Scientist Stories: Phil Baran, Studies in Natural Product Synthesis

Now, our speaker for today is probably someone who needs no introduction. Professor Phil Baran is a professor of chemistry at Scripps Research, California. He graduated from New York University with a Bachelor of Science with honors in chemistry. Professor Baran did his PhD under Professor Casey Nicolaou in the Scripps Research Institute, followed by a postdoctoral fellowship under Professor P.J. Corey at Harvard University. He was elected a Fellow of the Royal Society of Chemistry in 2013 and became a member of the National Academy of Sciences in 2017. Prof. Barron has also received numerous awards, including the MacArthur Fellowship in 2013, the Blavatnik National Laureate in Chemistry Award in 2016, and the Inhofen Medal in 2019. Today, we are very excited to have Professor Barron deliver his lecture on Studies in Natural Product Synthesis. Professor Barron, please.

Thanks so much, Bram. Can, can you hear me okay?

I can hear you fine, thank you.

Okay, all right, great. Yeah, thanks so much for the invitation. It's great to be giving this talk from my guest room, and thank you also for letting our friends from Argentina attend as well. It's great to give you a little overview what we've been doing over the past few years, and I wanted to kind of make this talk a little bit broadly accessible at the beginning, but then dive deep into some pretty hardcore natural product synthesis to give you an idea of what exactly we do. But if you can't stay awake for all of that, this is a pretty quick summary of our research, and if you want to sign off after this, that, that is understandable. This is pretty much what we do in our lab. This interesting quote from Elon when he was not getting himself in trouble on Twitter. It is really hard to find a simple solution to a problem in natural product synthesis, and it is actually easier to find a really complicated solution. And that truly summarizes quite a bit of studies in natural product synthesis for the past century of making more and more complicated natural products, and sort of as a badge of on or showing how many steps it took to get there. So when we started the group back in 2003, we realized that that approach probably would not be sustainable, and that we'd really need to find a different way to inspire students to work on natural products. And so we turned to a different way of thinking about this, namely from the lens of ideality, and thinking about really what is the core of making a molecule, how do you get to it in the least number of steps as possible, but doing so in a way which actually makes sense and forces a chemist to become an inventor. So if you remove, delete the concession steps of a synthesis, namely the protecting group manipulations, the redox fluctuations, the functional group interconversions, and you end up with really what is it that you want to be doing, it's just making skeletal bonds like CC, CN, and CO bonds. And so if we take a step back and sort of opine on the age of feasibility, it's that age of chemistry where people were sort of uncertain that complicated natural products could even be made at all, and thus people weren't really concerned with issues of efficiency because it was an era where it wasn't even clear that you could make such things. And so the contributions to this era were really enabled by some of the legends in the field, and E.J. Corey's book on the logic of synthesis, which really taught people, you know, how to make molecules. It is the, the bible of our field. And so when we started our group, we thought maybe there could be a slightly different way to approach retrosynthetic analysis to arm students with the ability to do things in a little bit more efficient fashion. And so in 2007, we published this sort of manifesto that was exemplified with a synthesis of various marine natural products showing the list of 10 different rules that, if followed, could lead to decidedly shorter ways of accessing complicated natural products. Today will be a case study in really how this is executed and the advantages it has towards synthesis, and more importantly, the fact that this approach requires students to become inventors. Equally important is the fact that the approach is agnostic as to structure types. It doesn't matter whether it's a peptide or an alkaloid or a terpene or even something rather crazy and bizarre. For example, we became interested in oligonucleotide chemistry very recently because it is another area of chemistry. It's natural product chemistry, which has been overlooked by the organic chemistry community. So Spinraza is a very popular medicine for the Western world, curing the largest ailment that afflicts young infants, spinal muscular atrophy. And this is given as injections to those infants as a mixture of 130,000 stereoisomers, although it's obvious to any organic chemist practicing in the field that probably they shouldn't be stereorandom, these linkages, these phosphorothioate linkages, and instead they should be diastereomerically pure. And so we teamed up with Bristol-Myers Squibb a few years ago to invent chemistry that would allow for the stereospecific incorporation of phosphorothioate linkages And we were guided by the rules of ideality. And indeed, it led to some really interesting inventions that we're following up on constantly, even as of now. So today I want to talk about 4 little topics and show you how they either inspired completely new directions in our laboratory or simply were a means to an end of showing how one could end up at a pretty short synthesis through an inventive approach that was hinged around aiming for ideality. So the first one is palaomine, interesting marine natural product we worked on starting in the very beginning of our laboratory. And because this is kind of like archaeology, I'm not going to go back through the synthesis of this compound, which was reported back in 2010. Instead, I just want to focus on the one transformation that I remember from the synthesis quite distinctly. Namely this direct oxidation of a methylene position with a molecule that's decorated with all these unprotected functionalities that only dissolved in water. And this reagent was both chemoselective in that it only touched that one position, but it was also what I call redox intelligent in that it knew to oxidize once and then stop, because if it went too far, of course, that would be the end of the story. You can't reduce that lactam back to the hemiaminal. And so that silver chemistry was presumably operating under a single electron transfer mode and got us into radical chemistry, and we thought, well, what else could you do with, with radicals that would be useful to people that are not working on phthalamine, which is pretty much the entire world? So we turned to Minisky chemistry, and we wondered, could we extend the age-old Minisky reaction, the decarboxylation of an alkyl carboxylic acid, instead to the deborylation of an aryl boronic acid? And in doing this with Minisky-type conditions, we could and under the catalysis of silver, of course, we could get to arylated heterocycles, and this chemistry has been used quite a bit in industry. Turning to our industry friends, they said, you know, Minisky is great, the borano Minisky looks useful, but we really want a way of adding alkyl cargo that are more difficult to access than the canonical Minisky can provide, and we turned our attention to zinc sulfonate salts because these compounds have a weaker R-S bond, and in principle could be coaxed into releasing a sulfur unit and delivering a radical. And we thought that might be useful. And the only precedent we had out there was from the Langlois group, who showed many years ago that this chemistry could only work in the presence of electron-rich aromatics, not electron-deficient. It required various transition metal additives. And so we reinvestigated this chemistry and found that, in fact, no metal is required. TBHP is enough of an initiator and it is actually far more general than people had thought, in that it doesn't require electron-donating groups. In fact, electron-deficient heterocycles work just fine. And so an entire library of these salts is now available from Aldrich under the name of Diversinates. And I think what got people in pharma kind of excited, and maybe got people in academia to roll their eyes a little bit, was the fact that we showed you could do this rather outlandish experiment, namely to go to the store and buy some tea and see if we could simply functionalize the caffeine found natively in the tea. Weighing out the things with a plastic spoon, a simple syringe to add in the TBHP, and leaving it on a broken stirring plate overnight, you can see the TLC shows that in fact, wow, you can functionalize caffeine in tea. And this was a little bit crazy because people in academia said, are you advocating, you know, using tea as solvent? But the people in industry, not surprisingly, they got the point right away, which is to that it is highly chemoselective and robust. So if you can do this in tea, you can probably do this on your late-stage intermediates and functionalize them without going to the very beginning and making them from scratch. And so here are some examples of this type of chemistry in action, and if you compare it to the two-electron approach, it's rather striking. So intermediate A here, which was required by Novartis because the, the interesting trifluoromethylcyclopropane is an isostere that's stable to metabolism instead of a t-butyl group, And this can be made instead of the two-electron chemistry you see here using Buley and Tebbe reagent and diazomethane and palladium. It's rather lengthy. Instead, a one-electron disconnection gets you right there in a completely ideal fashion. So you simply purchase intermediate D, you purchase intermediate B, you purchase the diversinate, you dump and stir the diversinate and D together. It tolerates supronic acid, and then in the same pot you just do your palladium Suzuki coupling to get to the product. Similarly, here are some examples from the patent literature. This is from a Merck patent where they show in the patent both the 2-electron approach and the 1-electron approach, where the 2-electron approach is sort of undergraduate level. You know, you take the aryl iodide and you convert it to the corresponding aldehyde ester, and then you reduce that down to the alcohol, and then you oxidize back up to the aldehyde, and then you treat it with a deoxyfluorinating agent, or you just take the compound without the iodo because that's non-strategic. Now you simply put the trityl group on that you need for later downstream manipulations and do the direct difluoromethylation using DFMS. So since 2012, the sales of these compounds has been quite high. They've sold millions of dollars of these reagents, and they've been found in numerous patents. And so we wrote a review last year summarizing a lot of these patents that have used this chemistry. And the surprising thing to us is that although it was designed to be used at a late stage in a that this has actually found quite a bit of use in the early stage as well. So it really became like a way of doing any stage functionalization of a heterocycle for divergent access to analogs or for late stage modification. So that turned out to be quite interesting. We turned our attention then to other precursors to radicals, and one of them that came to mind was olefins. So if you take a look at this example here, the way two-electron chemistry forges that bond noted in red, is through a 2-electron disconnection that requires taking a glycal, converting it to the anomeric chloride, then an anomeric tin, then the anomeric lithium, and then finally quenching with Michael acceptor. Alternatively, if there were a way to get a radical directly from an olefin, you could sidestep all of that. And by using hydrogen atom transfer chemistry with iron and— catalytic iron and silane, you can achieve that. So the same starting material just a couple of hours open to the air, and you get the exact same product with no cryogenic conditions, no tin, no HCl, no lithium reagents. It's quite convenient. This chemistry can also be used to forge N-C bonds, not just C-C bonds. So here's an example from the MedChem patent literature where the ways that they make these hindered N-C bonds is rather like a hammer and tongs. So you start off on the top with an aziridine, you open it up, from the nitro group. Same thing on the bottom, you take that nitro group, you reduce it down to the aniline, and then you alkylate, and then use Raney nickel and copper to reduce that down. And the, the sort of striking thing to us was how people abuse and don't really appreciate the reactivity of an aryl nitro compound. Essentially, those compounds are the orphan stepchild of the functional group literature, and all you do with the nitro compound is simply reduce them down to an aniline, but you ignore in many cases, their innate reactivity. And so we thought, well, what if a radical could directly attack the nitro and you can get the exact same product directly? And indeed, you can take an olefin, a radical can be derived from it, and it adds directly to the nitro compound to make the exact same product. So one step to make the exact same compounds from olefins that are commercially available, in the case of 2-methyl-2-butyne on the bottom, are just pennies per mole. And the chemistry is also rather practical, just like diversity chemistry. So this hydrogen atom transfer chemistry using iron to make C-C bonds and C-N bonds has now been used innumerable times. And it works in some solvents that graduate students can particularly appreciate. Turns out that this fully substituted carbon center you see here can be made in all of the solvents shown here with the lowest yielding one being in red wine, presumably because of the high number of antioxidants and radical traps that are found in, in red wine. But it is a rather robust reaction, doesn't require any kind of specialized technique to make really powerful C-C and C-N bonds. So if you look, you know, retrospectively at what is it that people do in drug discovery, you'll notice that they really do the most, largely the most boring reactions that exist. They do a lot of reductive aminations, Suzuki couplings, Buchwald-Alman, Goldberg-Hartwig type couplings, and of course some amide bond formation. And the question is, why is this? Is it because medicinal chemists are somehow less creative? That is clearly not the case. Some of the most talented chemists on the planet go into medicinal chemistry. It turns out that the reason for this is the most used reactions really comply to Barry Sharpless's rule of what makes a useful reaction. Namely, it has to be incredibly wide in scope, extremely easy to run, and a transformation that has starting materials that are literally ubiquitous. So they have to be the most widespread starting materials you can find, and that is a really good recipe for a useful reaction. One striking thing from this collection of various surveys in the literature is that there really are no radicals, and so we wonder, well, why is it that people are not using more radical chemistry in, in industry. And if you look at that, it might stem from the basic undergraduate and graduate education you get. So these are notes taken from various places on the internet. I'm sure Cambridge and Argentina have the same curriculum wherein you teach retrosynthetic analysis in a way which is easily understood by an undergraduate, and that is through using polar bond disconnections. And that really is because we're taught from the time we're an infant child that opposites attract. Plus and minus is a really easy concept to understand, and you look at a functional group and assign what the partial polarity is, and then you make disconnections based on partial polarities. It works. It's really great. The problem is it just completely ignores the radical-base disconnection, which can sometimes be far more direct than the polar bond disconnection. Maybe that's why the radicals are ignored in the modern era in large pharma. And so we thought, well, maybe a way to fast-track this problem and get people looking at radicals more would be to intercept the exact same intermediates they already use for the most boring and used reaction in all of organic chemistry, which is making amide bonds. And so that is a chemistry which has been known for, you know, 70 or 80 years, involving the activation of an acid to a so-called activated ester, which is then subjected to a nucleophile to give you a net dehydration. But if you look at many of those peptide reagents that are out there, those intermediates are not only activated for loss of water, they're also redox active. And so we thought, well, maybe in the presence of a proper transition metal and some sort of cross-coupling condition, you could intercept it not by losing water, but instead by losing CO2. And that pretty much summarizes 3 or 4 years of research in our laboratory to try to take carboxylic acids, activate them just like you would for amide bond formation, but at the end don't make the amide, instead make a carbon-carbon or a new carbon-heteroatom bond. And here's an example of how this works. So decarboxylative borylation is an attractive technique because alkyl or, you know, various boron-type compounds are found in medicines such as the one shown here, and Normally when people, medicinal chemists, look at a molecule like this, the sole attention is on the carbon-boron bond. The rest of the compound is extremely boring. As you can imagine, if this was a simple peptide with the BO2H being instead CO2H, you could make a million analogs of that in a week. So decarboxylative borylation allows you to take a native peptide, simply activate it, turn it into a radical, and then recombine to give you a transformation of CO2 to BO2. It is a rather chemoselective reaction, so you can take things like vancomycin and you can turn the carboxy group into a boronic acid. Now, I would hate to be the graduate student that would work 5 years on a total synthesis of boronovancomycin only to find out it's a rather mediocre antibiotic. Instead, we could do this just in a day using decarboxylative borylation.. Now the original nickel conditions are pretty useful, but we later on came out with some copper-based conditions which are so quick to do that we can just do it right here on the screen. The reaction is done, you can tell by color, in about 4 minutes as shown here, and it actually takes place faster than you can write out the reaction in your notebook. So if you don't want to use nickel, you can use copper, and this is a very scalable procedure as well. Okay, so, you know, another favorite functional group is of course olefins, and the way people make olefins is normally by breaking the olefin C-C bond, and that requires that you have an intimate knowledge of all the mechanisms and methods that are available for making olefins and the geometric preferences that are a result of those choices. And so you see here this example, the steroidal molecule that people made, and it requires 7 steps. Really, only one is making a C-C bond. The rest are completely complete waste. And so we thought, well, maybe decarboxylative olefination would solve this problem. And indeed, now it can be made in a completely ideal fashion, simply by taking the olefin and doing a cross-coupling with the same exact starting material, instead of taking a hammer and tongs, using diazomethane, and tons of protecting group chemistry, and redox chemistry to get to the same ends. And you can program in whatever olefin you might want in that. You can do the same thing for alkylation. So decarboxylative alkylation works just fine. The previous conditions here, as you see, we're using Apple-type conditions where you're doing SN2s of alkylnyl nucleophiles that are protected, but the decarboxylative alkylation allows you to basically use acetylene gas as your precursor to this useful alkylnylated amino acid in a single step from a glutamate starting material. Okay, so some of the key people behind this who have now all departed the lab. Tian has began his independent career at UT Southwestern. Rohan has begun his career at Merck. Ji has begun his independent career at SIIM. Pep is off at Max Planck. Jacob is at Bristol Myers, and Joel is at UF. So all on to great careers. Really miss all of them. You know, the radical age of cross-coupling is probably upon us. There's a lot of great reviews out there, a lot of practitioners in the area making amazing contributions. We're just making some small contributions to a very large field. And so we realized that, you know, it might be useful to have new functional groups that people didn't associate with making radicals, and maybe if they could be accessed, it would be useful as well. And so we thought, what about one of the functional groups that's associated with two-electron chemistry more than anything else? If we could turn that into a radical, an alkyl sulfone certainly came to mind. And so these are stable modular building blocks that can be used in a variety of different 2-electron processes? What if you could sort of invert all that reactivity? At the end of your 2-electron process, just do a 1-electron process. And so here that is, where you take the HAT chemistry I showed earlier. You can take this olefin on the top, you can turn it into a radical, or you can take the acid, turn it into a radical. It then does a Giese-type reaction onto this acceptor, and that redox-active sulfone that results can then be used in a cross-coupling. So giving you basically the equivalent of an ethylene diradical. So the key to this chemistry was really identifying a redox-active sulfone that could be used to really dramatically simplify a lot of molecule synthesis. So here's an example from the Schering patent where all the red steps here are really non-ideal, they're concession steps, and they could all be done away with simply by using a redox-active sulfone shown here, which is commercially available, doing your cross-coupling directly, and then getting out your desired product. In the same way, you can make a variety of different medicinally important compounds, such as the ones shown here. I'm going to go through this rather quickly just because of lack of time. And in general, when people think about these fluorinated residues, usually the fluorine, like a boron, changes your entire retrosynthetic analysis. So what this chemistry really allows you to do is treat a fluorine just like a methyl group, or if it was a methyl group, you wouldn't change your retrosynthetic analysis. But if it's a fluorine, you do. Look at all the crazy steps you do here, all these reagents you really wouldn't want to use in the real world. And now you can just simply remove all of those and use some very, very simple building blocks, extremely cheap, to get the exact same product with no concession steps at all. Totally ideal. So the radical world of cross-coupling is sort of upon us, and it is orthogonal to the two-electron world of cross-coupling. So you can take these two groups, as you see here, a redox-active sulfone, a redox-active ester, you can do the decarboxylative coupling first, a 2-electron Suzuki or Sonogashira or whatever you like secondly, and then finally end up with a desulfonolative cross-coupling as shown here to make a library of different compounds that I think would be difficult to make in a straightforward way in a library fashion in any other way. And so some of the people behind this, I mentioned Jacob before, Rohan Tian as well, but Cheng is now a graduate student with us. He was a visiting student during this work, and Monica was a visiting student as well. And then finally, to round out this part, just to mention that two-electron chemistry of course is not bad, but if you can turn it into one-electron chemistry, you get a lot of inherent chemoselectivity advantages. So late last year, we published a paper where we could take the canonical reactivity of Grignards and instead access them through radical chemistry to make alcohols, ketones, and amines. So you can take 2 carboxylic acids, for example, mix and match, and turn it into a ketone through a net decarboxylative process. And that already has found quite a bit of use. This was spearheaded by Julian, who spearheaded many of the things we do in our lab. He's a very talented staff scientist who will be— he's on the job market right now, but incredible guy. And we got an email recently from this medicinal chemist who had saw this paper, and they were trying Grignards off at WuXi. Most of the chemistry these days is done there. And the Grignards were not working, but this decarboxylative ketone synthesis worked for them on the first shot. So this could be pretty useful. Okay, the next topic is maximycin, switching gears from palaomine, another interesting little alkaloid. And this was spearheaded by one graduate student, Kyle McClement, who's about to graduate. Maximycin is an interesting compound that was isolated from a crowdsourcing expedition where many people went out and collected fungi and this interesting compound was identified. And what's most interesting about it is its sort of structure that stems from 3 different biosynthetic lineages, where you've got a shikimate pathway on the left, a tyrosine in the middle, and a polyketide on the right. And it is for that reason, not at the bioactivity, that we decided to pursue it, because structure is really so interesting. And of course, that 3,7 bond is hindered, and you've got a mixture of atropisomers that exist at steady state room temperature. Now, the previous ways of making these hydroxylated pyrone alkaloids are interesting, but not stereoselective, and they're rather stepwise ways of constructing these that wouldn't necessarily be useful for making large quantities of these types of compounds. So the two strategies shown here are alternative ways of accessing such structures, but we wanted to do something a little bit more direct. And guided by ideality, we made disconnections that were rather bold. So the first disconnection was just simply breaking the molecule in half. We thought the left-hand side could come from meschicomate. We didn't want to do N-hydroxylation, which we know is a very low-yielding reaction, so we instead bought our N-O bond by installing it in the starting material. On the right-hand side led to a malonate derivative, and that malonate derivative is very special because of the inherent symmetry found in it. So the left-hand side comes from a shikimate, but the right-hand side comes from a decarboxylate of guillouzet. Now notice what I've done here is that in that formation of that bond, the stereochemistry has been inverted. Because it is a radical reaction, it will give you the thermodynamic product, which should be trans relative to the substituents there, giving rise to the all-cis compound shown here. Now that all-cis compound could in principle be accessed from a simple benzoic acid derivative, if you could find a way to hydrogenate, desymmetrize, and do the CH activation. So radical retrosynthesis is really the simplifying disconnection that breaks that first bond, and then CH functionalization sets 4 stereocenters by desymmetrizing a mesoacid in one fell swoop. So if you think about the power of the Diels-Alder reaction, it sets a lot of stereocenters through a cycloaddition, but CH activation has not been known for something that can install multiple stereocenters in one step. And this is what this would do if we could figure out a way to desymmetrize this meso compound. We would set all these stereocenters in the natural product in one step. That was exciting, and that was the idea. And so we proceeded forward with the hydrogenation, which gave the all-cis product. And after that, Kyle screened for quite a long time, directing groups to see if he could figure out a way of activating that one methyl group to give a good diastereoselectivity, and I found out that the PIP-type directing group functioned best by switching that methyl group out to a t-butyl and then optimizing the substituents on the pyridine ring, the oxidant, the base, and the additive. He found that he had a decagram-scale procedure for getting the desired product as one diastereomer in a scalable fashion through the proposed stereo model shown here, although we have yet to successfully grow a crystal of the stoichiometric inserted product. Okay, so with that key desymmetrization in hand, all we needed to do now was a decarboxylative Giese-type homologation. So we started with the chemistry we had developed a few years ago using the NHPI redox active ester, and that vinyl sulfone is the one-carbon homologation unit. And unfortunately, it gave a mixture of these two products. And at the beginning, we were pretty depressed by this because it is impossible to take the byproduct on the right and convert it to anything productive, until we realized exactly what was going on. So the initial radical addition gave rise to a radical which, instead of stopping, went forward and did another 1,5 hydrogen atom transfer and then recombined with another molecule of vinyl sulfone. And that was a great hint that we could do something even, even better. So this problematic 1,5 gave a mixture of these two products. We could take the desired product, turn it into the aldehyde. But we wondered, what if we could take the radical intermediate that we're getting and just directly oxidize it in the same pot? If you could do that, boy, you would avoid the deprotection maneuver. You would avoid a redox fluctuation. You could do it all in one pot. And so we turned our attention to trying to do just that. And we could access this acid very quickly, as shown here. And our first proof of concept that this would work came from this electrochemical reaction shown here, although the yield was pretty low. So we switched over to simple Minisky conditions and found that the yield was now a whopping 6%, which is a great starting point for optimization. And we didn't need the protecting group at all. We could simply hydrolyze the lactone, get to the free, uh, seco acid shown here. And simply adding iron to this, catalytic amount of iron, increased the yield initially to 71, but as you'll see in a moment, pretty much quantitative yield in this reaction. Presumably the iron is acting as a selective oxidant for the alpha-alkoxy radical to give you the oxonium delivering the aldehyde. And this could be done in a one-pot reaction where you take that lactone, you hydrolyze, you add in your vinyl sulfone, the Minisky mixture, and this is the crude NMR. You simply have product, so you don't even need to purify this crude. You can simply take it on. It just has some solvent in there. Okay, so the completion of the diacid fragment could be accomplished by a simple oxidation and acylation sequence confirmed by X-ray crystallography. We could try as a model the late-stage pyridone synthesis to show that it was feasible, albeit in only 14% yield, and then make this interesting natural product pyridoxidine as a proof of concept. The desired oxime ether fragment was accessible, but not without quite a bit of struggle at the beginning, because Kyle's initial attempts at making these, the desired oxime, led to a variety of selectivity issues stemming from regiochemical problems and N versus O attack. And eventually settled upon a procedure where you could access the oxime at a little bit higher yield. And again, the problem was the desired coupling with the malonate was not working at all. We only recovered the oxime ether. And this didn't change regardless of the protecting groups that were employed. So he reasoned that he could enhance the oxime reactivity by using an effect that was really pioneered at Cambridge, or at least understood by Fleming, the beta-silicon effect for activation that has been used by, for instance, the venerable Hosomi-Sakurai type reaction. And so by using this beta-silicon effect, now we actually started to see some measurable quantities of product. So the silyl derivative was far more reactive, and by adding a little bit of silver to promote the reaction, we got even higher yield. And so the silyloxime could eventually be accessed in a much more efficient way using the hydroxylamine Boc protected, verified by X-ray, converted into the corresponding oxime as shown here. And in the event the fragment union worked out really well, going via the intermediacy of the bisacyl triflate. So the silver triflate used did a salt exchange with the— you got the silver chloride plating precipitating out, and then that added in really nicely to give a 41% yield of the desired product that could then be deprotected to give maximycin. I think Kyle made over 100 mg of this interesting natural product, and you could also see the comparison. It is the same mixture of atropisomers that, uh, exist in nature. So the overall route is pretty short. It's about 60% ideality, and it features some really interesting maneuvers such as installing all the stereocenters of the core of the molecule using CH activation, and then this very bizarre decarboxylative homologation radical translocation cascade. Okay, so the next topic is dexamycin, interesting marine natural product that was isolated back in 2010, and it was interesting because the dimeric form showed quite a bit more activity than the monomeric form. So there was a real need, a desire in our lab to figure out how to make that N-N bond, and it really is the beginning of a big story in our laboratory, which was a turning to electrochemistry, because when we tried making this N-N bond under any conditions that were chemical in nature, either the conditions were way too harsh to ever be used in a real system, or they simply led to nothing. The students really tried thousands of things until, out of desperation, which is when most people turn to electrochemistry, they turn to electrochemistry, did acyclic voltammetry of the starting material and realized that there is a very narrow window, and within that window you can get productive dimerization, and outside of that window you get clean decomposition, so to speak. So we realized then that there might be something pretty interesting to do with electrochemistry that maybe a lot of mainstream groups were not thinking about. Every time you make a bond, you're really— you can be thinking about electrochemistry, because a chemical reaction of course requires the reorganization of electrons and is by default almost always some type of redox reaction, unless you're doing a pericyclic reaction. And so electrochemistry offers an interesting, you know, untapped potential, we thought. And there's a few reasons why it's been held back in mainstream synthesis. One is a perception People think it's alien reactivity. It is ET, but not this ET. It's just electron transfer. So if you understand how electrons move and arrows can be moved around in a circle, you can easily understand the reactivity of electrochemistry. The other reality was non-standardized equipment. And I have opined on this in other lectures that you can view online. But essentially, we viewed this as the biggest problem holding back electrochemistry in the modern era. And that is nobody wants to get a PhD in arts and crafts to build an apparatus, and you're never going to convince anybody at Pfizer or Merck or wherever to do electrochemistry if they have to put half of their hood and turn it into this kind of contraption. It's never going to happen. So we work with ICA to develop a device called the Electrosyn, which is now sold well over 1,000 units around the world, and it basically offers standardization and very rapid uptake. So if you're electro-curious, so to speak, you can order this device and be doing your first reaction within minutes. It will teach you how to do it, has a smart assist feature. And if you invent a reaction, you can give the procedure on ElectroSyn and rest assured that other people around the world, the reaction invented in Cambridge can be reproducible in Argentina because the electrodes are the same, potential stat is the same, the distance between electrons is the same, everything is consistent., and therefore it is not anecdotal but rather standardized. And so the other problem with electrochemistry is the sort of perception that the reactivity is just kind of boring. We wrote a review on a lot of the reactions that are out there. Some of them are interesting, but honestly, it's going to be a hard press to tell someone they should be doing their alcohol oxidation with electrochemistry. It's just going to be difficult to convince someone to do that. You're going to need to show them reactivity that they actually care about. So some things that might be interesting to process chemists is that you can do things without canonical chemical oxidants, such as the diversity chemistry I showed earlier. Instead of using TBHP, you could just do it electrochemically. You can also achieve new transformations, such as the one we worked on with Bristol Myers Squibb a few years ago, where we did allylic oxidation. A lot of ways of doing allylic oxidation, but this one is extremely clean, inexpensive, no transition metal salts at all, and can literally be done in a bucket. That is a bucket in that picture of an allylic oxidation being done on hundreds of grams scale. And then new selectivity. A few years ago, we unveiled a way of doing methylene oxidation. It's been used by others in total synthesis at this point, using quinuclidine as a mediator that allows for very, very simple and cheap scalable CH oxidation of unactivated methylene positions. And then finally, we showed a few years ago, and followed it up with a full paper in second-generation conditions, ways that you could use nickel, because of the fluctuating redox states, to achieve Buchwald-Alman-Hartwig-Goldberg type C-N bond formation, which has found use in industry. And Yu Kawamata, our other staff scientist who has spearheaded so much of the electrochemistry in our lab, is really due so much of the credit for, for this work. And yeah, it's found use in industry already. Here's an email from a person at Novartis who, as you can see, they tried many different canonical conditions and they did not work, but with the amination protocol it worked pretty well, and the coupling partners had a lot of acidic NH and basic heterocyclic nitrogen. So that's good. Some other interesting techniques that we've reported are a really simple scalable way of doing Birch reduction that even an undergraduate can do, and making hindered ethers using really strongly oxidizing conditions. So I want to show a little bit of unpublished data. So here's a, a sort of application to total synthesis of electrochemistry that really came to the rescue, based upon some old chemistry from Perishan's group, where he showed before reductive cross-coupling became popular, people used to do that chemistry using electrochemistry. So here's reductive cross-coupling, um, using electrochemistry back in 2000. That has been extended by others now, and even taken on to use redox-active esters more recently in the context of electrochemistry. So with that backdrop, we decided to work on a problem in terpene chemistry, namely progesterone, which is one that if you give this product to an undergraduate or graduate student with a few lectures, it doesn't take them long to come up with the retrosynthetic precursor. But quite surprisingly and paradoxically, if you ask them to come up with a way of making that polyunsaturated precursor as a geometrically pure product, it takes them longer to come up with the route to that than it does to come up with the route to the initial disconnection, which is bizarre. And if you look at the ways of making these compounds, there's really a lot of hammer and tongs. And here's the sort of Johnson synthesis, and nobody pays any attention to all of the red, red steps here, which are not strategic. Because that final step is so beautiful, everybody is willing to ignore the imperfections of the way you actually make that starting material. So if we turn to a radical approach, we end up with something that's a lot more simple. However, it is still plagued with these pyrophoric materials, such as t-buoyli, and that really kept me up at night worrying that Steve would and Max would hurt themselves with the t-buoyli. And so we turned— the other thing that bothered me, of course, was the TBS groups that we wanted to get rid of. And so we turned to electrochemistry, and a modification of Perishan's method by actually coating the electrode with something special that we can't reveal at the moment enabled this reaction to take place without any kind of protecting group chemistry at all, and no pyrophoric material, and in a scalable way. So basically all the key bonds are made through electrochemical radical type methodologies. Here's another example. Of this polyunsaturated precursor that can be used to make celastrol in 2015. Very beautiful chemistry, although it is kind of, you know, stepwise, and a lot of concession steps are used to make this intermediate. Alternatively, you can use a radical-based disconnection. Again, you've got t-butyli, but if you excise the t-butyli and instead exchange that for electrochemistry, you now end up something, something that is protecting group-free and pyrophoric-free as well. And finally, isosteviol, a very interesting molecule that has been made by the Snyder group in 1998. And again, all these red steps here, you don't really need to look at the details except to know that those red steps don't contribute to making the carbon skeleton, and therefore they are non-strategic. Instead, you can try doing decarboxylative couplings with t-butyli, which unfortunately fails because when you take intermediate B and try to lithiate it, that compound does not undergo successful cross-coupling and instead reacts with itself. But electrochemically, we don't have that problem. So the same starting materials can be used without any protecting groups. And in this case, electrochemistry is not a gimmick. It's not just removing t-buoyly or protecting group. It actually is enabling. So to summarize, you know, nature makes terpenes, as I'll talk about in just a moment, through these polyunsaturated precursors. Topologically using E.J. Corey's book, you would come up with the disconnection shown here. But if you think about biosynthesis in the cyclase phase, a modular approach is the way to go. And so this way of making terpenes is modular and very intuitive, and hopefully this will be coming out later this year. If you're interested in some of the other little goodies we have in the pipeline in electrochemistry, one of them is non-traditional nucleophilic additions. So for instance, the transformation shown here normally took 7 steps, and now we can achieve it in 1. Unsaturation is something that is still found in the pages of very high-profile journals, and yet the really simple solution to this would just be directly removing hydrogen gas from a compound to get unsaturation. And finally, back in 2017, we showed how quinuclidine could be used for CH oxidation of unactivated methylenes, but are there other mediators to be discovered that are unusual structures that can be used to control the site and selectivity of oxidation of complex natural products and medicinally important compounds. So in the final— looks like I have 15 or so minutes— I'd like to talk about our 13-year journey to make Taxol. So I guess it's about 1 minute per year. So we started off many years ago back with the, at the time graduate student Tom Maimoni thinking about how nature makes terpenes, and it kind of led to a revelation to us of how maybe one should approach, or an alternative way of approaching, the synthesis of terpenes. You know, nature makes 55,000, and the number keeps growing every year, terpenes by an approach which is just mesmerizing in terms of its beauty. She starts off with the same two building blocks, and she weaves them together into these little threads. And then these little threads are then woven together, they're kind of like the way your grandmother would knit together a sweater, by what enzymologists refer to as a cyclase phase. And then kind of like the way you would decorate a Christmas tree every year, nature finds a way to take enzymes and decorate these little hydrocarbon frameworks with oxygen atoms all over the place. And in many cases she's selective, and in many cases she just makes a wild mixture of compounds that evolves over time to favor one or the other. And through this approach, she achieves 100% ideality. There's no wasted steps, there's no kind of protecting group chemistry, and everything is compartmentalized. So meaning that you take these complicated molecules and you simply reduce the problem into two smaller problems. One How do you make carbon skeletons really quickly? And then the other problem is, how do you put the oxygen atoms in, in a selective or non-selective way and let bioactivity guide the evolution to favor one product or another? So it is the ultimate way of doing both process chemistry and medicinal chemistry at the same time. And so that was why we thought, well, maybe two-phase terpene synthesis could be mimicked in the lab and have some inherent benefits. One benefit might be it could be much more efficient, maybe. Another benefit could be, if it's not more efficient, at least it could be more useful for medicinal chemistry. And really, the first description and proof of concept for this, after our initial 2009 report of UDESMAINS, but in a complex setting, was really Ingenol. And that stemmed from LEO Pharma. Gotta give a shout out to them, one of the first companies that really took a took a risk with us and believed in us and funded us. And the idea was they called me up and said, hey, could you make a gram of ingenol with, you know, we'll fund one postdoc in your lab. And I just sort of laughed and said, well, it's going to take more than one postdoc, probably going to take more like 5 people to do this, and expected never to hear from them again. When they called the next day and they said, okay, let's do it, which was, it was great, but it was also terrifying because we actually had to then figure out a way to do it. So ingenol was eventually made in just 14 steps, and it was perfectly symmetrical. So the first 7 steps made carbon-carbon bonds, and the last 7 steps made carbon-oxygen bonds. And one of the key steps here was this rearrangement from a, from a sort of farball-like skeleton to an ingenane skeleton through a rearrangement that on paper was not supposed to work. So you can check out the full paper in JACS to understand computationally why this reaction works. Even though the literature said this is never going to work. Forbal was the next target that Leo wanted us to work on, and that was also a really fun one, which was almost perfectly symmetrical. A lot, a lot of the carbon-oxygen bonds were made at the end of the synthesis, and one of the key steps that I remember from Forbal was the difficulty we had in installing one of the hydroxyl groups that were embedded onto the cyclopropane necessitated a kind of Cirque du Soleil in a round-bottom flask, where you have a cascade reaction here from this diketone to the cyclopropane, wherein there's 13 discrete transformations taking place in one flask on a gram scale. So it is a robust reaction. 2 new C-C bonds, 3 new ring systems, 3 C-O bonds, and then you cleave a bunch of things. And you get out the product in really good yield. So that was fun. But how about Taxol? That was really for us the kind of flagship of the two-phase approach because, you know, let's face it, it's the most famous terpene on the planet, and it's probably the most complicated terpene, complicated oxygenated terpene out there that has found commercial success. And it's kind of like if you took a normal molecule and you slammed it against the wall a million times until it was this sort of really densely functionalized and disorderly compound, where, as you'll see, or as we learned, anything you do to one part of the molecule, no matter how small, affects the rest of the molecule's personality. Now, historically, people have used semisynthesis to actually make a societal impact. That's the way the compound was initially discovered, commercialized, and to this day Now people use plant cell culture, but until recently people were using this enzymatically grown 10-acetoxibacatam-3, and then they were making a late-stage ester bond to put on the side chain at C13. The total syntheses have attracted hundreds of chemists over the years. There's been 10 total syntheses. Many of the strategies are shown here. A lot of students were educated making Taxol. It's one of the sort of, I would say, Manhattan projects of organic synthesis when it happened. Altogether, there's been about 39 milligrams of it synthesized, some of it racemic, some of it antiopure, and there's even been folks who've made it using a robot. So you can take one of the routes, you can automate it with a robot, and it'll spit out just enough micrograms for an NMR spectrum. So it doesn't— the robot doesn't necessarily shorten your route, at all. And Hudlicky has this really nice book, and he mentions in it that an academic type synthesis was a major challenge likely to be solved by preparation under 30 or 40 steps in length. Now our, our goal here was not to come up with a process route. I mean, we wouldn't mind it. We certainly tried to have something that could— you could make a kilogram of Taxol with one student in a week. That would be great. We wanted to do that. But we thought we have to be realistic. It's going to be very difficult to compete with plant cell culture, which makes metric tons of this compound every year through chemical synthesis. But we did think that at minimum, the two-phase approach would provide a blueprint for a hypothetical medicinal chemistry campaign towards Taxol. This was very fundamental in nature. We didn't expect there'd be any applications, and we had shown many applications already for two-phase synthesis that had actual medicinal use. Taxol is not one of them. It's mainly like a mathematical theorem that we were trying to solve for organic chemistry to see, is there a different way of making Taxol, one that would mimic nature, that would have some benefits, either efficiency, maybe not, or divergency, maybe not. Let's see. So we started our route to Taxol 13 years ago with— first we had to arrive at a cyclase phase endpoint, and so the cyclase phase was developed that could be used to make decagram quantities of the compound shown here called taxodinone. We then went about figuring out the various oxidation barcodes that were needed to functionalize all the different positions. So you've got to install 10 different oxygenations in Taxol to get this right. Nature uses about 20 different enzymes to make taxol, and 18 of those enzymes is just oxidation. So the allylic oxidase barcode was figured out, published about 6 years ago. We found in there that we could achieve the C5 oxidation using Ackermann-Bakewell-type oxidation, C13 through a chromium(5)-based oxidation, and then C10 through radical-based oxidation. We then went forward to Taxobacatin, and we had to learn the techniques of how we got the stereochemistry right at C2 and how you install that pesky C9 oxygenation. We could do that, uh, either through a Moof-type oxidation or another means, which you'll see in just a little while. And the remaining challenges were basically— we thought when the graduate student who eventually finished Taxol, Yuzuru Kanda, when he first joined, we thought this would be rather quick. It would be maybe, oh, quick summer project to just put two more auction atoms in, and that would be the end of it. It turned out to be a 5-year-long campaign where Yuzuru probably completed the amount of work that a normal graduate student would take 10 years or more to complete. And that was not because of me pressuring him. So here is the sort of overall oxidation choreography that we had to look at. If you take a look at this little chart here, you'll see biosynthesis does 5, 10, and 13. All the allylic CH bonds are at the beginning, then in the middle 9, 7, 2, and then at the end 1, 4, and 20. And, uh, for texaunane, we finished 5, 10, and 13. Those methods were critical to finishing the final product. And for Taxobactetin, we also mimicked pretty well the synthesis choreography. And our original plan here to install the oxygen atoms are shown at the bottom, and this was very close to what, what actually took place, although at the end one of those carbons needed to be oxidized at a later stage. So here are the key people behind this work. As I mentioned, Yuzuru, the star, graduate student who has graduated and he's off to do a postdoc at MIT. He was joined at a very late stage with a very talented postdoc, Yu Nakamura, and at the early stages we were lucky to have Shige with us, who is now a professor in Japan. And finally, this work would not have been possible without the collaboration we had with Chemvida, an amazing company that works on organic chemistry. They're based in India and they also have sites in the U.S. If there's anybody in pharma that's listening, and you want to work with someone on a difficult chemical problem, Chembita is quite amazing. So here is the synthesis, and I don't have time to go through every individual step. The full paper is out in JACS a couple days ago. You can certainly read it. I just want to focus on some key highlights of this route. First, let's talk about the cyclase phase. This is where Chembita came in to save the day, because we needed hundreds of grams of the starting material. And, uh, the first generation cyclase phase, while efficient, was not something you could actually make hundreds of grams of. And so they shortened the route by a step, they made it more process friendly, they changed the ligand that was used for the conjugate addition, and they also came up with a very simple recrystallization protocol. And so Yuzuru very rarely would have to go back and do enormous columns and scale up. Instead, Chembita would supply him with this key cyclase phase endpoint. The other key thing mentioning in the two-phase approach we took was that we made the strategic decision to install C2 and C13 oxidation because it's known that all taxanes have the C2 and C13 oxidation that are bioactive. And so those are the minimum two oxygenations you need to retain bioactivity. And so we actually make taxanes at step 9. After step 9, everything can be regarded as a taxane. Compare that to other syntheses where usually they don't get to the taxane until well past step 30 of their synthesis. So this allows for maximum divergency in accessing multiple different analogs. The C1 oxidation was something that took Yuzuru more than a year to figure out. And this was difficult because of not only the steric hindrance of C1, but also the competitive oxidations that could take place at the bridgehead olefin, and also the oxidation of C2, which was squelched by the use of the deuterium isotope effect. And finally, the peculiar conformation of ring C, which changed the reactivity at C2 and C1. And so Yuzuru was able to figure out that using a tertiary alcohol at C4 was critical, and as well, having the olefin adjacent was perfect because it would allow for an eventual redox relay. So we achieved both of these oxidations in the same pot using DMDO, and then the C7 oxidation was achieved using an oxidation relay through a carefully choreographed one-pot reaction that relied on everything being in one pot because of the instability of all the various intermediates in this cascade, wherein first we took that epoxide, opened it up to a very sensitive iodohydrin, then oxidized the iodine, leading to an elimination, then spontaneous oxidation of the eliminated iodine olefin to the epoxide. And that then set the stage, of course, for reductive opening of the epoxide using the Nugent-type conditions. So the completion of Taxol is shown here. The individual steps we can talk through at the end if you're interested for question and answer, but I refer you to the full paper, which goes through it. The final material we made, we got radiocarbon dating on to show that it was derived from crude oil rather than being derived from natural sources. The final oxidation choreography is very close to the biosynthesis you see. Really, we just exchanged the— order of oxidation of a couple of different, uh, of the oxidations, but very closely mimics that of biosynthesis. It takes place through the intermediacy of about 22 isolated intermediates, roughly 23 steps. We made 35 mg of this, and we would have made more if not for the coronavirus, because then Yuzuru could have processed one last batch we had. We were hoping to make about 100 mg of this, but we still have the intermediates in the lab. To do that if we ever want to go back and make more over time. So that was a fun story. So we should think about, you know, this key question, you know, would a complex terpene like Taxol be a viable starting point for a medicinal chemistry campaign if in the hypothetical case you couldn't get any from nature to do a semi-synthesis? Now before I answer that question, I will say for the compounds shown here, the answer is a resounding yes. So all of these compounds, if you needed to make a kilogram with synthesis, we've shown you could do that. And if you needed to make a number of analogs, you could also do that. In the case of Taxol, however, this is not a viable way to make multiple kilogram quantities unless you were really, really desperate and the activity of Taxol were picomolar level, and maybe you can convince a pharma company to adapt this route. To make that large of a batch. Really what the two-phase approach succeeded in doing for Taxol was the idea of a divergent path that could represent a really viable medicinal chemistry campaign. And who knows, maybe there is still something hidden in the Taxol framework, maybe a new drug in there somewhere that could be uncovered using the route shown here. That can access oxygenation patterns that are not easily accessed through plant cell culture and that are not at all accessible using convergent bond disconnections of the past. So I think I'm right on time, exactly 1 hour. Today I talked about a few different topics. You'll note that palauamine and maximycin really benefited from radical-based retrosynthetic strategies. Dexamycin is a fun natural product because I'll always remember it as a natural product that got us invested and interested in electrochemistry, an area that is really burgeoning in our lab. And then finally Taxol represented the culmination of more than a decade of work thinking about an alternative way of making terpenes that is predicated upon the biosynthetic logic that nature uses. And the most important slide is to thank all the super talented students and postdocs that really do all the work. I'm just their feeble spokesperson. And to them, all the credit is due. Also to these agencies for their financial support over the years. Really fun collaborating with so many folks from industry that give an amazing perspective. And the huge advantage of working with industry is that you can see your science have an impact in society in a very rapid fashion. So with that, I will be happy to answer any questions you may have.

Thanks so much. Thank you very much, Professor Barron, for your exciting talk about natural product synthesis. It has definitely opened our eyes to new approaches towards making pharmaceutical compounds, such as radical electrochemistry methods. So we will now open the Q&A to the audience. Due to time constraint, we can only allow one question per person, so kindly state your name before a question. Thank you very much.

Can I ask a question? Yes, please. Yep.

Hello, Professor Barron.

My name is Guillermo, and I have a question about the electrochemical reaction. And could it be possible to do this electrochemistry in an anti-selective way? Let's say, set up chiral centers in the electrochemical step?

Yeah, it's a great question. It is something that there's very limited precedent on. I think a great example of some recent work in this area is from Song Lin's group at Cornell has shown some of the very first, you know, useful electrocatalytic types of reactions, and I believe he achieves that redox chemistry with copper and a very cool ligand. There's another example that just came out in JACS a few days ago from the May group at SIOC, where they make biaryl linkages with a reductive nickel-mediated method. So this is beginning to happen. You know, there's no— I think what's special about electrochemistry is its heterogeneous nature. It has variables that are optimizable that don't exist in regular organic chemistry. And of course, there's the inherent benefit of sustainability. But often, as we wrote a review on recently, sustainability is not enough to convince people to use electrochemistry. You really got to show interesting new reactivity, and I think that's what we're beginning to see in the community right now. But yeah, the latent selectivity question is a good one, and I direct you to those two examples for the most recent contributions in the area. Right, thank you so much.

Yep.

Can I ask another question? Sure. So in, in the synthesis of ingenol, you mentioned this step that it was kind of crazy, this kind of semi-pinacle rearrangement, and Then you said that you use computational chemistry to explain how this chemistry was happening. However, how about the other way around? Can computations actually lead the way to find out new reactivity without the need to test 1,000 different substrates or reagents?

Yeah, it's a great question. I think the best example of Someone doing that is actually a former student, Tim Newhouse at Yale. He, he, he has a really provocative strategy for doing exactly what you just said, which is to de-risk key steps. In the case of Ingenol, I think you could use Tim's strategy to compute a variety of substrates for, for the most likely rearrangement substrate. That would have been great to do. I think that that's a great idea. In the case of something like Taxol, however, that, that might work. I mean, it's certainly— with the data that we have now shows it being so substrate dependent and so idiosyncratic based on tiny little changes. I'm not sure if the computation would be able to predict that, but I think that is a, uh, the to me, maybe the most exciting direction in computational chemistry for organic synthesis is to de-risk complex molecule synthesis so that you don't need to spend a year to make a substrate only to find out that it fails in something that you could find out by computation. The only risk associated with that is that if your computation is not done exactly right and properly, you could be missing out on some really cool discoveries because the computer told you it's not going to work. So it needs to be done in a, you know, I think in a careful way that you don't end up stifling your creativity.

Right, thanks so much.

Sure. It's okay if I ask a question? Sure. So thanks for a great talk, amazing chemistry. You talk a lot about radical chemistry and the future of radical chemistry.

Another big player in the field is photoredox catalysis that's been massively popular over the last decade.

Do you think that electrochemistry will in the long run outrun photoredox? What do you think are the potential pros and cons of the two? And from a sustainability point of view, definitely with photoredox relying on iridium and ruthenium. Do you think that the sustainability element of electrochemistry will outlast photoredox? I haven't given it much thought. I don't think they're competing. I mean, I think they're just complementary methods. And I mean, I think that they're both great because they enable allow things to take place that would sometimes be more difficult to take place otherwise, and they often have new reactivity associated with them. And so I think they're great. They can coexist happily. And then it's context-dependent when you scale up. For electrochemistry, there's already a lot of the largest industrial processes that are chemical in nature on Earth are actually electrochemical. Most people don't realize that. So the industry loves doing electrochemistry scale-up. For organic molecules, it's also known you can do it in flow. You can even do it in large batch reactors. There's a large facility at BASF that makes fragrances and does it electrochemically. And they, you know, the product is made— I visited it. It's made in like these water towers, store the final product, and it comes out through a faucet. I mean, it's just obscene amounts of material are made through electrochemistry. So for, you know, photo-induced electron transfer, or whatever you want to call it, you know, I think the future is bright for that. And certainly a lot of cool things you can do there that you can do electrochemically. And there's certain things you can do electrochemically you can't do photochemically, and vice versa. So I think the two fields are great, and they have, uh, you know, I don't think there's any competition for sure. I think they're just great, you know, great methods. Hi, Professor Baran, can I ask a question? Sure. Cool. So, I— your group is obviously working on a lot of different technologies to aid synthesis, so you work on electrosyn, and you've done a couple of developments in that sort of field. What's your opinion on other sort of technology-facing developments towards total synthesis? So, I'm specifically thinking about nanoscale, like, high-throughput experimentation, or like the applications of machine learning to organic synthesis? Yeah, I think for the first one, the nanoscale one, I mean, places like Merck have done it, a lot of companies do it. I think it's great. I think it just needs to be democratized. That's the key thing. For electrochemistry, it was not democratized. So you couldn't have anybody just do it, even though some people said, oh, I use a cell phone adapter. I use a lantern battery. That's not a sustainable way to have field, because you can't have— imagine if all of organic synthesis was simply stories around a campfire. Nobody would get anything done. So the point of electrosyn was make something the price of a stir plate that would give it accessibility to everybody on the planet, and make it a real science where you invent something and you know someone else can reproduce it. The same thing with the nanoscale. I think it's awesome, and certainly for every project in the lab, if we had something the price of a stir plate that would allow students to do 150 reactions an hour and analyze them all, we would be all over that. So it's not a question of whether we like it or not. We obviously like anything that can prevent students from, like Yuzuru, having to do a million reactions to get a PhD. If we could just simplify that and do everything nanoscale and have analytical techniques that would tell us what the products are and in a day get the data you would normally get from a month or a year, I mean, who wouldn't want that? So the question is just, how do you scale it? How do you get the cost down? How do you make it something which doesn't break all the time, that you have to have a full-time PhD running it? So if you can do that, which I'm sure the future will allow you to do, then I think that that's going to be great. With regards to the machine learning approach to synthesis, I think all of these things are a matter of time. And so as I mentioned, I think Tim Newhouse has got a great angle on this where you use machine learning computation to de-risk a synthesis of a complex natural product, for example, or maybe even a medicinally important compound that's complicated. That's in the short term, I think, the best use of machine learning, AI, whatever. There's programs out there that advertised to propose synthetic routes. And I would say that the days are still kind of early on that before they actually compete with even a first-year graduate student and SciFinder. And, you know, machine learning and robots or whatever to organic synthesis, I say bring it on. Bring all of it on. Bring your robots. Come like Terminator. It would be great. I'll take all the robots on. Bring them on. Because I think it can only be good, right? I mean, eventually, if the field becomes so much of an engineering that robots can do our job, you know, heck, I'll just go work at SpaceX. Hi, Professor Baran.

Yeah.

I have a question about the CH functionalization or electrochemistry also. I think, but from my point of view, the chemistry, the sugar chemistry or the glycosidic chemistry has far away of the age of the acidity, ideality, no? Because a lot of proteins are used. You're breaking out a little bit.

Yeah.

It's hard to— can you repeat the question because I don't know whether the organizer heard all of it, but a lot of it broke out.

Yeah, I am.

From my— for your point of view, the CH functionalization chemistry or electrochemistry also techniques is applicable also to glycosidic compounds or monosaccharides or sugar chemistry? Well, that's a great question. I mean, when we published our paper back in 2007 about, you know, advocating for less protecting groups, people was just sort of like, well, you know, how are you going to do that with oligonucleotides or oligosaccharides? Right, and the work we've been doing on oligonucleotides shows that there's actually a way of reducing the amount of chemistry associated with making the diphosphate linkages, so that, that is our contribution there, and in the case of sugar chemistry functionalization that doesn't require protecting groups, so it is a— the problem with sugars is it's context-dependent, it's very difficult to have general methods for, let's say, glycosylation that work on every single thing with the same catalyst, everything is protecting group free. And so it might be a case with sugar chemistry that, you know, some of the stuff like Chi-Wei Wang has been doing where you use enzymes to achieve those site-specific modifications is the way to go, or maybe in other cases some of Scott Miller's stuff where you use PEP to direct the site of acylation on a oligosaccharide, but it is a long problem. And I think, look, even if you can't solve it, the fun thing of working on a problem like that is it's going to require you to invent something. So it may not be solved in a year or even a decade, but I think as more and more people in the community work on such problems, it leads to valuable things. And then maybe 100 years from now, people look back and say, wow, sugar chemistry back in 2020 was like Neanderthal chemistry, and now look we can do this, this, this, this. We don't need any protecting groups. We can do everything in high yield with great stereo control. So it's just a matter of time, I think. Okay, thank you.

Good afternoon, Professor Barron. I'm Jihan and I have a question.

Can I ask? Hi.

So I'm quite concerned about the future of total synthesis because I'm really glad that you are doing such great work in the field of total synthesis, but as you know, a lot of chemists sort of think that total synthesis is not actually very applicable because, I mean, it's more of like a theoretical sort of our approach to synthesizing these molecules, but they take 20 steps, 30 steps, and a lot of times this actual synthesis, the elegant synthesis that we design, are not able to be put into industrial-scale productions. So I'm not sure what you think is the future of this.

You're breaking in and out. I'm not— I'm up. If you could rephrase the question and with a better internet connection, because I don't know. Bram, are you hearing the entire question? I've heard—

I think I've heard part of it.

Could you repeat it again? I'm sorry about the internet. So the question was mainly about what do you think is the future of total synthesis moving forward? Because a lot of chemists think that total synthesis is more of a theoretical sort of study because it cannot be easily translated into industrial use. For example, the 20 or 30-step synthesis, which is really elegant and beautiful things, but it's not immediately applicable. I feel like a lot of chemists are actually moving away from total synthesis.

Yeah, so, yeah, I think that, um, that's a good question. What, what, what, what is the point of doing this? Uh, I mean, we wrote about this on our blog a few times. You can check that out. Uh, I think I tried to make the point in the lecture today that Total synthesis is just, is a means to an end to get you to invent reactions that are useful, number one. So a lot of the stories I told you, in the case of Taxol, it led to tons of different methods that were developed over the course of the past 13 years that we published in a variety of different contexts. And the final, the final synthesis of things like polalumine, maximycin, dexamycin, Taxol, certainly didn't make their way to cancer patients to help them. That's not the point. And if you use that as a criteria to judge any piece of science, I will challenge you to find anything that takes place scientifically outside of a company which has that rapid of a rate of translation, where you can go from publication to cancer patient immediately. It's very rare, if at all, ever happens. And so I think that the, the scrutiny should be— before you go to total synthesis, the scrutiny should be put on a variety of other fields that I won't name, but you can certainly guess yourself, uh, that have really no application at all beyond just interesting fundamental science. And so the— at the very, very minimum for total synthesis, the sort of baseline is that you get students and postdocs that are educated better than anybody on the planet in terms of chemical reactivity. This is an irrefutable fact that any pharma company or any chemical company on the planet would like to hire those folks more than any other organic chemist that exists for that very reason. That's a fact. You can ask anyone in a company. So the baseline level, let's imagine you do boring chemistry, you make a target, it wasn't that inventive, it didn't really give you a new method or whatever, at the minimum level you have a student that will go on to reap huge societal benefit in terms of making new medicines, materials, agrochemicals. And what we aspire to is using total synthesis as a way to invent methods that have, you know, immediate impact in large pharma, because, you know, pharma industry, these are the places that actually make the translation to the patient. And we've also opined on this on our blog. You can read about it. I wrote an essay about it, of the many, many things that society gets by investing this infinitesimally small amount of money on total synthesis, which surprisingly are still the most read papers in all the journals. So for an area which, as you say, seems to be going down in interest, boy. It seems like a lot of people are really interested in it, and as an editor at JACS, I still get a lot of papers in total synthesis. So if the interest is going down, I'm not aware of it.

Right, thank you very much for your answer. I mean, I'm very happy to hear that it's still such a— I mean, it's such a good field because I'm personally also very interested in it.

Yeah, thanks.

Thanks.

Hello, is it alright to ask a question now? Sure.

I have a question about your reactions in different alcohols. A little bit of a kind of tongue-in-cheek what you were showing. You had this very cool hypothesis about the antioxidants in red wine could affect your activity and make it more efficient.

Did you find it less efficient?

Or less efficient, yes. Did you try and compare it against, for example, different wines where they would have different anthocyanins, or the white wine versus red wine where you wouldn't have that component?

It's a good— I think those studies have been done in more of a food chemistry world. It's a good question. I mean, we could probably use that radical chemistry as a litmus test to understand the degree of antioxidant or anti-radical character or radical scavenging character of various wines. I mean, I bet that could be done, but— and I haven't done the search on this— I'm pretty sure that people have studied wine to the point where they know the different varietals and the percentage of anthocyanins or different polyphenols that exist in them. But yeah, I mean, the main point for that was just to show you, hey, you could you know, make carbon-carbon bonds, make fully substituted carbon centers, uh, not in degassed THF in a glove box, but in like, you know, household solvents you can, you know, get at the supermarket or in your kitchen already.

Cool, thanks.

I mean, it could be cool to like look for counterfeit products this way as well or something. Yeah, yeah, you're right, you're right.

Thank you all for your questions. I think we still have time for one more question.

Hello, Professor Barron again. So I was quite interested about the industrial point of view that you said about scaling up reactions and also about these two-phase cyclase oxidase stages. But I mean, because that's what's happening in nature, but I mean, we don't see every day a plant or a marine sponge running column chromatography, which perhaps is the most time-consuming, tedious, and painful process when we are in the lab. So both in an industrial or in a future perspective, How do you see this, let's say, purification stages, which we don't really seem to, let's say, care about? We only are concerned with how efficient the synthesis is, but we are not really thinking about how many silica we are using, or how many solvent we are using, or how these can be reused or even substitute.

Yeah, it's a good question. I mean, I think in the case of Taxol, we were definitely cognizant of that, and that's why we did to the best of our ability, a lot of these reactions in one pot. You know, we only had 22 isolated intermediates, and that's because we were, you know, trying to do as many reactions orthogonally planned that would be compatible. Part of the route was designed with that in mind, where we could achieve— if we move one step from one location of synthesis to another, would allow us to merge them together because they would be mutually compatible. And so it's thinking about lowering your the number of isolated intermediates, you also lower the amount of chromatography. Now I think for industrial applications and for thinking about how people in modern pharma work, they obviously want to reduce step count because that obviously will reduce their chromatography load. And generally in pharma, you will avoid all chromatography and simply get to a waypoint that it can be crystalline. For example, in the landmark synthesis of irubulin that ACI commercialized, it's a, you know, 50, 60-step synthesis of that natural product. That work is accomplished by a series of dozens of crystallizations where compounds are taken on, intermediates are telescoped to next reactions, and then they get to a compound which can simply crash out of solution, and things are purified by recrystallization. So when you get to the level of process scale, These days they are so talented at finding crystal intermediates, making additional steps if needed to get to a crystal intermediate, and then purifying through that.