ACS DOC Webinar

追加: 2026-05-04

[00:00:00]acidic treatment and close that ring uh by ammonic reaction and this proceeds with high diysterio selectivity and pretty decent um yield over four steps.

[00:00:13]We can then take this compound and proceed onward.

[00:00:18]Now the next big challenge was the reaction I've showed below particularly forming that six-membered ring and in particular this C9 stereo center I've shown in red. We spent a lot of time um trying to con construct that stereo center.

[00:00:39]Now not surprisingly this um will not just close by a Michael edition. this is presumably reversible and um doesn't favor the closed product. We've looked at a lot of irreversible cycizations um pile chemistry and others and these always favor uh the incorrect ditherium at C9. So here's a case and and this is not unique to that lian. We've screened many lians and we would always get the wrong diaromer and many different um electrophiles at that spot were always uh forming the wrong uh stereo center including radical chemistry. So we could generate that tertiary radical with hat and then do a giza reaction uh exclusively the wrong stereo center. So really the solution to this problem uh proved to be somewhat simple. um we just needed to constrain that electrophile uh into a ring and so basically closing uh the ring prior to cycization is what we thought might might solve this and this proved to be true.

[00:01:51]So to get there we first did a birch reduction um which took takes the anosol down to this ketone. We can kealize and then uh protect both of those alcohols as acetates.

[00:02:07]We can then selectively reduce one of them uh with insitue generated Schwarz reagent and then oxidize uh with stalls condition uh that primary alcohol um to the aldahhide.

[00:02:24]We weren't able to selectively oxidize the primary alcohol in the presence of the secondary. So we had to selectively cleave the primary acetate. So that's how this sequence came about. And then there's all the carbons we need. We just need the acetate to cylize on there. And this uh proceeded smoothly with LDA and Bachan hydide to promote the E1CB elimination.

[00:02:51]So now we have our key um intermediate to close that ring. And what Shan found is a very I think somewhat elegant solution to this problem using a Mukayyama Michael type reaction. So when this compound is exposed to TMS tripflate what we see happening is the keel being removed uh CC bond being formed and we generate um two adjacent stereo centers.

[00:03:22]So we get um the correct diaromer that we've been looking for here. And kind of an added bonus is this quadinary center is also set uh with the correct configuration.

[00:03:35]So presumably the the rotimer that cycizes is this one where the where the dying is kind of uh pointing towards you and that leads to the compound with the enown um pointing towards you as well which is important because this is where uh the sul the sulonic acid needs to be attached. So this preceded as largely a single diyisterimer.

[00:04:01]Now when to form the 331 was of course on our mind.

[00:04:07]As you might suggest, we've tried using this esther to form that and this this didn't work. And it's not surprising based on the early precedent I showed from the isolation group. And we also have to find a way to put on that uh sulfonic acid which I mentioned is really quite a unique functional group um in synthetic chemistry sulfonic acid enown. We were inspired by the biosynthesis of the natural product.

[00:04:38]though after the oxidative rearrangement and you form Tuan sulfonine A there's presumed to be an E1CB elimination uh to generate that moy now we weren't really pursuing a biomimetic route but we thought we could gain access to a similar intermediate uh by opening an epoxide uh with some sort of uh sulfur source basically to get access to a kind of quasi biomimetic um intermediate which then could undergo an E1CB elimination.

[00:05:16]And so the first thing we did is open up the lacone uh with nitrogen and uh particularly analin and trimethyl aluminum. So this gives us an opportunity to then test protect the primary alcohol. So this can't close back down. We can then epoxidize the olifen under careful conditions and this doesn't oxidize the amine. I should note Shong basically carried out this whole synthesis with a free tertiary amine which is uh very bold move on his part but we figured out a lot of this chemistry. Um so this is um an epoxidation and then we could open that uh with triddle protected theol and in the same pot with base this will undergo an E1CB elimination now to give us uh this compound shown here.

[00:06:15]We can then form the lithium enolate which upon addition of bacan hydide will activate the amide and we can now get the deman cycization and this is not reversible because um the test protected secondary alcohol prevents that that retroman. So these are stable compounds. And then in the final step, uh kind of another risky oxidation, we just treat this thing with H2O2.

[00:06:48]Um in the presence of TFA, which presumably proteinates the tertiary amine, protecting it from oxidation. it cleaves all of the the acid labile um silicon groups and oxidizes the sulfur up to the sulfonic acid oxidation state um giving us the natural product. And so we're currently working on uh rendering this asymmetric and also exploring um some of those early precursors for um divergent access into different during alkaloids. So we think um this could be a a general strategy to access a number of project um natural products and this is ongoing work in the group.

[00:07:36]Now I'll switch gears and mention um a natural product that really has no bioactivity. I know that's usually reported as the driver for for our field, but this was one where the structure was just so unusual we had to we had to take a closer look at this and ask you know why is this thing made and how is it made? So this is this natural product gremison. This is isolated from a stptoyces uh found in China.

[00:08:11]Um I think many of you who have looked at a lot of natural products will realize suloxides are quite unusual.

[00:08:19]Sometimes we see them in peptidebased natural products where they're clearly a result of a methionine oxidation or something similar. But this one of course doesn't look like a peptide. In fact, it's clearly an angiocycan.

[00:08:36]This is a tetrayclic family of natural products. I think that needs very little introduction. It's the largest family of um aromatic type 2 PKS derived natural products. So there's hundreds of of different angiocycles that have these four rings and most of them have anti-tumor or antibacterial activity.

[00:09:02]Now angiocycles that have sulfur atoms are incredibly rare. There's probably fewer than 10 known and this is a assortment of them.

[00:09:13]Now after greamy was isolated another conjuner which bears clear resemblance but has a triculfide bridge um was also isolated by the same group um along with um groups at scripps Florida. And so this is uh clearly a result of something similar and I'll I'll discuss that. And then there's a couple more which for quite a while had mysterious uh biosynthetic origins.

[00:09:47]Now the natural product uh group who isolated gremison proposed that the angiocycan cage um under goes a dehydration which gives you those two ether linkages.

[00:10:02]And then there was some illdefined self-hydrolation step. So this was a a bit problematic to us how this would work. There's not really a biosynthetic reaction that would do kind of this uh stereo retentive displacement of a hydroxal group. And so this seemed a bit unusual to us. But the how this was made in nature was unclear when we started.

[00:10:31]Now Ben Shen's group um to their credit has shed a lot of light on how these thioangiocycles may be arising in nature.

[00:10:42]So he has studied this um um in a number of streptoyces and what he's found is that the biosynthetic gene cluster basically ends at the stage of these tac epoxides.

[00:10:57]So there appears to be no enzyme that makes that carbon sulfur bond. And what's proposed is that these undergo um non-enzyatic reactions with low molecular weight theols.

[00:11:11]And he's proposed that this is part of uh basically an innate detoxification system in bacteria. So if hydrogen sulfide adds in, you can basically rupture that epoxide generating this compound here. And this is set up to lose water and form a thofphenol.

[00:11:35]This molecule can of course attack another starting unit and that makes um basically these large dimers after aromatization.

[00:11:46]And this in this case would be essentially how donghe sulfan is made.

[00:11:53]Now if hydrogen tulfide adds in there is then the opportunity for the other end of the sulfide to essentially mic back in and form that ring. And then you can envision uh the final two hydroxils engaging in a Michael and keilization step.

[00:12:17]So this is presumably how neo griem is made.

[00:12:22]So back to our target. Um we're still weren't sure exactly what the biosynthetic precursor of grie would be.

[00:12:31]Note that greamy is in a lower oxidation state than neogomy. Right? Because at this carbon shown in red we have a CH bond and not a CS bond. So presumably the biosynthetic precursor unless there's a redux adjustment would be a lower oxidation state um derivative and so we speculate that this um denone is actually the precursor to griezy in nature.

[00:13:02]Interestingly, this was isolated by the same group who isolated griezy as along with um some re related um angiocycles name known as the kiamy.

[00:13:18]Now this compound looks like something you wouldn't want to carry through a long synthesis, right? It's set up to aromatize um at several positions. And so this was something we would we think would be pretty fragile.

[00:13:35]This work was spearheaded by Conincaid who took on making this pre biosynthetic precursor. He was joined by uh two very talented undergrads at the time who are now uh firstear graduate students uh Dylan who's in my group and Carlilele who is in the Baron lab at Scripps. So uh this team was really all the work was was their own and I owe them a lot of credit for pulling off a lot of uh very challenging reactions.

[00:14:05]So to mask that enown, we had hoped to use um this 222 BIC cycle basically as a mask which could then be unveiled late stage by an E1CB elimination.

[00:14:23]So this actually looks kind of like the natural product, right? We've got the 222 cage. We have this proton here uh with squiggly stereochemistry. So we didn't think that would be important because we're going to deproinate there and then unveil the Michael acceptor at a late stage.

[00:14:41]This we had hoped to be able to um synthesize using a barbier type reaction to form that six-membered ring along with an olifin hydration.

[00:14:54]And then one of the key steps was a Fredelcraft's reaction. So we had hoped that if we had an electronrich aromatic ring here, this could attack this oxonium ion and make the central sixmembered ring.

[00:15:09]We like this disconnection because the precursor is basically a lacone and this specific lacone looks like a dealer reaction between a hydroxy pyone and an activated denaphile and there's of course rich literature in hydroxy pyone cyclloitions. So this is where we set out uh our investigations.

[00:15:35]We could prepare this hydroxy pyone in three steps and evaluated it as a competent dying in in cycllo editions.

[00:15:45]Not surprisingly, we would see the endo product uh with malleic andhydride.

[00:15:51]When we switch to a more representative denapile to our synthesis, we would start to see uh the product we wanted.

[00:16:02]This might go by a step-wise mechanism.

[00:16:05]We we see that if we use the Z configured um alken we get the trans product. So it could be isomerizing or a step-wise Michael closure.

[00:16:17]One of the other benefits of hydroxy pyone as a starting material is that there's some really beautiful chemistry by Lee Deng's group uh using cona alkaloids as bifunctional catalysts for uh catalytic asymmetric uh pyone cycllo additions. So he has several different um catalyst architectures that can promote this. What we found is this uh QD2 that he had developed was a very competent catalyst for the merger of these two pieces.

[00:16:54]So with 10 mole% of this just simple stirring at room temperature we can get uh pretty good yield and very high ees of this uh 222 B cycle. So this was a a very nice opening sequence for us. And then we could reduce this with sodium borohydide and this would um generate a stereo center here by aerocetone reduction correctly if the tertiary alcohol was first protected. So TMS tripflate goes on here. This presumably prevents a directed reduction which gives the wrong stereochemistry.

[00:17:36]So that's why the the TMS group goes on.

[00:17:39]So that generates this initial sodium um aloxxide. We find that this swings over transactizes again with the esther below it and then more hydride reduces that to give us now this compound shown here.

[00:18:00]And we are set up to explore that freed crafts reaction.

[00:18:05]Now a number of luis acids um including tint tetrachloride which was optimal would uh promote ionization to the oxonium ion but we would see formation of a CO bond uh rather than the desired CC bond of the freedcraft's product. So many acids would basically get stuck at this point.

[00:18:29]We found that this could then be telescoped by addition of uh titanium tetrachloride in lithium perchlorate which presumably reionizes that CO bond and then allows the aine to u participate in the freed crafts reaction.

[00:18:50]And this gives us u a workable yield and 3:1 regio selective um ratio relative to where the positioning between these two groups.

[00:19:02]You can see that we need to use basically a blocking group here. So if we just use a simple anosol ring which would be ideal for the synthesis we get essentially the para isomer in the freed craft. So that was something we couldn't control. So we had to put on this extra group here which we'll need to burn off at the end. So this was one of our concession steps for this approach.

[00:19:31]We can then generate the alyic iodide from the alyic chloride and then treat this with indium zero to do a barbier type addition. And this is a rare example of an alyic indium attacking an esther. So this chemistry is usually seen uh with aldahhide or ketone electrophiles. So this is situated right underneath it and so probably proximity helps this closure but gives us this enown uh in very short order. We can then use a copper catalyzed boral addition which comes from the desired diyisterase.

[00:20:12]This gives us a BPIN group there which we can then oxidize and then then under very carefully controlled conditions sodium hydide at minus30° we can elicit the sight selective E1CB elimination.

[00:20:30]Okay. So that's deproinating here and breaking open this uh ring to generate our um ketenone. And this is obviously a sensitive reaction, right? If it if you go it gets too warm, we can obviously enalyze on the other side and then we're basically out of phenol here and it's it's game over. So this was a really um concaid's uh diligent work to get this to to go cleanly.

[00:21:02]Okay. So we're now set up for the biomimetic cascade. And I will redraw here what we're hoping to to elicit.

[00:21:10]This is a 16 addition of sulfur followed by a 14 addition and the keilization.

[00:21:18]Now we initially used uh ethane theol because it's it's not a gas like methane theol. So we were conducting our model studies initially with the ethyl group.

[00:21:29]Um our first hits came when we went to polar solvents.

[00:21:34]So with ethane theol and dmf we saw initial um6 addition and then water actually helped to pro um promote those second steps. So those Michael and keilizations. So we can then get the full ring system and ultimately we had to then find a suitable methane thile equivalent.

[00:21:59]And what worked nicely is this uh thio thio methylated um isothioua.

[00:22:08]So these have been used in medcm a bit kind of sporadically as theolate equivalents for SNAR reactions. Right?

[00:22:17]So in the presence of aquous base this will break down uh to release um the theolate and presumably this releases it at a a good rate for this um cyization right it doesn't just immediately give us a gas which bubbles out. So with this salt we can get 60% yield of our cascade product which was great.

[00:22:43]We then explored the oxidation of this compound. So at minus 80 with MCPBA we get uh diysterio selective um oxidation favoring the natural configuration.

[00:22:58]We then can exchange the O group for a methanol equivalent. We think likely that greamy's um keel here is an isolation artifact.

[00:23:10]So neogy has an O and the extraction of this natural product used a lot of methanol. So we think that this is probably um where that's coming from in nature. We can then convert the TIPS group to the tripflate and do a palladium catalyzed reduction to generate um the natural product. And we're working to um use this towards the kiamy as well.

[00:23:39]Okay, I'll try to speed through this last one cuz I see I've only got about 5 minutes left. Uh the furanosoids I think are natural products that really need very little introduction. You can see at the bottom almost every synthetic group that I'm aware of has dabbled in at least one of these at at some point. So there's been a lot of interest particularly with this molecule in the top right corner which has a very interesting cage polyyclic topology. So this went as an unsolved uh problem in synthesis for quite some time.

[00:24:18]It was uh initially made in a semiynthetic process by patinence groups. So he found that epi singleptapalide could serve as a potential uh biosynthetic uh precursor. So if this molecule is acetylated you can do basically a CC bond formation between these two red circles kind of an addition elimination and then these orange circles can connect by another Michael addition and then in 2022 uh John Wood's group performed the first total synthesis um of this natural product using a a bio inpired appro approach and this really opened the flood gates to a lot of um subsequent work a really nice synthesis by Brian Stoultz's group and then the following year two syntheses by the first ner and then SARA groups which were important because they also established that ineliganalide can come from scablide B which is an an isomer um in this lacone which then a pimemerizes.

[00:25:27]So this offers additional biosynthetic insight into this natural product. Now our groups worked in sbronoids for at least six years. A number of students have and the work I want to show you today was really spearheaded um by this talented postoc Quan as well as Danny and Katarina two very uh exceptional grad students in my group.

[00:25:54]And so our retroynthetic disconnection was to just um add an additional H uh additional O which we could then remove.

[00:26:04]And this opened up a deio type disconnection where we can basically cut the molecule in half by a 2 plus2 in fragmentation route. And this um chromophor we would need for the 2 plus2 is kind of an unusual one. It's a alpha beta unsaturated carbonal but also an enol of a dichetone. So we weren't really sure how this would would react in 2 plus2s and we were hoping to get the transcllo adduct. So um 2 plus2 reactions with cyclopentinones if if you're familiar often have mixtures of stereoisomers at these two positions indicative of stepwise uh diradical type chemistry. So um the initial excited enone probably adds in and there's a radical there that can then reclose with either stereochemistry. So this was something we wanted to try to actually take advantage of. So we do an alilation that gives us a stereo center. We can then use stalls um or sigman's sorry um uh var modifications using spartene and then we proceeded to add in a lithate and this reaction was not very diio selective we would get about 1.5 to one if we used chelating groups we could use keilate control to very um to favor the wrong one but if we silated the secondary alcohol we could get the compound we wanted. So we have a way of getting either stereoisomer. This is important because um some furanosids namely bechowsky have the opposite configuration at that tertiary alcohol.

[00:27:56]So we think we can get either one. For in analide we do a posaction which gives us this compound here with good dr. And then really a challenging step was to make that chromophor. So we have the correct oxidation state in the posroduct.

[00:28:15]We just need to shift a few things around. So what we did is silate that and then when we work this up with HCl and fluoride it hydraizes and we get back to this compound here.

[00:28:29]We then explored some photochemistry and nicely we found that we could use recemic um cylohexenol and only one of the anantimers reacted and under optimized conditions we can form now the transadduct uh with good yield.

[00:28:49]And um what was really nice is that if we just take that photochemical adduct which was somewhat unstable and add HCl it fragments to generate the ineliganalide ring system in a single step.

[00:29:04]So we think that this goes by desilation first that allows then a first asterification which then allows the U pi star system to be aligned for the fragmentation.

[00:29:21]So this presumably doesn't fragment until this laces made. We thought it made a diet ketone, keilized, tattoized.

[00:29:30]What we weren't so sure is why the cis isomer, which was another thing in some of those photochemical reactions, this did not undergo the desired uh rearrangement, right? This could this could potentially work because at the end this will become a ketone and this could be a pimmerized.

[00:29:52]And so this was kind of puzzling to us.

[00:29:54]But Katarina sheds some light on this by DFT. So we see that both the cis and trans isomers can fragment.

[00:30:04]The trans one though appears to break the CC bond and kealize in the same step. So basically it goes all the way to product in a single uh transition structure. The cis isomer undergoes fragmentation and then reveals this dicetone. So we think that this thing is prone to decompose. Uh it can basically eliminate or retrodol while the transis isomer basically avoids this intermediate altogether and go straight to the product.

[00:30:41]Then in the final steps we just need to take off the O group. This uh proved to be quite straightforward. Um we activated it. Um and we were hoping to make a cyclic sulfate between the two O's. But what we found is that actually a tertiary chloride was made which could then be um reduced with samarium and then a single uh um global oxidation formed first a carbonal and then an enown and then in the last step we add in uh the isopropenal group which comes from the correct face uh leading to a a 10-step uh asymmetric synthesis of an eleanalide.

[00:31:27]and we think aspects of this route um are definitely amendable to other members of the family and this is ongoing work in the group.

[00:31:37]Okay, I'm sorry for going a bit over, but I want to end with of course the most important slide, the funding agencies, particularly the NIH who funded a lot of our synthetic work particularly in the tarpine space and then the very talented uh group of graduate students and postocs I've been blessed to work with and I I tried to point out their contributions along the way and I would be happy to take any questions you might have. Thank you >> Tom. That was some claps uh from me but on behalf of the entire audience uh really a fantastic talk. I loved hearing about the strategy and reactions. I think because of time we're going to pause for pause questions and have them be addressed um on the using the text forum.

[00:32:31]Um so with that I will turn the mic over to Alex.

[00:32:36]Yes. And in the meantime, artist, if you could begin sharing your slides. Um, so our next speaker of the day, um, is Artist who joins us for merc research and development. Uh, prior to his very successful career at Merc, artist completed his PhD at the University of Wisconsin Madison under the leadership of Professor Edwin Viday. This was followed by a post-doal research fellowship at MIT with Steven Bookwald.

[00:33:04]Aside from an extremely impressive publication and patent list, artist is a recipient of the ACS Young Investigator Award as well as the US Environmental Protection Agency green chemistry challenge award. So today, Artist is going to tell us about Enlistide, which if you are unfamiliar with, I think you will be in for a real treat because this molecule is truly beautifully complex.

[00:33:29]So on that note, artist uh take it away.

[00:33:33]Yeah, thank you so much Alex for the for the introduction and then for an opportunity to present here. Yes, I'll uh be presenting on enolicide today. Uh enlistide is a PCSK9 inhibitor and that is the uh validated target for the treatment of uh cardiovascular disease and I won't go into much details of how the mechanism is complex. Basically PCSK9 regulates the um LDL receptor which is uh responsible for capturing uh the LDL cholesterol particles basically the uh the bad cholesterol.

[00:34:15]uh so and there are already uh several approved PCSK9 inhibitors on the market but they are delivered they're large molecules delivered by injection and that's uh significantly limits their use. So our um goal would be to uh find an orally available cholesterol-linging pill. What makes this um complicated is that uh the protein protein interactions in general are very difficult to um uh target and uh so so in this case as well the interaction between PCSK9 and LDL receptors a large uh flat surface. So it provides a a great challenge for designing a a small molecule. So uh fortunately enlistide is indeed an orally bavailable inhibitor and uh at only 1% molecular weight of uh antibbody PCSK9 inhibitors and uh we have uh recently completed um phase three uh clinical studies with uh positive results and FDA has uh granted uh the commissioner's national priority voucher for enlistide.

[00:35:39]So, we're all very excited about this molecule.

[00:35:43]Uh, obviously there's no such thing as a free lunch. And, uh, shown here, you're not supposed to see all the details here. My apologies for the very fine uh, print here. Uh, I just wanted to illustrate that this this is a very complex synthesis. And this is not like the discovery synthesis, not the first generation. This is already chemistry that was highly optimized and performed on over 100 kilo scale. Uh one thing I just wanted to point out on this slide although there is some convergence. So you can see we're bringing two small things here on the slide together. It is then followed by about 10 steps where molecular weight basically doesn't change. And this is really bad for the efficiency. But as I said this is already kind of state-of-the-art synthesis we can achieve with the existing chemical methods and this also involves high volume microcyizations as well.

[00:36:42]So uh shown here is kind of the progress we made as we develop this compound. So discovery we start over 70 synthetic steps and then we go towards 100 kilo synthesis. You can see kind of we're approaching a uh a flat line uh basically plateauing out and um half of the steps are actually protections deproctions. So so it's not very um um sustainable either but most importantly even though we're able to reduce some cost initially we're nowhere near uh meeting commercial availability using this uh chemical approach. So we need a breakthrough to enable a better route and uh basically if you look at the retroynthesis clearly much simpler retroynthesis than we saw in the previous talk uh it's a microcyclic peptide. What's unusual it's a tricyclic microyclic peptide. It it's unusually heavily cross-lin uh but you can still break it down to 13 small pieces and two natural amino acids. um uh six unnatural and uh three uh linker fragments. The first thing that caught our attention when you look at just these simple fragments uh is that three of them are complex unnatural amino acids are not really readily available and uh as shown here the known chemistry is highly inefficient. So for just this really small amino acid one needs seven steps 6% overall yield and also a cary separation at the end uh to provide an anti-opure compound and the other amino acids may be a little better but still either expensive or very long steps and this is where um bicatalysis came to our rescue. Uh for example, one can just take simple natural Loline and use an engineering hydroxilase that will selectively hydroxilate in the desired three position which is uh different from their preferred natural selectivity in the four position and uh we were um uh if you're more interested more on this this was done in collaboration codexis we published the work in organic or PRD recently and the other two amino acids.

[00:39:11]Similarly, we can use uh u enzymes that are involved in natural biosynthesis amino acids and engineer them to provide uh the high selectivity for the desired products. uh in this case we're using tripsynthetase which was pioneered by um the Arnold group and here we're using phenylammonas and running it in the opposite direction in the synthetic direction.

[00:39:42]So with these uh building block problems solved we have 13 building blocks. So it's about 6 billion possible routes one can imagine. We need to set some priorities how we are going to uh put molecule together and we really want to maximize the convergency and u have very efficient macrocyizations. That's where we had a lot of inefficiencies in the previous route and we want to avoid protecting groups whenever possible and obviously no chromatography. You don't want to use any chromatography.

[00:40:16]And the first disconnection that we really wanted, we kind of the the biggest complexity in this molecule is we have these three fused macro cycles and we felt that uh the greatest simplification, the most powerful disconnections would be to break off these three rings.

[00:40:35]Uh that would simplify the structure but also it would provide convergence. And the first one we wanted to break off was this uh uh western ring containing quatronium ammonium salt.

[00:40:49]Mainly because it's not easy to carry through these quatermonium salts to the synthesis. They're highly polar uh water soluble. They're kind of basically surfactant like it's kind of comp. You don't want to work with like a a soap solution throughout the 20 or 30 steps. So, so that's kind of like the last one, last fragment that we want to introduce and then we are left with the two fused macro cycles and we want to break the larger macro cycle again to simplify the structure as much as possible and finally we are left with this smaller macro cycle that we could uh imagine assembling in uh convergent convergently assembling in one step if possible. So obviously these are very aspirational uh transformations mostly because we have a lot of functional groups and we already agreed we don't want to use protecting groups wherever possible. So traditional chemical methods would really struggle with the selectivity here. So again as we already did in the case of the monomeic amino acids we turned for help to uh uh the nature and uh bactus can enable routes that are very difficult chemically with extremely high selectivities. And we can also combine multiple biocatalysts to work together to accomplish cascade transformations.

[00:42:18]And more importantly we can involve these bio catalysts from natural starting points. There's only one challenge with that. Uh if we uh often times especially in our case our desired function is very far removed from the natural function. So if you want just look at the enzyme sequence there there's a lot of modific need to make a lot of modifications of the wild type enzyme. And the problem with that is that if we screen the wild type enzyme for the desired activity, we don't see any activity. So basically we cannot really uh evolve or select the right variant. And that's where we need to rely on substrate walk. In the substrate work, we uh select a substrate that's maybe closer to the uh natural substrate, but maybe has some features of the desired substrate and then we evolve it for higher activity until we uh reach some activity with the next substrate that is even closer to the desired. So basically repeat the cycle until we get uh to desired endpoint.

[00:43:28]And u the first cascade where we wanted to do that is this um basically assembly of the cyclic tetropeptide from a deptide and uh two amino acids unnatural amino acids. And clearly chemically without protecting groups this would be impossible because we have all these competing nucleophiles that would basically form a intractable olygomeic mixture.

[00:43:53]uh in nature there are multiple ways how one can make amide bonds. Uh obviously we don't want to use ribosomal peptide synthesis just because these amino acids are so much removed from natural ones.

[00:44:07]Uh what got our attention was um ATP grasp liases.

[00:44:13]uh those are um enzymes that can uh perform ami bond um formation using ATP activation. Basically they proceed through asil phosphate intermediates and uh one can also use them to elongate the peptide chain.

[00:44:36]So obviously we also need u to since ATP is not very uh cheap we also need a recycling system for that. The problem with the uh grass libase liases is that they only work with small asil donors.

[00:44:52]So we cannot really use them for this final cycization reaction here. We need a different enzyme.

[00:45:00]And here we looked for um using an eststerase uh in a function of a macrocyase.

[00:45:08]So the way it works is an eststerase takes an esther. Essester provides the thermodynamic driving force for the overall transformation and it uh basically transillates it to serene in the enzyme active site. And normally this would react with water to form the hydraysis. But um we we thought that we could also have an amine acting as a nucleophile instead of water. And indeed we were able to pick up some activity on wild type enzymes for the desired uh amide bond formation versus uh uh hydraysis. Obviously we needed uh further evolution to improve this selectivity.

[00:45:52]And uh I'll just illustrate the substrate work for the uh ATP grasp ligase for one of them. So as I said we wouldn't be able to see reactivity with the uh intended substrates with the wild type enzymes even if we screen large collections of them. So we started with a model system basically two tryptophans we just uh and we do pick up activity on that and then we gradually walk over to the nucleophile being a natural deptide of proline 3anine and then we start introducing some um uh steric bulk in the uh proline until we uh arrive at um the right substitution pattern.

[00:46:38]And then the only thing and we also add some bulk on the uh tryptophan nucleophile until we arrive at the uh desired uh product. So this is kind of a very uh simplistic uh explanation of how substrate walk works in this case. And as I mentioned, we also need u to recycle the ATP co-actor because uh we want to make it more efficient and uh in this case we're using kynise to recycle it and uh sodium polyphosphate as the uh stochometric uh source of u the u uh to drive the reaction. And obviously we all of these enzymes also had to be evolved for selectivity activity uh lowering loading and also higher concentration. And in this case since we're using salt sodium polyphosphate also for inorganic phosphate tolerance. But we were able to do in over 20 rounds for all three enzymes and uh enable them to work all in the same pot.

[00:47:46]And uh at the end of the reaction what we are getting is a a relatively usual looking white slurry. Nothing special but we have a peptide that we need to get out of this uh aquous suspension. And the way we dealt with it is the first thing is we um do a um pH swing to denature the enzyme and precipitate it out so that we can filter it off. And then we use uh nbutinol a green solvent to uh extract it out. And then uh basically we have a solution in nbutinol. Even though nbutinol is a green solvent, it's a drawback. It's hyolic point. So we don't want to concentrate it. We we were very lucky to find a um uh crystallin oxilate salt that was really critical for this process that enabled us to directly precipitate the uh uh bisoxilate salt of this uh cyclic peptide uh from the organic layer and uh this basically completed the synthesis of this northern fragment and we have run it on several hundred kilo scale already.

[00:49:01]Uh so with this accomplished we uh needed uh to look at the next transformation. We needed to add the eastern fragment. And for eastern fragment we uh actually could use chemical synthesis. Chem the advantage of chemical synthesis it is uh very it's it's flexible and also portable. So it's very easy to scale up. So in this case uh it it it was much more preferable than using an enzyatic route and uh notably here we also can use an unprotect an amino acid without protecting groups. Basically in this case we're using carbon carbonil dimeazol as the activating agent and reaction proceeds through naroxy and hydide which um allow us to avoid protecting groups in in this transformation. We do have a bar group here but it's a relatively low cost to to pay for the overall advantages.

[00:50:03]So um obviously challenges in bringing the two larger pieces together.

[00:50:11]First challenge is that we have this uh isopropylster here and it is a little bit um sensitive to hydrarolysis. It's the background hydrarolysis. So we decided to use this as the first transformation to join the asil group here with the isopropyl with the amino group uh so that we don't have background hydraysis and then we would think about how to close the ring afterwards and obviously we have again the um problem of competing nucleophiles. So we look for um biocathalysis as a uh solution for this synthetic problem and uh we decided to use the thsterase domain of the non ribosop non ribosomal peptide synthetase uh for the um formation basically for the performing per performing the uh peptide liation naturally uh this is the C terminal um uh domain of the NRPS and usually it performs the cycllization but there are also some examples in the nature where uh they perform uh loation of uh two fragments. So we were optimistic on using it and also there was some groundbreaking groundbreaking work by uh the Walsh group where the um leading group in the uh protein pepidil carrier protein could be replaced with a much simpler analog anacetylcyamine.

[00:51:46]So building on this uh prior knowledge, we screened the um uh cyamine est thioester as a starting point and we did get the desired reactivity with the amine. However, the reactivity with the alcohol predominated. So we need to perform a substrate walk where we temporarily block the alcohol as an aid and uh then we were able to um evolve for the desired um uh selectivity at the amino nucleophile and uh we were then able to switch back at some point to the required alcohol and uh at some point uh later in the evolution to the isopropylster.

[00:52:36]And we selected isopropyester because it's synthetically easy to prepare and it also provides it's a small uh group uh and also provides um the optimal hydraytic stability as opposed to let's say methyl or ethylester.

[00:52:56]Uh so obviously this was the most difficult of all the transformations in the synthesis. So we needed 34 round ohms uh to develop it. And uh then we turned to the next problem. How do we close this uh large microcycle here? And we decided to use um a redactive amination in an enzyatic fashion. So first we would oxidize the alcohol to the aldahhide and then it would spontaneously be in equilibrium with an amine. That amine reductase enzyme would then take and reduce it down to the desired diamine and amine reductase would provide us the selectivity for the macrocyclesization versus olymerization that would happen if we for example did it chemically.

[00:53:52]And um one thing that caught our attention is that um keto reductase and amino reductase can use the same co-actors.

[00:54:01]So we could basically couple the two uh reactions in uh one cycle uh providing a hydrogen borrowing cascade. And uh we're indeed uh pleased to see that it works and it worked up to 77% as a yield.

[00:54:20]Unfortunately, um there's a thermodynamic limit to this equilibrium and uh 20% starting material remained.

[00:54:30]That's a thermodynamic limit. There's not much we can do about changing it.

[00:54:35]And uh you would think okay 20% is maybe not so bad, but in our business uh every percent is money. So we really wanted to maximize the yield for all of our reactions. So we decided to uh separate the two en enzyme recycling systems and basically for the uh keto reductase we would use NAD and for the amine reductase uh we would use NADPH and then we would have a dual recycling system. For NAD we would use lactate dehydrogenase and alpha ketogluterate to recycle it. And for the NADPH we would have alcohol dehydrogenase and isopropyl alcohol as the hydride donor. And after uh further evolution we were able to uh evolve it for very high selectivity and uh much improved yield over the thermodynamic ratio of this transformation.

[00:55:40]And um we were also able to find a crystal in Bis HCl salt in this case. So again we're able to crystallize out the diamine salt with great upgrade in purity without using chromatography.

[00:55:55]Um and again this this has been demonstrated on multi 100 kilo scale already.

[00:56:02]the uh this leaves us with the final problem here which is how do we attach that western fragment and the synthesis of western fragment again we're using um chemical synthesis and in this case I would even argue that chemical synthesis perhaps uh more optimal because these are all polar intermediates if we let's say made them biocatalytically through bio ami bond formation it would be very difficult to extract them from water so having this run This chemically actually provides uh practical advantages.

[00:56:37]And um one another important uh feature of this western fragment is that this only has one caroxyic acid and all the other functional groups are basically protected. There's no free amines.

[00:56:52]So we could in this case consider um highly selective chemical coupling as well, not just via catalytic.

[00:57:00]uh and uh the only problem is that the other coupling partner is a damine. So it has a primary amino group and a secondary amino group and unfortunately the undesired coupling to the primary amine predominates.

[00:57:16]We still didn't want to give up on this so easily because this was a difficult transformation to perform biocatalytically. It's a secondary amine. It's a non-natural caroxyic acid.

[00:57:28]uh and uh we were thinking is there any way to selectively block the primary amine so that we can do the isolation of the uh secondary amine and one uh temporary insitue protecting group like that would be an amine shift base. We did find that benzaldahhide was able to revert the selectivity to the desired always some issues of course it also as you form an amine which is sometimes used in phase transfer alkalations as well it acidifies the alpha proton to the nitrogen. So we screened other aldahhides that wouldn't provide the erosion of um kyality and found that the more electron-rich solicahhide gave us uh even improved yield and selectivity and more importantly also much higher carl purity. So we went forward with this and uh we're able to form the uh substrate that we're only we need to form one more bond between this ail group and amino group. And for this one since um we're at the end of the synthesis uh we don't really we want to have a transformation that's extremely high yielding and highly selective. we you have invested so much in the synthesis so far. We don't want to uh basically lose all all of the all of the work that we've invested in the synthesis so far.

[00:59:05]So we want to go with the most selective and most high yielding transformation and again we turn to biocatalysis and uh in this case it turns out that there are examples of other thioesterases that are standalone thysterases uh piece uh penicellin binding protein type thieststerases that in nature do similar transformations.

[00:59:32]So these were our um evolution starting points for for this and again we used the u cyamine as the uh surrogate for the pepidil carrier protein and um we uh were able to evolve the uh enzyme that selectively cycllizes even if there's any epimerization it would selectively cycllize only the desired epimer at either the amino or the uh ASIL end and we were able to evolve it for the isopropylster which as I mentioned already has optimum stability and um uh ease of preparation and we didn't really see any detectable after evolution esester hydraysis or ligomeration and we're able to run this reaction at a 10 volume concentration which is 100 grams per liter which is uh quite unusual for macroycization s of this complexity.

[01:00:34]Um yeah, so overall uh what's left was basically how do we isolate the final product out of the mixture and for small molecule we would um normally always crystallize them out for purification. For large molecules usually they're purified by chromatography. For these cyclic peptides I mean there aren't really that many of them on the market. So we already had used chromatography on the 1 kilo preparation and basically what we found out that chromatography was more than onethird of the cost time and uh also basically it's not scalable because they're very specialized uh uh plants that can do uh chromat chromatographic purification. And so so it's very difficult to scale up on for the commercial volume that we would expect for this product. So we also wanted to crystallize the this final microcyclic peptide. And here uh since this is the final active pharmaceutical intergian there there are a lot of uh criteria that one needs to uh meet in order to make the um uh make a best uh uh drug possible. And uh what what complicated in our case this u crystal form search is that the molecules a quarter ammonium salt. So so it's not easy to exchange these annions for for the other peptide intermediates. These were both uh amines actually diamines. So we could just make a free base and then add the acid of your choice and we make crystallin salts. And um this is actually another advantage of using biocatalysts for these peptide bond formations is that since we're not using many protecting groups for nitrogens, we have a lot of these free amines. So we we have a larger choice of finding um crystallin salts than if the molecule was neutral.

[01:02:48]uh and also these enzyatic transformations can tolerate a larger variety of let's say caroxyic acids as the counter ions as opposed to chemical coupling methods where they would couple so so there so this almost this biocatalyic strategy helps the isolation as well and for the final API we decided that um we want we didn't want to use ion exchange chromatography because it's unpractical on large scale. So we uh decided to basically do a simple extraction where we um use potassium bicarbonate solution and we can actually get phase separation with set nitral to the salting out. And then uh due to the differences in partitioning and also the um uh concentrations of the ann ions we we can get the uh enlistide bicarbonate salt in the cetonitra layer and then we just can take that organic layer and add the uh acid of our choice and then it's basically traceless salt formation because we generate CO2 and that way we're able to form uh um multiple crystal in salts and able to select the best one and this process clearly is a major uh uh costsaving over let's say uh ion exchange or chromatographic purification.

[01:04:20]So in the end uh we basically have demonstrated this on uh again mult multiund kilo scale and u no protecting groups and product isolated through crystallization and uh if you look at overall synthesis um it's it's really a chematic synthesis it's it's the fusion of the best of the two worlds. So the chemical synthesis provides us uh flexibility with manufacturing and portability and uh biocatalysis uh provides us with unparallel selectivity and avoids and basically lets us avoid um protecting groups. So it really works extremely well in combination and uh why would one need uh to have this highly convergent synthesis? It's basically uh illustrated by the process mass intensity. So for the first kilo uh scale preparations where we used chromatography the PMI was in the order of 60,000 just by crystallizing the API and also optimizing a few of the steps we are able to reduce the uh PMI 10fold and combining this with biocatalysis which enables much higher conver uh and also avoids protecting group steps protect protection deproction steps we are able to reduce uh the PMI 100fold and um sustainability is not just a uh idealistic quest there's a dollar value attached to every liter or kilo of solvent reagent that we use and also factory equipment so this definitely translates into commercial benefit as Well, and in our case, this is really what allowed commercial viability of this product. We wouldn't be able to really manufacture it using the existing chemistry at the time. So, uh overall, I think this is a good example how um innovation enables both sustainability and commercial supply of a complex molecule. And uh clearly uh uh this this is the most complex uh project I worked on in my career. And there are a lot of people involved who made uh enormously important contributions to this project.

[01:06:59]I would uh like to thank all of them for their work and thank you for attention.

[01:07:08]>> All right. Thank you so much artists.

[01:07:10]really great talk and really complicated molecule. Um there's a couple of questions in the chat already, so I think we'll start there.

[01:07:19]Um the first one is from Dave Martin.

[01:07:22]When the desired substrates give 0% yield for the first 10 to 20 rounds of evolution, how do you know you are going in the right direction? How do you choose from one of many potential paths you could take in the optimization since you don't know what the final outcome will be for route selection?

[01:07:41]Yeah. So I I guess there are two questions. So for the first one we we cannot really select if there's we don't see anything even by we use mass spectrometry to pick up very low levels of product formation. But if you cannot pick up the product even by mass spectrometry then we cannot really evolve. So that's the reason why we use the substrate walk where we use basically we evolve for a substrate that's I would say halfway between the natural substrate and the um and the uh desired one. So that that's kind of how we um work around that.

[01:08:22]And sorry what was the second question?

[01:08:26]uh how do you choose from one of the many potential paths you could take in optimization since you don't know what the final outcome will be. So with respect to route selection, >> yeah, there's a lot of risk obviously there. So um maybe maybe we were lucky here as well or maybe we just I think we had extremely good people working on the project. So um we do make prioritizations and uh usually by the time we select the route we already have proof of concept for some of these uh enzyatic transformations. So let's say we don't have 100% yield 99% selectivity but we have some activity detectable activity on the desired substrate and that is when we say okay we will bet our on ourselves that we'll be able to protein engineers to evolve the enzyme obviously there's experience working with in the past with protein engineering and there's some kind of feeling what what's realistic what's not I'm not an expert in that so I won't be able to provide details but but yeah so so we do look at other routes as well the the route that eventually was selected is very close to one of the two main original designs so it wasn't like we had 100 different routes from the beginning so this is one of the two main uh approaches that we originally thought and then obviously as we get these uh initial hits with the enzymes we basically lock in the route uh in about a year or so.

[01:10:02]>> Thank you. Uh the next question is from Alex Argaru.

[01:10:07]Uh he says, "Great work and thank you for your talk. Why not screen for antibbody binders that can be expressed in E.coli and isolate those instead of undertaking such an extensive synthetic route."

[01:10:22]Um, so what I'm I'm I would really need a clarification for that. Is it um for make synthesizing the molecule or or designing the drug? I'm just curious what this question refers to.

[01:10:47]>> Yeah, Alex, if you're online and you can clarify. To me it kind of sounds like discovering the molecule.

[01:10:54]>> I see. Yeah. Yeah. Yeah. Yeah. So this this goes back to the um uh again I'm not a discovery chemist but this goes back to the oral bailability. So we want to have a relatively small molecule that uh has uh basically uh enough permeability GI permeability that we could uh administer orally. So that's the main reason for you for trying to use a smaller molecule and in this case we need to use a relatively large small molecule just because as I mentioned this is a difficult I guess they used to be called undruggables nowadays nothing's undruggable anymore uh undruggable because because they're I guess largely flat surface again I'm not a discovery chemist so so you don't you need a relatively large um molecule to cover to to have a specific interaction with sufficient um activity.

[01:11:54]>> Thank you. And then I think Alex had a follow-up question and he was just wondering uh whether or not it was expensive to use s such large amounts of ATP even with the recycling systems.

[01:12:07]>> No. So actually yeah. So so so this really unfortunate I don't don't have much time to talk about all the details.

[01:12:15]I could talk the whole day about this project. So we actually used a monofaspate AM and AM is much cheaper and you can find suppliers for AM uh that are very reasonable prices and our loading is very low. It's like percentage level and uh this kynise can recycle uh AM to ADP and ATP. So so basically all all three of those are competent substrates for that. So basically we use polyphosphate. Polyphosphate is cheap.

[01:12:48]Polyphosphate is basically um food additive. I guess things like ice cream have it.

[01:12:56]Awesome. Thank you artists. So uh one more time let's thank our speakers artists and Tom for giving us such great talks today about such complex molecules. Um and of course thank you to uh my co-host Alex. um everyone else on the committee um the ACS ACS doc Emory uh beyond CCHF and Pharma Block and of course to the audience members for attending and listening today. So see you all in July. Have a great day everyone.

[01:13:45]Nice job everybody. That was great.

[01:13:48]>> Oh shoot. Sorry Scott. I just saw >> No, no, no. Don't worry about it. You You were over time. Um artist, I have a I have a question. Um >> what when you talk about evolution, is it directed or random? And if it's directed, like how do you figure out what to change in the DNA sequence you want to?

[01:14:07]>> Yeah. So those are unfortunately things I a I'm not an expert in that and b I I won't be able to provide the details anyways. So we we do plan to publish this soonish.

[01:14:23]So hopefully there will be more more details there. But I mean we don't use anything extraordinary. It's basically standard um >> what everybody else is using >> more or less.

[01:14:33]>> Again I'm not an expert. So >> I one of my colleagues who was watching on his own computer just came running over to say wow. So great job.

[01:14:42]>> Yeah. Thanks again. I'm not a protein engineer so >> Oh, it's okay. No, me neither. I'm not even a process chemist.

[01:14:54]>> Right. Anyone else have any other questions while we have the two speakers still?

[01:15:00]Oh, >> all right. Well, then thank you again so much. Um, I think this is a really great seminar. We had like over a hundred participants sustained throughout the entire you know duration of the virtual symposia. So really happy about that.

[01:15:21]Clearly some you know >> we had 158 at one point.

[01:15:28]Amazing.

[01:15:31]Clearly a testament of, you know, the topics today of the speakers today. So, thank you again, >> Steve. You and I should have quit before.

[01:15:41]>> Nice job, Alexis. Thanks, Er, and uh and um Tom, great job.

[01:15:46]>> Thanks.

[01:15:47]>> Thank you.

[01:15:48]>> This was great. Thank you again. Nice to meet you both.

[01:15:52]>> Same here.

[01:15:53]>> You as well. Thanks again for organizing.

[01:15:56]and Allan, great job.

[01:15:58]>> Yes, Alan, thank you.

[01:16:00]>> Sorry. No, sorry. There was some issue with the YouTube at the beginning. My computer was just doing that and that and that and then I restarted YouTube and then it seemed to work. So, I don't know what was going on.

[01:16:12]>> Yeah, we ended up with about 20 or so there on top of the 160 we had here.

[01:16:17]Nice job. Yeah.

[01:16:19]>> Right. See y'all.

[01:16:21]>> Thank you. Bye.