T. Ryckmans et al. / Bioorg. Med. Chem. Lett. 19 (2009) 4406–4409
4407
R1
O
R1
N
N
N
N
N
O
N
N
HN
b
a
R
R
R
N
2a-2b
7
8
e
c
d
O
O
O
R1
S
N
N
N
N
N
H
N
N
N
R
R1
O
R
R1
R
N
N
9
10
11
Scheme 2. Library design around templates 2a–b. Alkyls 7: (a) Aldehydes, DCM,
NaBH(OAc)3, 30 °C, 16 h success rate 46%. Carbamates 8: (b) DCM, TEA CDI 30 °C
16 h, then alcohols 30 °C 16 h success rate 77%. Ureas 9: (c) DCM, TEA, Triphosgene
30 °C 16 h, then amines 30 °C 16 h success rate 35%. Amides 10: (d) Acids, DIEA,
HATU, DMA 30 °C 16 h success rate 70%. Sulfonamides 11: Sulfonyl chlorides,
Dichloroethane, TEA, DMAP 30 °C 16 h success rate 71%. All final compounds were
purified by preparative HPLC.
was performed using PtO2 affording the fused piperidines 2a and
2b in 35 and 50% yields respectively.
Scaffolds 2a and 2b were functionalized by reductive amination
and formation of carbamates, ureas, amides and sulfonamides by
parallel chemistry using standard procedures. Overall 126 com-
pounds distributed in 5 classes were prepared (Scheme 2).
There is general consensus in the medicinal chemistry commu-
nity that high molecular weight and high lipophilicity are corre-
lated with poor oral drug-like properties. For example within a
series, increased microsomal clearance and pharmacological pro-
miscuity8 are often associated with higher cLog P values, while lim-
ited cell permeation and absorption are linked with low
lipophilicity. To achieve a compromise between absorption and
first-pass clearance, a cLog P value between 2 and 3 is often consid-
ered optimal in an oral drug program.
On the other hand, protein–ligand binding is partially driven
by lipophilic interactions, and the optimal cLog P for a class of li-
gands will depend on the nature of target protein. Since the
endogenous ligands for the CB2 receptor are highly lipophilic fatty
acid derivatives (endocannabinoids9), a higher range of cLog P val-
ues was considered, despite the increased risk of metabolic
instability.
Monomers were selected based on similarity with previously
established SAR, and our library design criteria included cLog P
and molecular weight filters (cLog P between 0 and 5, mean value
3.0 and molecular weight (MW) below 500, mean value 377).
Figure 2a shows the distribution of cLog P versus MW for each class
of compounds, and Figure 2b the distribution of compounds by
cLog P bins. 84 compounds (66%) had a cLog P between 2 and 4.
Distribution across bins was series dependent, for example, most
alkyl derivatives 7 (yellow) had a cLog P above 4 and most sulfon-
amides 11 (blue) had a cLog P value between 2 and 3.
Figure 2. (a) cLog P versus molecular weight distribution of compounds, colored by
class. (b) Distribution of compounds in each cLog P bins. Yellow: alkyls 7; black:
carbamates 8; green: ureas 9; red: amides 10; blue: sulfonamides 11.
LipE is increased when the pIC50 is increased by more than the
increase in cLog P. For example, a potent (EC50 9) compound with a
cLog P of 2 will have a LipE of 7. This value of cLog P is often con-
sistent with reasonable in vivo clearance, solubility and protein
binding. These factors contribute positively to the chances of such
a compound achieving good efficacy and duration in vivo through a
combination of good PK and potency.13 High LipE compounds
should outperform low LipE compounds as these will be compro-
mised by reduced potency or increased lipophilicity. Furthermore,
keeping the log P or log D of clinical candidates low will reduce the
chances of seeing toxicity or side effects.8 A recent analysis of ani-
mal safety studies14 showed an increased risk of adverse outcome
for compounds with cLog P > 3. In fact, we have found that it is very
common for a clinical candidate to have the highest, or near high-
est, LipE for the series.12 Whilst achievable LipE values are target
and series specific, compounds with LipE above 5 are usually con-
sidered highly optimized.
High in vitro potency is a desirable attribute in drug candidates,
as it reduces the risk of non-specific, off-target pharmacology.
Associated with low clearance, high potency also allows for low to-
tal dose and thus lower risk of idiosyncratic toxicity.10,11 Given the
aforementioned link between lipophilicity and clearance there is a
greater likelihood of achieving good in vivo performance when po-
tency can be increased within a series by making changes that keep
log P or log D fixed or even reduced.
Modestly potent CB2 full agonists were found in the alkyl, car-
bamate, urea and amide classes. Representative examples from
each class are represented in Figure 3. The nature of the alkyl side
chain in the imidazole 2-position had a moderate if unpredictable
impact on binding affinity (7b and 7c vs 9b and 9c). LipE values
were higher for amides 10a–b (Table 1).
Lipophilic Efficiency (LipE) has recently been introduced8,12 as a
parameter that combines both potency and lipophilicity (cLog P,
cLog D or measured Log D, if appropriate), and is defined as
LipE ¼ pIC50ðor pEC50Þ ꢀ cLogP
ð1Þ