In our fragment-based drug discovery projects, we constantly think about new ways of stitching together heterocycles. It is particularly satisfying if we find a way to make completely new, previously unexplored, chemistry matter. As you might imagine, herein lies a dilemma. On the one hand, we want high ligand efficiency, which goes implies relatively small size. On the other hand, try to plug something completely new (for instance, a 6-membered ring) that is composed of common heteroatoms into SciFinder. Good luck with that. Chances are, your composition of matter has been disclosed in a patent or, worse yet, published in peer-reviewed literature. Challenges notwithstanding, we recently bravely embarked on a journey that has sought the discovery and development of new kinds of small and medium rings. One of our favorite examples is the so-called boromorpholinone, which is obtained by plugging a boron atom into the framework of a well-familiar morpholinone scaffold. You see a model of a representative member of this class of compounds in the graphic below (boron is green). Earlier this year, Aleksandra Holownia (pictured below) joined our efforts as a Summer undergraduate student. Under the direction of my graduate student Adam Zajdlik, Aleksandra not only helped us understand the boromorpholinone area better, but earlier today won one of the AstraZeneca poster awards. I applaud her dedication to this project. As you might imagine, the boron atom in these sorts of rings is not going to be an innocent bystander. There are, in fact, many applications that await these compounds, which is something we are engaged in together with Ben Cravatt at Scripps. Speaking of Adam, he just came back from a one month internship at Anacor in California, where he further learned about the use of boron. Anacor is the pioneer in the deployment of boron-containing molecules in drug design. There are myriad reasons to like boron and I will give you just one: boric acid, the main metabolite of organoboron therapeutics, has LD50 of 2.7 g/kg. This toxicity is comparable to that of table salt. Once again, congratulations, Aleksandra!
In tonight’s post I will attempt to bridge hydrophobic interactions and life as we know it. It might get philosophical, so bear with me. As many of you probably know, not long ago, Romesberg and colleagues at Scripps created the first organism that can grow and replicate with a completely unnatural base pair in its DNA. The DNA of this organism can, in principle, code for up to 172 amino acids. On a molecular level, the “glue” that holds the novel base pair together is purely hydrophobic in nature (the pair comprises a substituted methylisoquinoline and a methoxynaphthalene). There are no hydrogen bonds there at all:
The most interesting part of this research is how the authors managed to get their cells to replicate. In order to pull this off, they needed to figure out a way to smuggle the new bases inside the E. coli bacteria. The team found a unique protein transporter that was able to specifically take up the aforementioned synthetic bases. As a result, the new base pair was incorporated into DNA and was later found in replicated plasmids.
While there are many hurdles that prevent this system from being truly efficient, there are as many ethical questions that can be posed here. Apparently, the Romesberg study has already resulted in renewed calls to halt research in synthetic biology (for instance, by the ETC group: http://www.etcgroup.org). I am not an expert in these kinds of debates and it is difficult to speculate on the dangers of this science at this point. But I do want to note something else that might be interesting. Without going into a theological discourse, I note that anti-religion zealots such as Richard Dawkins are (ironically) not that far from some of the key principles they try to attack at all costs. I refer to how Dawkins looks back in time to find our ancestors, assuming that evolution is progressive, culminating with us. He does it in his book The Ancestor’s Tale. The amusing feature of this logic is that he ends up with a chain of being that is very similar to what religion promotes, namely that man is the culmination of it all. There is an excellent piece published by Sean Nee in Nature close to 10 years ago (I urge you to read it) that discusses this central issue and presents an alternative view of evolution. This essay offers a conjecture. Nee reminds us that over the past 600 million years a great variety of Bacteria, Archaea and microbial Eukarya have been evolving. And (I quote): “One of the huge species, Homo sapiens, got remarkably self-important. But when, to his surprise, a virus wiped him out, most of life on Earth took no notice at all.”
The lesson here is that we tend to assume too much about our relative significance in the great chain of being. But how does this relate to synthetic biology? We have curious minds. What if we do indeed create a form of life that will prove Nee’s conjecture that we might have gotten somewhat self-important?
I was looking at an old paper collection that I have, trying to find methods that enable clean monomethylation of amines (we rely on this reaction in our macrocycle work) and came across a brilliant old paper by Grieco. I have always been very fond of this reaction. As you might imagine, many studies in organocatalysis have this 1987 report as one of their cornerstones. The question of amine monomethylation may sound simple, but it is anything but trivial. Low chemoselectivity is usually quoted as the main obstacle towards high yields in this chemistry. The trouble is typical of amine transformations: the product is more reactive than the starting material. Even reductive amination cannot offer a decent monomethylation solution, which is due to the fact that the corresponding imine is not easy to isolate and that formaldehyde is quite reactive. The workaround developed by Grieco involved in situ Diels-Alder chemistry that rapidly generates the azanorbornene skeleton shown below. Once this is accomplished, the adduct can be carried through various steps and can then be cleanly reversed to the iminium ion through retro-Diels-Alder reaction. In the presence of triethylsilane as the hydride source, the desired methylated amine is generated. Temporary masking of reactive functional groups using reversible processes is a really useful trick. We need to use this process more often in our own peptide synthesis efforts.
What is the best heterocycle out there and how do we find it? This is a dumb question because it is not clear what the purpose of the query is. Let’s say we are talking about drug discovery. I really doubt that there are yet-to-be-discovered “silver bullets” that will beat everything else from the standpoint of ligand efficiency (I blogged about this concept in the past). It is actually quite amusing to talk to biologists, who expect that our heterocycle-containing inhibitors can be magically turned into something that is way better (by several log units – this is what they want) by some simple tweak that should be obvious to us based on our years of training… Something like this might become possible if we discover a handful of new elements on Mars, bring them down here, and plug them into pyridine. But not before… If we consider drug discovery a bit further, though, there are reasons to rank heterocycles in terms of their metabolic liability. In fact, there are efforts aimed at comparing heterocycles from the standpoint of binding to human serum albumin, CYP450 inhibition, etc. One recent study carried out at GlaxoSmithKline suggests that pyridazine is one of the most “developability”-friendly heterocycles as it presents the fewest downstream issues. This is quite interesting and you can read the details of analysis here:
As luck would have it, I got some interesting “pyridazine” insights today. Earlier I had a phone conversation with Dr. Herdewijn of Galapagos in Belgium. I am not at liberty to disclose the subject of our conversation at this point, but I went online to read about some chemistry developed at Galapagos and was excited to come across a really cool pyridazine construction that came out of their labs. The sequence that you see below is a fantastic way to these scaffolds; it employs a rarely used process – the diaza-Wittig reaction (not to be confused with the aza-Wittig reaction, which is a very well established way to make imines).
First of all, you might say: “Forgotten by whom? Not by me!”… You might be right, I admit that it is rather presumptuous of me to come up with post titles boasting the words “forgotten”, “neglected”, etc. But I would still venture to say that, given the wealth of high tech machines a synthesis laboratory might have these days, the majority of us do not resort to the older, more obscure purification methods. Today I want to bring to life silver nitrate (SN) impregnated chromatography as a method par excellence in really tricky separations. Below you can see two pinene isomers that are rather difficult to separate using conventional silica gel chromatography, yet are differentiated by the delta Rf of 0.2 using SN chromatography. I am including an excellent review by Mander that goes into the details of this great method. Some of you wonder if this would work best on alkene-containing molecules and you are right (due to silver-alkene coordination), but the method is not limited to separating on this basis alone. In Mander’s review, you will notice some intriguing examples of heterocycles that have been successfully separated using SN chromatography. So let’s keep these old methods in our lore of separation tricks, ladies and gentlemen.
I visited Alphora Research about 10 days ago. Alphora is a company that specializes in organic synthesis and leverages their tremendous process research expertise in projects that involve production of pharmaceuticals (under GMP standards). I think this is a great place to practice one’s synthetic skills because pretty much everything that is remotely interesting in the pharmaceutical process research is nowadays outsourced from places such as Pfizer (who are busy doing acquisitions) to smaller companies such as Alphora. As a result, students who are trained in synthesis may get jobs in such smaller companies and practice what they are passionate about – making molecules. I was really impressed with the kinds of projects Alphora scientists get to work on. They really solve important problems. For a representative recent example I direct you to a paper they put out earlier this year in Tetrahedron Letters (see the link below). The target of synthesis here is Eisai’s Eribulin molecule, which is a truncated version of the natural product Halichondrin B. I am amazed that molecules of this complexity (there are 15 chiral centers here!) have reached the market. The innovation in Alphora’s approach was to introduce nitrogen early in synthesis, which is something you typically want to avoid as the risks outweigh the benefits: according to conventional wisdom, one is better off introducing nitrogen atoms towards the end of synthesis. But, under the leadership of Dr. Boris Gorin, Alphora scientists took the risk of the “early nitrogen game plan” and, as a result, reaped the benefits of dealing with crystalline intermediates along the way.
There was something else during this trip that attracted my attention. It was one of those Chemtrix instruments that perform flow synthesis. I have seen a ton of flow synthesis machines in recent times at various venues, but Chemtrix really caught my eye. My beef with flow synthesis has always been about what happens next. It is ok to make a few mg’s really well. Or a gram. But what if you need to make 100 kg’s or 100 tons? Apparently, this Chemtrix instrument is a result of a ton of work on behalf of a very dedicated team of engineering geeks who promise linearity in scale-up. In other words, they have figured out an appropriate scaling algorithm that enables one to take the results of flow synthesis on a very small scale and have close to a guarantee that it will produce the same yield on a multi-kg or a multi-ton scale. All you need are fancier pumps and larger surface area in your tubes and reactors (which they produce and supply). None of this is random, but is calculated with utmost precision taking into account flow dynamics, mixing times, turbulence, etc. I need to buy one of these instruments one day.
Recent efforts to chart novel classes of molecules beyond the co-called “rule of 5” space have consistently pointed at macrocycles as privileged scaffolds. A lot has been said about their conformation and capacity to hide hydrogen bonds. It is not easy to evaluate the significance of complex conformational ensembles, let alone extract useful rules that might have predictive power. I have been longing for a reductionist approach that could hopefully unambiguously demonstrate the effect of ONE hydrogen bond on lipophilicilty and other drug-like properties in a cyclic molecule. In fact, there is a recent study that does just that. In their J. Med. Chem. paper, Kihlberg and co-workers evaluate a diastereomeric series of T. cruzi growth inhibitors and showcase vastly different solubility, lipophilicity, pKa, and cell permeability for two sets of four stereoisomers. Intriguingly, all it takes is a switch in chirality of one of the stereocenters to improve the chances for intramolecular hydrogen bond and concomitant pKa difference. The authors carefully analyzed the conformations of their 8-membered rings by NMR, considered the differences in chemical shifts, their temperature dependencies, and obtained solid evidence pointing toward the formation of intramolecular hydrogen bond in the trans (not cis) diastereomer shown below. This is a very thought-provoking case as it highlights a fairly small perturbation that results in a substantial difference in properties. While I am impressed with the results, I keep reminding myself that obtaining this sort of data is not trivial and we are unlikely to see easy fixes as there are so many different scaffolds out there. The main reason there is a higher degree of difficulty in macrocycles compared to small molecules is that functional groups are much more interrelated through limited rotation (and the one you see isn’t even a macrocycle… it is a medium-sized model).