4310 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 25
Letters
effectively discriminated between both sn-1 and sn-2
binding alternatives since transposition of the NOH
group to replace the lactone oxygen resulted instead in
a very weak ligand. This situation contrasts with
compounds where both chemical groups are esters (side
chain ester and lactone) which bind in a combination
of sn-1 and sn-2 modes.7 In the present work, we propose
that the N-hydroxyl amide group is able to direct the
binding of DAG-lactones into the C1 domain of PK-C
exclusively in the sn-2 mode.
Dr u g Design Con sid er a tion s a n d Biologica l Re-
su lts. The gradual steps taken in the structure-activity
analysis leading to the potent N-hydroxyl amide DAG-
lactone 6a are described in Table 1. For the sake of
simplicity, only the geometric Z-isomers will be dis-
cussed in the analysis, although both Z- and E-isomers
were synthesized and tested. The DAG-lactone 1a ,
selected as the parent compound for this approach, was
identified in a previous study as a high binding affinity
ligand for PK-CR (Ki ) 2.89 nM) in a competition assay
that measures the ability of the ligand to displace bound
[20-3H-PDBU] from the enzyme.7 This molecule contains
two similar branched alkyl chains, one acyl (R1) and one
R-alkylidene (R2). The corresponding E-isomer 1b was
also a potent ligand for PK-CR, but only the Z-isomer
1a displayed potent and selective antitumor activity in
the NCI 60-cell line in vitro screeen.7 Since we also
discovered that DAG-lactones bearing linear acyl chains
were devoid of cellular activityspresumably caused by
cleavage of the acyl group by cellular esterasessit was
important to maintain R1 as a highly branched alkyl
group. Thus, the first step toward reducing log P
involved trimming four carbons off the alkyl group of
1a to give the smallest possible and highest branched
R1 group (i.e., pivaloyl). Relative to compound 1a , such
change lowered the lipophilicity of 2a by almost 2 log P
units without compromising binding affinity, which was
only reduced ca. 3-fold. Since the pivaloyl group was
considered an ideal small group stable toward hydroly-
sis, it was maintained constant. To recover the 3-fold
loss in potency caused by the smaller pivaloyl group,
the branched R-alkylidene (R2) group was enlarged by
two additional carbons to give compound 3a . This
compound equaled 1a in potency and was still 0.86 log
P units lower (more hydrophilic). Support for the
pivaloyl group as the smallest and most effective acyl
group was provided by compound 4a , which contained
the original larger branched acyl group (R1) of 1a and
the same R-alkylidene (R2) chain of 3a . This change was
not advantageous since the resulting compound (4a ) was
not only less potent but more lipophilic. Having reached
what appeared to be the ideal set of branched acyl and
R alkylidene chains in compound 3a , we turned our
attention to the ester moiety. Changing the ester moiety
into an amide (X ) O f NH) was desirable, since for
the same set of branched acyl and R-alkylidene chains
the log P was reduced by ca. 1 log P unit. Unfortunately,
the affinity of compound 5a for PK-C plummeted.
However, when the amide hydrogen in compound 5a
was replaced by a hydroxyl group (X ) NH f NOH),
potency was restored, and, relative to the initial DAG-
lactone 1a , the new compound (6a ) was 2.3 log P units
less lipophilic! This change has produced the most
potent and least lipophilic DAG analogue known to date.
F igu r e 1. H-bonding interactions of DAG-lactones at the C1
domain of PK-C in alternative binding modes sn-1 (left) and
sn-2 (right).
modeling is limited to examining only binary complexes
between the C1 domain and the DAG-lactone ligands,
since the phospholipid component of the ternary com-
plex is missing in the experimentally available X-ray
structure.8
In our previous studies with DAG-lactones, we com-
bined linear and branched R1 and R2 alkyl chains, but
were only able to guess at the preferred binding mode.
In fact, in most instances it appeared that both binding
alternatives coexisted.7 To increase our chances of
designing C1 domain, isozyme-specific DAG-lactones, we
felt that it was imperative to know which specific
binding mode, sn-1 or sn-2, was operating.
While the role of the alkyl chain in DAGs has been
principally correlated with providing adequate lipophi-
licity to facilitate partitioning into a lipid-rich environ-
ment, docking of the DAG-lactones into the C1 domain
of PK-C revealed potentially important hydrophobic
contacts with the protein that could provide the means
of achieving discrimination between sn-1 and sn-2
binding modes.7 Thus, an increase in binding affinity
due to the branched alkyl chains would probably result
from a combination of two factors: adequate membrane
partitioning and specific hydrophobic contacts with the
protein. We anticipated that by reducing the lipid
dependency required for membrane distribution, the
compounds would be more effectively discriminated by
the protein in terms of the two possible binding alterna-
tives (sn-1 versus sn-2), since this process of recognition
would allow protein-ligand interactions to become the
dominant forces.
While working with a set of DAG analogues, we
recently showed a practical way to minimize lipid-
dependency as a way to increase interaction of the
ligand with the protein through the gradual reduction
of log P.9 In the present work, the application of this
approach to DAG-lactones started with the progressive
reduction of log P from the potent, symmetrically
branched DAG-lactone (1a , Table 1).7 We will demon-
strate that this approach, coupled with other structural
changes, reduces lipophilicity by more than 2 orders of
magnitude without sacrificing binding affinity for PK-
C. A critical change that allowed this reduction in log
P was the isosteric replacement of the side chain ester
group with an N-hydroxyl amide group [RC(O)O f RC-
(O)NOH], which lowered the log P to an unprecedented
calculated value of 3.58, almost matching the value of
the prototypic phorbol ester, phorbol-12,13-dibutyrate
(PDBU, calculated log P ) 3.43). In addition, the N-OH
group provided a fourth pharmacophoric group that