From noncovalent to covalent fragment for BTK

The approved anti-cancer drug ibrutinib is a poster child for covalent modifiers, with projected 2018 sales of more than $1.2 billion. The molecule reacts with a cysteine residue in the kinase BTK as well as several other kinases, forming an irreversible bond. However, it is also a potent noncovalent inhibitor of multiple kinases, leading to various side effects. This is because the so-called “hinge-binding” moiety is quite promiscuous. To find a more selective and potentially safer molecule, researchers at EMD Serono, Constellation Pharmaceuticals, and Hoffmann-La Roche turned to fragments, and describe their results in two recent Bioorg. Med. Chem. Lett. papers.

In the first, Richard Caldwell and collaborators disclose Fragment A. Although no details are provided as to how this was discovered, the researchers were able to determine a crystal structure of the fragment bound to BTK, revealing that the carboxamide forms interactions with the hinge region. The binding mode also suggested how an acrylamide warhead could be positioned to react with the nearby cysteine residue, and indeed compound A7 turned out to be a nanomolar inhibitor. Further fragment growing ultimately led to compound A17, with subnanomolar biochemical activity and nanomolar cell activity. Unfortunately, this compound had poor oral bioavailability in mice.

The second paper, by Hui Qiu and collaborators, picks up the story. Hypothesizing that the number of nitrogen atoms in compound A17 could be deleterious, the researchers swapped the added portion of the molecule with the phenoxyphenyl moiety present in ibrutinib. Compound B7 did indeed show good permeability, albeit at a cost in potency. However, a crystal structure of this molecule bound to BTK revealed the potential for improving hydrophobic interactions.

The best molecule reported, B16, has picomolar activity in a biochemical assay and nanomolar activity in cells. Moreover, it is orally bioavailable in rats. While ibrutinib inhibits 35/270 kinases at 1 µM, compound B16 only inhibits 4. However, the compound does inhibit hERG, which can cause cardiac complications, so more work needs to be done. A crystal structure reveals that B16 (gray) binds in a similar manner to the initial Fragment A (cyan).

We have previously described examples of covalent fragments being used to target kinases, but these new papers are a useful reminder that it is also possible to start with ordinary non-covalent fragments and introduce a warhead later. Or not – as we highlighted in 2015 regarding a noncovalent BTK inhibitor from Takeda. The possibilities are limited only by your creativity.

Fragment chemistry roundup

Our current poll (please vote on the right-hand side of the page!) asks about commercial fragment libraries. However, there is much to be said about making your own fragments: you have total control over the quality, and it is easier to get into novel chemical space. We highlighted one example of custom fragments in 2016; here are a few more. Please feel free to share others in the comments.

James Bull and collaborators at Imperial College London and Eli Lilly describe a nice divergent approach to cyclopropane-containing compounds in a Eur. J. Org. Chem. paper published last year. As they note, cyclopropane is the tenth most common ring found in small-molecule drugs: common enough to be validated, but sufficiently rare as to quickly yield novel compounds. They start with an efficient cobalt-catalyzed cyclopropanation that can be conducted on gram-scale to produce scaffolds A1 and A2. The sulfur atoms can be oxidized to sulfoxides or sulfones, and the esters can be converted to amides. Perhaps more interestingly, the sulfoxides can be converted to Grignard reagents that can be reacted with electrophiles or used in cross coupling reactions, ultimately generating diverse molecules such as A24 and A25.

The researchers also calculated various parameters of the 56 molecules they synthesized, and found that many of them were quite “shapely” as assessed by calculating principal moments of inertia.

A more recent paper, by Adam Nelson and colleagues at the University of Leeds and published in Bioorg. Med. Chem., also looked at more “three-dimensional” fragments based on bridged bicyclic lactams found in certain alkaloids. Intermediates such as compound B11 could be rapidly assembled and diversified at multiple points to generate very different molecules such as B17b and B19. All in all, 22 fragments with < 17 non-hydrogen atoms and clogP < 2.5 were generated.

Finally, Nicola Luise and Paul Wyatt at the University of Dundee describe the synthesis of semi-saturated bicyclic pyrazoles in Chem. Eur. J. As the researchers point out, at least five fragment-derived drugs that have gone into the clinic contain pyrazole moieties, and 4-bromopyrazole seems to be a universal fragment. Although many pyrazoles are commercially available, the number drops considerably with partially aliphatic bicycles, and these may also have improved physicochemical properties.

As we noted in 2016, Paul Wyatt is having students build libraries of novel fragments. Given the range of chemistries explored in this paper, this seems like good training. Starting from 3-bromopyrazole, the researchers generated 25 different molecules, all conforming roughly to the rule of three, and also characterized whether they met criteria for purity, stability, and solubility at 2 mM in phosphate buffer. Only a single molecule dropped out, supporting the design criteria, and the molecules have been added to the Dundee fragment library.

Papers like these do not get as much attention as they deserve, in part because the biological properties of new molecules are by definition unknown. Still, it is refreshing to see chemists coming up with creative new classes of fragments. Hopefully we will revisit some of them as hits in future posts!

Library vendor poll!

A good fragment library is essential for generating good fragment hits. Earlier this year we asked about fragment size and library size, and summarized the results here.

Some folks make their own fragments, but this is expensive, and it’s probably fair to say that most fragment libraries contain a large fraction of commercially available compounds. Unfortunately, what you buy is not always what you get: sometimes the wrong compound is sent, or the compound is not as pure as advertised. As this post makes clear, different vendors have different track records.

Two years ago we highlighted a number of fragment library vendors. The current poll first asks whether you’ve bought fragments in the past several years. (Please answer this question as otherwise I have no idea how many people actually vote.)

If the answer to the first question is yes, the second question asks which vendors you would recommend buying from – presumably because you’ve had good experiences with them in the past few years. You can vote for as many as you’d like.

Finally, the third question asks which vendors you would avoid.

Please vote on the right-hand side of the page, and feel free to leave comments below, particularly if you've used vendors not on the list.

And on the subject of voting, if you are eligible to vote in the United States, please make sure to do so before the polls close on November 6.

Democracy atrophies when citizens don’t exercise their rights.