Classic galactosemia
is caused by loss-of-function mutations in
galactose-1-phosphate uridylyltransferase (GALT) that lead to toxic
accumulation of its substrate, galactose-1-phosphate. One proposed
therapy is to inhibit the biosynthesis of galactose-1-phosphate, catalyzed
by galactokinase 1 (GALK1). Existing inhibitors of human GALK1 (hGALK1)
are primarily ATP-competitive with limited clinical utility to date.
Here, we determined crystal structures of hGALK1 bound with reported
ATP-competitive inhibitors of the spiro-benzoxazole series, to reveal
their binding mode in the active site. Spurred by the need for additional
chemotypes of hGALK1 inhibitors, desirably targeting a nonorthosteric
site, we also performed crystallography-based screening by soaking
hundreds of hGALK1 crystals, already containing active site ligands,
with fragments from a custom library. Two fragments were found to
bind close to the ATP binding site, and a further eight were found
in a hotspot distal from the active site, highlighting the strength
of this method in identifying previously uncharacterized allosteric
sites. To generate inhibitors of improved potency and selectivity
targeting the newly identified binding hotspot, new compounds were
designed by merging overlapping fragments. This yielded two micromolar
inhibitors of hGALK1 that were not competitive with respect to either
substrate (ATP or galactose) and demonstrated good selectivity over
hGALK1 homologues, galactokinase 2 and mevalonate kinase. Our findings
are therefore the first to demonstrate inhibition of hGALK1 from an
allosteric site, with potential for further development of potent
and selective inhibitors to provide novel therapeutics for classic
galactosemia.