I want to talk about some unexpected and counterintuitive findings that run against what we might anticipate as chemists: that the introduction of a highly electronegative fluorine atom into a molecule necessarily increases the H-bond properties of adjacent functional groups. I refer to the work by Linclau and co-workers, which was published in Angewandte a couple of years ago. The cyclohexane-bound hydroxyl group was investigated in this study. The focus was on H-bond property of the OH functionality, which was measured using a really nice method: by looking at the decrease in absorbance of the hydroxyl substituent’s IR stretching band upon complexation with N-methylpyrrolidinone (NMP). Very cool stuff. Importantly, the paper casts doubt on the assumption that fluorination always increases H-bond acidity. There is a simple explanation of the observed effect – just take a look at the two representative examples shown below.
The main lesson here is that fluorination can attenuate alcohol H-bond acidity in unanticipated ways. The intramolecular F···HO interaction can in fact be responsible for a decrease in H-bond acidity. Clearly, this intramolecular interaction can effectively outcompete the electron-withdrawing effect of fluorine which is expected to lead to increased acidity of the adjacent polar groups. The Linclau study opens doors for rational modification of acidity through site-selective fluorination and should have many applications in the design of bioactive molecules.
The track record of man-made enzyme mimetics is not impressive. I previously commented that, time and again, we get beaten by nature’s enzymes in terms of catalytic efficiency. If we consider hydrolases, it is all about the spatiotemporal relationship between their catalytic amino acid residues and the peptide bond that is being cleaved. Things happen at the right place and at the right time. There have been many attempts to prepare synthetic hydrolase mimetics that were claimed to be close to their natural congeners in terms of catalytic efficiency. None of these examples have withstood the test of time: each was shown to be the product of sample contamination with some hydrolase. However, there is one really excellent example from the past that employs spatiotemporal arguments and proves that, once proximity has been secured, amazing efficiency for activating strong bonds can be achieved. A classic JACS paper by Fred Menger from 1988 describes a very curious amide that hydrolyzes fast at neutral pH with a t1/2 of 8 minutes, which corresponds to an effective molarity of at least 10l2M. Given the strength of the prolyl amide linkage, this classic example is remarkable.
Is there an immediate use for something like this? It is difficult to think of an application, but there is a lesson here to those of us who are driven by small peptides that fold into compact shapes. First and foremost, I refer to cyclic peptides and peptidomimetics. The question about cyclic peptide stability has been raised time and again, yet one often assumes that it is all about stability against proteolysis. The proteolytic degradation is difficult for a cyclic peptide (to the extent that it is not even worth pointing out this feature as a cool attribute of cyclic peptides – it is obvious that they do not adopt the necessary extended conformations. Here is a great reference that discusses this property, by the way: http://pubs.acs.org/doi/pdf/10.1021/cr040669e). But chemical stability is another issue and I would predict that there are cases where amide bonds in complex macrocycles are unstable for reasons that operate in the case described by Menger 25 years ago.
People often equate the presence of fluorine in organic compounds with increased stability and other unique properties. The trifluoromethyl group is sometimes erroneously equated with some “magical” stabilization. Honestly, I think we tend recall Teflon way too much. Tonight I want to talk about two fluorinated groups: one is well-known (CF3), while the other one is rather obscure (SF5). The reality is that the trifluoromethyl (CF3) substituent is not that stable (I did my whole PhD on it…). For instance, in some cases, you can quantitatively hydrolyze it under aqueous alkaline conditions into the corresponding carboxylate. But this is only part of today’s story. In terms of stability, I want to point to a lesser-known group – pentafluorosulfanyl (SF5) – which is poised to become a very interesting moiety with which to design materials and pharmaceuticals. At first glance, this thing just does not look right as there seem to be way too many fluorine atoms. But it is what it is: sulfur in the oxidation state “+6”. The hydrolytic stability of SF5-containing molecules makes trifluoromethylated analogs pale in comparison. Take a look at the reactions shown below! In addition, the van der Waals volume occupied by the SF5 group is slightly larger than that of the CF3 substituent. The SF5 group is also more lipophilic. This property, along with improved stability, ought to make SF5–containing compounds more and more interesting in drug design. In fact, there are examples of biologically active molecules that derive their properties from the unique characteristics of the SF5 substituent. There will be more on that in my future posts.
Some of you might wonder what this “Bro5” acronym stands for… Bro5, or “Beyond the rule of 5“, is meant to represent a vast chunk of chemistry space that is heavily populated by molecules that behave as drugs, yet do not satisfy the so-called Lipinski rule of 5 for drug-like properties. Here are the famous commandments from Lipinski:
As it typically happens, exceptions prove the rule. We don’t have to go very far: consider the case of taxol. This outlier is perfectly bioavailable, yet violates some of the Lipinski rules. The post-Lipinski period is characterized by trying to rationalize the behaviour of other outliers that emerge from drug discovery campaigns. More importantly, can we predict fairly large molecules with favourable oral bioavailability? When it comes to my lab’s research, the Bro5 considerations hit close to home as peptide macrocycles are among our favourite targets. The following paper, published several years ago by a team from Pfizer in the UK, contains a very nice discussion of Bro5. This manuscript also proves that it should be possible to rationally design orally bioavailable cyclic peptides. As you have seen in my older posts, formation of intramolecular hydrogen bonds is the main driver behind oral bioavailability of cyclic peptides. These weak interactions cooperatively shield polar functional groups and facilitate membrane permeability and intestinal absorption. One tantalizing question is whether or not it is possible to run simple predictions of intramolecular hydrogen bonds through modeling of the low energy gas phase macrocycle conformations. I have always thought that some of the coveted folded conformations of cyclic peptides might be fully predictable using such simple tools. The Pfizer work in MedChemComm supports this notion. So… Why painstakingly make a ton of macrocycles and screen them? Let’s make just one, baby…
In terms of basic science, control over cis– vs trans- amide bond geometry is one of the ongoing research areas pursued by my lab. We think that this problem is important for many reasons that range from fundamental physical organic chemistry to one’s ability to dictate conformations in complex polyamide macrocycles. I have already blogged about some elements of amide cis / trans interconversion. Recently, my lab has uncovered an interesting case that points to the possibility of kinetic selection between these rotamers. On the heels of our findings, I started to think about the smallest possible ring where a clear-cut cis/trans interconversion can be observed. Below is an old and very thought-provoking paper by North and Zagotto. Apparently, amide geometry in the 8-membered ring that you see is determined by the relative configuration of the two chiral centers. Strikingly, the two cyclic diastereoisomers have different preferences for the amide bond geometry. It is highly unlikely that the cis-amide (case B) is present in the starting linear dithiol before oxidative cyclization. What likely happens in the case B is a thermodynamically controlled cyclization that involves product isomerization into a more stable (NB: in this particular instance!) cis-amide. I will leave it up to you to wonder why the cis-amide is preferred in the cyclic diastereoisomer corresponding to B. I do think that disulfide’s flexibility might be playing a role in allowing the final isomerization to take place. To my knowledge, the 8-membered ring shown below is the smallest cycle that shows such interesting cis/trans amide behavior. If you now of a smaller system, please let me know.
I am virtually certain that, as I type this post, there are reactions involving sodium azide being run in my lab. We use this versatile reagent to generate organic azides that are subsequently converted into a gamut of nitrogen-containing building blocks. We are obviously not the only ones who rely on this silver bullet of a nucleophile (to paraphrase my mentor, Barry Sharpless). In fact, the vast majority of synthetic and biological chemists probably use sodium azide in order to make organic azides and later run click reactions with alkynes. While my lab does not run these cycloaddition processes, we like to reduce our azides to amines. In the course of this reaction we lose two nitrogen atoms off an azide and gain two protons. What if we want to “chop off” only one nitrogen atom? I was thinking about a reaction that would correspond to such a transformation earlier today and recalled to mind a really cool process reported by Professor Ron Raines of the University of Wisconsin several years ago. In it, an organic azide interacts with a carefully designed phosphine reagent shown below. The leaving group attached to the carbonyl group enables intramolecular formation of a five-membered intermediate that collapses to release the diazo product. The authors refer to this reaction as “deimidogenation”. Azides are indeed very versatile and multifaceted intermediates. I think we should put this reaction to good use…
As I prepare my lectures for the second year organic chemistry class, I can’t help but wonder about a good way of teaching Lewis acid/base chemistry. When our students learn about the likes of AlBr3 for the first time, they get used to the idea that Lewis acids are, by their very nature, water-sensitive compounds. There is no doubt that the vast majority of “traditional” Lewis acids do not respond well even to trace amounts of water. In fact, the notion of water-tolerant Lewis acids would have been a heresy even 20 years ago. But things have changed. When I talk about the “next-gen” Lewis acids, I always point to the classic work of Kobayashi, particularly to his insightful JACS paper from 1998. In it, Kobayashi has taught us all a valuable lesson in achieving the balance of thermodynamic and kinetic factors in regards to Lewis acidity.
If the key parameters (the so-called water exchange rate constant (WERC) and hydrolysis constant) are chosen properly, a metal-based salt could turn into an excellent Lewis acid even in the presence of water. What takes place between a given metal salt and water is substitution of inner-sphere water ligands. To have a Lewis acid catalyst in water, all you need to do is tone down your metal’s affinity to water, while keeping water exchange rate constant high. If the stars are aligned properly, the resulting compound will have a chance to be a water-tolerant Lewis acid. It makes perfect sense: in an aqueous environment, metal salts would undergo exchange reactions of their water ligands. If the substrate you want to activate exists in the system, it can coordinate to the metal cation instead of the water molecule, resulting in Lewis acid activation. Rational, parameter-driven choice of reagents is relatively rare in synthesis, which makes this classic Kobayashi’s paper particularly important.