Chemistry Reference
In-Depth Information
Since solubility prediction from Clog
P
or Clog
S
w
is well known to be unreliable for
many compound classes, we tested our library compounds experimentally for solubility at
500 M concentration under aqueous assay conditions. It was found that that the major-
ity of compounds were soluble and that many of the precipitate-forming compounds still
showed significant partial solubility. Protein-observed NMR assays can tolerate such par-
tially soluble compounds, which can still give rise to strong signals. In consequence, the
NMR technology permits the screening of fragment libraries which reflect the Clog
P
profile
of known drugs.
3.3.5 Physicochemical Parameters
The above-mentioned constraints on MWand Clog
P
already set the framework for H-bond
donor (HBD) and H-bond acceptor (HBA) filters. We found that the majority of compounds
passing our MW and Clog
P
filters also met our HBD and HBA filters, which we defined
as
7 HBAs, resulting in average values of 1.2 HBDs and 3.6 HBAs. Very few
compounds (
<
0.4%) were allowed to exceed these limits.
The number of rotatable bonds (rotB) reflects the rigidity of compounds, which is thought
to correlate with a low entropic penalty of binding and improved target selectivity. By
contrast, a few rotatable bonds allow fragments to flexibly adapt to the protein pocket,
thereby sampling a wider range of 3D chemical space. As a compromise, we set an upper
limit of five rotatable bonds and targeted an average of three rotatable bonds.
≤
4 HBDs,
≤
3.3.6
Immediate Validation of Bioactivity
Ultimately aiming at
in vivo
active target modulators, we designed our fragment library
so that the most efficient fragment binders can be immediately tested for bioactivity in
biochemical or cellular assays at low throughput. Filtering fragment binders immediately
for bioactivity avoids misallocating valuable F2L chemistry and structure determination
resources on bioinactive binders. We have validated this fast-into-cells approach in several
anti-infective discovery projects and found that bioactivity is indeed a very stringent filter
for fragment hits, showing drop-out rates of 80-90% for antibacterial targets. Furthermore,
by testing the entire NMR fragment library in a robust biochemical HTS assay, the feas-
ibility of finding several biochemically active fragments with 2-100 MIC
50
values was
experimentally confirmed.
[
49
]
3.3.7 Fast SAR Follow-up
Fast and cost-efficient hit follow-up with respect to SAR is an important advantage of
working with fragments. We made sure that we had sufficient chemical matter in stock for
confirmative secondary assays. Further, we laid emphasis on timely commercial availab-
ility of
10mg follow-up material for co-crystallisation and cellular assays by choosing
reliable vendors. Likewise, quick commercial access to analogues enables fragment hit val-
idation by comparatively low-cost, preliminary SAR studies. As the next step, we required
fragment scaffolds to be accessible by one or very few synthetic steps and to contain suf-
ficient functionality for fragment elaboration. For these reasons, we have avoided chiral
compounds and complex natural products.
≥
