Differential binding modes of diacylglycerol (DAG) and DAG lactones to protein kinase C (PK-C)

Article abstract of DOI:10.1021/jm020476o

Diacylglycerol lactones (DAG lactones), analogous to highly potent diacylglycerols (DAGs) were synthesized to demonstrate the ability of PK-C to discriminate between two differential binding modes, sn-1 and sn-2. While both sn-1 and sn-2 binding modes are

Full text of DOI:10.1021/jm020476o

J . Med. Chem. 2003, 46, 1571-1579
1571
Articles
Differ en tia l Bin d in g Mod es of Dia cylglycer ol (DAG) a n d DAG La cton es to
P r otein Kin a se C (P K-C)
Dina M. Sigano,† Megan L. Peach,† Kassoum Nacro,†,‡ Yongseok Choi,† Nancy E. Lewin,§ Marc C. Nicklaus,†
Peter M. Blumberg,§ and Victor E. Marquez*,†
Laboratory of Medicinal Chemistry, Center for Cancer Research, National Cancer Institute at Frederick,
National Institutes of Health, Frederick, Maryland 21702, and Laboratory of Cellular Carcinogenesis &
Tumor Promotion, Center for Cancer Research, National Cancer Institute, National Institutes of Health,
Bethesda, Maryland 20892
Received October 24, 2002
Diacylglycerol lactones (DAG lactones), analogous to highly potent diacylglycerols (DAGs) were
synthesized to demonstrate the ability of PK-C to discriminate between two differential binding
modes, sn-1 and sn-2. While both sn-1 and sn-2 binding modes are allowable in terms of
hydrogen bonding, it has been found that in general, DAGs prefer to bind sn-1, while the
corresponding analogous DAG lactones prefer to bind sn-2. However, this binding orientation
can be directly influenced by the disposition and nature of the acyl substituent, particularly if
it is highly branched. When the “binding driving force” (i.e., the larger branched acyl chain) is
in the sn-2 position, a dramatic increase in binding affinity is observed in the DAG lactone as
compared to its open chain DAG counterpart. As these analogous DAGs and DAG lactones
have almost identical log P values, this difference in binding affinity is a direct result of the
entropic advantage of constraining the glycerol backbone.
In tr od u ction a n d Ba ck gr ou n d
sociation is strongly facilitated by the lipophilic second
messenger, DAG, which is generated as a result of a
stimulus-initiated activation of phospholipase C.11 The
binding of PK-C to the plasma membrane is transient
and regulated by the association of its C1 domain with
DAG in the membrane.12
Protein kinase C (PK-C) is a family of serine/threo-
nine specific isozymes that are regulated by phospho-
rylation, by calcium, and by association with phospho-
lipids and 1,2-diacylglycerol (DAG).1-4 Its discovery5 in
the early 1980s as “the receptor” for the tumor-promot-
ing phorbol esters led to a barrage of studies revealing
the central role of PK-C in cell signaling. The PK-C
family comprises 10 isozymes grouped into three
classes: conventional (R, â, γ), novel (δ, ꢀ, η, θ), and
atypical (ú, ι/λ). In addition, PK-C µ and ν are considered
by some to constitute a fourth class and by others to
comprise a distinct family called protein kinase D.6 All
members contain a C-terminal kinase domain with
serine/threonine specific kinase activity and an N-
terminal regulatory domain. The regulatory domain
contains two key functionalities: an autoinhibitory
sequence and one or two membrane-targeting domains
(C1 and C2).
The classical and novel PK-C isozymes each contain
two small (∼50 residues) zinc fingerlike domains (C1a
and C1b), consisting of two â-sheets and a small R-helix.
The crystal structure of an isolated C1 domain from PK-
Cδ in complex with phorbol-13-O-acetate13 showed that
when the ligand is bound in the pocket between the
â-sheets, a continuous hydrophobic surface is created
over the top third of the domain, which in turn promotes
its insertion into the membrane.
Previously, we have found that DAGs containing
branched acyl groups exhibit a high affinity for PK-C
with the presence of these branched acyl groups being
of critical importance to reach binding affinities in the
nanomolar range.14 This affinity is found to be partially
based on a lipophilic requirement, measured as log P
(the octanol-water partition coefficient), provided by the
acyl groups. While the absolute log P value cannot be
used to predict binding affinity, it can be used to
compare the relative lipophilicity of a compound and,
therefore, to describe its degree of hydrophobic and
hydrophilic binding. Aside from this receptor interac-
tion, log P values are also important for membrane
permeability and disposition before the drug even
reaches the receptor. For binding to PK-C, optimal
compounds provide hydrophobic contacts that bind to a
group of conserved hydrophobic amino acids located
The accepted dogma has been that these isozymes are
cytosolic in the inactive state and, as part of the
activation process, translocate to the inner leaflet of the
cellular membrane.7,8 Classical (R, â, and γ) as well as
novel (δ, ꢀ, η, θ) PK-C isozymes become activated as a
result of the association of the cytosolic enzyme with
membranes containing acid phospholipids.9,10 This as-
* To whom correspondence should be addressed. Tel: 301-846-5954.
Fax: 301-846-6033. E-mail: marquezv@mail.nih.gov.
† Laboratory of Medicinal Chemistry, Center for Cancer Research.
‡ Current address: Albany Molecular Research Inc., P.O. Box 15098,
Albany, NY 12212.
§ Laboratory of Cellular Carcinogenesis & Tumor Promotion, Center
for Cancer Research.
10.1021/jm020476o This article not subject to U.S. Copyright. Published 2003 by the American Chemical Society
Published on Web 03/29/2003

1572 J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 9
Sigano et al.
in the C1 domain. Thus, the contrast in Ki values
between 1 and its transposed analogue (2) reflects the
ability of PK-C to discriminate between the different
orientations of the two acyl branches.
We have also observed highly potent PK-C ligands
based on the DAG structure where the glycerol back-
bone is constrained into a five member lactone ring
(DAG lactone).14 It was discovered that the Z isomer of
the DAG lactone 3 exhibited a 10-fold increase in
binding affinity when compared to its open chain
counterparts (4 and 5). Because these compounds, 3-5,
have almost identical log P values, this 10-fold increase
in potency is believed to be the direct result of the
entropic advantage of constraining the glycerol back-
bone.
F igu r e 1. Hydrogen-bonding interaction of DAG (top) and
DAG lactone (bottom) at the C1 domain of PK-C in the
alternative binding modes of sn-1 (left) and sn-2 (right).
Numbering of the PK-C residues is internal to the C1 domain
(e.g., Gly 23 corresponds to Gly 253 in PK-Cδ C1b, Gly 59 in
PK-CR C1a, and Gly 124 in PK-CR C1b). Numbering of the
DAG carbon backbone is shown in red.
around the rim of the binding pocket at the top of the
C1 domain where phorbol esters bind.
Docking studies using the crystal coordinates of the
C1b domain of PK-Cδ in complex with phorbol-13-O-
acetate13 revealed two possible binding modes, sn-1 and
sn-2, when DAG or DAG lactone15 was docked into the
empty C1b domain. Although the carbonyl moieties are
not equivalent, they are able to display a similar
network of hydrogen bonds to Thr12, Leu21, and
Gly23smatching those seen with phorbol-13-O-acetates
in two distinct orientations with participation from the
primary OH and only one of the carbonyl groups. The
sn-1 binding mode in either DAG or DAG lactone is
defined as that in which the sn-1 carbonyl is directly
bound to the protein; while the sn-2 binding mode is
defined as that in which the sn-2 carbonyl is directly
bound to the protein (Figure 1).
Previously, we successfully undertook the task of
identifying a DAG candidate with the lowest log P value
that did not compromise binding affinity.16 In doing so,
nonspecific binding interactions between the DAG and
the membrane were minimized and the differences in
Ki values reflected changes in specific interactions
between the ligand and the protein. Included in this
series is 1, the most potent, hydrophilic DAG ligand
known to date. It is 1.5 log units less lipophilic than
1,2-dioctanoyl-sn-glycerol (diC8), activates PK-Ca ca.
2-fold better than diC8, and is capable of translocating
the full-length protein efficiently to the membrane.16
From this study, the significance of hydrophobic inter-
actions between ligand, receptor, and phospholipid in
the ternary complex has become more evident. The
observed differences in the binding affinities within this
DAG series were attributed to the particular orientation
of the large and small branched acyl groups and their
ability to interact with specific hydrophobic amino acids
Resu lts a n d Discu ssion
SAR Design . An exercise similar to the one described
above using the corresponding DAG lactones derived
from these more recently discovered,16 less lipophilic
DAGs was expected to be equally informative. DAG
lactones (6-9), conceptually generated by sn-2 lacton-
ization of the DAG analogues (1 and 2), were selected
as targets to study the effect of acyl group predilection
on a constrained glycerol backbone (Scheme 1). It should
also be noted that despite the extra carbon required to
form the lactone ring, the DAG lactones are less
lipophilic than their open chain counterparts. Therefore,
to keep the lipophilicity consistent between the DAGs
and the DAG lactones, additional carbons were added
to the sn-2 acyl group to increase the log P closer to the
value of 3.9 calculated for the DAGs.
Syn th esis. Lactones 6, 7, and 9 were synthesized
according to previously published procedures.15,17 Aldol
condensation of lactone 1018 with acetone gave 11, which
was not isolated but directly subjected to elimination
conditions to give 12. Deprotection followed by monoa-
cylation gave 6 (Scheme 2). Alternatively, condensation
of 10 with the desired aldehyde gave a mixture of
diastereomeric alcohols, which were not isolated but
immediately subjected to elimination conditions. A
mixture of E and Z olefin isomers (16 and 17) was
obtained in ca. 2:1 ratio and was readily separated by

Binding Modes of DAG and DAG Lactones to PK-C
J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 9 1573
Sch em e 1
Sch em e 2
column chromatography. Deprotection followed by
monoacylation gave the desired DAG lactones, 7 and 9
(Scheme 3).
competition curve. The Ki values for inhibition of bind-
ing were calculated from the ID50 values. The octanol/
water partition coefficients (log P) were calculated
according to the fragment-based program KOWWIN
1.6320 (Scheme 1).
Interestingly, DAG lactones 6 and 7, analogues of the
most potent DAG (1), did not show the expected 10-fold
increase in binding. In fact, while the Ki values of
lactones 6 and 7, with the larger acyl branch at the sn-1
position, were slightly lower, they were within a factor
e2.5 of that for DAG 1. While DAG lactone 8 was not
synthetically accessible, when we compared DAG lac-
tone 9, derived from DAG 2, with the larger acyl branch
at the sn-2 position, we saw a dramatic increase in
Biologica l Activity. Binding affinity assays were
done with intact recombinant PK-CR, an isozyme where
both the C1a and the C1b domains are involved in
ligand binding and membrane translocation.19 The
interaction of the DAGs and DAG lactones was assessed
in terms of the ability of the ligand to displace bound
[20-3H]phorbol-12,13-dibutyrate (PDBu) in the presence
of phosphatidylserine. The inhibition curves obtained
for all ligands were of the type expected for competitive
inhibition, and the ID50 values were determined by fit
of the data points to the theoretical noncooperative

1574 J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 9
Sigano et al.
Sch em e 3
Ta ble 1. Hydrogen-Bonding Interaction Energies (kcal/mol)
binding. Indeed, the more potent Z isomer, 9Z, showed
a 33-fold increase in binding when compared to 2. Even
the less potent E isomer (9E) exhibited 10-fold better
binding. It is also worth noting that for the sake of
synthetic expediency, the DAG lactones synthesized are
racemic; hence, the potency for all known active (R)-
DAG lactone enantiomers should be approximately
doubled.21
Molecu la r Mod elin g. For computational docking
studies, homology models of the C1a and C1b domains
of PK-CR were built based on the crystal structure of
the C1b domain of PK-Cδ.13 The open chain DAGs 1 and
2, along with the E and Z isomers of the corresponding
DAG lactones 7 and 9, which have comparable log P
values, were docked into both the C1a and the C1b
domains using the docking programs FlexX22 and
GOLD.23
Surprisingly, the docking results from FlexX and
GOLD were not consistent. FlexX docked the DAG
ligands exclusively in only one of the two possible
orientationsssn-1 for the open chain DAGs and sn-2 for
DAG lactones. GOLD, however, produced a mixture of
sn-1 and sn-2 orientations in all cases, in a manner
similar to previous docking studies with DAG ligands
using the program AutoDock.15,16,24 In some cases, the
highest ranked GOLD structure of a DAG lactone was
docked sn-1, but all of the FlexX structures were docked
in the sn-2 orientation for all of the DAG lactones.
Therefore, in an attempt to determine which orientation
is preferred, the docked sn-1 and sn-2 structures for all
of the DAG ligands were analyzed in further detail.
The highest scoring conformation of each ligand
docked in a specific orientation into either the C1a or
the C1b receptor was selected and refined with 100
cycles of simulated annealing to generate an ensemble
of similar low-energy structures for each docked con-
formation, and the average hydrogen-bonding interac-
tion energy between PK-C and DAG for each ensemble
was calculated (Table 1). Because small changes in
receptor-ligand conformation can produce very large
differences in electrostatic energy, averaging over an
ensemble smoothes out any spurious energy differences
and Buried Surface Areas (Å2) between DAG Ligands and the
C1 Domains of PK-CR
H-bonding energy buried surface area
C1a
C1b
C1a
C1b
DAG 1 bound sn-1
DAG 1 bound sn-2
DAG 2 bound sn-1
DAG 2 bound sn-2
DAG lactone 7Z bound sn-1
DAG lactone 7Z bound sn-2
DAG lactone 7E bound sn-1 -53.34 -83.51 567.01
DAG lactone 7E bound sn-2 -94.84 -145.13 540.79
DAG lactone 9Z bound sn-1
-120.33 -115.30 547.59
-61.22 -102.35 533.31
-119.59 -107.60 548.92
-87.52 -100.52 553.62
-61.21 -88.76 554.71
-97.49 -147.27 511.66
563.74
547.11
555.47
562.27
572.41
558.38
542.95
532.20
523.70
618.23
522.83
536.45
-57.19 -83.88 514.20
DAG lactone 9Z bound sn-2 -123.80 -175.14 520.91
DAG lactone 9E bound sn-1 -53.37 -82.08 559.36
DAG lactone 9E bound sn-2 -80.49 -143.06 501.62
and allows more accurate analysis and comparison of
the relative energies of the two binding modes for each
ligand.
For each ensemble, the average difference in solvent
accessible surface area between the bound complex and
the unbound ligand and receptor was also calculated
(Table 1). This number is related to the nonpolar
component of the solvation energy (i.e., the hydrophobic
effect). Thus, the size of the surface area that is buried
upon ligand binding is a good measure of the strength
of the hydrophobic interactions between ligand and
receptor.
For the open chain DAGs (1 and 2), in both the C1a
and the C1b domains of PK-CR, the hydrogen-bonding
energy is stronger (lower) with sn-1 binding (Table 1).
This agrees with our previous study,16 where we ob-
served shorter hydrogen-bonding distances in the sn-1
orientation for a series of DAG compounds. When the
larger acyl side chain is next to the sn-1 carbonyl, as in
DAG 1, the size of the buried surface area upon ligand
binding is also larger with sn-1 binding, suggesting that
this compound favors the sn-1 orientation (Table 1).
However, when the larger acyl side chain is next to the
sn-2 carbonyl, as in DAG 2, although the hydrogen-
binding energy is stronger with sn-1 binding, the size
of the buried surface area is larger with sn-2 binding.

Binding Modes of DAG and DAG Lactones to PK-C
J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 9 1575
This antagonism of forces is mirrored with the DAG
lactones 7 and 9. Hydrogen bonding for the lactones is
stronger in the sn-2 orientation (Table 1), and this
binding mode is reinforced by the increased size of the
buried surface area when the bulkier side chain is next
to the binding (sn-2) carbonyl as in 9Z. For 7E and 7Z
however, although the hydrogen bonding is stronger for
the sn-2 orientation, the buried surface area is increased
in the sn-1 orientation due to the bulkier side chain
being on the sn-1 carbonyl.
adjacent to the lactone carbonyl engaged in binding (i.e.,
sn-2), the binding affinity increases at least ca. 10-fold
and is also sensitive to the E and Z stereochemistry of
the double bond as the presence of the R,â-unsaturated
system in the lactones creates restricted rotation near
the Gly23 binding site.
A surface representation of the empty binding site
shown from above (Figure 2A) offers a structural
explanation for the importance of the disposition of the
larger branched acyl chain. In the center of the binding
pocket is a deep cavity, where the primary OH group
binds to the backbones of Thr12 and Leu21.25 The rim
of the binding site is lined with hydrophobic residues
and contains three “pockets” through which the acyl
chains of DAG and other ligands can extend into the
membrane. One small, shallow pocket is located be-
tween Pro11 and Leu20; a second, medium-sized pocket
is located between Gly9 and Leu24; and the third,
largest pocket is located between Leu20 and Leu24,
across Gly23. This glycine residue at position 23 is
absolutely conserved across all C1 domains in the PK-C
isozyme family, suggesting that it is vital to the struc-
ture and function of these domains.
When the larger acyl chain is next to the carbonyl
that binds to Gly23, as in the sn-1 orientation of DAG
1 (Figure 2B) or the sn-2 orientation of DAG lactone
9Z (Figure 2E), the ligand fits neatly into the large
pocket between Leu20 and Leu24. The acyl chain of the
ligand is buried, and the gap in the hydrophobic surface
of the receptor is filled. However, when the smaller acyl
chain is next to the binding carbonyl, as in the sn-2
orientation of DAG 1 (Figure 2C) or the sn-1 orientation
of DAG lactone 9Z (Figure 2F), it is not sufficiently large
to fill this large pocket. In this orientation, the bulkier
acyl chain is forced into the smaller gap between Gly9
and Leu24, where it is not completely buried.
DAG lactone 9E, on the other hand (Figure 2D),
cannot conform to the geometry of the binding site due
to restricted rotation about the double bond. When
binding in the sn-2 orientation, the bulkier acyl chain
packs into the small, shallow pocket near Pro11, instead
of fitting into the pocket between Leu20 and Leu24. This
isomer may in turn prefer to bind in the less efficient
sn-1 mode, which explains its lower binding affinity
relative to the Z isomer (Table 1).
DAG lactones 7E and 7Z, derived from the lactoniza-
tion of 1, most likely bind sn-1. This mode of binding in
DAG lactones appears to be driven by the presence of
the bulkier branched chain on this carbonyl. Because
there is no restricted rotation of the branched alkyl
chain in the environment of the sn-1 carbonyl, there is
no difference in the binding between E and Z isomers;
because the carbonyl involved in binding (sn-1) is not a
component of the lactone ring, there is not a significant
entropic advantage (<3-fold in the best case) due to
lactonization.
For the DAG ligands 1 and 9Z, the modeling results
unequivocally show which binding mode is more favor-
able, whereas for the ligands 2 and 7E/Z, deducing the
correct binding mode depends on whether binding is
driven more by hydrogen-bonding interactions or by
hydrophobic interactions between receptor and ligand.
The hydrophobic effect is generally considered to be the
major driving force for protein folding and is likely also
to be the driving force for interactions between PK-C
and its DAG ligands, especially since it is the formation
of a continuous hydrophobic surface over the top of the
bound complex that drives membrane translocation and
activation of PK-C. This suggests that DAG 2 will bind
in the sn-2 orientation to maximize its hydrophobic
interactions with PK-C, whereas the DAG lactones 7E/Z
will probably bind in the sn-1 orientation for these same
reasons.
These results also give an explanation for the dis-
crepancy between the docking results from FlexX and
those from GOLD. The docking algorithm in FlexX
begins by placing a core fragment into the binding site,
and then, the rest of the ligand is built up piece by piece.
Because hydrogen-bonding interactions are stronger in
the sn-1 orientation for the open chain DAGs and
stronger in the sn-2 orientation for the DAG lactones,
this determines the orientation in which FlexX will
always place the core fragment. GOLD, on the other
hand, uses a genetic algorithm to place the whole ligand
in the binding site at once, and its internal scoring
function is weighted to favor hydrophobic van der Waals
interactions, thus giving a mixture of sn-1 and sn-2
binding.
Some interesting patterns that differentiate the mode
of binding between DAGs and their corresponding
lactones have begun to emerge. The open chain DAGs
appear to prefer to bind in the sn-1 mode.16 This binding
mode can be reinforced by the presence of a highly
branched alkyl chain next to this carbonyl, such as in
1. However, the disposition of the branched alkyl chains
can alter this preference and force the less efficient sn-2
binding mode, resulting in a weaker ligand, such as 2.
From the present work, we learn that the situation with
the DAG lactones seems to be the reverse. It appears
that the sn-2 binding mode is favored for the DAG
lactones (6-9) and this reversed binding mode hypoth-
esis is confirmed by the binding results. We observe that
lactonization of the most efficient DAG leads to a less
efficient DAG lactone, while the less efficient DAG leads
to a more efficient DAG lactone. Here, too, the location
of the branched alkyl chain drives the binding mode. If
the branched alkyl chain is adjacent to the sn-1 carbo-
nyl, the DAG lactone binds such as a DAG (sn-1) and
no entropic benefit is derived from the cyclization.
However, if the location of the branched alkyl chain is
However, DAG lactone 9Z, derived from the lacton-
ization of 2, most likely binds sn-2, and because the
carbonyl engaged in binding is the lactone carbonyl, a
ca. 10-fold increase in binding affinity is seen relative
to its parent DAG. Here, too, the binding mode appears
to be driven by the bulkier side chain now adjacent to
the lactone carbonyl. Because the lactone (sn-2) carbonyl
is the one engaged in binding, we observe a 2-3-fold

1576 J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 9
Sigano et al.
F igu r e 2. (A) Empty C1B domain of PK-CR; (B) DAG 1 bound sn-1; (C) DAG 1 bound sn-2; (D) E-DAG lactone 9 bound sn-2; (E)
Z-DAG lactone 9 bound sn-2; (F) Z-DAG lactone 9 bound sn-1. The DAG ligands are shown in CPK representation and colored by
atom type (C ) green; H ) white; O ) red). The Connolly surface of the receptor was generated using Insight II (2000) and is
colored by residue hydrophobicity (blue ) hydrophilic; red ) hydrophobic). Residue numbering is internal to the C1 domain.
C1b are 42% identical and 54% similar, those of PK-Cδ C1b
and PK-CR C1b are 62% identical and 80% similar, and the
zinc-binding residues and residues essential for ligand binding
are highly conserved. This suggests that the backbone struc-
ture will be very similar for all three C1 domains. Side chains
for the homology models were constructed using the program
SCWRL,27 which uses a backbone-dependent rotamer library
to place residues in their most likely conformation given the
backbone φ-ψ angles at that position. Residues homologous
to PK-Cδ were left unchanged from their crystallographic
positions. The homology models were refined with a small
energy minimization, with phorbol-13-O-acetate left in position
from the crystal structure to prevent the binding site loops
from closing during the minimization. Harmonic positional
restraints on the backbone atoms were gradually relaxed over
the course of the minimization to eliminate steric clashes in
the side chains without inducing deformations in the backbone.
Dock in g. FlexX was run through its SYBYL module,28 with
default options throughout. The binding site was defined as
residues 8-13, 20-24, and 27. Standard default settings were
also used in GOLD, with the binding site defined by atoms
within a 10.0 Å radius of the Nꢀ atom of residue Q27. To
overcome GOLD’s tendency to dock the DAG ligands with one
hydrophobic chain buried in the binding site, a template
similarity constraint was added to the hydrogen bond acceptor
atoms in the crystal structure of bound phorbol-13-O-acetate.
difference between the Z and the E isomers due to the
restricted rotation of the alkyl chain in the environment
of this carbonyl and its effect on the interaction with
the C1 domain. This result suggests that such an
interaction is with the C1 domain and not with the lipid
where the stereochemical disposition of the alkyl chain
would be expected to be less important.
Obser va tion s a n d Con clu sion s
The ability of PK-C to discriminate between the
different orientations of the acyl branches is clearly
demonstrated by this study. We observe that in terms
of hydrogen bonding, both sn-1 and sn-2 binding modes
are allowed. In general, the open chain DAGs prefer to
bind sn-1, while the corresponding DAG lactones prefer
to bind sn-2. However, the disposition and nature of the
acyl chain, particularly if it is highly branched, can have
a direct influence on the binding orientation. When the
“binding driving force” (i.e., the larger branched acyl
chain) is in the sn-2 position, a dramatic increase in
binding affinity is observed in the DAG lactone as
compared to its open chain DAG counterpart. As these
analogous DAGs and DAG lactones have almost identi-
cal log P values, this difference in binding affinity is a
direct result of the entropic advantage of constraining
the glycerol backbone.
Refin em en t. The highest scoring structure for the sn-1 and
sn-2 conformations of each ligand in each C1 domain (from
either the FlexX or the GOLD results) was selected for further
refinement. All simulations were run in CHARMM29 version
28n1 or 27b3, and the all atom CHARMM 22 force field,30
which includes parameters for general organic compounds as
well as amino acids, was used. Missing parameters and partial
charges for the DAG ligands were generated in QUANTA 97.27
No explicit waters were included, and the nonbonded interac-
tions were calculated without cutoffs and with a constant
Exp er im en ta l Section
Com p u ta tion a l Mod elin g. Hom ology Mod elin g. Homol-
ogy models of the C1a and C1b domains of PK-CR were built
on the backbone coordinates of the crystal structure of the C1b
domain of PK-Cδ.13 The sequences of PK-Cδ C1a and PK-CR

Binding Modes of DAG and DAG Lactones to PK-C
J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 9 1577
dielectric of 1.0 to mimic the low-dielectric environment of the
membrane. Only the binding site residues (7-13, 20-24, and
27) were free to move along with the ligand; all other residues
were held fixed. Gentle constraints holding the distance
between the atoms involved in the three hydrogen bonds
between the DAG ligand and the protein (Figure 1) at less than
3.0 Å were included to keep the ligand in the binding site
during high-temperature molecular dynamics and to preserve
its sn-1 or sn-2 orientation. The docked structures were first
minimized with the Adopted Basis Newton-Raphson algo-
rithm until the RMS value of the derivatives reached a
convergence of 0.001, then heated in 10° increments every 200
steps from a starting temperature of 50-300 K, and equili-
brated at 300 K for 10 000 steps. During equilibration, the
atomic velocities were reassigned according to a random
Gaussian distribution every 200 steps. Simulated annealing
was performed by rapidly heating the system from 300 to 500
K at a rate of 1° per step, maintaining it at 500 K for 500
steps, and then slowly cooling by 5° every 100 steps back to
300 K. This cycle was repeated 100 times, and the structure
at the endpoint of each cycle was minimized to convergence
as before to yield an ensemble of 100 low-energy structures
for each docked conformation.
An a lysis. The hydrogen-bonding energy for each structure
was calculated in CHARMM by first defining four atom subsets
consisting of ligand donor atoms, ligand acceptor atoms,
receptor donor atoms, and receptor acceptor atoms. Each
potential donor-acceptor pair was tested to see whether it was
within allowable hydrogen-bonding geometry (i.e., with an H‚
‚‚A distance of 2.7 Å or less and a D-H‚‚‚A angle greater than
90°). If so, the nonbonded (electrostatic and van der Waals)
interaction energy between the two atoms was calculated. The
sum total of these interaction energies for all of the structures
in the ensemble was then divided by 100 to give the average
hydrogen-bonding energy for the ensemble. The solvent ac-
cessible surface areas (SASA) for the ligand alone, the receptor
alone, and the bound complex were also computed in CHARMM,
using the Lee and Richards algorithm,31 a probe radius of 1.4
Å, and van der Waals radii for all atoms. The size of the buried
surface area upon ligand binding was calculated as the SASA
of the bound complex subtracted from the sum of the SASA of
the unbound receptor and ligand. The average buried surface
area was calculated for each ensemble.
5,5-Bis[(2,2-d im et h yl-1,1-d ip h en yl-1-sila p r op oxy)m e-
t h yl]-3-(m et h ylet h ylid en e)-4,5-d ih yd r ofu r a n -2on e (12).
POCl3 was added dropwise to a 0 °C solution of 11 in pyridine
(15 mL), which was allowed to reach room temperature
overnight. The reaction was diluted with Et2O, washed with
10% aqueous CuSO4 solution (3×), dried over Na2SO4, and
concentrated. Isolation by silica gel chromatography gave 12
(2.07 g, 97% from 10), which was used without further
purification. 1H NMR (300 MHz, CDCl3): δ 1.08 (s, 18 H, CH2-
OSiPh2tBu), 1.92 (s, 3 H, CdCHMe2), 2.34 (s, 3 H, CdCHMe2),
2.82 (s, 2 H, lactone-CH2), 3.78-3.80 (m, 4 H, CH2OSiPh2tBu),
7.41-7.51 (m, 12 H, SiPh2tBu), 7.71-7.68 (m, 8 H, OSiPh2tBu).
13C NMR (75 MHz, CDCl3): δ 19.18, 19.70, 24.33, 32.33, 66.37,
83.14, 127.62, 132.66, 132.84, 135.49, 148.26, 169.54.
Gen er a l P r oced u r e for Con d en sa tion /Deh yd r a tion for
Com p ou n d s 16E/Z a n d 17E/Z. Under argon, LDA (1.4 equiv,
2 M solution) was added to a -78 °C solution of lactone (1
equiv) in THF (6 mL/mmol). After it was stirred at -78 °C for
2 h, a solution of aldehyde (4 equiv) in THF (6 mL/mmol) was
added dropwise. The reaction was kept at -78 °C and
monitored by thin-layer chromatography (TLC). After the
designated time, the reaction mixture was quenched at -78
°C with aqueous NH4Cl (6 mL/mmol), warmed to room
temperature, extracted with Et2O (3×), dried over MgSO4, and
concentrated in vacuo to give the hydroxy adduct, which was
used without further purification.
The crude product was then dissolved in CH2Cl2 (7 mL/
mmol) containing pyridine (4 equiv) and cooled to 0 °C. MsCl
(2 equiv) was added dropwise, and the reaction was stirred at
0 °C for 30 min and then at room temperature for 90 min.
The reaction mixture was then cooled again to 0 °C, DBU (5
equiv) was added dropwise, and the reaction reached room
temperature overnight. The crude mixture was then filtered
through a thin pad of Celite, concentrated, and purified by
column chromatography to give both the E and the Z isomers.
5,5-Bis[(2,2-d im et h yl-1,1-d ip h en yl-1-sila p r op oxy)m e-
th yl]-3-[1-h yd r oxy-4-m eth yl-3-(m eth yleth yl)p en t-2-en yl]-
oxola n -2-on e (15). This compound has been previously
reported.15
(Z)-5,5-Bis[(2,2-d im eth yl-1,1-d ip h en yl-1-sila p r op oxy)-
m e t h yl]-3-(3-m e t h ylb u t ylid e n e )-4,5-d ih yd r ofu r a n -2-
on e (16Z). Yield: 0.806 g, 27%. IR (neat): 2957 (CH), 2870
(CH), 1759 (CdO), 1671 (CdC) cm-1 1H NMR (400 MHz,
.
1-(3-Isop r op yl-4-m et h ylp en t a n oyl)-2-(3-m et h ylb u t -2-
en oyl)-sn -glycer ol (1). This compound has been previously
reported.16
CDCl3): δ 0.94 (d, J ) 6.6 Hz, 6 H, CdCHCH2CHMe2), 1.02
(s, 9 H, CH2OSiPh2tBu), 1.72 (apparent septet, J ) 6.7 Hz, 1
H, CdCHCH2CHMe2), 2.63 (tt, J ) 7.2, 2.0 Hz, 2 H, Cd
CHCH2CHMe2), 2.86 (m, 2 H, lactone-CH2), 3.72 (AB quartet,
J ) 10.6 Hz, 4 H, CH2SiPh2tBu), 6.14-6.19 (m, 1H, C )
CHCH2CHMe2) 7.34-7.45 (m, 12 H, SiPh2tBu), 7.62-7.65 (m,
8 H, SiPh2tBu). 13C NMR (100 MHz, CDCl3): δ 19.21, 22.34,
26.69, 28.60, 33.10, 36.22, 65.99, 84.30, 127.74, 129.78, 132.71,
132.90, 135.57, 142.21, 169.30. FAB-MS (m/z, relative inten-
sity): 633 (MH+-C4H10, 11), 135 (100). Anal. (C43H54O4Si2) C,
H.
1-(3-Met h ylb u t a n oyl)-2-(3-isop r op yl-4-m et h ylp en t -2-
en oyl)-sn -glycer ol (2). This compound has been previously
reported.16
(Z)-{2-(H yd r oxym et h yl)-4-[4-m et h yl-3-(m et h ylet h yl)-
p en tylid en e]-5-oxo-2-2,3-d ih yd r ofu r yl}m eth yl-4-m eth yl-
3-(m eth yleth yl)p en ta n oa te (3). This compound has been
previously reported.14
1,2-Di-(3-isop r op yl-4-m eth ylp en ta n oyl)-sn -glycer ol (4).
This compound has been previously reported.16
(E)-5,5-Bis[(2,2-d im eth yl-1,1-d ip h en yl-1-sila p r op oxy)-
m e t h yl]-3-(3-m e t h ylb u t ylid e n e )-4,5-d ih yd r ofu r a n -2-
on e (16E). Yield: 1.8 g, 52%. IR (neat): 2930 (CH), 2857 (CH),
1760 (CdO), 1685 (CdC) cm-1. 1H NMR (400 MHz, CDCl3): δ
0.98 (d, J ) 6.6 Hz, 6 H, CdCHCH2CHMe2), 1.04 (s, 18 H,
CH2OSiPh2tBu), 1.84 (apparent septet, J ) 6.7 Hz, 1 H, Cd
CHCH2CHMe2), 2.06-2.10 (m, 2 H, CdCHCH2CHMe2), 2.80-
2.81 (m, 2 H, lactone-CH2), 3.75 (AB quartet, J ) 10.7 Hz, 4
H, CH2SiPh2tBu), 6.78 (tt, J ) 7.6, 2.9 Hz, 1 H, C ) CHCH2-
CHMe2) 7.36-7.47 (m, 12 H, SiPh2tBu), 7.63-7.66 (m, 8 H,
SiPh2tBu). 13C NMR (100 MHz, CDCl3): δ 19.17, 22.48, 26.64,
28.07, 29.73, 39.22, 66.10, 85.16, 127.74, 129.78, 132.59,
132.84, 135.53, 138.61, 170.42. FAB-MS (m/z, relative inten-
sity): 633 (MH+-C4H10, 29), 135 (100). Anal. (C43H54O4Si2) C,
H.
1-(3-Isop r op yl-4-m et h ylp en t a n oyl)-2-(3-isop r op yl-4-
m et h ylp en t -2-en oyl)-sn -glycer ol (5). This compound has
been previously reported.16
5,5-Bis[(2,2-d im et h yl-1,1-d ip h en yl-1-sila p r op oxy)m e-
th yl]-3,4,5-tr ih yd r ofu r a n -2-on e (10). This compound has
been previously reported.18
5,5-Bis[(2,2-d im et h yl-1,1-d ip h en yl-1-sila p r op oxy)m e-
t h yl]-3-(1-h yd r oxy-isop r op yl)-3,4,5-t r ih yd r ofu r a n -2-
on e (11). Under argon, LiHMDS (3.5 mL, 1 M solution) was
added dropwise to a -78 °C solution containing 10 (2.0 g, 3.2
mmol) in THF (8 mL) and stirred at -78 °C for 1 h, after which
acetone (250 µL, 3.5 mmol) was added slowly. After 21 min,
the reaction was diluted with Et2O (25 mL) and quenched with
saturated aqueous NH4Cl. The phases were separated, and
the aqueous phase was extracted with Et2O (3 × 25 mL). The
combined organic phases were dried over MgSO4 and concen-
trated to give 11, which was used directly without further
purification.
(Z)-5,5-Bis[(2,2-d im eth yl-1,1-d ip h en yl-1-sila p r op oxy)-
m eth yl]-3-[4-m eth yl-3-(m eth yleth yl)p en tylid en e)-4,5-d i-
h yd r ofu r a n -2-on e (17Z). This compound has been previously
reported.15

1578 J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 9
Sigano et al.
(E)-5,5-Bis[(2,2-d im eth yl-1,1-d ip h en yl-1-sila p r op oxy)-
m eth yl]-3-[4-m eth yl-3-(m eth yleth yl)p en tylid en e)-4,5-d i-
h yd r ofu r a n -2-on e (17E). This compound has been previously
reported.15
Gen er a l P r oced u r e for Desilyla tion . Under argon, HF/
pyridine (4.0 equiv, 70% solution) was added dropwise to a 0
°C stirring solution of lactone (1.0 equiv) in CH2Cl2 (2 mL/
mmol) and the reaction was allowed to warm to room tem-
perature overnight. The reaction was quenched by adding solid
NaHCO3, filtered, concentrated, and purified by column chro-
matography.
5,5-Bis(h yd r oxym et h yl)-3-(m et h ylet h ylid en e)-4,5-d i-
h yd r ofu r a n -2-on e (13). Yield: 0.549 g, 100%, white crystals;
mp 89-90 °C. IR (neat): 3379 (OH), 2882 (CH,), 2826 (CH),
1713 (CdO), 1660 (CdC) cm-1. 1H NMR (300 MHz, CDCl3): δ
1.95 (s, 3 H, CdCMe2), 2.31 (s, 3 H, CdCMe2), 2.79 (br s, 2 H,
CH2OH), 3.28 (s, 2 H, lactone-CH2), 3.78 (app q, J ) 12.0 Hz,
4 H, CH2OH). 13C NMR (100 MHz, CDCl3): δ 19.89, 24.52,
31.61, 64.72, 83.57, 119.47, 151.47, 170.16. FAB-MS (m/z,
relative intensity): 187 (MH+, 100). Anal. (C9H4O4) C, H, O.
(Z)-Bis(h yd r oxym eth yl)-3-(3-m eth ylbu tylid en e)-4,5-d i-
h yd r ofu r a n -2-on e (18Z). Yield: 0.306 g, 78%. This compound
was obtained as a colorless oil, which contained a small
amount of the E isomer. Because the E and Z isomers were
inseparable, the following step was carried out with this
mixture.
Cl2 (8.6 mL) containing pyridine (129 µL) to give isolated 7Z
as a colorless oil (0.041 g, 22%) along with an unseparated
mixture of 7E/Z as a pale oil (0.077 g, 41%) and diacylated
product (ca. 25%, not isolated). 7Z: IR (neat): 3465 (OH), 2958
(CH), 2872 (CH), 1740 (CdO), 1670 (CdC) cm-1. 1H NMR (400
MHz, CDCl3): δ 0.80 (d, J ) 6.8 Hz, 6 H, C(O)CH2CH-
(CHMe2)2), 0.89 (d, J ) 6.8 Hz, 6 H, C(O)CH2CH(CHMe2)2),
0.93 (d, J ) 6.6 Hz, 6 H, CdCHCH2CHMe2), 1.58 (app pentet,
1 H, CdCHCH2CHMe2), 1.67-1.77 (m, 3 H, C(O)CH2CH-
(CHMe2)2), 2.19 (d, J ) 5.7 Hz, 2H, C(O)CH2CH(CHMe2)2), 2.38
(t, J ) 6.8 Hz, 1H, CH2OH), 2.59-2.63 (m, 2 H, CdCHCH2-
CHMe2), 2.69-2.75 (overlapping dtd, J gem ) 16.6 Hz, 1 H,
lactone-CH2), 2.88-2.94 (overlapping dtd, J gem ) 16.4 Hz, 1
H, lactone-CH2), 3.62 (dd, J ) 12.1, 6.6 Hz, 1 H, CH2OH), 3.70
(dd, J ) 12.2, 7.0 Hz, 1 H, CH2OH), 4.19 (AB quartet, J )
11.9 Hz, 2 H, CH2OC(O)CH2CH(CHMe2)2), 6.26 (tt, J ) 7.8,
2.3 Hz, 1H, C ) CHCH2CHMe2). 13C NMR (100 MHz, CDCl3):
δ 18.67, 21.30, 22.26, 22.28, 28.59, 29.34, 32.75, 39.10, 36.32,
46.84, 64.66, 65.34, 82.30, 124.47, 144.42, 168.68, 174.72. FAB-
MS (m/z, relative intensity): 355 (MH+, 47). Anal. (C20H34O5)
C, H.
(E)-[2-(Hyd r oxym eth yl)-4-(3-m eth ylbu tylid en e)-5-oxo-
2-2,3-d ih yd r ofu r yl]m eth yl-4-m eth yl-3-(m eth yleth yl)p en -
ta n oa te (7E). Yield: 0.045 g, 54%, white solid; mp 32-33 °C.
IR (neat): 3462 (OH), 2958 (CH), 2873 (CH), 1741 (CdO), 1681
1
(CdC) cm-1. H NMR (400 MHz, CDCl3): δ 0.79 (dd, J ) 6.7,
1.1 Hz, 6 H, C(O)CH2CH(CHMe2)2), 0.88 (d, J ) 6.8 Hz, 6 H,
C(O)CH2CH(CHMe2)2), 0.94 (d, J ) 6.6 Hz, 6 H, CdCHCH2-
CHMe2), 1.54-1.60 (m, 1 H, C(O)CH2CH(CHMe2)2), 1.68-1.86
(m, 3 H, CdCHCH2CHMe2 and C(O)CH2CH(CHMe2)2), 2.04-
2.08 (m, 2 H, CdCHCH2CHMe2), 2.18 (d, J ) 5.9 Hz, 2 H,
C(O)CH2CH(CHMe2)2), 2.50 (t, J ) 6.4 Hz, 1H, CH2OH), 2.62-
2.68 (overlapping dtd, J gem ) 17.0 Hz, 1 H, lactone-CH2), 2.79-
2.85 (overlapping dtd, J gem ) 17.2 Hz, 1 H, lactone-CH2), 3.64
(dd, J ) 12.1, 6.4 Hz, 1 H, CH2OH), 3.72 (dd, J ) 12.2, 6.7 Hz,
1 H, CH2OH), 4.20 (AB quartet, J ) 11.9 Hz, 2 H, CH2OC-
(O)CH2CH(CHMe2)2), 6.76-6.81 (m, 1H, C ) CHCH2CHMe2).
13C NMR (100 MHz, CDCl3): δ 18.67, 21.29, 21.30, 22.40,
28.09, 29.33, 29.34, 29.91, 32.74, 39.31, 46.85, 64.81, 65.48,
83.07, 126.62, 140.83, 169.87, 174.68. FAB-MS (m/z, relative
intensity): 355 (MH+, 93), 57 (100). Anal. (C20H34O5) C, H.
(E)-Bis(h yd r oxym eth yl)-3-(3-m eth ylbu tylid en e)-4,5-d i-
h yd r ofu r a n -2-on e (18E). Yield: 0.043 g, 80%, white crystals;
mp 64-65 °C. IR (neat): 3303 (OH), 2957 (CH), 2883 (CH),
1748 (CdO), 1677 (CdC) cm-1. 1H NMR (400 MHz, CDCl3): δ
0.88 (d, J ) 6.6 Hz, 6 H, CdCHCH2CHMe2), 1.74 (apparent
septet, J ) 6.7 Hz, 1 H, CdCHCH2CHMe2), 1.98-2.02 (m, 2
H, CdCHCH2CHMe2), 2.64-2.65 (m, 2 H, lactone-CH2), 3.60
(dd, J ) 12.1, 6.2 Hz, 2 H, CH2OH), 3.69 (dd, J ) 12.3, 6.1 Hz,
2 H, CH2OH), 4.09 (t, J ) 6.2 Hz, 2 H, CH2OH), 6.62-6.67
(m, 1H, C ) CHCH2CHMe2). 13C NMR (100 MHz, CDCl3): δ
22.36, 28.01, 29.43, 39.20, 64.65, 85.77, 127.24, 140.72, 171.15.
FAB-MS (m/z, relative intensity): 215 (MH+, 100). Anal.
(C11H18O4) C, H.
(Z)-5,5-Bis(h yd r oxym eth yl)-3-[4-m eth yl-3-(m eth yleth -
yl)p en tylid en e]-4,5-d ih yd r ofu r a n -2-on e (19Z). This com-
pound has been previously reported.15
(Z)-{2-(H yd r oxym et h yl)-4-[4-m et h yl-3-(m et h ylet h yl)-
p en t ylid en e]-5-oxo-2-2,3-d ih yd r ofu r yl}m et h yl-3-m et h -
ylbu ta n oa te (9Z). Under argon, Bu2SnO (0.075 g, 0.30 mmol)
was added to a solution of 19Z (0.054 g, 0.20 mmol) in PhMe
(5.0 mL) containing 4 Å molecular sieves and heated to reflux.
After 3 h, the reaction was cooled to 0 °C, isovaleryl chloride
(27 µL) was added, and the reaction was allowed to warm to
room temperature overnight. The reaction mixture was then
quenched with pH 7 buffer solution, extracted with CHCl3
(3×), dried over MgSO4, and concentrated. Purification by
silica gel chromatography gave 9Z as a colorless oil (0.013 g,
18%). IR (neat): 3545 (OH), 2958 (CH), 2927 (CH), 1737 (Cd
O), 1671 (CdC) cm-1. 1H NMR (400 MHz, CDCl3): δ 0.83 (dd,
J ) 6.7, 1.5 Hz, 6 H, CdCHCH2CH(CHMe2)2), 0.87 (dd, J )
6.8, 1.4 Hz, 6 H, CdCHCH2CH(CHMe2)2), 0.92 (d, J ) 6.6 Hz,
6 H, C(O)CH2CHMe2), 1.08 (pentet, J ) 5.5 Hz, 1 H, Cd
CHCH2CH(CHMe2)2), 1.69-1.80 (m, 2 H, CdCHCH2CH-
(CHMe2)2), 2.00-2.10 (m, 1 H, C(O)CH2CHMe2), 2.18 (d, J )
6.8 Hz, 2 H, C(O)CH2CHMe2), 2.64-2.71 (m, 3 H, CdCHCH2-
CH(CHMe2)2 and lactone-CH2), 2.82-2.88 (m, 1 H, lactone-
CH2), 3.62 (AB quartet, J ) 12.2 Hz, 2 H, CH2OH), 4.18 (AB
quartet, J ) 11.9 Hz, 2 H, CH2OC(O)CH2CHMe2), 6.20-6.26
(m, 1H, C ) CHCH2CH(CHMe2)2). 13C NMR (100 MHz,
CDCl3): δ 19.61, 21.82, 22.54, 25.80, 26.53, 29.54, 33.41, 43.27,
51.39, 64.89, 65.29, 82.37, 112.81, 148.08, 168.62, 173.12. FAB-
MS (m/z, relative intensity): 355 (MH+, 100). Anal. (C20H34O5)
C, H.
(E)-5,5-Bis(h yd r oxym eth yl)-3-[4-m eth yl-3-(m eth yleth -
yl)p en tylid en e]-4,5-d ih yd r ofu r a n -2-on e (19E). This com-
pound has been previously reported.15
Gen er a l P r oced u r e for Acyla tion for Com p ou n d s 6
a n d 7E/Z. Under argon, the acid chloride (1.5 equiv) was
added to a 0 °C stirring solution of lactone (1.0 equiv) and
pyridine (3.0 equiv) in CH2Cl2 (16 mL/mmol). The reaction was
monitored by TLC and then concentrated in vacuo. Purification
by column chromatography gave the desired product along
with a minor amount of the diacylated product.
[2-(Hyd r oxym eth yl)-4-(m eth yleth ylid en e)-5-oxo-2-2,3-
d ih y d r o fu r y l]m e t h y l-4-m e t h y l-3-(m e t h y le t h y l)p e n -
ta n oa te (6). Yield: 0.059 g, 56%. IR (neat): 3470 (OH), 2959
(CH), 2875 (CH), 1740 (CdO), 1666 (CdC) cm-1. 1H NMR (400
MHz, CDCl3): δ 0.79 (dd, J ) 6.7, 0.9 Hz, 6 H, C(O)CH2CH-
(CHMe2)2), 0.88 (d, J ) 6.8 Hz, 6 H, C(O)CH2CH(CHMe2)2),
1.55 (pentet, J ) 5.8 Hz, 1 H, C(O)CH2CH(CHMe2)2), 1.69-
1.77 (m, 2 H, C(O)CH2CH(CHMe2)2), 1.86 (s, 3 H, CdCMe2),
2.19 (d, J ) 5.9 Hz, 2 H, lactone CH2), 2.24 (app t, J ) 2.2 Hz,
3 H, CdCMe2), 2.61-2.67 (m, 2 H, CCH2OC(O)CH2CH-
(CHMe2)2 and OH), 2.79-2.85 (m, 1 H, C(O)CH2CH(CHMe2)2),
3.61 (dd, J ) 12.0, 4.6 Hz, 1 H, CH2OH), 3.69 (dd, J ) 11.9,
5.3 Hz, 1 H, CH2OH), 4.18 (AB quartet, J ) 11.9 Hz, 2H, C(O)-
CH2CH(CHMe2)2). 13C NMR (100 MHz, CDCl3): δ 18.66, 19.91,
21.26, 24.54, 29.33, 32.03, 32.80, 46.87, 64.85, 65.54, 81.19,
119.00, 151.54, 169.05, 174.72. FAB-MS (m/z, relative inten-
sity): 327 (MH+, 100). Anal. (C18H30O5‚0.25H2O) C, H.
(Z)-[2-(Hyd r oxym eth yl)-4-(3-m eth ylbu tylid en e)-5-oxo-
2-2,3-d ih yd r ofu r yl]m eth yl-4-m eth yl-3-(m eth yleth yl)p en -
ta n oa te (7Z). According to the general procedure, 4-methyl-
3-(methylethyl)pentanoyl chloride15 was added to a solution
of 18 (0.114 g, 1.6 mmol, mixture of E and Z isomers) in CH2-
(E)-{2-(H yd r oxym et h yl)-4-[4-m et h yl-3-(m et h ylet h yl)-
p en t ylid en e]-5-oxo-2-2,3-d ih yd r ofu r yl}m et h yl-3-m et h -
ylbu ta n oa te (9E). Under argon, Bu2SnO (0.433 g, 1.7 mmol)
was added to a solution of 19E (0.314 g, 1.2 mmol) in PhMe
(20 mL) containing 4 Å molecular sieves and heated to reflux.
After 2 h, the reaction was cooled to 0 °C, isovaleryl chloride

Binding Modes of DAG and DAG Lactones to PK-C
J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 9 1579
measure of the effect of constraining the glycerol backbone in
DAG lactones. Bioorg. Med. Chem. Lett. 2000, 10, 653-655.
(15) Nacro, K.; Bienfait, B.; Lee, J .; Han, K.-C.; Kang, J .-H.; Benzaria,
S.; Lewin, N. E.; Bhattacharyya, D. K.; Blumberg, P. M.;
Marquez, V. E. Confromationally Constrained Analogues of
Diacylglycerol (DAG). 16. How Much Structural Complexity is
Necessary for Recognition and High Binding Affinity to Protein
Kinase C? J . Med. Chem. 2000, 43, 921-944.
(156 µL) was added, and the reaction was allowed to warm to
room temperature over the weekend. The reaction mixture was
then quenched with pH 7 buffer solution, extracted with CHCl3
(3×), dried over MgSO4, and concentrated. Purification by
silica gel chromatography gave 9E as a colorless oil (0.085 g,
21%). IR (neat): 3447 (OH), 2958 (CH), 2873 (CH), 1740 (Cd
1
O), 1676 (CdC) cm-1. H NMR (400 MHz, CDCl3): δ 0.81 (d,
(16) Nacro, K.; Sigano, D. M.; Yan, S.; Nicklaus, M.; Pearce, L. L.;
Lewin, N. E.; Garfield, S. H.; Blumberg, P. M.; Marquez, V. E.
An optimized protein Kinase C Activating Diacylglycerol Com-
bining High Binding Affinity (Ki) with Reduced Lipophilicity (log
P). J . Med. Chem. 2001, 44, 1892-1904.
J ) 6.9 Hz, 6 H, CdCHCH2CH(CHMe2)2), 0.87 (dd, J ) 6.9,
0.9 Hz, 6 H, CdCHCH2CH(CHMe2)2), 0.91 (d, J ) 6.5 Hz, 6
H, C(O)CH2CHMe2), 1.14-1.21 (m, 1 H, CdCHCH2CH-
(CHMe2)2), 1.70-1.79 (m, 2 H, CdCHCH2CH(CHMe2)2), 1.97-
2.11 (m, containing 1 H, C(O)CH2CHMe2 and 2 H, Cd
CHCH2CH(CHMe2)2), 2.17 (d, J ) 0.9 Hz, 2 H, C(O)CH2CHMe2),
2.60-2.66 (overlapping dtd, J gem ) 17.0 Hz, 1 H, lactone-CH2),
2.76-2.82 (overlapping dtd, J gem ) 17.1 Hz, 1 H, lactone-CH2),
3.65 (AB quartet, J ) 12.2 Hz, 2 H, CH2OH), 4.19 (AB quartet,
J ) 11.9 Hz, 2 H, CH2OC(O)CH2CHMe2), 6.75-6.80 (m, 1H,
(17) For compound 6: Hanessian, S.; Abad-Grillo, T.; McNaughton-
Smith, G. Synthesis of (4S)-hydroxymethyl-(2R)-(2-propyl)bu-
tyrolactone:
a quest for a practical route to an important
hydroxyethylene isostere chiron. Tetrahedron 1997, 53, 6281-
6294.
(18) Sharma, R.; Lee, J .; Wang, S.; Milne, G. W. A.; Lewin, N. E.;
Blumberg, P. M.; Marquez, V. E. Conformationally Constrained
Analogues of Diacylglycerol.10. Ultrapotent Protein Kinase C
C ) CHCH2CH(CHMe2)2). 13C NMR (100 MHz, CDCl3):
δ
Ligands Based on
a Racemic 5-Disubstituted Tetrahydro-2-
19.49, 19.54, 21.76, 21.82, 22.53, 25.79, 28.90, 29.40, 30.15,
43.24, 50.51, 65.01, 65.48, 83.20, 125.04, 144.17, 170.19,
173.05. FAB-MS (m/z, relative intensity): 355 (MH+, 100).
Anal. (C18H30O5‚0.5H2O) C, H.
furanone Template. J . Med. Chem. 1996, 39, 19-28.
(19) Bo¨gi, K.; Lorenzo, P. S.; AÅ cs, P.; Sza´lla´si, Z.; Wagner, G. S.;
Blumberg, P. M. Comparison of the roles of the C1a and C1b
domains of protein kinase C alpha in ligand induced transloca-
tion in NIH 3T3 cells. FEBS Lett. 1999, 456, 27-30.
(20) Meylan, W. M.; Howard, Ph. H. Atom Fragment Contribu-
tion Method for Estimating Octanol-Water Partition Coef-
ficients. KOWWIN 1.63, Syracuse Research Corp.; http://
esc.syrres.com.20. J . Pharm. Sci. 1995, 84, 83-92.
Refer en ces
(1) Newton, A. C. Protein Kinase C: Structure, Function and
Regulation. J . Biol. Chem. 1995, 270, 28495-28498.
(2) Quest, A. F. G. Regulation of protein kinase C: A tale of lipids
and proteins. Enzyme Protein 1996, 49, 231-261.
(3) Mellor, H.; Parker, P. J . The extended protein kinase C super-
family. Biochem. J . 1998, 332, 281-292.
(4) Ron, D.; Kazanietz, M. G. New insights into the regulation of
protein kinase C. and novel phorbol ester receptors. FASEB J .
2000, 13, 1658-1676.
(5) Castagna, M.; Takai, Y.; Kaibuchi, K.; Sano, K.; Kikkawa, U.;
Nishizuka, Y. Direct activation of calcium-activated, phospho-
lipid-dependent protein kinase by tumor-promoting phorbol
esters. J . Biol. Chem., 1982, 257, 7847-7851.
(6) Newton, A. C. Protein Kinase C: Structural and spatial regula-
tion by phosphorylation, cofactors, and macromolecular interac-
tions. Chem. Rev., 2001, 101, 2353-2364.
(7) Sakai, N.; Sadaki, K.; Ikegaki, N.; Shirai, Y.; Ono, Y.; Saito, N.
Direct visualization of the translocation of the γ-subspecies of
protein kinase C in living cells using fusion proteins with green
fluorescent protein. J . Cell Biol. 1997, 139, 1465-1476.
(8) Oancea, E.; Teruel, M. N.; Quest, A. F. G.; Meyer, T. Green
fluorescent protein (GFP)-tagged cysteine-rich domains from
protein kinase C as fluorescent indicators for diacylgylcerol
signaling in living cells. J . Cell Biol. 1998, 140, 485-498.
(9) Newton, A. C.; Keranen, L. M. Phosphatidyl-L-serine is necessary
for protein kinase C’s high affinity interaction with diacylglyc-
erol-containing membranes. Biochemistry 1994, 33, 661-6658.
(10) Newton, A. C. Regulation of protein kinase C. Curr. Opin. Cell.
Biol. 1997, 9, 161-167.
(11) Nishizuka, Y. Intracelluar signaling by hydrolysis of phospho-
lipids and activation of protein-kinase-C. Science 1992, 258,
607-614.
(12) Oancea, E.; Meyer, T. Protein kinase C as a molecular machine
for decoding calcium and diacylglycerol signals. Cell 1998, 95,
307-318.
(13) Zhang, G. G.; Kazanietz, M. G.; Blumberg, P. M.; Hurley, J . H.
Crystal structure of the cys2 activator-binding domain of protein
kinase C delta in complex with phorbol ester. Cell 1995, 81, 917-
924.
(14) Nacro, K.; Bienfait, B.; Lewin, N. E.; Blumberg, P. M.; Marquez,
V. E. Diacylglycerols with lipophilically equivalent branched acyl
chains display high affinity for protein kinase C (PK-C). A direct
(21) Lee, J .; Shaomeng Wang, S.; Milne, G. W. A.; Sharma, R.; Lewin,
N. E.; Blumberg, P. M.; Marquez, V. E. Conformationally
Constrained Analogues of Diacylglycerol. 11.1 Ultrapotent Pro-
tein Kinase
C Ligands Based on a Chiral 5-Disubstituted
Tetrahydro-2-furanone Template. J . Med. Chem. 1996, 39, 29-
35.
(22) Rarey, M.; Kramer, B.; Lengauer, T.; Klebe, G. A fast flexible
docking method using an incremental construction algorithm.
J . Mol. Biol. 1996, 261, 470-489.
(23) J ones, G.; Willett, P.; Glen, R. C. Molecular recognition of
receptor sites using a genetic algorithm with a description of
desolvation. J . Mol. Biol. 1995, 245, 43-53.
(24) Benzaria, S.; Bienfait, B.; Nacro, K.; Wang, S.; Lewin, N. E.;
Beheshti, M.; Blumberg, P. M.; Marquez, V. E. Conformationally
constrained analogues of diacylglycerol (DAG). 15. The indis-
pensable role of the sn-1 and sn-2 carbonyls in the binding of
DAG-lactones to protein kinase C (PK-C). Bioorg. Med. Chem.
Lett. 1998, 8, 3403-3408.
(25) Numbering of the PK-C residues is internal to the C1 domain
(e.g., Gly 23 corresponds to Gly 253 in PK-Cδ C1b, Gly 59 in
PK-CR C1a, and Gly 124 in PK-CR C1b).
(26) Insight II and QUANTA 97 are products of Accelrys Inc., 9685
Scranton Road, San Diego, CA 92121-3752 (www.accelrys.com).
(27) Bower, M.; Cohen, F. E.; Dunbrack, R. L., J r. Side chain
placement from a backbone-dependent rotamer library: A new
tool for homology modeling. J . Mol. Biol. 1997, 267, 1268-1282.
(28) FlexX and SYBYL are products of Tripos Inc., 1699 South Hanley
Road, St. Louis, MO 63144-2913 (www.tripos.com).
(29) Brooks, B. R.; Bruccoleri, R. E.; Olafson, B. D.; States, D. J .;
Swaminnathan, S.; Karplus, M. CHARMM: A program for
macromolecular energy, minimization, and dynamics calcula-
tions. J . Comput. Chem. 1983, 4, 187-217.
(30) Momany, F. A.; Rone, R. Validation of the general purpose
QUANTA 3.2/CHARMm force field. J . Comput. Chem. 1992, 13,
888-900.
(31) Lee, B.; Richards, F. M. The interpretation of protein struc-
tures: Estimation of static accessibility. J . Mol. Biol. 1971, 55,
379-400.
J M020476O