Going retro

Chemists like to coin new terms, which is particularly meaningful when there is a connection to the place of discovery of a given molecule or process. Graduate students are usually well aware of munchnones. These intriguing intermediates are named after Munich, where Professor Huisgen spent a good portion of his career and where he coined the term. Comparatively less known are sydnones, which differ from munchnones by the presence of nitrogen. As a matter of fact, sydnones predate munchnones. They were developed in Sydney by Earl and Mackney back in 1935. Below is an awesome paper by Harrity and co-workers I came across. It showcases the use of sydnones in heterocyclic synthesis. Image


While the paper is hardly new, I marvel at this way of introducing boron atoms into heterocycles. Cycloaddition/retro-cycloaddition does the trick here. By the way, there is a great review by Mike Willis that contains a lot of other, equally intriguing, cases that employ retro-cycloadditions. Here is a link to Mike’s review:



On AstraZeneca, macrocycles, and logD

Earlier today I was in Boston, where I gave a talk at AstraZeneca (in fact, the exact location was Waltham, Mass). My former student Tim Rasmusson is working there now, so it was great to see him again. Dr. Scott Cowen was my host and he was responsible for bringing me to their site. Throughout the day I met a lot of people and heard a very coherent message, which indicates that macrocycles are among the coolest commodities in today’s drug discovery. One of the reasons Scott brought me over had to do with an ongoing collaboration between AstraZeneca and Encycle Therapeutics (http://www.encycletherapeutics.com), the company I started together with MaRS Innovations in 2012. Encycle now has 8 employees and is in hot pursuit of several proprietary targets in addition to collaborations with AstraZeneca, Merck, GSK, and Pfizer. We will be updating our website soon, so stay tuned! I was encouraged by the warm reception our platform has received at AstraZeneca. This interest provides further impetus for trying hard to reach our milestones. I want to stress that we are working together with an awesome team led by Professor Eric Marsault of the University of Sherbrooke. Together, we are amassing an impressive collection of macrocycles that are being evaluated using a wide range of techniques.

I don’t intend to talk about any of Encycle’s accomplishments today, although I can say a lot in praise of the hard work of our team. This will be mentioned in detail once we make some corporate announcements (I hope soon). Instead, I want to discuss an issue that continues to present challenges in the general area of peptide macrocycles. These molecules are substantially larger than their small molecule counterparts. Macrocycles have high polar surface areas due to the presence of multiple amide bonds. This makes it difficult for them to traverse cellular membranes, which is something I noted in the past when I said that control over intramolecular hydrogen bonds is key to reducing the polar surface area of these molecules. logD, or the logarithm of the ratio of the concentration of a given molecule in octanol to the concentration of it in water at a given pH, is one of the metrics that is widely used to assess lipophilicity. logD can be measured experimentally and can be reliably computed, with the caveat that there are big discrepancies between experiment and computation when it comes to molecules that can form intramolecular hydrogen bonds. Incidentally, this problem affects even fairly simple compounds. While visiting AstraZeneca, I was reminded that the calculated logD of the following two molecules is identical, although the experimental values are vastly different due to the presence of an intramolecular hydrogen bond in the first case. Image

If folks are having trouble with such simple stuff, just think about the nightmares you’ll have when it comes to giant rings with rich conformational space. We really need to improve computational approaches in order to enumerate virtual libraries of macrocycles. I guess this is today’s message. Once again, today was a great one-day visit to Boston, so thanks to Tim and Scott (I loved that Venezuelan restaurant!).

A hidden side of azides

I just saw a cool Angewandte paper from Professor Chiba’s lab at the Nanyang Technological University in Singapore. This manuscript serves as a reminder to all of us: think about the role of resonance structures on reactivity! The authors were able to extract some impressive benefits from the enamine character of the vinyl azide functional group. Vinyl azides can be generated using the method developed by Professor Alfred Hassner (I met Alfred in the 90’s when he was visiting Olah’s lab… He is a class act!). Hassner’s method calls for in situ generation of IN3, which is electrophilic enough to react with olefins. Upon base-induced elimination of HI, vinyl azides can be isolated in good yields. Chiba and colleagues found that vinyl azides react with a range of electrophiles to generate iminodiazonium ions. Following a Schmidt- type rearrangement, nitrilium ion formation ensues. The nitrilium ion is rapidly hydrolyzed in the presence of water to give the coveted amide linkage. There are a number of finer details here, such as the use of hexafluoroisopropanol to improve isolated yields, temperature optimization, etc. Needless to say, what’s shown below is an unsymmetrical substrate where selectivity towards migration is due to the presence of an aryl group. Not surprisingly, two alkyl substituents around the C=N bond would lead to comparable migratory aptitude of the two substituents (and lower selectivity). Regardless, this is a very clever method that builds on the fundamentals and one’s ability to employ what others might have seen, but have not put to good use.



To bind some place else

Let’s say you have an enzyme. It has an active site, a known substrate, and all that jazz. If you know the activity assay for this enzyme, you can design experiments aimed at inhibitors by measuring substrate turnover. While plenty of tools for targeting competitive active site inhibitors (both covalent and non-covalent) exist, the situation is much bleaker for the so-called allosteric inhibitors. These are in a league of their own as they bind some place other than the active site. Identifying allosteric sites and designing inhibitors that bind there is an interesting undertaking. Earlier today I attended a talk here at the University of Toronto by Wolfgang Jahnke of Novartis, Switzerland. The gist of one of his projects is described in the JACS paper I want to talk about tonight. In it, Jahnke and colleagues interrogated the allosteric site of Bcr-Abl kinase. Myristic acid is the natural ligand for this site (see below – I have made these images using PyMol). As it turns out, there are two possible conformations. When myristic acid binds, one of the helices gets bent (shown). Without myristic acid, the helix is unstructured. The authors designed a clever NMR assay: they first prepared the kinase labeled with 15N at Val525 (see the pinkish “V” underneath the myristic acid molecule?). When myristic acid is bound to the enzyme, Val525 is structured and gives a weak signal in the 15N/1H HSQC spectrum, whereas the unbound state is characterized by a strong Val525 signal. With this in hand, the Novartis team developed an assay in which they added a known inhibitor first (PD166326 was one of them, identified as “inhibitor” in the graphic below). The role of this molecule is to bind in a canonical fashion (at the ATP site) and stabilize the inactive kinase conformation. Subsequently, the authors screened small molecule fragments, monitoring Val525 HSQC peak. The “hits” were molecules that weakened the Val signal. This is a beautiful experiment that enabled the authors to discover fragments that acted in concert with PD166326 (one of them is shown on the right hand side of the graphic below). Given the fact that point mutation is the main mechanism of resistance to ATP-competitive kinase inhibitors, you can see the benefits of this approach to overcome resistance against current Bcr-Abl inhibitors. Administered in concert, cocktails might represent a compelling therapeutic approach to battle cancer. I loved the fundamental science that has gone into assay development here. Incidentally, one of these combinations apparently goes into phase II next week. Good luck, Novartis!



B-O bond activation

I enjoyed reading Suzanne Blum’s full paper on intramolecular alkoxyboration of alkynes in JACS. The reaction is quite remarkable and amounts to addition of B-O bonds to alkynes under gold catalysis. The authors quote the fact that B-O bond is significantly stronger than B-C bond, which is one of the reasons why there have been comparative lack of success in the area B-O bond activation. I am showing a brief part of the overall process reported by Blum (the [B] fragment undergoes trifluoroacetate-assisted departure and is reincarnated as the electrophile in the next step): Image


Au(I) turned out to be the optimal catalyst which, in its typical “alkynophilic” fashion, renders the triple bond susceptible to the attack by the nucleophilic phenoxide oxygen. It is at this point that boron electrophile is generated, which is the most intriguing part of the paper. Due to my lab’s recent interest in the effect of MIDA group on boron transfer reactions, the nature of boron electrophile here certainly piqued my interest. At this point, it is a bit difficult for me to envision B[MIDA] trifluoroacetate (that would be an analog of compound 16) as an effective electrophile. There are some thought-provoking lessons here, in my view.

Trouble with elimination

I am at the ACS National meeting in Dallas today, where I took part in the Symposium on Macrocycles. We had a good line-up of speakers, including W. van der Donk, S. Lokey, Allyn Londregan, Keith James, and myself. David Price of Pfizer organized this event, which was a great success (thanks David!). I am writing this post while waiting for David and Scott to go to a steakhouse.

Our event took place in the afternoon, but I woke up bright and early and went to the total synthesis symposium, where lectures were presented by graduate students. I always like to hear these kinds of talks because you get to see what actually happened. Today was no exception and there was a very well selected line-up. The talks were both informative and well presented, despite the fact that only a few people were in attendance this early in the morning (perhaps folks had too many margaritas the night before?). In every synthesis endeavour, there comes a point that defines one’s problem-solving abilities. I am making a subfield-agnostic statement, but such defining points are perhaps best illustrated by way of total synthesis. I was intrigued by Dennis Wright’s synthesis of frondosin A, presented by his student Michael VanHeyst. The paper describing their approach recently appeared in JACS and contains a number of clever mechanistic twists. I will comment on just one of them, as it goes well with the theme of today, which is about problem-solving. A case in point is the attempted cleavage of the oxabicycle shown below. The failed attempt illustrates the capricious nature of the 7-membered ring, which leads to complete lack of selectivity during elimination. After months of trying, a reasonable solution came in the form of a phosphine-mediated rearrangement that “relayed” bonds in a super clever way that culminated in elimination of phosphine oxide. This paper also features a fascinating explanation of stereochemistry inversion that has put to rest some previous controversies.





The title of today’s post is reminiscent of Stallone’s old movie, but I want to talk about cliffhanging of a different kind. A recent paper in J. Med. Chem. by Bajorath is a follow-up to his previous review in that journal, covering the so-called activity cliffs in drug discovery. This concept relates to pairs of structurally similar compounds with a significant difference in potency. There are many kinds of activity cliffs (hydrogen bond, stereochemical, lipophilic), and they come from many different target classes. Here is a typical example:


As you can imagine, some target classes are notorious for their association with scaffolds that lead to cliffs. The presence of more than one activity cliff in a given series might complicate compound optimization and turn it into a complete nightmare. This is particularly true at later stages when different molecular properties beyond potency must be optimized. Paradoxically, although activity cliffs are associated with structure/activity information, the corresponding knowledge might not be readily exploited when a given compound needs to be optimized. I think that activity cliffs really challenge the capabilities of computational approaches because the latter rarely provide good answers as far as cliffs is concerned. In broader terms, we should apply this concept in other disciplines such as catalysis. I don’t think we emphasize the cliffhanging culprits enough when we study the structure/activity relationships in ligand design.