Three years ago we
described the
discovery of LYS006, an inhibitor of leukotriene A4 hydrolase (LTA4H) from
Novartis currently in phase 2
clinical trials. Companies often pursue multiple
chemical series for important targets, and in a recent
J. Med. Chem.
paper Gebhard Thoma and colleagues describe another fragment-derived lead
against LTA4H.
A biochemical high-throughput
screen yielded compound 2, which is quite potent for a fragment-sized molecule.
However, despite good
ligand efficiency, the LipE (or
LLE) was less impressive
due to the high lipophilicity of the fragment. (Note that throughout the paper LipE
is calculated based on measured logD rather than logP.) A co-crystal structure revealed
that it bound in a similar fashion to other previously characterized LTA4H
inhibitors such as compound 1, derived from LYS006 and reported in a
J. Med.
Chem. paper last year. Adopting elements from these led eventually to
compound 12, which though less potent was also much less lipophilic and more
soluble while still remaining fragment-sized.
Continuing to borrow from the
rich literature around this target, the researchers added a basic amine group
to get to the very potent compound 14. This was metabolically unstable, but
further optimization led to compound 3.
Compound 3 was profiled
extensively in a battery of tests. In addition to good biochemical potency, it
showed mid-nanomolar activity in a human whole blood assay and was also active in
other assays, including a mouse arthritis model. Other attractive features
included a clean profile against a plethora of off-targets, good oral
bioavailability in mice, rats, and dogs, and a predicted human oral dose of 40
mg once daily. However, a two week toxicology study in rats and dogs was “slightly
less favorable” than compound 1.
This is a lovely example of property
and structure-guided drug design, and the researchers are refreshingly open
about borrowing elements from other molecules, even from outside Novartis. Interestingly,
a crystal structure of compound 3 bound to LTA4H revealed that while the
overall binding mode was similar to compound 1, which contains the same left-hand
portion, the pyrazole and pyridine rings rotated 180º
to make different hydrogen-bond interactions. Another reminder that despite our
leaps in predictive capability, molecules can still provide many surprises.
