ISIS Lectureships. And before I introduce our distinguished speaker and guest, let me take half a minute or so to thank Dr. Eric Swayze and his associates. Eric is VP research at Isis Pharmaceuticals for their continued support. Like I said, this is the 5th year not only of supporting these fantastic lectureships where they allow us to bring in the most eminent and tremendous chemists from all over the world. The list is on the back. That you can see, and today is no exception. So thanks again, Eric, and your associates, the whole Isis Pharmaceuticals gang, to allow us to do this. It's a great personal pleasure to introduce our speaker, Dr. Bertozzi-Carolin. I hope you've had a chance to look at her very concise and brief CV. A very modest person, and her— for her tremendous accomplishments, many, many awards, international awards, and also her dedication to teaching. I think this is something that's exceedingly important. Just by way of just a few words about this whole fascinating area that Carolyn is interested in, it used to be that we would say— nucleosides or nucleotides for control, proteins for recognition, and carbohydrates for fuel and lipids. That was a long time ago. This is the age of the dinosaurs when I did some carbohydrate chemistry. But things have actually changed, and Professor Bertozzi, through her really seminal work, her tremendous insights into the field, has come in and really revolutionized the whole area of glycobiology. One of the most difficult things for a young starting academic, or even not so young— I go back to her starting days— is really to choose a topic of research where you'll have some impact, you'll make some impact down the line as the years go by, in addition, of course, to working with fantastically talented young people like you. And Carolyn has been exceptionally successful at doing that and staying the course, choosing an area that had— that was really unmet, if you wish, chemical and biological need and has done fantastic work in there. So the title of her talk is Chemistry in Living Systems. This is very important, not of living systems, but that very appropriate title. So without further ado, Carolyn, welcome to UCI.

Thank you so much, Professor Hanesian, Stephen, as I call him now. Unlike the first time we met, which I'm sure you don't remember, but I was like a second-year graduate student and you were our student invited speaker at UC Berkeley. This was like in the late 1980s. And I think I might have introduced you then. So who knew back then? I introduced you as one of the founding fathers of modern carbohydrate chemistry, and you earned that title. So let me also join in thanking ISIS for their generous support of this lecture, and thanks for coming up here. And I look forward to seeing you all and your colleagues tomorrow as well. So what I thought I would do today is basically put together a collection of stories from the lab that all share this common general theme of chemistry in living systems. And hopefully this collection of stories, which range from, you know, efforts to do protein chemistry for the goal of making biotherapeutics, all the way to in vivo imaging problems that we're interested in— but hopefully you'll see kind of the opportunities that are out there if in fact it becomes possible to reliably do chemistry in living systems, which I think is a reality today. Okay, so I think understanding the challenges that are inherent to chemistry and living systems, it helps to kind of first take stock of chemistry historically, which has been in pretty much the opposite of living systems. I mean, you know, I was trained as a synthetic organic chemist, and most of us who are practitioners of this art are most familiar with reaction environments like this one. And what's implicit in this complicated device, is that it affords you very rigid control over all the parameters of the reaction. And something you kind of take for granted because, you know, we do these reactions all the time in our fume hoods and so on. But, but it's that control that gives you access to the diversity of organic chemical transformations that we enjoy today. And what I mean by that is that, you know, if you have a flask like this, you can control all the details like the solvent's identity, you can control the temperature, you can control the atmosphere, you can even control the frequency of the stir bar, right? You can control the order of addition of reagents and the rate of addition. You can subtract byproducts that might interfere with the reaction. And all of that actually is quite important for a majority of chemistries that we enjoy. And so the problem with doing chemistry in biological systems, on biological molecules, in biological settings, is that you have to forfeit a majority of that control. And when you do that, all of a sudden, you know, the thousands of pages from the organic reaction compendia that sit on our shelves all of a sudden become essentially useless. And it's within those boundary conditions that I think there's an opportunity to develop really new classes of reactions, where the reaction now has to have selectivities and biocompatibilities that compensate for your loss of control. And you lose a lot of control even when you just start choosing molecules that are not canonical small organic compounds. Like, for example, if you want to do chemistry on a protein, all of a sudden you've lost a lot of control over solvent choice. Your proteins are probably only soluble in water, aqueous buffers of certain ionic strengths and certain pHs. You've lost control over temperature options. If you heat it up too high, you might denature and precipitate, and you'll never get that thing back in solution. Right, so right away life changes dramatically for the organic chemist who wants to do synthesis using protein substrates. And then the boundary conditions really clamp down on you if your goal is to do that chemistry in what I would think of as a physiologically authentic environment. Like, for example, perhaps you'd like to do chemistry on this protein when the protein is in its native habitat. Maybe you'd like to image that protein and follow it around, or maybe you'd like to change its properties in situ. And study in a dynamic fashion what happens to the cell after that. And to do chemistry in a system like this, in a living cell, all of a sudden restricts you to the use of non-toxic reagents, and you're even more limited with respect to temperature ranges and pHs and so on. And then ultimately, you know, if you'd like to do that chemistry in the setting of an intact organism, like a model organism like a zebrafish, which is something we work with in our lab, Now you have to worry about metabolism and circulation and serum half-lives and, and all the issues that medicinal chemists have to contend with that add so many years to the process of drug development. And so it's the challenges that are embodied in these kinds of systems that basically prompted people in my lab and many other labs, like here, the lab of Jen Prescher, to spend time thinking about how do you develop reactions, how do you invent chemistries, or repurpose chemistries so that you can actually do transformations in these living systems without stepping on a mine. I mean, these are minefields, right? And it all comes down to selectivities and biocompatibilities and interestingly kinetics. So there's a term that the field now uses that kind of encompasses all of these boundary conditions, and we often refer to these kinds of chemistries as bio-orthogonal. Literally, what we meant when we coined that term is that bio-orthogonal chemistry is chemistry among functional groups that neither interact with nor interfere with a biological system. So, you know, there is a certain biological reactivity space that you can define, and, you know, it's huge, right? There's a lot of different functionalities in a biological system, but it's not infinite. It does have boundaries. And you can think of that as a galaxy over here. And our assumption was that in the universe of chemistries, there might be other galaxies where you have chemical functional groups that really have no counterparts in the biological setting and also don't interact with those biofunctionalities. And so this is the space that the field has been trying to explore now for really what's almost 15 years. So with that, very general introduction. This is kind of the menu of the talk. And the first example I'll give you, which was kind of a problem in bioorthogonal chemistry that then drove a reaction development program, has to do with site-specific protein chemical modification and the goal of making biotherapeutics. And then the second two topics are kind of related in that they both deal with our interests in in vivo imaging. In the first case, vertebrate glycans. And this is, you know, kind of like the direct descendant of my first exposure to carbohydrate chemistry back in my graduate school days. And then more recently, we've become interested in in vivo imaging of bacterial cell wall structures with the hope that we can shed some light on the role of the cell wall during microbial pathogenesis. Okay, so starting with protein chemical modification, the biotech industry has been chemically modifying proteins for actually quite a while. And many chemically modified drugs are on the market today. Probably the most recognizable are pegylated proteins, right? So these are recombinant proteins that have been chemically conjugated to a synthetic polymer, which is polyethylene glycol. And the reason people pegylate proteins generally is because that can increase their serum residence time and also prevent their proteolysis. So proteins that have poor pharmacokinetic profiles can be improved dramatically by, by pegylation. More recently, I would say that the most recognizable chemically modified protein drugs are the antibody drug conjugates, and these have become kind of the darlings of the last few years in biotech and also in big pharma. These are right now mostly targeted towards cancer indications, and the idea is one that is conceptually simple and conceptually been around for many decades, is that, you know, if you can chemically conjugate a cytotoxic drug to an antibody that is specific for a tumor-specific antigen, then not only can you concentrate the drug on the tumor and in the tumor and kill the tumor cells, but you can also prevent unwanted side effects that would be the result of systemic toxicity. These molecules, people have been talking about these, and the industry has been working on these for several decades. Oh, it's so rude when a cell phone goes off. That's my cell phone, actually. Excuse me. Oh my goodness, that's just really embarrassing. Jen, somehow that's your fault. Kind of brings you back, doesn't it? All right, anyways, so antibody drug conjugates were basically, you know, on everybody's mind for many years, but it took until 2012, you know, for the first of these to be clinically approved in a way where one felt confident that this would have lasting clinical impact, and that was a molecule generated by Seattle Genetics. A few months after that, Genentech had their first approval, and now, you know, Genentech's pipeline is rich with antibody-drug conjugate candidates. And so, you know, who's interested in chemically modified protein drugs? Everybody. So these are like all the logos of the companies where they said something on their website about either pegylated proteins or antibody drug conjugates, which means that there's a lot of people in the industry who are thinking about how to do chemistry on proteins. So, you know, some of these molecules already are in the clinic, so clearly, you know, some problem has been solved at a certain level. But let me just cut to the chase and tell you that it's a mess, okay? And I thought a good illustration of how messy the chemistry is that's been used to date is to actually show you how Genentech's compound called Cadxyla, that's its commercial name. It's a Herceptin cytotoxin conjugate, so it's for basically HER2-positive breast cancers. So what people in the industry have been doing is kind of old-school bioconjugation chemistry, which means that they basically will modify a natural amino acid side chain, and typically it's either lysine or cysteine. In the case of Cadxyla, it's lysine residues. So they take the antibody molecule, which has probably around 100 surface-accessible lysines on it, and they chemically conjugate the toxic drug, which is armed with an NHS ester, which gives selective acylation reactivity for lysines, but it cannot pick out one or a few lysines over the 100 that are there. And generally, they're kind of targeting a certain stoichiometry that empirically has been identified as optimal. So Genentech, you know, aims to get maybe 3 or 4 at most on average drugs conjugated to their antibody. But since they can't control the sites of modification or the stoichiometry of modification, they just basically kind of tune these reagents and the conditions to get a mixture that on average might have 3.5 drugs per molecule. But if you actually look at the analytics behind this clinically approved molecule, it's— you'll see how complicated it is. By the way, so this is a mass spec of basically a preclinical sample of CadSila, where each of these peaks corresponds to a different antibody-drug ratio, right? So down here, this is like an antibody molecule with 3 drugs. Here's one with 4, 5, 6, 7, 8, and so on and so forth. That was presented on a poster that their scientists gave at an AACR meeting up in San Francisco, and I had my cell phone camera, and I kind of took a picture of it. So Hope that's okay. Don't tell anybody. It was a public meeting, and this is being recorded, so I should be quiet now. Anyway, so, you know, right away you can see that this is a very heterogeneous mixture of molecules. And if you pick any one of these stoichiometries, like let's say D6, okay, so there's actually many, many different species within this peak because there's lots of ways you can get 6 drugs on an antibody if you're targeting 100 lysines, right? So you can do the combinatorics and the predictions on this. My guess is that clinically approved capsyla is probably a mixture of upwards of 10,000 different molecular species. 10,000 compounds are in there. So you know as chemists that some of these molecules are better than others. Some are more active and efficacious than others. Some are more toxic and are bigger liabilities than others. But if you can't separate them out, you'll never be able to identify the best possible structure, and you'll never be able to optimize it either. And everyone kind of knows that this is a problem in the industry, but until now, there really has not been a viable approach to controlling the chemistry of the protein modification so that you could systematically vary the stoichiometry and get homogeneous species and test them so that you could move the sites of the drug modification around the antibody and find what's the best regiochemistry, right? Things that small molecule medicinal chemists do routinely but has never really been translatable to these chemically modified protein therapeutics. So this is why there's kind of recently been a big push in the industry with a lot of support from academic labs to develop methods for site-specific protein modification that allows you to get homogeneous, chemically defined molecules and test them and optimize them. And, you know, the approach that everyone's thinking about quite generally is to introduce orthogonal chemical capability site-specifically onto the protein. And the idea is that, you know, if you could identify a functional group that is orthogonal to all of the natural amino acid side chains, You could then direct the chemistry to that site, and to the extent that you can control where the X group is, you can now control the modification and make all the compounds you want and do SAR, just like a medicinal chemist would do with small molecules. So this is kind of a classic problem in bioorthogonal chemistry where the orthogonality is defined by the protein's intrinsic functionality. We're certainly not the first people to think about this problem and to try and offer a solution. You know, in academic settings, there's a good two decades of work that has focused on introducing new chemistries, new functionalities into proteins, and then modifying those proteins. And just to highlight a few that will be recognizable, particularly to the students of Chem 219 here at Irvine— I understand that you just had a lecture on this, is that right? So everyone's like, we did? Oh my god. Well, Professor Prescher assures me this will be on the midterm, so I can just give you that right now. Okay? Here it is. So these are two methods that originated from academic settings, and the one on the left you might recognize as what's called unnatural amino acid mutagenesis. This is a technique that was pioneered by Peter Schultz, and it was spun out in various interesting directions by David Terrell and Jason Chin and Peng Chen, and there's been a lot of contributors. But basically, in a nutshell, what you do is you re-engineer the entire protein translational machinery so that you can encode an unnatural amino acid opposite to what normally would be a stop codon. OK, so this is actually now being used in a commercial setting to generate antibody drug conjugates as well as other site-specifically modified protein drugs. So this has actually progressed quite a way since its academic infancy. Over here we have what's called the native chemical ligation. This is where you express a protein fragment recombinantly and another one you make synthetically, and the synthetic fragment you can introduce any modification you want because you're making it yourself. And then you link them together using this really interesting thioester rearrangement chemistry. So far, to my knowledge, this has not been translated into a commercial setting, but, you know, that's something that might be contemplated by its various founders like Steve Kent, Tom Muir, and Phil Dawson. So, you know, people have tried to address this problem in a number of ways. I would argue though that these methods are pretty complicated technically. You know, it took a good decade of work just to get this methodology to the point where you could make these proteins in cells as opposed to in vitro, and whether this can be scaled to the level of biofermentation for a clinical product has yet to be proven, but obviously they're going to work hard on it. So we've tried to make a contribution as kind of latecomers to the business, and our goal was to try and come up with a way of doing this that was simple, operationally very simple and low technical barrier, didn't require a lot of cell engineering, and scalable. And we kind of had an insight into how this might be accomplished in the course of a completely unrelated project on on a family of enzymes called sulfatases. So I won't go into the details of why we were interested in sulfatases. There's a whole interesting biology around these enzymes. But they have a kind of interesting mechanistic idiosyncrasy, which is that they rely on a non-canonical amino acid called formylglycine, which resides in their active site for catalysis. The thought process is that this formylglycine is required because it gets hydrated in the active site of these enzymes, which produces a gem-diol. And then one of these hydroxy groups is a catalytic residue, a nucleophilic catalytic residue that facilitates sulfate ester hydrolysis much in the same way that a serine protease hydrolyzes an amide bond, or a serine esterase. And so this is something that the kind of enzymology community has been curious about for a few years now. But what got us excited about these sulfatases and the formylglycine is the fact that formylglycine has an aldehyde group, and that the chemical reactivity of an aldehyde is fundamentally orthogonal to the reactivities of the canonical amino acid side chains. Aldehydes are electrophiles. You don't see electrophiles on proteins, pretty much, at least not reactive ones. And the organic chemists in the audience think of an aldehyde as a versatile synthetic building block. There's lots of different chemistries you can do to conjugate things to aldehydes. So this is a discovery that was made probably around 1995. And so it was biochemically identified in sulfatases just by digesting the proteins and looking at the amino acids. But it was not known where formaldehyde comes from until 2003. And what happened that year is two papers were published back-to-back in Cell on the discovery of a single enzyme that generates the formylglycine residue on these sulfatases. And it was quite descriptively, but rather uncreatively, given the name formylglycine generating enzyme, okay, or FGE for short. So what FGE does is it recognizes a conserved amino acid sequence. Which is found in all sulfatases, which is a mere 5 residues, very short peptide sequence. And it recognizes specifically the cysteine side chain of this residue and does an oxidation reaction on that side chain to generate the aldehyde. And this is another really interesting beast of enzymology. FGE, for those of you who are kind of enzymology buffs, like to think about enzyme mechanisms, FGE does not use any cofactors. It's an oxidase, but there's no flavin, no copper, no NAD+, right? None of the kind of usual suspects that you associate with oxidation chemistries. FGE has no cofactors, and it uses molecular oxygen as the stoichiometric oxidant. It activates oxygen with no metal and no flavin. Pretty cool. So, you know, think about that if you're interested. But knowing the consensus sequence and and the enzymatic machinery now enables you to think about ways that you might be able to harness this system as a technology for doing site-specific protein modification. Because it occurred to us, if you— it might be possible to take that sequence and clone it into heterologous proteins, right? What if you could lift that sequence, put it in your protein of interest, you know, just doing basic molecular biology, and if FGE is agnostic to the surrounding sequences and really is only looking for this, conceivable that you could express any protein with a formylglycine residue and use the aldehyde to do chemistry rather than catalysis. So that all came to life in my lab kind of around the 2008 period. And, you know, we demonstrated in a number of model systems that in fact you can clone the sequence into really lots of different kinds of proteins. We really haven't found a limit on this. And as long as you express those proteins in cells, that also express kind of an abundance of FGE, enough FGE to keep up with the protein expression that you're inducing, then all you have to do is get these two genes expressed at the same time, and the cells will spit out your protein with an aldehyde group at the site encoded by the cysteine. So that's the hard part. Now, of course, there's actually a literature on aldehyde condensation chemistries that are already known to be protein compatible. For example, it's well known that you can conjugate aldehydes to amino oxy groups and generate oximes. And that is an aqueous chemistry that's very mild and selective, and that people had been using for bioconjugations in other systems for a long time. So that was kind of like the first chemistry that we employed in conjunction with this thing that we call the aldehyde tag to do site-specific protein modification. And by the way, in terms of technical complexity, it really couldn't be much simpler than this. So introducing an aldehyde tag into your protein is as simple as introducing a histag. It's 5 residues instead of 6, in fact. OK, so that's how you should think of this. That is pretty simple. And this chemistry is very simple too, but for some of the applications to biotherapeutics, we found it to be non-ideal, primarily because oximes are actually quite hydrolytically labile. So this is a chemistry that at higher concentrations will favor the adduct, but if you dilute oximes into aqueous media, they will start to hydrolyze. Over time. And so one of the methodology projects that this opportunity sparked in the lab was an effort to develop a new aldehyde-specific bioorthogonal reaction that would produce a stable product, an irreversible product. And if you're an organic chemist, irreversibility, sometimes you think of as, well, let's make a carbon-carbon bond. I mean, that should do it. So there's lots of ways of making carbon-carbon bonds with aldehydes. Most of them, however, you would never do in water or with a protein, right? Like the 1,000 undergrads that I teach that do a Grignard reaction every year. OK, that's one example. But there are a few that look promising, like you could potentially render them aqueous compatible. And one that stood out in the mind of my student Paresh Agarwal was the classic Pictet-Spengler reaction. So this is one of the named reactions that many of us memorize when we take our graduate synthesis class. And it dates back to like 1905. So like an oldie but goodie. And in its classical rendition, what you do is condense with your aldehyde a tryptamine nucleophile, where the first step is the formation of a reactive iminium ion intermediate. And this, of course, would be a reversible reaction, the other product of which is water. And then this is set up for an intramolecular electrophilic aromatic substitution. So once you set up this iminium ion, you can close this with a carbon-carbon bond and make basically irreversibly a product, a stable product. In its classic form, the Pick-Tet-Spengler reaction wasn't ideal for what we wanted to do. First of all, it's actually pretty slow. And this is something that you kind of get calibrated on if you work in this field over time. But let me just tell you that the rate constant of the classic reaction is around 10 to the minus 4 per molar per second, second order rate constant. And what that translates to in a laboratory setting is if you're doing a classic Pick-Tet-Spengler reaction, you're probably going to do it refluxing benzene, like for many hours with a Dean-Stark trap, pull off the water and drive it forward. And so that's too harsh for most proteins. And so we knew that we probably had to boost the rate constant by several orders of magnitude in order to get this reaction to go in a few hours at room temperature. And we also didn't want to have to drive off the water because water is going to be our solvent. So Paresh went through kind of a series of mechanistic tweaks. And what he finally arrived at was— kind of a rendition that we call the Pictet-Spengler ligation. The important modifications he made versus the classic reaction is first he replaced the tryptamine, this primary amine, with an N-methylamino oxy group. And so just having an oxygen bound to the nitrogen increases its nucleophilicity by several orders of magnitude due to the alpha effect. And so the first step, the condensation of this nitrogen with the aldehyde to make this oximinium ion, is much faster. Than formation of the iminium ion. It's also a more stable adduct, and so you accumulate more of this intermediate even when water is present as the solvent. And because there's a methyl group here, there's no opportunity to deprotonate. For example, if this were a proton, you would imagine just deprotonating and generating an oxime, and that would be a thermodynamic trap. But as long as the methyl group is here, this is trapped in its positively charged state. You have a good electrophile. And now cyclization occurs. You might also notice that in the classic Pictet-Spengler reaction, the amine is linked to the 3-carbon of the indole rather than the 2-carbon. We moved that over, and that's because the 3-carbon is actually the most nucleophilic carbon on the indole. We wanted that carbon to participate in ring cyclization. That also contributes to the overall kinetics of the reaction. So add it all up, and now we have a chemistry that has a second-order rate constant that's 4 to 5 orders of magnitude faster. Than the classic reaction. So this Pictet-Spengler ligation goes at room temperature in aqueous buffers in a matter of a few hours. And then that enabled Paresh to do some antibody chemistry. So what he did is he got his hands on the genes encoding Herceptin, Genentech's cancer drug, cloned in the aldehyde tag near the C-terminus, expressed the proteins, heavy chain and light chain, in CHO cells, and then basically took it to the test with a variety of different Pictet-Spengler ligation substrates. So generically, these are substrates where we have this indole moiety— that's kind of the business end— to which Peresh conjugated a linker and a variety of different cargoes. Like, this could be a fluorescent dye, and he's also done it with a variety of different drug-like molecules. And then using a variety of analytical tools like gels and fluorescent gels and Western blots, you can show quite definitively that the chemistry occurs exclusively on the heavy chain, and there's no side reactivity with other parts of the protein, which is great. 150-kilodalton protein. So there's a lot of functionality there. And importantly, after doing the chemistry, there's no ill effects on the ability of this antibody to bind to its targeted antigen, which is the HER2-neu. And that is something Parash demonstrates by just labeling cells that are HER2-positive and doing flow cytometry analysis. And I won't go through the details. It's all published. But the bottom line is— What I'm meaning to showcase here is how a problem in, in this case, biotechnology, you know, has a solution that requires bioorthogonal chemistry and how that can drive a methodology project that's a lot of fun for an organic synthetic chemist to work on. OK, so now I'll jump ahead and tell you about different kinds of chemistries that we've developed now because of the demand to take the chemistry in vivo into living systems. And so now we finally get to talk about sugars. Steve, you were wondering, probably. OK. He's like, no. It's fine. So like I mentioned earlier, when I was in grad school, I first learned about sugars and the challenges and the art of sugar chemistry and became interested in the biology of sugars, which was very undeveloped back then. And it's still relatively underdeveloped. The more we learn about sugars, the more excited I get about them, because it's become very clear that they are a very sophisticated integrated data set. So this cartoon is meant to illustrate the fact that in vertebrates, in fact most eukaryotic systems, sugars are kind of residents of the cell surface, and they're usually found on cell membranes, either conjugated to proteins, and we call them glycoproteins, or conjugated to lipids. And there's a lot of different varieties of glycoproteins with lots of different architectures. There are glycoproteins that are little globular structures. There are glycoproteins that are like giant redwood trees, and the glycolipids are kind of like the grass, grassy meadows. So there's a very rich topology on cells as a consequence of all these sugars. And these sugars, they're like a really information-rich barcode. And we don't really know how to read it, but we're pretty sure it's trying to tell us something, okay? What we do know is that if you think of the cell surface glycans in their totality as a collection, they're basically integrating data from gene expression, from nutrient availability, from central metabolism. They're very sensitive to the microenvironment of the cell, okay? And they're dynamic. So we know that the collection of cell surface glycans changes when cells undergo a physiological change. It's quite clear, for example, that the collection of glycans— and by the way, nowadays we call that the glycome. You didn't need another -ome, I know, but we have one that I'll offer you. And the glycome of a pluripotent human embryonic stem cell is quite different from all the differentiated cell types. So there are markers. Differentiation, markers of stem cell pluripotency that are based in sugars. We know that the glycome of a healthy cell transforms when a cell undergoes a malignant transformation. There are cancer-specific patterns of glycosylation, okay? And these things are actually quite stereotypical and quite predictive of pathology. So, you know, why don't we know more about what the glycome is trying to tell us? In part, it's because we really don't have tools to read it. And so when I started my lab at Berkeley after doing a postdoc in the field of glycobiology and PhD in carbohydrate chemistry, you know, one of my mission statements was to try and bring chemistry into this field in a way that would allow people to read something about the glycome. And so we focused on reading the glycome through imaging technologies, which could conceivably be done on live intact cells and even in organisms and maybe someday in human patients. So our strategy is conceptually quite straightforward, and it all begins with what we call metabolic labeling. And the idea was to kind of tap into the cell's metabolic systems so that you could deliver an unnatural, chemically functionalized sugar into those cell surface glycans by simply adding that sugar to cell culture media, injecting it into a laboratory animal, maybe even feeding it to a human. And then, you know, if one could accomplish that step, there would be a chemical functionality embedded within that unnatural sugar that allowed a second step, which was a chemical reaction step, so that a probe molecule that was suited for a particular imaging modality— so a fluorescent probe for optical imaging, a radiolabeled probe for clinical imaging— and that the reaction of these two would now allow you to use the intensity of this probe as a proxy for something about the composition of the glycome. So if you see this probe, that means that sugar was there, and now you can say something about at least that sugar and its contribution to the glycome. And this, although it was easy to draw this cartoon on paper for my first-year graduate students when I first started my lab, took a long time to actually realize this in the laboratory. And the fundamental problem was to figure out what should this chemistry be? Well, it can't be aldehyde chemistry. Because unlike a protein where an aldehyde is orthogonal, if the aim is to do this on cells or organisms— now, there's plenty of aldehydes in your cell and plenty of aldehydes in organisms. They're metabolites all over the place. Glucose is an aldehyde. It's small percentages. So the demands of orthogonality were much higher for this problem. And it basically got us thinking very deeply about what are the constituents of a cell, and how do you maneuver around all of that functionality, and What's not in the cell that you could bring in there and introduce novel chemical potential? So fast forward, there's actually quite a few things that are not in cells, actually. And our group and other groups, many groups now over the years, have done a lot of really interesting experiments with functional groups like these, not all of which I'll talk about. But, but I'll say that we have certainly gotten the most mileage in terms of imaging sugars out of This one right here. That's the azide. So you guys are mostly chemists, I think. So I don't have to talk a lot about the properties of the azide to you. I know if there's any biologists in the audience, I'll preempt the frequently asked question. But oh my goodness, sodium azide is a metabolic poison. Are you crazy? Azides are toxic. That's true for azide anion. But the cool thing is when you link an azide to a like a carbon-azide linkage to some kind of an organic substrate, now the azide becomes a relatively inert functional group that is actually very well tolerated in living systems, including in human patients. And we know this because of decades of literature around drugs like AZT, which have azides in them, and it's totally harmless. Azides are abiotic, so we don't normally have them in our bodies. They're inert, they're small. That's very helpful if you're trying to sneak an azide into a sugar and get all the enzymes in the cell to process that sugar. And most amazingly, for its tiny little size, the azide packs a real punch of reactivity. It has not one but two modes of reactivity that are very rare in nature. So azides are soft electrophiles. They can react with soft nucleophiles. And they're also 1,3-dipoles, which means that they can undergo cycloaddition chemistries. With pi systems. And both of those modes of reactivity have now been capitalized on in the context of bioorthogonal chemistry. So just to summarize what is now a very large body of work for many labs, in chronological order, the first bioorthogonal reaction of the azide was called by us the Staudinger ligation. And we named it that because it was based on the classic Staudinger reduction of azides by triarylphosphines. But it turns out that if you arm a triarylphosphine with an ester substituent ortho to this nucleophilic phosphorus, a rearrangement occurs during the reaction that affords, at the end of the day, an amide-linked adduct. And this turned out to be a very well-tolerated, very biocompatible chemistry for conjugating probes and other various molecules to azides. So bio-friendly that we showed many years ago and in fact this was some of the early work of Jen when she was in my lab as a grad student, that you can inject azides into mice and you can inject phosphine counterparts into mice, and those two reagents will find each other in the body of that animal and do this reaction and ignore all of the functionality in that mouse's body as if it were anhydrous THF and argon. It's crazy, actually. So we've used this chemistry in a variety of different ways. To make a long story short, it turned out to be not so great for imaging applications for the simple reason that it's not a very fast reaction. It has kind of intrinsically slow kinetics, at least slow by the standards of bioorthogonal chemistry. So what that means in practice is that it takes many, many, many hours for these reagents to react to completion at the concentrations that one is kind of bounded by. When doing the chemistry in animals. And for imaging applications, that's too slow. It can be outcompeted by metabolism. Also, you can't really study things that change on a rapid time scale if the chemistry takes like a whole day. So, you know, after a few years of working with this chemistry, we kind of decided that in order to make progress for in vivo imaging, we would have to develop something faster, a couple of orders of magnitude faster. Now, at that same time, other labs were developing very fast reactions with azides that are quite bioorthogonal. And by far the most recognizable will be the copper-catalyzed azide-alkyne cycloaddition, which was coined click chemistry by Barry Sharpless and also reported by Morten Meldahl. And this chemistry basically takes advantage of the dipolar cycloaddition capabilities of the azide. So what happens here is that a terminal alkyne reacts with a copper catalyst, copper(I), to generate a copper acetylide intermediate. And then that reacts with the azide to generate a triazole product. And this chemistry is fast. And it's a beautiful, beautiful chemistry with a spectacular mechanism that I, to this day, cannot believe that this works in these biological systems with cells and so on. It's really fantastic. It's not great, however, for in vivo imaging because the copper catalyst is quite cytotoxic. So where this chemistry has been used most artfully has been for in vitro bioconjugation reactions, for doing imaging experiments on azide-labeled molecules in fixed cells, or in cells where you're going to sacrifice them immediately anyways. It doesn't matter if they get sick. But to do this in live organisms, you're really going to suffer from the toxicity. Many labs are trying to fix this problem. And some actually have some very exciting recent results, probably not even published yet. And what they're focusing on is basically finding better ligands, ligands that can sequester the copper and keep it away from biological systems that relate to toxicity, but at the same time salvage its reactivity and so on. So I think some really clever ligand engineering might solve the toxicity problem soon. But at the time we needed a chemistry, There was nothing close to that. So back in the kind of early 2000s, we realized that this was not going to solve our problems. But there was another approach that did. And that was what we now call the strain-promoted azide alkyne cycloaddition. So all we did here was basically dig through the old physical organic literature from the 1950s and '60s on the classic azide alkyne cycloaddition, which was characterized in all of its glory by Rolf Wiesgen, very famous chemist from the University of Munich who's retired but still has an office there and is there quite frequently. And his contemporaries at that time demonstrated that what is usually a very slow reaction can be accelerated dramatically by constraining the alkyne substrate in an 8-membered ring in the context of a cyclooctyne. And the strain associated with that system is kind of in the range of 20 kilocalories per mole, so pretty strained. So it turns out that azides will react with these cyclooctynes to form triazole products quite rapidly. And we thought that this probably was going to be the most productive avenue for us, because we thought we could probably get the kinetics we wanted without relying on a toxic catalyst. So in the kind of early 2000s to mid-2000s, we We put a lot of work into figuring out how to make substituted cyclooctynes, trying to understand their kinetic potential, like what was the range of kinetics that you could achieve in that space, in that reactivity space. And I have to say, I'm still really amazed at what these molecules can do. Cyclooctynes are quite stunning functionalities. First of all, I'll tell you that you can tune the reactivity of cyclooctynes over quite a broad range, like orders of magnitude, which turns out to be quite useful for a number of applications. And so here's an example of some cyclooctynes that are out in the literature now, some from my lab, some of these were made in other labs. And I'll just highlight a few examples from this chart just to show you kind of what the field has learned over the years. So let's just start with one of our very early generation cyclooctynes up here in this corner. At the time we made this molecule, it was like maybe 2004, 2005. We named it ALO because the undergraduate who made it was Anderson Lowe. OK, like J-Lo, ALO. I don't know. You had to be there. But this compound reacts with azides in a model reaction with a second-order rate constant around 1 × 10��³ per molar per second. And just to calibrate you, that's kind of on par with the Staudinger ligation, the phosphine chemistry, which we had already judged to be kind of too slow. So we knew that this wouldn't cut it on its own. But we thought we could probably accelerate the kinetics by thinking like a physical organic chemist. And our first success story was in the form of this compound called DIFO, which stands for difluorocyclooctyne. And what we learned is that if you introduce two fluorine atoms, these electronegative atoms, at the propargylic position, there is an orbital perturbing effect. That accelerates the kinetics by almost 2 orders of magnitude. So DIFO was the first compound that we felt was reactive enough that we were confident we could take it into an in vivo imaging setting. So we actually did a lot of imaging with molecules in this class. Then, through the work of Geert-Jan Boons and others, it became clear that you could achieve a rate acceleration comparable to DIFO by, instead of putting fluorine atoms on, fusing the ring to benzene substituents on either side. And that kind of enhances the ring strain. And so we kind of added another little kick by fusing not just two benzene rings, but also putting an amide bond in the ring, which has sp2 character. That even kind of strains it even a little more. And that led to this very reactive compound that we quite cleverly named Barack. You had to be there for that too, actually. And this compound has a second-order rate constant that's almost 1,000 times higher than where we started. And I'll mention BARAC in another slide as well. And then I should say, kind of at the stage of BARAC and sort of analogs of BARAC, we kind of hit the glass ceiling. We were really not able to substantively boost the reactivity of cyclooctynes beyond this level without suffering from a lot of instability problems. So these compounds at some point are so reactive that they start doing things you don't want them to do. They polymerize. They're not stable. They don't have a good shelf life. In order to kind of get a substantive increase in rate, you kind of have to switch to a different ring system. And so we discovered that if you kind of close the ring down by one more atom, so from an 8-membered ring to a 7-membered ring, well, first of all, if it's all carbon ring, you generate a compound that's unstable, cannot be isolated. But if you introduce a sulfur into that ring, the carbon-sulfur bond is a little bit longer than the carbon-carbon bond, so a little breathing room. Still pretty hot, so you've got to put 4 methyl groups, right? You're going to methylate propargylic positions. And this known compound— this was in the literature— called TMTH is actually a very reactive substrate for a dipolar cycloaddition with azides. It's got a second-order rate constant around 4,000 times what we started with. But it's actually kind of hard to functionalize. So we haven't done a lot of work with molecules in this class in terms of applications. But we've done a lot with Barack. And it turns out Barack has some really nice qualities. First and foremost, it's actually quite easy to make. The postdoc who developed this compound is John Jewett. He's now on the faculty at University of Arizona. And by the way, it took us like— it took him almost as long to make the compound as it took him to figure out how legitimately to call his compound Barack. Because the truth is that John was from Chicago. It was 2008, you know, when he joined my lab, and he was very excited about Barack Obama's first campaign. Campaign, and at the time we really needed fast molecules to give us hope and change and all this kind of stuff. I don't know. So John chose to name his compound Baroque. And here's his scientific rationalization for that, so that the reviewers of the JACS paper would leave us alone. But he came up with a really nice synthesis of Baroque that was only 7 steps from commercial starting materials. These are really simple, cheap commodity materials, indanone and phenylhydrazine. And both of those starting materials can be purchased with a lot of different substitution patterns, so that gave us access to analogs of Barac. We were never really able to analog the other cyclooctynes to the same extent because the syntheses just didn't lend themselves. Barac had some nice pharmacokinetic properties compared to some of the all-carbon ring systems we had looked at. And we think that we can actually integrate BARAC into some interesting probe molecules in a way that will render them fluorogenic, which is a very nice property to have if you're doing optical imaging, fluorescence imaging. It's great if you have a probe that's kind of off before it finds its target but turns on after it does, because you get better signal-to-noise, signal-to-background. And so it's a step in that direction. John was able to show right before he left that you can make molecules in which you kind of make a chimera of baraque and borrow one of the phenyl— one of the benzene rings from baraque and kind of integrate it into a fluorescent dye, like a coumarin, for example. And as it happens, it was already known that if you have an alkyne substituent at this position on a coumarin, that is relatively quenching of fluorescence. But after the reaction with an azide to form a more electron-rich triazole, that shifts the electronics so that you actually get a— what's in this case a 10-fold fluorescence enhancement. So that's a nice property to have. And I didn't put it on the slide because it was too embarrassing, but John actually gave this compound a name. He called it Coombaroc. Anyways. OK. So that hopefully gives you a flavor for the kind of methods development and probe design that we spend a lot of time thinking about in the lab. But none of this would be useful at all if we couldn't get azides into those sugars and image them. And that actually turned out to be the easy part. We were quite shocked at how many biosynthetic pathways, multi-enzyme pathways inside vertebrate cells are amenable to unnatural sugars. We'll just take unnatural sugars and process them. And the first kind of lucky break we had here was our work tapping into the pathway for the biosynthesis of sialic acid. In glycosircles, there's a lot of excitement about sialic acid. Because it appears in some really interesting physiological circumstances. It's often involved in molecular recognition. It's part of epitopes that are important in immunology. It's developmentally regulated. It's often overexpressed on tumors, so there's a malignancy relevance. And what we discovered early on is that we could introduce an azide metabolically off of this acetyl side chain of sialic acid by simply feeding cells an azidoacetyl derivative of a sugar called mannoseamine. So normally, the N-acetyl mannoseamine would be a substrate for this biosynthetic pathway. And we just put an azide on the acetyl group. There's 6 enzymes that take what we call mannase to the azidosialic acid. And every single one of those enzymes seems to tolerate this addition to the structure, which is— there was no reason to think that should be the case. It was kind of a lucky break. So that basically gave us the azide in a position to image the sialic acids. And others have shown that you can put azides into fucose through a salvage pathway, just feeding cells in azido-fucose derivative. One can label N-acetyl galactosamine with azides using this N-azidoacetyl derivative. There's a salvage pathway that allows that too. And so these are three structures that, you know, there is motivation because of the interesting glycobiology around targets that integrate these sugars. Sugars, which I won't go into for the interest of time. But let me just show you the nature of the experiments that were going on in my lab. Early on, of course, we first validated all of these reagents in cell culture models. So for example, Jeremy Baskin and Scott Laughlin showed that one can just take cells— like, these are HeLa cells, kind of a run-of-the-mill cell line that are easy to grow in the laboratory— incubate them with the sugar we call mannose, So that puts azides into the sialic acids. And then after the metabolism has reached some stage of completion, do a chemical reaction on the cells with a reagent like DIFO-488, which has this structure. So the DIFO part is down here. That's the difluorocyclooctyne. 488 stands for Alexa Fluor 488. That's the fluorescent part. And so hopefully you can see that there's a nice concentration of green fluorescent labeling. That's the —this Alexa Fluor dye—on the membranes, which is where the sialylated glycoproteins and glycolipids are concentrated. And you don't see that labeling in control cells that have been treated with mannose, so that's the natural sugar with no azides. So these cells have no azides, so when you react them with DIFO-488 under the same conditions as in this panel, you don't get that chemical labeling of the membrane glycoproteins. That's kind of a testament to the orthogonality of the chemistry. No azides, there's nothing for the DIFO to react with. So the first in vivo system we treaded into was the zebrafish for a variety of reasons. Zebrafish are a wonderful model of vertebrate development. They're probably developmental biology's most favorite model organism these days. They're actually increasing in popularity as models for human disease, which used to be the exclusive purvey of the mouse. Mice are really difficult to image because they're opaque, if you haven't noticed. But by contrast, zebrafish are translucent, so you can see their internal organs. You can actually watch them develop from the single-cell stage embryo all the way into adulthood in vitro in a test tube, and they're translucent, and you can see it all unfold using optical microscopy techniques. So they really lend themselves to these kinds of chemical labeling. Technologies. So we asked the question whether we could label cell surface glycoproteins in developing zebrafish and actually look at spatial and temporal changes in glycosylation that are associated with development. And this was the beautiful work of two former students, Scott Laughlin and Jeremy Baskin. Scott is now on the faculty at Stony Brook, and Jeremy is on the job market next year, so, you know, keep your eyes peeled for him. So what they showed is that all you have to do is fertilize zebrafish eggs in vitro and initiate development, and at that moment of conception, add an azido sugar of your choice to the media surrounding them, just in a petri dish. And during development, they take up sugars from the media and they process them, metabolize them, and then at different time points, one can take those embryos and chemically label them with the cyclooctyne dye like DIFO-488, and then take those embryos and rinse them off a little bit, and you can look at them in a microscope, and they're fine. This doesn't seem to harm them at all. Okay, this is actually quite benign for these embryos. Scott and Jeremy did a variety of multiple labeling experiments where they would take those very same embryos that had already been labeled with one fluorescent dye, do a quick wash with a reagent called T-SAP— if you're interested, we can talk about what that does later— but, and then take those same embryos and Introduce them again to azido sugar, the same or a different one. Let them metabolically label again, so a new population of glycoproteins will now possess azides. And those azides can be labeled with a different color dye, like a red dye linked to DIFO. So we call this DIFO-555, for example. So now at the end of the day, you've got zebrafish with two different colors, right? And the two different colors represent populations of glycoproteins that were made at different time points during during the developmental program. You can see how they've moved around and where they've gone. So those kinds of experiments led to images like these. This is the head of a 5-day-old zebrafish embryo. It's a larva at this stage. And it was labeled at 3 different time points with 3 different colored dyes. So there's red, green, and blue. The red dye was the labeling color right before the image was taken. So that was kind of like the most recent population of glycoproteins— whereas the green and the blue were older populations from 10 and 12 hours earlier. And just to orient you, this is— you're sort of looking at a ventral view of the zebrafish. These are the nostrils of the olfactory organs of the head, and this is the jaw. And, you know, you can actually walk through this image cell by cell using a confocal microscope and see that the red glycoproteins are predominantly on the surface of cells, because they just were labeled at that moment, took a snapshot of them. Whereas the green and blue glycoproteins, which had been chemically labeled almost a day earlier, had already been internalized or had moved around somewhere else. So individual cells have different dynamics of membrane turnover. You could actually see that in these intact organisms. Karen Daynard was a student that followed Scott and Jeremy, and she did a lot of work with early embryos. So this is a 10-hour-old embryo. It just looks like a ball of cells. —there's a lot of undifferentiated cells here. There's stem cells that are still making a cell fate decision, in fact. And a lot of cell division is going on. It's a very active structure. And she made some interesting observations. So for example, one thing she noted was that every time cells underwent a round of cell division, mitosis, which you could see in real time, there was a redistribution of the glycoproteins that she had labeled in this experiment. And in this case, it was mucin glycoproteins that she labeled with this azido-GALNAc derivative. And just to blow up one of those, so what you're looking at here, this is all the same cell division event where different species are marked. So in the panel on the left, the chromosomes are marked. It's actually chromatin that's marked with a histone marker. In this panel, it's the membrane that's marked, so with a membrane marker. And in this panel, this is the sugar, the azido-sugar labeling. So these are the mucin glycoproteins that we had labeled.. And all three panels are overlaid in this colored picture, and they're false colored so you can distinguish what's going on. But what she noticed is— and I'll wait for it to loop back again— so here's the mother cell, there it is. The chromosomes duplicate, they segregate, new membrane is formed between the two daughter cells. But interestingly, the mucin glycoproteins, they start to concentrate at the junction between the two cells. It's called the cleavage furrow. So they form like their own sort of concentrated structure, and they never really even fill in. The space between the two daughter cells, at least not in the time scale of this video loop. And so, you know, that's kind of an interesting proposal for a possible role of mucins during mitosis, which is to facilitate separation of daughter cells after the late stages of cytokinesis. So, you know, that's something that we followed up with a collaborator named Sharon Omaker, who's a big zebrafish jock at Berkeley. Now she's at Ohio State, actually, she moved. But, you know, the point here is that If you open the window to look at a different class of biomolecules, you see phenomena that you could never see if you were just focusing on proteins, nucleic acids, things that we already know how to image. So this is kind of an exciting observation for us. Okay, so with these technologies in place and having demonstrated that you can use these to image glycosylation and even the dynamics of it in invertebrate systems, I had a postdoc who joined the lab who came from a microbiology background, and she wondered whether you could translate some of these technologies into microbial systems where there's a lot of really interesting unanswered questions about the roles of structures that are sugar-based and very difficult to image using any other method. The postdoc's name is Sloan Seagrest, and Sloan was particularly interested in a structure called peptidoglycan, which is ubiquitous in bacteria. It's in Gram positives and Gram negatives. Which I learned about when I was an undergrad. But back then we were taught that peptidoglycan, it's the major component of the bacterial cell wall, and as so its purpose is just like kind of like a structural one, sort of like keeps the cell from exploding or something, you know, just sort of like the walls of this room. That was in the '80s. You know, today I think we now understand that peptidoglycan is very dynamic, actually. It's heavily regulated, it turns over, it's remodeled continuously, and— and its nature is different when microbes are in a pathogenic environment than when they're not. There's evidence for this from all kinds of studies, mostly based in electron microscopy. So Sloan wondered whether we could use metabolic labeling and bioorthogonal chemistry to actually image peptidoglycan in real time and try and understand its dynamics in living systems, in infection systems, in microbes that are infecting hosts. Rather than just in static images from electron microscopy. And although peptidoglycan certainly has a sugar component, it's made up of a repeating disaccharide, it was this other component that Sloan focused on, which is the pentapeptide. So the pentapeptide moiety within peptidoglycan serves a cross-linking function. So generally we think of the structure as you kind of line up these repeating disaccharide chains, then link them together with this pentapeptide, and you get this kind of three-dimensional mesh that way.. And it's well known that a highly conserved feature of bacterial peptidoglycan is its composition of D-alanine residues. And this is actually an interesting distinguishing feature of bacteria versus, for example, their mammalian hosts, if they're a pathogen. We, as humans and vertebrates and mammals, don't use D-amino acids in our primary metabolism. We do some things with them in the brain and so on, But we don't use them as building blocks for proteins or for other biomolecules. We use L-amino acids for that. Bacteria have this sort of unique and orthogonal metabolism where they require D-alanine residues in their peptidoglycan. And Sloan thought maybe you could exploit the uniqueness of D-alanine as a way to target metabolic labels to the bacteria, even if they are inside a human cell. And so she thought by feeding bacteria D-alanine analogs, maybe you could label them selectively. Selectively, orthogonally to their host. So here was the idea, quite straightforward. Sloan's concept was to basically take derivatives of D-alanine in which she added a bioorthogonal chemical group. And it could be an azide, that would be an obvious choice for our lab. It could also be an alkyne, right, because you can kind of switch the chemistries around. And if you could integrate the metabolic label into the bacterial cell selectively, then you'd be poised to conjugate an imaging probe and image something about the peptidoglycan in the living context. And there was ample reason to think that this would work because there's actually quite a bit of literature on the ability of the peptidoglycan biosynthetic enzymes to take unnatural D-amino acids. It was already known that if you deprive bacteria of D-alanine, but instead you supplement their media with any one of these other D-amino acids, they'll use them as a fallback. So they'll integrate, for example, D-phenylalanine, big benzene ring into peptidoglycan if they don't have D-alanine. So that suggests a certain promiscuity in this biosynthetic system. So we're pretty confident that bacteria can process D-phenylalanine, probably they would be able to process these things. Now Sloan was a microbiologist, so she'd never done any chemistry. And I said, oh, this is your chance, Sloan, I said. Because you can make these, right? That's not too hard. It's sort of a straightforward synthesis, a good chance for you to get your chops when you could have some credentials as a chemist. Wouldn't that be great? She's like, oh yeah, I don't know about that. So the next day she came in the next day to my office. She's all excited. I said, oh, this is great. She's all pumped up. She's ready to go. She's like mapped out her synthetic route. And she said, look, I found them in a catalog. I was like, oh, really? Yeah. So it turns out that Jaina Bioscience actually sells D-azidoalanine and D-alkynoalanine. They were just sitting in this catalog. I couldn't believe no one had done this experiment, actually. So Sloan did it. And it turns out that lots of different bacteria, everything she's looked at so far, including Gram negatives and Gram positives and even mycobacteria, are quite happy to metabolize azido-dealanine or alkynyl-dealanine and integrate those unnatural amino acids into their peptidoglycan. And these are just a range of bacteria she looked at. And the bacteria are fluorescent, which just shows that the unnatural amino acid was in there, and then we clicked on— fluorescent So E. coli, Bacillus subtilis, Listeria monocytogenes, which is a pathogen that infects human cells that lives intracellularly. This is a Corynebacterium. This is a Streptomyces. And here's Mycobacterium tuberculosis. So we thought, oh, this is a pretty interesting phenomenon and potentially a useful tool for the microbiology community. We now know that the tolerance of the biosynthesis is beyond certainly the simple azido and alkyne DNA. Galanine derivatives, but these are other derivatives we've made in the laboratory. And all of them are metabolized by many bacterial types and integrated into peptidoglycans. So cyclooctyne, this thing's called quadricyclane, this is a norbornene, which the pressure group uses for some of their chemistries. So you can get away with quite a bit of stuff here. And if you think that's crazy, look at this. So Yves Bruun and his collaborators published this paper in Angewandte just about a year and a half ago showing that you can even put really big fluorescent dyes on D-amino acids, and these things also get incorporated into peptidoglycan. So this is basically a fluorescein and a rhodamine derivative of D-lysine. You could feed these to not all bacteria, but some— many bacteria, they've shown, will take these things up. So D-alanine pathway, I think, is like a goldmine of weird metabolic labeling experiments. So what have we been doing? Well, you know, we've been using this methodology to understand how bacterial peptidoglycans are regulated. You know, what are their dynamics? How are they spatial and temporally regulated as a function of cell cycle, as a function of pathogenesis? This is an experiment that Sloan did which I really like. So she did a pulse chase study labeling mycobacterial peptidoglycan to see basically where new peptidoglycan is biosynthesized and how it distributes across the cell as the cell matures. And so in this experiment, she took bacteria and she basically labeled them metabolically with— I think in this one it was alkylnaldealanine for 30 minutes. And then she stopped and kind of took away the alkylnaldealanine and just added regular media with regular dealanine. That's called the chase after the pulse. And chased for these different time points, and then stopped the experiment and did a chemical reaction with an azide dye, a click chemistry reaction.— and so, you know, kind of visually you can see that at early time points there's a concentration of fluorescence at the poles of the bacterial cell— those are the two ends— and at the septum, the part in the middle. And as a function of time, that fluorescence redistributes itself, and it's kind of hard to see how that occurs by the eye, but actually we were able to quantitate this. So an undergrad that works with her wrote an imaging— an image processing algorithm that kind of goes through a field like this and quantitates the fluorescence across the long axis of each individual cell and then maps it out on a chart like this. So here's the cell length and here's the fluorescence intensity. So like if a cell like this one that has concentration of fluorescence at the poles and a little bit in the middle, it looks like this. So there's a lot of fluorescence here, a little bit here, and then a lot there. Okay, and so, you know, using this kind of graphical, it's a little clearer to see how the fluorescence redistributes as a function of time and how asymmetric that is. But that's actually what we expected based on a paper that Sarah Fortune's group had published just around the same time, where they were studying the elongation of Mycobacteria after a round of cell division. And what they found in this study was that the two poles of the Mycobacteria don't elongate at the same rate, that there's a pole that elongates quickly and another that elongates slowly. They call this the polar asymmetric growth model, and we were able to replicate that quite precisely using this method. Metabolic labeling techniques. So that was a really nice validation. So what Sloan's really interested in doing is this kind of thing, where she wants to study how peptidoglycan biosynthesis progresses in vitro versus in vivo. So in other words, if you take pathogenic bacteria, introduce them into a human host cell where they're in an infection environment, how does that affect the dynamics, the abundance, the density, spatial distribution of peptidoglycan? And it should be possible to do this because of the orthogonality of D-alanine metabolism. In the context of a human host. So, you know, she'll load bacteria into a host, introduce one of these, you know, synthetic analogs, in principle selectively label the bacterial cells, and then do a chemical reaction to introduce the fluorescent dye. And it looks pretty promising so far. This is some very recent data that she obtained using Mycobacterium tuberculosis, which lives inside human pathogens. It's kind of hard to see, I realize, but basically what you're looking at down here is a brightfield image of macrophages that are infected infected with mycobacteria, kind of just looks like a blob. But up here are— this is the fluorescent image of the same field, and hopefully you can see that these are mycobacterial cells that are selectively labeled at peptidoglycan inside a human macrophage. So a big push in the lab right now is focusing on trying to understand how things change in vivo versus what you might observe in vitro, and how the cell modulates peptidoglycan in response to cues from the host. OK. So with that, I'm well over time, so I'll just wrap up by pointing out that even though I gave a very long talk, I didn't talk about a lot of chemistry. I mean, I talked about a few reactions that have proven to be very powerful for us and for others that enabled some what I think are really interesting experiments. But it's not like we have a huge compendium of bioorthogonal chemistries to choose from at this point, right? So if you go to our chemistry library at Berkeley and you walk down the aisle for organic synthetic transformations, there's hundreds and hundreds of books. Oh, I was told in a recent talk that for younger students I should explain what a library is. Is that true? It's a place where books live. And what are books? Books are these— it's like paper, and you hold it in your hand and you open it. It's three-dimensional, actually, is what's cool about it. But anyways, there's just a huge literature in the conventional organic chemistry space. But I don't know whose bookshelf this is. It's from Google Images. But there's not much to choose from yet in the bioorthogonal chemistry space, which means that there's a lot of room for contributions, for new contributions. And they're happening rapidly. They're happening right here at home in the lab of Jen Prescher. She published some really brilliant work on the use of strained alkenes in a chemistry called the tetrazine ligation. Which is an absolutely phenomenal reaction. It was actually first reported using other kinds of alkene substrates by Ralph Weisslader's lab and also by Joe Fox's lab. So these guys kind of started it, but now the Pressure Lab has repurposed this chemistry in a way that makes it very amenable to metabolic labeling and for in vivo use. And what's so phenomenal about this chemistry is its kinetics. I don't know, for those of you who aren't familiar with this, the second-order rate constants of these tetrazene ligation reactions, they really blow away anything I talked about today by orders of magnitude. Something quite special about this. And then we've been doing some work. We're continuing to try and improve upon something we call the quadricyclane ligation, which also kind of falls under the general category of strange, weird, strained hydrocarbons from the previous century whose utility was thought to be limited to rocket fuels. But actually, you can do bioorthogonal chemistry with these things too. And I won't go into the details. And then finally, since I'm on the subject of the pressure lab, at lunch your students told me that you work really hard and you have no fun, that you never go out, you have no social life, and it all sounds really bleak and all that stuff. Let me just— I'm just telling you, none of that is true. This is like— this is pretty much how we spent 5 years together, like hanging out at bars and eating exotic cocktails. I don't even know what that thing is. This is Jen with some of her labmates. This guy, I don't know, that's Scott Laughlin of zebrafish imaging fame. It looks like he's trying to stab his dinner with like a lethal weapon or something. Anyways, okay, but you know, if you remember only one thing from this talk, it should be hopefully that bioorthogonal chemistry, you know, which I like to think is still kind of in its fledgling days of at least reaction development, has the potential to enable really different kinds of experiments that have no obvious counterpart in conventional biology. So hopefully this will allow biologists to ask questions that they previously thought were unaskable, and that's kind of like the mission of our field. And with that, I'll thank the people— this is my lab as it presently exists, and that microbiology work, peptidoglycan imaging, is in the very capable hands right now of Sloane Seagrist, who who is also on the job market as we speak, in fact. She's interviewing here and there. And on the heels of Sloan come postdocs like Justin Kim, who's working with these kinds of reagents in mycobacterial pathogenesis systems, as well as very talented graduate students, including Chelsea Gordon, who's done some really nice mechanistic work with cyclooctynes in collaboration with Ken Hauck's group, and many others, some of whose names I've mentioned. I don't really have time to tell you what they're all doing, but none of it would be possible without them. These are the people that support the research we do in the beautiful location of UC Berkeley. And thank you very much, ISIS, once again for sponsorship, and you for your attention. Thank you.