From HTS to Market: the Discovery and Development of Maraviroc, a CCR5 Antagonist for the Treatment of HIV - Accounts in Drug Discovery Case Studies in Medicinal Chemistry - page 186

9.4 Identifying Leads with Anity for CCR5
Since the natural ligands for CCR5 are known, it may be possible to develop
drugs from these. However, such an approach offers two significant challenges.
Firstly, the ''conversion'' of peptidic leads into drugs with good pharmaco-
kinetic properties continues to be dicult, as it requires designing compounds
that mimic the key interactions of an unknown conformation of a flexible peptide
while simultaneously improving metabolic stability and intestinal permeability.
Secondly, these chemokines interact with CCR5 via extracellular loops through a
series of complex interactions which it may not be possible to replicate with a
small molecule. 20 A more attractive route to small-molecule CCR5 antagonists
stems from CCR5's membership of the class A (rhodopsin-like) GPCRs, a class
which has been well-researched. As a consequence, many pharmaceutical com-
panies' compound collections contain drug-like compounds designed to target
related GPCRs, including muscarinics, 5HT 1A and the a1-adrenoceptor. 21 Like
many other pharmaceutical companies, we embarked on a strategy of high-
throughput screening (HTS) to identify suitable leads for a drug discovery
programme. We chose a screen based upon stably expressed CCR5 on HEK-293
cells. 22 As noted above, the one known chemokine selective for CCR5 was MIP-
1b; consequently, the assay was configured to measure inhibition of radiolabelled
MIP-1b0 binding. While we were confident that this assay would select ligands
that block the binding of MIP-1b, we were less certain if these hits would also
block binding by viral gp120 protein. As a consequence, additional assays were
developed during the course of the programme to more accurately assess inhi-
bition of viral protein binding to the receptor.
At that time, the Pfizer compound collection contained 0.5 million com-

pounds. Following screening, the ''hits'' were triaged based upon inhibitor
potency. The most potent hits from HTS are often the largest and most lipo-
philic compounds. However, these frequently tend to be the least tractable hits
for conversion into drugs. Consequently, ligand eciencies were calculated for
hits from the CCR5 screen to re-bias hits towards those showing the most
e cient binding for their size. Ligand e ciency (LE) is quickly obtained by
scaling the binding energy by the number of heavy atoms in the molecule (eqn
9.1). 23 More recently, other scaling parameters have been developed, including
lipophilic ligand eciency (LipE) 24,25 (which tries to compensate for lipophi-
licity-driven binding) and ''fit quality score'' 26 (which further scales ligand
eciency to prioritise smaller fragments which may otherwise be overlooked).

B

DG ð binding Þ
P heavy atoms ¼

RT log ð Ki Þ
P heavy atoms Ligand efficiency ¼

ð 9:1 Þ

1:4 log ð Ki Þ
P heavy atoms

ð kcalmol Þ

E

Two of the leading compounds to emerge from the HTS were 1 (UK-107543,
binding IC 50 ¼ 0.34 mM, LE ¼ 0.29) and 2 (UK-179645, binding IC 50 ¼ 1.1 mM,