Halo Library says hallo to crystals

Last week we wrote about the magical properties chlorine can impart to molecules. More generally, halogen atoms can be helpful for a variety of reasons beyond new types of interactions with proteins and improved metabolic stability. For example, fluorine NMR can be used to rapidly identify ligands, and we’ve written about a custom fluorinated library. Heavier halogens can be particularly useful for screening by crystallography, and we’ve written about two libraries (HEFLib and FragLites) containing chlorine-, bromine,- or iodine-bearing fragments. Now Francesc Ruiz, Eddy Arnold, and colleagues at Rutgers bring us the “Halo Library,” described in a new (open access) J. Med. Chem. paper.

 

The researchers assembled a library of 46 halogenated fragments. In contrast to the libraries above, which focused on either fluorine or heavier halogens, this one is multi-purpose, with about half the compounds containing fluorine, half containing bromine, and a handful of molecules containing chlorine or iodine. Most of the fragments came from their internal collection and had been screened against several targets, and the rest were commercial compounds that had been reported to bind to at least one target.

 

The Halo Library is similar in terms of molecular properties to HEFLib and FragLites, though with a slightly lower average molecular weight (172.5 Da). Like the two earlier libraries, the Halo Library is relatively “flat,” with Fsp3 = 0.2.

 

Ten years ago we highlighted work from the Arnold lab in which a library of 775 fragments was screened crystallographically against HIV reverse transcriptase (HIV-1 RT), resulting in a 4% hit rate. The researchers returned to this protein with their Halo Library, soaking crystals with individual fragments at 20 mM. This resulted in 12 hits, an impressive hit rate of 26%. Admittedly some of these halogenated fragments had been identified as binders previously, so it will be interesting to see how the library behaves on other targets.

 

In addition to the high hit rate, fragments bound to six sites not occupied by ligands in the 2013 study, and two of these sites had never been reported to be ligand binding sites, despite extensive work on this protein. (Roughly half of anti-retroviral drugs for HIV target RT, and ART regimens typically include two or three separate inhibitors.)

 

Eight of the fragments inhibited the biochemical activity of the enzyme by at least 50% at 5 mM, and three of them gave IC₅₀ values in the low mM range with ligand efficiencies as high as 0.47 kcal/mol/atom. Among these three, two bound to a single site, while the third (4-amino-3-bromopyridine) bound at four separate sites. Another library member, the “universal fragment” 4-bromopyrazole previously identified by the researchers, bound to eight sites but showed only 22% inhibition at 5 mM.

 

The binding modes of all twelve fragments are described in some detail and show the standard range of hydrogen bonds and van der Waals interactions. Halogen bonds were surprisingly rare, unlike the case of FragLites against different proteins. It would be interesting to see a summary of the types of interactions, and how many involved the halogen atoms. The identities of all the library members are provided in the Supporting Information, so you can build your own library. 

 

And on that note, this is the last week to take our survey on fragment libraries, so please make sure to vote!