2-(3-oxo-1,3-diphenylpropyl)malonic acids as potent allosteric ligands of the PIF pocket of phosphoinositide-dependent kinase-1: Development and prodrug concept

Article abstract of DOI:10.1021/jm3010477

The protein kinase C-related kinase 2 (PRK2)-interacting fragment (PIF) pocket of phosphoinositide-dependent kinase-1 (PDK1) was proposed as a novel target site for allosteric modulators. In the present work, we describe the design, synthesis, and structure-activity relationship of a series of 2-(3-oxo-1,3-diphenylpropyl)malonic acids as potent allosteric activators binding to the PIF pocket. Some congeners displayed AC₅₀ values for PDK1 activation in the submicromolar range. The potency of the best compounds to stabilize PDK1 in a thermal stability shift assay was in the same order of magnitude as that of the PIF pocket binding peptide PIFtide, suggesting comparable binding affinities to the PIF pocket. The crystal structure of PDK1 in complex with compound 4h revealed that additional ionic interactions are mainly responsible for the increased potency compared to the monocarboxylate analogues. Notably, several compounds displayed high selectivity for PDK1. Employing a prodrug strategy, we were able to corroborate the novel mechanism of action in cells.

Full text of DOI:10.1021/jm3010477

Article
pubs.acs.org/jmc
2‑(3-Oxo-1,3-diphenylpropyl)malonic Acids as Potent Allosteric
Ligands of the PIF Pocket of Phosphoinositide-Dependent Kinase-1:
Development and Prodrug Concept
Adriana Wilhelm,† Laura A. Lopez-Garcia,‡ Katrien Busschots,‡ Wolfgang Frohner,† Frauke Maurer,§
̈
Stefan Boettcher,† Hua Zhang,‡ Jorg O. Schulze,‡ Ricardo M. Biondi,‡ and Matthias Engel*,†
̈
†Pharmaceutical and Medicinal Chemistry, Saarland University, Campus C2.3, D-66123 Saarbrucken, Germany
‡Department of Internal Medicine I, University of Frankfurt, Theodor-Stern-Kai 7, D-60590 Frankfurt a.M., Germany
̈
§Organic Chemistry, Saarland University, Campus C2.3, D-66123 Saarbrucken, Germany
̈
S
* Supporting Information
ABSTRACT: The protein kinase C-related kinase 2 (PRK2)-
interacting fragment (PIF) pocket of phosphoinositide-dependent
kinase-1 (PDK1) was proposed as a novel target site for allosteric
modulators. In the present work, we describe the design, synthesis, and
structure−activity relationship of a series of 2-(3-oxo-1,3-
diphenylpropyl)malonic acids as potent allosteric activators binding
to the PIF pocket. Some congeners displayed AC50 values for PDK1
activation in the submicromolar range. The potency of the best
compounds to stabilize PDK1 in a thermal stability shift assay was in
the same order of magnitude as that of the PIF pocket binding peptide
PIFtide, suggesting comparable binding affinities to the PIF pocket. The
crystal structure of PDK1 in complex with compound 4h revealed that
additional ionic interactions are mainly responsible for the increased
potency compared to the monocarboxylate analogues. Notably, several compounds displayed high selectivity for PDK1.
Employing a prodrug strategy, we were able to corroborate the novel mechanism of action in cells.
type II inhibitor diminished only anchorage-independent
growth and cell migration but not normal proliferation of
several tumor cell lines, suggesting that the mode of action of
an inhibitor can largely influence its potential therapeutic
effect.11 With respect to a potential application for the
treatment of diabetes type 2, the requirements for a PDK1
inhibitor would be to inhibit downstream activation of substrate
kinases responsible for desensitization of insulin signaling, in
particular S6K,12 while not affecting the activity of other
substrates that function as downstream mediators of insulin
action. Such a pharmacological profile can hardly be reached by
compounds directed to the ATP binding site, which are
expected to prevent the activation of all substrate kinases,
including AKT/PKB. Whereas evidence from knockout studies
suggests that inhibition of S6K-catalyzed phosphorylation of
the insulin receptor substrate (IRS)1 might lead to attenuation
of a negative feedback loop and sensitize cells to insulin,13−15
PKB plays a pivotal role in mediating glucose uptake and
storage (reviewed in ref 16). Knockdown of PKBβ caused
insulin resistance and severe diabetes.16 Similarly, systemic
RNAi-mediated knockdown of the PDK1 activity by 90%
induced hyperglycemia and hyperinsulinemia in mice, which
INTRODUCTION
■
The Ser/Thr kinase phosphoinositide-dependent kinase-1
(PDK1) is a pivotal regulator constitutively phosphorylating
some substrates and also acting downstream of insulin- and
growth factor receptor-activated phosphatidylinositol-3 (PI3)
kinase. PDK1 phosphorylates and activates at least 23 protein
kinases from the AGC family.1,2 The PDK1 substrates,
including all isoforms of p70 S6 kinase (S6K), p90 ribosomal
S6 kinase (RSK), serum- and glucocorticoid-induced protein
kinase (SGK), PKC isoforms, protein kinase C-related kinase 2
(PRK2), and PKB/AKT, are implicated in various cellular
processes such as metabolism, cell growth, differentiation, and
survival.2−4 Hence, constitutive activation of PDK1 can lead to
cancer and other disorders including inflammation, arthritis,
diabetes, and cardiovascular disease. In the past, PDK1 has
been extensively investigated as an oncology target, while other
potential indications have received little attention so far.
A lot of effort was dedicated to the development of inhibitors
targeting the ATP binding pocket.5−9 However, analysis of
several structural classes of small-molecule PDK1 inhibitors did
not reveal any selective compound, which can be attributed to
their purely ATP-competitive mechanism of action.10,11 Only
recently, an inhibitor binding to the inactive form of PDK1 by
stabilizing a DFG-out conformation in the active site was
reported, which displayed a high selectivity.11 Interestingly, this
Received: July 17, 2012
Published: October 29, 2012
© 2012 American Chemical Society
9817
dx.doi.org/10.1021/jm3010477 | J. Med. Chem. 2012, 55, 9817−9830

Journal of Medicinal Chemistry
Article
Figure 1. Design of the novel allosteric modulator 4a based on the previous scaffolds I and II.
might be due to repression of PKB activation.17 In a further
study, pancreas-specific PDK1 knockout mice developed severe
hyperglycemia and pancreatic hypoplasia.18
the PKCζ phosphorylation state in intact cells was not analyzed
yet.
Of note, the selective effects of PIF pocket blockage in cells
as outlined above are expected to occur with any PIF-pocket
ligands that are either allosteric activators or neutral antagonists
with respect to the influence on catalytic activity. By contrast,
allosteric inhibitors would affect the phosphorylation of all
substrates, including PKB.
In contrast to ATP-competitive inhibitors, allosteric ligands
binding to the so-called PRK2-interacting fragment (PIF)
pocket of PDK1 might allow a more selective modulation of
downstream activations and hence could be explored as
potential agents for the treatment of diabetes type 2. The
PIF pocket is a shallow hydrophobic pocket on the small lobe
of the catalytic domain, which is shared with all members of the
AGC kinase family.19 In PDK1, it serves two functions: (i) as
an allosteric site regulating the catalytic activity, and (ii) as a
docking site for the transient interaction with the C-terminal
hydrophobic motif (HM) peptides of the substrate kinases,
which share the consensus core sequence F-X-X-F-S(P)/T(P)/
D-Y (X: any amino acid; P: phosphate group).
Most inactive substrate kinases of PDK1 depend on binding
of their C-terminal HM peptide to the PIF pocket in order to
become phosphorylated and activated, according to point (ii)
above.2,20 Therefore, compounds directed to the PIF pocket are
expected to compete with the essential docking of the
substrate's HM peptides, thus preventing the activation of
many substrates in the insulin signaling pathway, including S6K.
In this regard, PIF pocket ligands are predicted to function as
inhibitors of protein−protein interactions between the PIF
pocket and the HM peptides. Interestingly, the PIF pocket
meets the criteria as identified by Bogan and Thorn for
druggable hot spots in protein−protein interfaces, which were
characterized as being enriched by hydrophobic aromatic
residues, and basic residues, in particular arginine.21
As an important exception in the PDK1 pathway, the
activation of PKB is not dependent on the PIF pocket.20
Rather, the phosphorylation of PKB is triggered by
colocalization with PDK1 at the membrane, mediated by
binding of their pleckstrin homology (PH) domains to
phosphatidylinositol (3,4,5)-trisphosphate.22,23 Thus, com-
pounds directed to the PIF pocket of PDK1 should not affect
the activation of PKB, compatible with a potential application
for the treatment of diabetes type 2.
Of concern is rather the potential inhibition of PKCζ, which
was reported to enhance glucose transport in skeletal muscle,
and its activation was shown to be impaired in diabetes type
2.24−26 PKCζ is phosphorylated by PDK1 at the activation loop
residue Thr410.27 In agreement with the postulated PIF-
pocket-dependent mechanism of activation, expression of
glutathione-S-transferase (GST)-PIFtide, which competes
with binding of the PKCζ HM peptide to the PIF pocket,
prevented the T-loop phosphorylation and activation of PKCζ
in HEK293 cells.28 However, the phosphorylation of the PKCs
by PDK1 is believed to initiate the maturation process after
biosynthesis, leading to a PKC species constitutively
phosphorylated at the activation loop, obviating the need for
PDK1 activity at later stages of the protein life span.29 So far,
the effect of a short-term blocking of the PDK1 PIF pocket on
In our previous work, we published the first small molecules
that were able to allosterically activate PDK1 by binding to the
PIF pocket.30 Furthermore, we described the structure−activity
relationship (SAR) of a larger series of 3,5-diphenylpent-2-
enoic acids as PIF-pocket-directed PDK1 activators, providing
evidence that the PIF pocket might be a druggable site.31 The
binding mode of one of these compounds and part of the
allosteric mechanism were characterized by methods in solution
and using X-ray analysis of a cocrystal structure with PDK1.32
Further compounds suggested to bind to the PIF pocket were
discovered by other groups employing different techniques
including NMR-based fragment screening, molecular docking,
and ultrahigh throughput screening.33−35 While most small
molecule PIF-pocket ligands were activators of the catalytic
activity, Sadowsky et al. reported small fragments that acted as
allosteric inhibitors after covalent disulfide linkage to a mutated
Cys residue at the PIF-pocket border.36
The affinity of the most potent HM peptide known so far,
PIFtide (a 24-mer deriving from the PDK1 substrate PRK2),
was determined in the range 43−90 nM (Kd value).30,37
However, the affinity of most reversible ligands described
hitherto was rather low, in the micromolar Kd range, which
appeared insufficient to effectively compete with binding of the
substrate HM peptides to the PIF pocket in cells. Wei et al.34
reported a series of PDK1 activators exhibiting an up to 5 times
greater potency in displacing PIFtide from PDK1 protein than
compound I (Figure 1). However, this finding was obscured by
unexpectedly high AC50 values toward the PDK1 preparation
used (>15 μM). Thus, the potential efficacy toward native
PDK1 in a cellular setting remains unclear. In addition to
strong binding affinity, high selectivity is another prerequisite to
perform detailed cellular and in vivo studies for a proof-of-
principle; selectivity data were only reported for a few PDK1
PIF-pocket ligands hitherto. The objective of the present study
was therefore to develop high affinity ligands for the PDK1 PIF
pocket that were able to compete with endogenous HM
peptides and that were not targeting the PIF pockets of related
kinases from the AGC family. As an additional prerequisite, the
effects of the PIF pocket ligands on the catalytic activity were
preferred to be activating or neutral, in order to prevent
inhibition of PKB.
In the following, we describe the design, synthesis, and SAR
of 2-(3-oxo-1,3-diphenylpropyl)malonic acids as a new class of
PDK1 activators with improved potency and high selectivity
even toward closely related AGC kinases. Furthermore, the
binding mode was elucidated by cocrystallography. In order to
9818
dx.doi.org/10.1021/jm3010477 | J. Med. Chem. 2012, 55, 9817−9830

Journal of Medicinal Chemistry
Article
a
Scheme 1. Synthetic Procedures Used to Prepare the Title Compounds
a
Reagents and conditions: (i) bromomethyl-methylether, NaH, DME, 0 °C, 1 h; (ii) method A: NaOH, EtOH, 1 h, rt; (iii) method B1: diethyl
malonate, MgO, toluene, reflux, 2 h; (iv) method B2: diethyl malonate, K2CO3, ethanol, reflux, 2 h; (v) method B3: diethyl malonate, NaH,
methanol, reflux, 2 h; (vi) 10% HCl, methanol, reflux, 2 h; (vii) method C: NaOH, EtOH, rt, 4 h; (viii) method D: 160−170 °C, 2 h; (ix) method E:
bromomethyl acetate, NEt3, DMF, rt, 4 h. For substituents “Ar” see Table 1.
enable cellular studies, some of the highly polar compounds
were converted into prodrugs. Finally, the effect of one prodrug
on the phosphorylation of the PDK1 substrates S6K, PKCζ,
and AKT/PKB was examined in cells.
Alkaline hydrolysis of the alkyl malonate analogueues under
reflux afforded the corresponding (3-oxo-1,3-diheteroaryl/
arylpropyl)malonic acids 4a−v. Finally, pyrolysis of the
corresponding malonic acid analogues at about 165 °C
provided facile access to the 5-oxo-3,5-diheteroaryl/arylpenta-
noic acids 6a−d, 6f−i, 6k, 6n, 6q−s, and 6u.41 To enable cell
permeation of the highly polar carboxylic acids, carrier prodrug
forms were synthesized. Treatment of the (3-oxo-1,3-
diheteroaryl/arylpropyl)malonic acids 4a−v with bromo
methylacetate and triethylamine in dimethylformamide af-
forded the bis- or monoacetoxymethylester derivatives 5a, 5e,
5g−i, 5l, 5n, 5o, 5q, 5r, 7g, and 7q in modest yields.42
Synthesis of 4m, starting from 4-hydroxyacetophenone,
required protection of the nucleophilic alcohol function. To
this end, the methoxymethyl (MOM) group was introduced
using bromomethyl-methylether with sodium hydride in
dimethylformamide.43 Finally, the MOM group was cleaved
under acidic conditions to give the desired dimethyl 2-(1-(4-
chlorophenyl)-3-(4-hydroxyphenyl)-3-oxopropyl)malonate 3m
followed by hydrolysis to furnish the target molecule 4m.43
Biological Activity. 1,3-Diaryl Malonyl Propanones
Potently Activate PDK1 via Binding to the PIF Pocket. We
analyzed the effect of the malonic acid analogues on the activity
of PDK1 using a radioactive kinase assay exactly as previously
described.30 For comparison, the effect of compound II was
measured in parallel, whereby the published data31 were
essentially reproduced (AC50: 8 μM; Amax: 3-fold). As can be
seen from Table 1, most of the dicarboxyl compounds
including the most comparable analogue 4a prompted a
stronger allosteric activation of PDK1 and displayed lower
AC50 values than II. Several analogues, comprising 4f, 4o, 4q,
4u, and 4v, were found to be even more potent under these
RESULTS AND DISCUSSION
■
Chemistry. In order to optimize potency of previous
compound series, we designed a chimeric compound retaining
the ketone function from compound I combined with the
shorter carboxyl side chain from compound class II (Figure 1).
While the keto group offered potential for additional H-bond
interaction, a further goal was to provide a better mimetic for
the phosphate group, since the natural ligands of the PDK1 PIF
pocket are mostly phosphopeptides. We assigned this role to
the malonic acid moiety, thus yielding compound 4a. Scheme 1
depicts the synthetic route for the new compound classes.
The malonic acid derivatives were prepared starting from the
previously described 1,3-(heteroaryl/aryl)-prop-2-en-1-ones
(chalcones)31 1a−v in three steps. These were accessible by
classical Claisen−Schmidt condensation of benzaldehydes or
heteroarylaldehydes with acetophenones or heteroaryl methyl
ketones. The chalcones were converted to the corresponding
dialkyl 2-(3-oxo-1,3-diheteroaryl-/arylpropyl)malonates via
base-catalyzed Michael addition of CH-acidic malonates. Due
to its mildly basic nature, magnesium oxide was used whenever
possible, ensuring facile product isolation and good activity of
most of the Michael adducts 2a, 2b, 2d, 2h, 2k, 2n, 2p−s, 2u,
and 2v.38 However, because of the limited activity depending
on the solvent and the chalcone, the remaining malonate
derivatives were obtained using alternative catalysts, potassium
carbonate for 2g and sodium hydride for 2c, 2e−j, 2l, 2m, 2o,
and 2t.39,40 Treatment with sodium hydride in methanol caused
transesterification leading to the dimethyl intermediates.
9819
dx.doi.org/10.1021/jm3010477 | J. Med. Chem. 2012, 55, 9817−9830

Journal of Medicinal Chemistry
Article
Table 1. Activity of the 1,3-Diaryl Malonyl Propanones and the Corresponding Monoacids towards PDK1 (Cell-Free Kinase
Activity Assay)
9820
dx.doi.org/10.1021/jm3010477 | J. Med. Chem. 2012, 55, 9817−9830

Journal of Medicinal Chemistry
Article
Table 1. continued
a
b
All compounds were tested as racemates except the 4h eutomer and distomer. Mean value of at least two independent experiments, standard
deviation <20%. Amax: maximum activation of PDK1 compared to DMSO control (=100%); AC50: concentration required for half-maximum
activation; n.e.: no effect.
assay conditions than the PIF pocket-binding peptide PIFtide
(Table 1).
for the interaction between the carboxylate oxygen and the
Gln150 carboxamide group.
To gain insight into the binding mode and the essential
interactions of the dicarboxylic acid compounds with the
protein, we cocrystallized 4h by soaking the compound with
the His-PDK1 50-556 crystal. X-ray analysis revealed that
binding solely occurred in the PIF pocket (Figure 2;
crystallographic details will be published elsewhere44). The
overall binding mode was very similar to the one previously
published for II in complex with PDK1 (PDB code: 3HRF)
(see overlay in Figure 2A). Both phenyl rings occupied the two
hydrophobic subpockets separated by Leu155, whereby ring B
exhibited T-shaped CH−π interactions with Phe157. The
protein crystal specifically selected the S-enantiomer of 4h,
probably driven by the more favorable carboxylic side chain
interactions. One of the two carboxylates occupied a position
similar to that found with II, interacting with Lys76, Thr148,
and Arg131.
The second carboxylate group formed an additional salt
bridge with Arg131 as the sole interaction with the binding site
(Figure 2B). Obviously, the ionic interaction (distance to
Arg131: 3.2 Å) was preferred over the also possible H-bond
formation with Gln150 (distance to Gln150-N: 3.5 Å) and was
mainly responsible for the increase in potency of the
dicarboxylic acid compounds. This observation highlighted
again the importance of Arg131 for high binding affinity.
While the X-ray structure substantiated that 4h exploited
most of the molecular interactions achievable for a small
molecule with the active conformation of the PIF pocket, it did
not provide direct evidence for H-bond formation with the
Gln150 and Thr128 side chains. To establish H-bonding with
at least one of these residues bordering the PIF pocket is
straightforward to further enhance binding affinity without
having to increase considerably the molecular mass and was
therefore another goal of our compound design. At least it was
noticeable that, in spite of the too large distance in the cocrystal
(3.6 A), the ketone of 4h and the Thr128-OH from helix α-C
already appeared appropriately oriented toward each other. It
should be considered that the cocrystal structure appears to
represent one of the most active states of the kinase44 and that
H-bond formation between the ketone and Thr128 may be
relevant at other conformational states. The same could be true
Structure−Activity Relationships of the 1,3-Diaryl Malonyl
Propanones. Although there was obviously a predominant
contribution of the dicarboxyl moiety to the affinity, the activity
could be further improved by optimization of the ring
substituents. A comparison of the test compounds 4a and 4b
(Table 1) suggested a somewhat higher potency of the para-
substituted compound, so we prioritized this position for
systematic variation of further substituents. Within the p-
halogen substituted series, the AC50 improved in the order Cl,
Br, I at comparable 4-fold activation efficacies. The iodine
analogue 4f showed a significantly reduced AC50 in the high
nanomolar range (0.4 μM). A particular enhancement of
potency by iodine was also observed in the homologous set 4c,
4e, and 4g. The increase in affinity according to estimates based
on the AC50 values was only 1.8-fold from 4a (4-Cl) to 4d (4-
Br), but 3.5-fold from 4d to 4f (4-I), and 6-fold from 4e (4-Br)
to 4g (4-I). Altogether, these SARs prompted our speculation
that halogen bonding might be involved, which is most strongly
mediated by iodine.45 It is conceivable that p-iodine might take
the place of the trifluoromethyl group from 4h in the
hydrophobic subpocket and interact with the Lys115 backbone
carbonyl. A similar strong effect of the iodine was observed
previously with allosteric inhibitors of PKCζ that are binding to
the homologous PIF pocket.46,47
Looking at the efficacy, we found that, interestingly, the para-
trifluoromethyl malonic acid derivative 4h activated PDK1 to a
higher degree than PIFtide (5.5-fold vs 3.4-fold activation,
respectively, Table 1). The particular effect of the trifluoro-
methyl substituent might be related to the multiple interactions
with the helix α-B area (Figure 2B): the hydrophobic
interactions with Val124, Ile 118, and Ile119 are complemented
by an orthogonal multipolar interaction with the backbone
carbonyl from Lys115. Although the CO···F distance in the
latter case (3.9 Å) is not among the short distance contacts
observed in several cocrystal structures (between 2.9 and 3.1
Å), it falls into the more abundant range of observed distances,
and the angles between the CF bond and the carbonyl plane
match the average of typical values (CO···F: 102.5°; C
F···O: 128.3°).48 Thus, the orthogonal interaction, albeit weak
9821
dx.doi.org/10.1021/jm3010477 | J. Med. Chem. 2012, 55, 9817−9830

Journal of Medicinal Chemistry
Article
benzoazepin-2-one analogues in the study of Wei et al., where
some compounds achieved a more than 10- to 14-fold
activation of the PDK1 construct used, which was however
accompanied by AC50 values estimated to exceed 15 μM.34
However, within a certain range of maximum activation, there
was no strict correlation with AC50 values both in our present
and previous series of PDK1 activators.31 For instance,
compounds 4g, 4h, and 4s displayed the same AC50 values,
while their activation efficacies were quite different. Presumably,
these compounds bound to and stabilized distinct conforma-
tions of the PIF pocket, while the complex formation in each
case proceeded with similar affinity constants. A comparison of
different PDK1−activator cocrystal structures (PDB codes:
3HRF, complex with PS48; 4AW1, complex with 4h) with the
apo form (PDB code: 3HRC) proves that different activators
indeed stabilize different conformations that possess distinct
catalytic capacities; while the less potent activator PS48 from
our previous study (II in Figure 1) induced a local movement
of the active site Lys111 and ordering of the activation loop,32
only 4h was able to stabilize a more closed active structure of
the catalytic domain of PDK1.44 On the other hand,
compounds 4a, 4f, and 4q activated PDK1 to the same degree,
while their AC50 values were quite different. Presumably, these
compounds allosterically stabilized PDK1 conformations of
similar catalytic activity but exhibited different binding affinities
to the corresponding conformations of the PIF pocket.
Since 4h was endowed with several favorable properties (see
also below), the racemic mixture was separated by chiral high-
performance liquid chromatography (HPLC) to analyze the
biological activity of the individual enantiomers. We noticed
that nearly all of the biological activity was confined to one
enantiomer, which had been eluted from the column with the
higher retention time. Due to limited amounts of material, it
was not possible to determine the absolute configuration for the
two peak fractions. However, the cocrystal structure suggested
that the S-enantiomer is the eutomer, since it was selected by
the crystallized protein from the racemate and presented with a
strong electron density, thus ruling out any mixed occupancy of
the PIF pocket by both enantiomers. The sharp discrimination
can be explained by the interactions visible in the cocrystal that
make clear that only the S-enantiomer of 4h is able to form two
salt bridges with Arg131 (Figure 2).
Additional introduction of a chlorine into phenyl ring B in
the 4-position of the above presented compounds 4a, 4d, 4f,
4h, and 4k abolished the activating effect on PDK1
considerably. Presumably, this halogen in the para position
impaired the edge-to-face interaction with Phe157. In contrast,
the activity did not decrease when introducing a chlorine atom
in the 2-position of phenyl ring B (compound 4j) suggesting
unfilled space around this position that could be exploited to
further increase potency by appropriate substituents. The
complete lack of activity of 4m showed that polar functional
groups are not well tolerated by the subpocket that
accommodates ring A. The 4-phenylthiophene analogue 4p
was synthesized in order to probe whether an elongated
aromatic ring system could be forced to enter deeper regions of
the pocket, for example, by pushing away the phenyl of Phe157,
which apparently did not happen. On the other hand,
condensed aromatic ring systems were generally found to be
more favorable for affinity, as seen with 4o (AC50 = 0.2 μM,
Table 1), and even well tolerated at both ends of the compound
as exemplified by 4s. The positive trend observed for the
naphthyl compound 4q was further corroborated by the
Figure 2. Cocrystal structure of compound 4h complexed with the
PIF- pocket of PDK1. (A) Superimposition of the PIF pockets with 4h
(yellow) and II (cyan) (PDB code: 3HRF) as bound ligands,
respectively. The binding modes of 4h and II present in the cocrystal
structures are largely identical. The phenyl rings of both compounds
bind to the same subpocket areas, and the carboxylate of II exactly
superimposes with one of the 4h carboxyl functions, exhibiting the
same set of interactions. Arg131 in the 4h cocrystal has moved to
establish additional interactions with the second carboxylate. (B)
Interactions of 4h (yellow) with the PIF pocket. Dashed lines in
magenta indicate directed polar interactions comprising ionic
interactions, H-bonding (with Thr148), and fluoro−carbonyl inter-
action (Lys115 carbonyl); distances are given in angstrom. Dotted
lines with distances indicated in black denote potential additional H-
bonding interactions in other conformational states. It is also
noteworthy that the depicted Thr128 rotamer with the hydroxyl
pointing to the 4h keto function is not present in the cocrystal with II
(cf. A). Phenyl ring B (see Figure 1) additionally contributes to the
binding energy by forming edge-to-face CH−π interactions with
Phe157. The image was generated using PyMOL (http://pymol.
sourceforge.net/).
by itself, might increase the affinity synergistically with the
hydrophobic contacts.
In comparison, the 4-ethylphenyl derivative 4k also exhibited
a pronounced activation; however, it was binding with 2-fold
lower affinity as assessed from the AC50. Interestingly, the
highest possible PDK1 activation was not yet reached by 4h
and 4k, since 4n was able to induce a 7-fold increase in PDK1
catalytic activity; however, significantly higher concentrations
were needed to achieve this high activation, as reflected by the
poor AC50 of 7 μM. A similar tendency was observed for the
9822
dx.doi.org/10.1021/jm3010477 | J. Med. Chem. 2012, 55, 9817−9830

Journal of Medicinal Chemistry
Article
phenanthrene derivative 4u and the carbazole analogue 4v,
resulting in potent activators with submicromolar AC50 values.
The second substance class presented here, the monoacid
derivatives (right portion of Table 1), generally exhibited
reduced potency compared with the corresponding dicarboxyl
analogues, corroborating the importance of the second
carboxylate for the enhancement of both affinity and efficacy
of activation.
Obviously, the single negative charge of one carboxyl moiety
was not sufficient to satisfy the two cationic residues
surrounding the PIF pocket, Lys76 and Arg131. Similarly,
phosphotyrosine mimetics had also been reported to benefit
from two negative charges, for example, provided by a malonic
acid residue or by a combination of carboxylate and tetrazole.
Cyclohexylmalonic acid was identified as a potent phospho-
tyrosyl mimetic in a nonpeptidic C-terminal SH2 domain
inhibitor of the Syk tyrosine kinase.49 In the case of growth
factor receptor-bound protein-2, dicarboxyl analogues bound to
the SH2 domain with a strength similar to the phosphonate-
based phosphotyrosyl mimetics.50 However, it remains unclear
whether the malonic acid moiety of the 2-(3-oxo-1,3-
diphenylpropyl)malonic acids 4a−v occupies the same position
as the phosphate of the natural HM phosphopeptides.
Moreover, the matched pairs of di- and monocarboxylate
analogues allowed us to analyze the relative contribution of the
second carboxylate in dependence on further variations at the
aromatic rings. Notably, fused aromatic rings appeared to
increase the CH−π and van der Waals interactions to a level
where the overall binding affinity became less dependent on the
ionic interactions, as exemplified by 6q, 6s, and 6u compared to
their dicarboxylate counterparts 4q, 4s, and 4u. These findings
suggested that it might be possible to combine a bicyclic
aromatic system, especially a more drug-like benzoheterocycle,
as ring B with less acidic moieties, such as sulfonamides,
without a substantial loss of potency.
The hydroxynaphthyl derivative 6r was designed to probe
potential H-bond formation with the Lys115 carbonyl (cf.
Figure 2B). However, the loss of activity compared to 6q
indicated that this either did not happen or did not provide
sufficient energy gain to outbalance the desolvation enthalpy of
the polar hydroxyl group in the mostly hydrophobic subpocket.
It should be noted that the AC50 values given in Table 1 can
only be a rough measure for the relative affinity of the
modulators described here. Although we observed a correlation
with the Kd values in our previous study, the AC50 values of
some of the compounds presented herein approached or went
below the concentration of PDK1 protein in the assay, which
was 0.25 μM. Thus, a linear relationship between AC50 and Kd
could not be expected. Our attempts to accurately measure Kd
values using isothermal calorimetry were hampered by the
strongly exothermic dilution reaction of the dicarboxylic acid
compounds that conferred a high error rate to the much lower
binding enthalpy. As an alternative to compare the binding
affinity of our small molecule ligands with that of PIFtide, we
analyzed the compound-mediated thermal stabilization of
PDK1 using differential scanning fluorimetry (DSF). It was
previously shown that the potency of a kinase inhibitor to
increase the denaturation temperature of the protein structure
was proportional to its binding affinity, both in the case of
ATP-competitive51 and non-ATP-competitive compounds.52
Analyzing the effect of our most potent modulators (AC50 < 1
μM) along with PIFtide on the thermal stabilization of PDK1,
we found that the EC50 values of 4o and 4v were comparable to
that of the 24-mer peptide PIFtide (13.0 and 30.2 μM vs 14.2
μM, respectively, Table 2). However, PIFtide raised the
Table 2. Potencies of Compounds 4f, 4h, 4o, and 4v to
Increase the Thermal Stability of PDK1 in Comparison to
a
PIFtide
b
c
compd no.
EC50 (μM)
ΔTm max (°C)
c(ΔTm 1.5 °C) (μM)
4f
60.0
85.8
13.0
30.2
14.2
5.2
3.9
2.4
4.9
7.3
30.9
67.6
18.9
18.2
3.1
4h
4o
4v
PIFtide
a
Values from one of two independent experiments are shown that
gave essentially similar results. The standard deviation was <15% for all
b
ΔTm values used to calculate the EC50. Maximum shift of thermal
c
stability. Concentration required to increase thermal stability by 1.5
°C.
denaturation temperature to a higher level than any of the
small molecules (ΔTm = 7.3 °C, Table 2), suggesting that
potencies should not be compared solely based on the EC50
values. Therefore, the concentration required to produce a ΔTm
shift of 1.5 °C was calculated as a further parameter. This ΔTm
was chosen because it was located in the linear slope of the
sigmoidal EC50 curve for each compound. Again, 4o and 4v
emerged as the most potent small ligands, both requiring 6-fold
higher concentrations than PIFtide to increase the ΔTm by 1.5
°C. Taken together, the potencies of our best modulators, 4o
and 4v, and PIFtide to enhance the thermal stability to PDK1
were in the same order of magnitude. This suggested that the
affinities to the PIF pocket were also comparable, especially
since the high binding affinity of PIFtide is only reached by
interacting with additional sites outside the PIF pocket. Thus, it
is reasonable to assume that the small molecule ligands are able
to effectively compete with PIFtide and HM peptides with
lower affinity by blocking the “hot spot” PIF pocket.
Compound 4h was included in the thermal stability shift
assay to assess whether the AC50 values might correlate with the
potency to protect against denaturation. Indeed, 4h was found
to be the weakest stabilizator of PDK1, in agreement with the
significantly higher AC50 value of 4h compared to the other
three compounds.
1,3-Diaryl Malonyl Propanones Are Highly Selective for
PDK1. The selectivities of the 2-(3-oxo-1,3-diphenylpropyl)-
malonic acids were evaluated toward a panel of AGC kinases,
which all share a homologous PIF pocket with PDK1. If at all,
there was only a weak influence on the other AGC kinases
examined. Some selectivity data is exemplarily given for 4a and
4h in Table 3. Only at concentrations of about 25-times the
AC50 value for PDK1, 4a and 4h appeared to inhibit S6K by
about 50% and had a marginal activating effect on RSK. These
results demonstrate that, despite the similarity of the
homologous PIF pockets, differences suffice to develop highly
selective modulators.
Since kinases not belonging to the AGC family do not
possess an equivalent binding pocket, it was expected that these
should not be affected by the allosteric modulators. Indeed,
compound 4h did not show any effect on 121 further kinases
representing all branches of the kinome in our parallel study.44
Cell Permeation Studies involving a Prodrug Strategy. All
dicarboxylic acid compounds were inactive in cellular assays
due to the high polarity of the carboxylate groups. In a first
9823
dx.doi.org/10.1021/jm3010477 | J. Med. Chem. 2012, 55, 9817−9830

Journal of Medicinal Chemistry
Article
a
Table 3. Selectivity Profile of Compound II and the Dicarboxyl Compounds 4a and 4h
compd no.
μM
PDK1
S6K1
PKCζ
SGK
PKBα
PKA
PRK2
RSK
MSK
II
2
n.e.
242
228
558
305
637
n.e.
84
84
47
82
46
n.e.
n.e.
n.e.
n.e.
n.e.
n.e.
n.e.
n.e
n.e.
n.e.
n.e.
n.e.
n.e.
n.e.
n.e.
74
n.e.
n.e.
n.e.
n.e.
n.e.
n.e.
n.d.
n.d.
n.e.
160
n.e.
140
n.d.
n.d.
n.e.
n.e.
n.e.
n.e.
50
2
4a
4h
n.e.
n.e.
n.e.
n.e.
n.e.
n.e.
n.e.
80
50
2
50
a
Values show the % catalytic activity of the DMSO-treated control (= 100%); n.e.: no effect; n.d.: not determined. Each value is representative of at
least two independent assays that essentially gave the same results.
Figure 3. Prodrug 5h is efficiently taken up by rat L6 cells and converted into the active compound 4h. (A) Dependency of intracellular formation of
4h on the prodrug concentration. L6 cells were incubated for 2 h with the indicated concentrations of the prodrug 5h and then extracted with
ethylacetate, and the content of 4h was quantified using HPLC-MS/MS. (B) Time course of the intracellular accumulation of 4h and concomitant
disappearance of the prodrug 5h. L6 cells were incubated with 20 μM of the prodrug 5h or of 4h for the indicated times, the cells were extracted, and
⧫
the compounds were quantified. Analysis revealed that 4h (▰) is released from 5h ( ) in a time-dependent manner, whereas 4h itself is not cell-
●
permeable ( ).
attempt to achieve cellular delivery, we utilized the correspond-
ing methyl and ethyl esters of the 2-(3-oxo-1,3-diphenyl-
propyl)malonic acids that were readily available as synthetic
precursors. However, these derivatives displayed no cellular
activity either, suggesting that they were resistant to hydrolysis
by esterases. Hence, we decided to synthesize esters with higher
enzymatic lability; in this regard, acyloxyalkyl esters represent
established prodrug moieties that are in therapeutic use; for
instance, the penicillin derivative pivampicillin exhibits
improved bioavailability, whereas simple alkyl esters of
penicillin lacked therapeutic potential .53−56
We synthesized the bisacetoxymethyl esters 5a, 5h, and 5q
starting from the corresponding malonic acid derivatives 4a, 4h,
and 4q (Scheme 1). In the cell-free assay, these ester derivatives
were completely inactive like all simple alkyl esters that we
tested. However, the para-trifluoromethyl-phenyl prodrug
analogue 5h showed the most promising activity in a
preliminary cellular screening, so we selected this compound
for a detailed analysis of the concentration and time
dependency of prodrug cleavage to identify optimal conditions
for cell assays. Initially, we checked the chemical stability of the
ester prodrug 5h in cell-free medium at 37 °C; under these
conditions, the half-life due to spontaneous hydrolysis was 4.8
h, suggesting that no significant degradation would occur in the
absence of esterases outside the cells.
monitored by mass spectrometry (MS) after extraction of the
compounds under acidic conditions using ethyl acetate.
First, we aimed to analyze the concentration dependency of
the prodrug conversion to the free malonic acid derivative in
the cells. As can be seen from Figure 3A, the rate of conversion
showed little concentration dependency over a wide range.
Only at concentrations higher than 40 μM, saturation of the
intracellular enzyme began to show. For the subsequent
experiment, we chose a prodrug concentration of 20 μM.
The bisacetoxymethyl ester 5h exhibited a high membrane
permeation rate as judged from its appearance inside the cells
already after 5 min (Figure 3B). However, it reached only
moderate concentrations and declined steadily thereafter, most
likely because of the rapid conversion to the free dicarboxylic
acid analogue 4h in the cytoplasm. Based on the measured rate
of 4h production, the rate constant of hydrolysis in the cells was
calculated to be 67 times higher than that of the uncatalyzed
reaction, reducing the half-life of 5h to 4.3 min.
The dicarboxylate compound 4h concomitantly accumulated
in the cells until it reached a plateau after approximately 1 h.
The main reason for this halt of increase was the depletion of
the prodrug compound from the medium. In a further
experiment, we observed that, after one hour, the prodrug
form was degraded to more than 90% in the cell medium. Since
this was considerably faster than the spontaneous hydrolysis,
the degradation was probably caused by esterases leaking from
the cells. The maximum levels of 4h reached in the cells can
roughly be estimated to be 80 μM, if an average cell volume of
3 pL is assumed.58,59 This suggested that the dicarboxylic acid
compound became trapped and was about 4-fold enriched in
the cells.
To evaluate the cellular conversion of the prodrug, the rat L6
skeletal muscle cell line was chosen as a model because it
derives from a potential target tissue for antidiabetic drugs and
displays acetoxymethyl esterase activity in the postfusional
form.57 The intracellular accumulation of the ester prodrug 5h
and the corresponding malonic acid derivative 4h was
9824
dx.doi.org/10.1021/jm3010477 | J. Med. Chem. 2012, 55, 9817−9830

Journal of Medicinal Chemistry
Article
In sharp contrast to the prodrug, the free malonic acid
analogue could not be detected, neither in the soluble nor in
the insoluble cell fractions under the same incubation
conditions (closed circles in Figure 3B, and data not shown).
Having established that 4h was released from its prodrug form,
we analyzed the gross intracellular distribution of 4h in a
further experiment. To this end, treated cells were disrupted by
sonication and centrifuged at high speed. Subsequent HPLC-
MS analysis showed that the product 4h of the enzymatic
conversion was only detectable in the soluble fraction,
excluding accumulation in the membrane (data not shown).
Effects of the Prodrug 5h on the Phosphorylation of S6K,
PKCζ, and PKB in Intact Cells. The effects of 5h on PDK1-
catalyzed phosphorylations were investigated in HEK293 cells.
To analyze the influence on the phosphorylation state of the
activation loops of S6K and PKCζ, the cells were transfected
with expression plasmids encoding GST-S6K[T412E] and
GST-PKCζ, and the recombinant proteins were isolated after
cell treatment with 5h or DMSO and stimulation by insulin-like
growth factor (IGF)-I. Probing of the western blots with
phospho-specific antibodies revealed that 5h decreased the
phosphorylation of the S6K activation loop, while there was no
significant effect on the PKCζ Thr410 phosphorylation (Figure
4A). This result indicated that 4h, which was generated in the
cells, bound to the PIF pocket, thereby preventing phosphor-
ylation of S6K at the activation loop residue Thr229. In
contrast, transient blockage of the PDK1 PIF pocket did not
significantly affect the activation loop phosphorylation of
PKCζ, confirming the hypothesis that PKC is constitutively
phosphorylated by PDK1. Only permanent blockage of the PIF
pocket, for example, as exerted by cellular overexpression of
GST-PIFtide in the study of Balendran et al.,28 would probably
prevent activation loop phosphorylation of newly synthesized
PKCζ. In another experiment, nontransfected HEK293 cells
were identically treated with the prodrug 5h or a DMSO
control followed by IGF-I stimulation, and the cell lysates were
directly used for immunoblotting. The antibodies used here did
not require prior overexpression or pull-down of the proteins.
Upon treatment with 5h, phosphorylation of the ribosomal S6
protein by S6K was strongly inhibited (Figure 4B), indicating
that the reduction of the activation loop phosphorylation by
PDK1 resulted in a diminished catalytic activity of S6K.
Importantly, treatment of the HEK293 cells with 5h did not
significantly inhibit the PDK1-dependent T-loop phosphor-
ylation of PKB, which is in line with similar experiments
performed by us using C2C12 myoblasts.44
CONCLUSIONS
■
The introduction of a second negatively charged carboxylate
resulted in a strong increase of potency of the new PDK1
modulator class presented in this study, yielding several
compounds that allosterically activated PDK1 at submicromolar
concentrations. According to our thermal stability shift assay
results, some of the 2-(3-oxo-1,3-diphenylpropyl)malonic acids
are expected to possess binding affinities to the PIF pocket in a
range comparable to that of PIFtide. Thus, further considering
that the PIF pocket might truly represent a “hot spot” due to its
interaction with the core motif of the HM peptide, the allosteric
ligands should be able to effectively compete with substrate
HM peptides for the binding to the PIF pocket in cells. Indeed,
our results indicated that blockage of the pocket is the major
cellular effect of the compounds binding to the PIF pocket,
which is independent of whether the compound is an activator
of PDK1 in the cell-free assay. The remarkable selectivity of the
compounds for PDK1 confirmed once again that, by targeting
less conserved allosteric sites, modulators even specific for a
single kinase might be developed.
To enable the cellular delivery of the highly polar
compounds, we applied an established prodrug strategy. The
prodrug 5h efficiently reached the cytoplasm and was readily
converted to the active malonic acid derivative 4h. Altogether,
we developed a new series of PIF pocket ligands that are potent
and selective enough to conduct the first detailed validation
studies in cells. The pharmacological profile of allosteric PIF
pocket ligands is expected to be novel and distinct from type I
and type II inhibitors directed to the ATP-binding site. In first
experiments, we analyzed the effects of our prodrug 5h on key
substrates of PDK1 that play major roles in insulin-dependent
signaling, namely, S6K, PKCζ, and PKB. The following short-
term cellular effects of 5h were observed: inhibition of S6K
phosphorylation and activation with no influence on PKCζ and
PKB. Hence, further studies will be designed to evaluate
whether targeting the PIF pocket could become a novel
approach for the treatment of diabetes type 2.
Figure 4. Effect of the prodrug 5h on the PDK1 substrates S6K,
PKCζ, and PKB in intact cells. (A) HEK 293 cells were transfected
with GST-S6K[412E] and GST-PKCζ, incubated with 50 μM 5h for 2
h, and stimulated by IGF-I for 20 min. The recombinant proteins were
isolated and analyzed by immunoblotting using antibodies detecting
the activation loop phosphorylyations. Signals from the same blots
probed with anti-GST antibody were used to normalize the quantified
phosphorylation levels. (B) Serum-starved HEK293 cells were treated
with 50 μM 5h for 2 h and stimulated by IGF-I for 20 min. Cell lysates
were analyzed by immunoblotting using phospho-specific antibodies as
indicated. Antiphospho-S6(Ser235/236) detects S6K-dependent
phosphorylations in the ribosomal S6 protein, and Thr308 is the
PDK1 phosphorylation site in the activation loop of PKB. Normal-
ization of the quantified phosphorylation signals was done after
reprobing the blots with antiactin antibody (not shown). In (A) and
(B), one representative experiment out of two separate experiments is
depicted.
9825
dx.doi.org/10.1021/jm3010477 | J. Med. Chem. 2012, 55, 9817−9830

Journal of Medicinal Chemistry
Article
mL of dichloromethane. The solvent was evaporated, and the crude
product was crystallized from diethyl ether and hexane.
EXPERIMENTAL SECTION
■
Chemical Methods. Solvents and reagents were obtained from
commercial suppliers and were used without further purification. Flash
column chromatography was carried out using silica-gel 40 (35/40−
63/70 μM) with hexane/ethyl acetate mixtures as eluents, and the
reaction progress was determined by thin-layer chromatography
(TLC) analysis on Alugram SIL G/UV254 (Macherey−Nagel).
Visualization was accomplished with UV light. 1H NMR and 13C
spectra were recorded at 500 MHz on a Bruker DRX-500
spectrometer. 1H shifts are referenced to the residual protonated
solvent signal (δ 2.50 for DMSO-d6 and δ 7.26 for CDCl3), and 13C
shifts are referenced to the deuterated solvent signal (δ 39.5 for
DMSO-d6 and δ 77.2 for CDCl3). Chemical shifts are given in parts
per million (ppm), and all coupling constants (J) are given in hertz.
The purities of the title compounds 4a−v, 5a, 5h, 5q, 6a−d, 6f−i, 6k,
6n, 6q−s, and 6u were determined by HPLC coupled with mass
spectrometry and were higher than 95% in all cases.
Mass spectrometric analysis (HPLC-ESI-MS) was performed on a
TSQ quantum (Thermo Electron Corporation) instrument equipped
with an electrospray ionization (ESI) source and a triple quadrupole
mass detector (Thermo Finnigan, San Jose, CA). The MS detection
was carried out at a spray voltage of 4.2 kV, a nitrogen sheath gas
pressure of 4.0 × 105 Pa, an auxiliary gas pressure of 1.0 × 105 Pa, a
capillary temperature of 400 °C, a capillary voltage of 35 V, and source
CID of 10 V. All samples were injected by autosampler (Surveyor,
Thermo Finnigan) with an injection volume of 10 μL. A RP C18
NUCLEODUR 100-3 (125 mm × 3 mm) column (Macherey−Nagel)
was used as the stationary phase. The solvent system consisted of
water containing 0.1% TFA (A) and 0.1% TFA in acetonitrile (B).
HPLC method was as follows: the flow rate was 400 μL/min. The
percentage of B started at an initial 5%, was increased to 100% during
16 min, kept at 100% for 2 min, and flushed back to 5% within 2 min.
All masses were reported as protonated parent ions. Melting points
(mp) were determined in open capillaries on a Mettler FP1 melting
point apparatus and are uncorrected.
Diethyl 2-(3-Oxo-1-phenyl-3-(4-(trifluoromethyl)phenyl)propyl)-
malonate (2h). Synthesized according to method B1 using (E)-3-
phenyl-1-(4-(trifluoromethyl)phenyl)prop-2-en-1-one (1.5 g, 5.42
mmol) and diethyl malonate (0.82 mL, 5.92 mmol) and magnesium
1
oxide (0.54 g); white solid; yield: 1.8 g (76%); H NMR (500 MHz,
(CD3)2SO): δ (ppm) = 0.85 (t, J = 7.2 Hz, 3H), 1.16 (t, J = 7.2 Hz,
3H), 3.46 (dd, J = 3.8, 17.0 Hz, 1H), 3.69 (dd, J = 3.8, 17.0 Hz, 1H),
3.83 (q, J = 7.2 Hz, 2H), 3.90−3.98 (m, 2H), 4.08−4.18 (m, 2H), 7.15
(t, J = 7.6 Hz, 1H), 7.22 (d, J = 7.6 Hz, 2H), 7.29 (d, J = 7.8 6z, 2H),
7.87 (d, J = 8.2 Hz, 2H), 8.05 (d, J = 8.2 Hz, 2H); 13C NMR (125
MHz, (CD3)2SO): δ (ppm) = 13.4, 13.7, 40.4, 56.7, 60.6, 61.2, 116.2,
3
116.4, 125.6 (q, JC−F = 3.7 Hz), 126.8, 127.9, 128.2, 128.5, 140.2,
167.7, 168.4, 197.2.
Method B2: Michael Addition Reaction, for the Preparation of
Diethyl 2-(1-(4-Chlorophenyl)-3-(4-iodophenyl)-3-oxopropyl)-
malonate (2g). (E)-3-(4-chlorophenyl)-1-(4-iodophenyl)prop-2-en-
1-one (2.0 g, 4.07 mmol) and diethyl malonate (0.68 mL, 4.47 mmol,
1.1 equiv) were dissolved in ethanol, and potassium carbonate (1.12 g,
8.14 mmol, 2 equiv) was added. The mixture was heated at reflux for 2
h. The reaction was quenched with H2O (50 mL) and extracted with
CH2Cl2 (3 × 20 mL). The organic layer was washed with brine, dried
with MgSO4, and evaporated under vacuum. The crude product was
crystallized from methylene chloride and hexane; pale beige solid;
yield: 0.86 g (40%); 1H NMR (500 MHz, CDCl3): δ (ppm) = 1.05 (t,
J = 7.2 Hz, 3H), 1.25 (t, J = 7.2 Hz, 3H), 3.35 (dd, J = 9.5, 9.5 Hz,
1H), 3.49 (dd, J = 4.1, 4.1 Hz, 1H), 3.76 (d, J = 9.8 Hz, 1H), 3.98 (q, J
= 7.2 Hz, 2H), 4.09−4.25 (m, 3H), 7.18 (d, J = 8.8 Hz, 2H), 7.22 (d, J
= 8.8 Hz, 2H), 7.59 (d, J = 8.8 Hz, 2H), 7.79 (d, J = 8.8 Hz, 2H); 13C
NMR (125 MHz, CDCl3): δ (ppm) = 13.8, 14.0, 40.1, 42.3, 57.2, 61.5,
61.8, 101.2, 128.6, 129.5, 129.6, 133.0, 133.9, 138.0, 138.8, 167.5,
168.1, 196.6.
Method B3: Michael Addition Reaction, for the Preparation of
2c, 2e, 2f, 2i, 2j, 2l, 2m, 2o, 2t. The corresponding chalcone (1
equiv) and diethyl malonate (1 equiv) were dissolved in methanol, and
sodium hydride (cat.) was added. The mixture was heated at reflux for
2 h. The reaction was quenched with H2O (50 mL) and extracted with
CH2Cl2 (3 × 20 mL). The organic layer was washed with brine, dried
with MgSO4, and evaporated under vacuum. The crude product was
crystallized from methylene chloride and hexane.
Dimethyl 2-(3-(4-Iodophenyl)-1-(naphthalen-2-yl)-3-oxopropyl)-
malonate (2o). Synthesized according to method B3 using (E)-1-(4-
iodophenyl)-3-(naphthalen-2-yl)prop-2-en-1-one (1.5 g, 3.90 mmol),
diethyl malonate (0.59 mL, 3.90 mmol), and sodium hydride (cat.);
pale yellow solid; yield: 1.10 g (55%); 1H NMR (500 MHz, CDCl3): δ
(ppm) = 3.47 (s, 3H), 3.53 (dd, J = 8.8, 16.9 Hz, 1H), 3.59 (dd, J =
4.7, 16.9 Hz, 1H), 3.73 (s, 3H), 3.95 (d, J = 9.1 Hz, 1H), 4.31−4.36
(m, 1H), 7.39 (dd, J = 1.9, 8.5 Hz, 1H), 7.41−7.45 (m, 2H), 7.60 (d, J
= 8.5 Hz, 2H), 7.67 (ds, J = 1.26 Hz, 1H), 7.74−7.77 (m, 3H), 7.78 (d,
J = 8.5 Hz, 2H); 13C NMR (125 MHz, CDCl3): δ (ppm) = 40.8, 42.2,
52.4, 52.7, 57.2, 101.1, 125.9, 126.0, 126.1, 127.6, 127.8, 128.3, 129.5,
132.6, 133.3, 136.0, 137.8, 137.9, 168.0, 168.7, 196.5.
General Synthetic Methods and Experimental Details for
Some Key Compounds. Method A: Claisen−Schmidt Condensa-
tion, for the Preparation of 1a−v. The corresponding benzaldehyde
(1 equiv) was dissolved in EtOH (2 mL per mmol of aldehyde), 3 M
NaOHaq solution (1 mL per mmol of aldehyde) and acetophenone (1
equiv) were added, and the resulting mixture was stirred at rt for 2 h
forming a yellow precipitate. The yellow solid was separated by
vacuum filtration and was washed three times with 10 mL of ice water.
The crude product was purified by recrystallization from MeOH.
(E)-3-Phenyl-1-(4-(trifluoromethyl)phenyl)prop-2-en-1-one (1h).
Synthesized according to method A using benzaldehyde (1.62 mL,
15.95 mmol) and 1-(4′-(trifluoromethyl)phenyl)ethanone (3.0 g,
1
15.95 mmol); pale yellow solid; yield: 3.05 g (70%); H NMR (500
MHz, CDCl3): δ (ppm) = 7.49−7.43 (m, 3H), 7.50 (d, J = 15.8 Hz,
1H), 7.67−7.65 (m, 2H), 7.77 (d, J = 8.0 Hz, 2H), 7.84 (d, J = 15.8
Hz, 1H), 8.10 (d, J = 8.0 Hz, 2H); 13C NMR (125 MHz, CDCl3): δ
1
3
(ppm) = 121.6, 123.6 (q, JC−F = 272.9 Hz), 125.69 (q, JC−F = 4.3
2
Hz), 128.6, 128.8, 129.0, 130.9, 133.9 (q, JC−F = 32.3 Hz), 134.5,
Method C: Hydrolysis, for the Preparation of 4a−v. A solution of
the corresponding malonic ester (2a−l, 2n−v, and 3m) (1 equiv) and
10 M NaOHaq (1−3 mL per mmol malonic ester, as indicated) in
EtOH was refluxed for 4 h. After completion of the reaction, the
cooled mixture was poured into water (30 mL), acidified to pH 2 with
10% HCl, and extracted with ethyl acetate (3 × 20 mL). The ethyl
acetate extracts were collected and further extracted with aqueous
sodium bicarbonate solution (4 × 20 mL). The pH of the bicarbonate
extractions was adjusted to 2 with 10% HCl. White solids formed that
were further extracted with ethyl acetate (3 × 50 mL). The organic
solution was dried over MgSO4 and evaporated to afford a residue
consisting of the free acid, which was purified by crystallization.
2-(3-Oxo-1-phenyl-3-(4-(trifluoromethyl)phenyl)propyl)malonic
acid (4h). Synthesized according to method C using diethyl 2-(3-oxo-
1-phenyl-3-(4-(trifluoromethyl)phenyl)propyl)malonate (1.76 g, 4.03
mmol) and 10 M NaOHaq (6.71 mL); white solid; yield: 1.25 g (82%);
mp 128−130 °C; 1H NMR (500 MHz, CD3OD): δ (ppm) = 3.53 (d, J
141.1, 146.1, 189.7.
(E)-1-(4-Iodophenyl)-3-(naphthalen-2-yl)prop-2-en-1-one (1o).
Synthesized according to method A using 2-naphthaldehyde (1.90 g,
12.19 mmol) and 4′-iodoacetophenone (3.0 g, 12.19 mmol); pale
1
yellow solid; yield: 3.90 g (83%); H NMR (500 MHz, CDCl3): δ
(ppm) = 7.51−7.56 (m, 2H), 7.58 (d, J = 15.6 Hz, 1H), 7.76 (d, J =
8.5 Hz, 2H), 7.76−7.80 (m, 1H), 7.82−7.89 (m, 3H), 7.88 (d, J = 8.5
Hz, 2H), 7.98 (d, J = 15.6 Hz, 1H), 8.04 (s, 1H); 13C NMR (125
MHz, CDCl3): δ (ppm) = 100.6, 121.6, 123.6, 126.8, 127.5, 127.8,
128.7, 128.8, 129.9, 130.8, 132.2, 133.4, 134.5, 137.6, 137.9, 145.5,
189.6.
Method B1: Michael Addition Reaction, for the Preparation of
2a, 2b, 2d, 2h, 2k, 2n, 2p−s, and 2u−v. The corresponding
chalcone (1 equiv) and magnesium oxide (0.1 g per 0.1 mmol of
chalcone) were dissolved in toluene, and diethyl malonate (1 equiv)
was added. The reaction was stirred at rt for 2 h. The magnesium oxide
was separated by vacuum filtration and was washed three times with 10
9826
dx.doi.org/10.1021/jm3010477 | J. Med. Chem. 2012, 55, 9817−9830

Journal of Medicinal Chemistry
Article
= 9.7 Hz 1H), 3.56 (d, J = 6.9 Hz, 1H), 3.83 (d, J = 10.9 Hz, 1H),
4.02−4.08 (m, 1H), 7.14 (t, J = 7.2 Hz, 1H), 7.21 (t, J = 7.6 Hz, 2H),
7.27 (d, J = 7.9 Hz, 2H), 7.75 (d, J = 8.2 Hz, 2H), 8.05 (d, J = 8.2 Hz,
2H); 13C NMR (125 MHz, CD3OD): δ (ppm) = 42.6, 44.3, 58.7,
HPLC-MS/MS Detection of Hydrolysis of the Prodrugs in L6 Cells.
L6 cells were seeded in 10 cm Petri dishes at a density of 106/mL in
Dulbecco’s modified Eagle’s medium (DMEM), supplemented with
penicillin/streptomycin and 10% fetal bovine serum (FBS). After
having reached confluency, the L6 cells were washed with phosphate-
buffered saline (PBS), the medium was replaced by DMEM without
FBS, and the cells were treated with the compounds for appropriate
time intervals (5−90 min). At the end of the incubation periods, the
supernatant was removed, and the cells were washed three times with
cold PBS. The cells were harvested by trypsin/EDTA treatment, the
cell suspensions were transferred to a flask, and the cells per volume
were counted using a Neubauer chamber. At that point, carbamazepine
was added to the suspension as an internal standard to determine the
recovery rate. One mL of 0.6 N HCl in water was added, and the
dicarboxylic acids were extracted using 3 mL of ethyl acetate with
shaking of the samples for 20 min. The samples were centrifuged, and
1 mL aliquots of the ethyl acetate supernatants were taken and
evaporated to dryness. The samples were resuspended in 500 μL of
methanol and used directly for mass spectrometric analysis (HPLC-
ESI-MS/MS). The amounts of the compounds were calculated using a
standard calibration curve. DMSO-treated cells were processed in
parallel and analyzed as a control. The disappearance of the ester and
the appearance of the corresponding dicarboxylic acid were addionally
confirmed by UV peak quantifications and spiking of the samples with
the purified compounds. In the HPLC runs, mobile phase A was water
containing 0.5% trifluoracetic acid, and mobile phase B was methanol
containing 0.5% trifluoracetic acid. From 0 to 8 min, a gradient of 70−
100% B was applied, followed by 100% B until the end; sample volume
was 25 μL, the flow rate was 0.5 mL/min, and the column temperature
was 40 °C.
Cell Transfections, Pull-Down of Recombinant Proteins, and
Immunoblotting. HEK293 cells (ATCC collection) were cultured in
DMEM containing 10% fetal bovine serum (FBS, Gibco)
supplemented with 1% (v/v) antibiotic/antimycotic solution
(Gibco). The cells were transiently transfected at 80% confluency in
6-well plates using pEBG2T plasmids coding for full length
S6K[412E] and PKCζ as GST-fusion proteins. In the S6K construct,
the phosphorylatable residue Thr412 in the hydrophobic motif was
mutated to glutamate in order to mimic a constitutive phosphor-
ylation, which promotes interaction with the PDK1 PIF pocket.
Transfections were performed using polyethyleneimine (PEI, Poly-
sciences Inc., USA), with a PEI/DNA ratio of 10:1 (μL/μg) and 2 μg
of DNA per well. The cells were serum-starved during 24 h before
treatment with compound 5h for 2 h. Then, IGF-I (10 ng/mL) was
added for 20 min, and the cells were lysed. Isolation of the
recombinant proteins was performed using glutathione sepharose
essentially as described previously.30 Aliquots of the isolates were
resolved by sodium dodecyl sulfate polyacrylamide gel electrophoresis
(SDS-PAGE), transferred to nitrocellulose membranes (Protan 0.2
μm, Whatman), and probed using phospho-specific antibodies: anti-
phospho-S6K(Thr229) (cat. no. NSB918, Novus Biologicals) and anti-
phospho-PKCζ/λ(Thr410/403) (cat. no. 9378, Cell Signaling). To
quantify the blotted amounts of the recombinant proteins, the
membranes were also developed using an anti-GST antibody (cat. no.
2622, Cell Signaling). In another experiment, serum-starved HEK293
cells were treated with 5h for 2 h, followed by IGF-I stimulation (10
ng/mL) for 20 min. Lysates from the cells were subject to western
blotting, and the blots were developed using phospho-specific
antibodies: anti-phospho-S6(Ser235/236) (cat. no. 2211, Cell Signal-
ing), which detects the phosphorylation specifically catalyzed by S6K
on the ribosomal S6 protein, and anti-phospho-PKB(Thr308) (cat. no.
9275, Cell Signaling) which is specific for the PDK1-dependent T-loop
phosphorylation of PKB. The quantified phosphorylation signals were
normalized after reprobing the blots with anti-actin antibody (cat. no.
A5441, Sigma−Aldrich).
1
3
125.2 (q, JC−F = 272.2 Hz), 126.7 (q, JC−F = 3.7 Hz), 128.0, 129.3,
2
4
129.6, 129.8, 135.2 (d, JC−F = 32.9 Hz), 141.3, 142.0 (d, JC−F = 1.8
Hz), 171.5, 171.8, 199.2; LC/MS (+ESI): m/z = 381.2 [MH+]; Rt =
4.35 (≥96%).
2-(3-(4-Iodophenyl)-1-(naphthalen-2-yl)-3-oxopropyl)malonic
acid (4o). Synthesized according to method C using dimethyl 2-(3-(4-
iodophenyl)-1-(naphthalen-2-yl)-3-oxopropyl)malonate (0.7 g, 1.36
mmol) and 10 M NaOHaq (4.4 mL); white solid; yield: 0.5 g (75%);
1
mp 150−151 °C; H NMR (500 MHz, (CD3)2SO): δ (ppm) = 3.41
(dd, J = 3.5, 17.02 Hz, 1H), 3.71 (dd, J = 10.4 Hz, 1H), 3.85 (d, J =
10.7 Hz, 1H), 4.02 (dt, J = 3.8, 10.4 Hz, 1H), 7.41−7.46 (m, 2H), 7.50
(dd, J = 1.6, 8.5 Hz, 1H), 7.74−7.77 (m, 2H), 7.79−7.82 (m, 2H),
7.86 (d, J = 8.5 Hz, 21H), 12.84 (s, 2OH); 13C NMR (125 MHz,
(CD3)2SO): δ (ppm) = 40.2, 40.5, 57.3, 101.7, 125.5, 125.8, 126.7,
126.9, 127.2, 127.3, 127.4, 129.5, 131.8, 132.6, 134.4, 135.7, 137.5,
138.7, 168.9, 169.6, 192.4; LC/MS (+ESI): m/z = 489.2 [MH+]; Rt =
5.05 (≥97%).
Method D: Esterification, for the Preparation of 5a, 5h, and 5q.
To a stirred solution of the corresponding malonic acid derivative or
(1 equiv) and NEt3 (5 equiv) in anhydrous DMF (5 mL),
bromomethyl acetate (3 equiv) was added. After stirring for 4 h at
rt, the mixture was hydrolyzed, extracted with ethyl acetate (3 x),
washed with brine, dried over MgSO4, filtered, and concentrated at
reduced pressure. The residue was purified by flash column
chromatography.
Bis(acetoxymethyl) 2-(3-Oxo-1-phenyl-3-(4 (trifluoromethyl)-
phenyl)propyl)malonate (5h). Synthesized according to method D
using 2-(3-oxo-1-phenyl-3-(4-(trifluoromethyl)-phenyl)propyl)-
malonic acid (4h) (0.2 g, 0.52 mmol), bromo methylacetate (0.206
mL, 2.10 mmol), and triethylamine (0.44 mL, 3.12 mmol); white
solid; yield: 0.17 g (62%); mp 89−90 °C; 1H NMR (500 MHz,
CD3OD): δ (ppm) = 1.98 (s, 3H), 2.09 (s, 3H), 3.50 (dd, J = 8.8 Hz,
1H), 3.60 (dd, J = 4.2 Hz, 1H), 3.96 (d, J = 9.1 Hz, 1H), 4.17−4.21
(m, 1H), 5.55 (q, J = 5.7 Hz, 2H), 5.76 (d, J = 5.7 Hz, 2H), 7.18−7.21
(m, 1H), 7.23−7.28 (m, 4H), 7.69 (d, J = 8.2 Hz, 2H), 7.98; 13C NMR
(125 MHz, CD3OD): δ (ppm) = 20.5, 20.6, 40.4, 40.5, 56.5, 79.4,
3
79.8, 118.0, 125.6 (q, JC−F = 4.6 Hz), 127.6, 128.1, 128.4, 128.7,
164.9, 165.8, 166.0, 166.5, 169.2, 169.3, 196.4; LC/MS (+ESI): m/z =
525.3 [MH+]; Rt = 5.31 (≥99%).
Method E: Decarboxylation, for the Preparation of 6a−d, 6f−i,
6k, 6n, 6q, 6s, and 6u. The corresponding malonic acid derivative (1
equiv) was subject to pyrolytic treatment in an oil bath at 160 °C for 1
h. Heating was stopped after the CO2 evolution ceased. The crude
product was dissolved in acetone/methanol and precipitated with
dichloromethane to afford the acid.
5-Oxo-3-phenyl-5-(4-(trifluoromethyl)phenyl)pentanoic acid
(6h). Synthesized according to method E using 2-(3-oxo-1-phenyl-3-
(4-(trifluoromethyl)-phenyl)propyl)malonic acid (4h) (0.10 g, 0.26
1
mmol); white solid; yield: 0.70 g (68%); mp 99−101 °C; H NMR
(500 MHz, CDCl3): δ (ppm) = 2.75 (dd, J = 7.0, 16.1 Hz, 1H), 2.87
(dd, J = 7.0, 16.1 Hz, 1H), 3.36 (dd, J = 6.8, 17.0 Hz, 1H), 3.42 (dd, J
= 6.8, 17.0 Hz, 1H), 3.83−3.89 (m, 1H), 7.20 (tt, J = 1.6, 6.9 Hz, 1H),
7.22−7.31 (m, 4H), 7.69 (d, J = 8.2 Hz, 2H), 7.90 (d, J = 8.2 Hz, 2H);
13C NMR (125 MHz, CDCl3): δ (ppm) = 37.2, 39.9, 44.8, 114.8,
3
115.3, 116.0, 116.5, 125.7 (q, JC−F = 3.7 Hz), 127.1, 127.3, 128.4,
128.8, 175.6, 199.4; LC/MS (+ESI): m/z = 337.2 [MH+]; Rt = 4.80
(≥99%).
Biological Assays. PDK1 Activity Assay. Determination of the
PDK1 activity was performed exactly as described previously.30 In
brief, T308tide (KTFCGTPEYLAPEVRR) was used as the substrate
peptide and the phosphorylation was started by addition of γ32P-ATP/
Mg2+. The phosphorylated peptides were captured on P81
phosphocellulose paper (Whatman) and detected by phosphorimag-
ing. The sequence of the reference effector peptide PIFtide was
REPRILSEEEQEMFRDFDYIADWC.
Differential Scanning Fluorimetry. Protein unfolding was moni-
tored by the increase in the fluorescence of the fluorophor SYPRO
Orange (Invitrogen) using a realtime PCR device (StepOnePlus,
Applied Biosystems) basically following the protocol described by
Niesen et al.60 PDK1 was diluted in 10 mM HEPES pH 7.5 buffer
9827
dx.doi.org/10.1021/jm3010477 | J. Med. Chem. 2012, 55, 9817−9830

Journal of Medicinal Chemistry
Article
containing 150 mM NaCl. Reactions were performed in a 10 μL final
volume in 96-well PCR microtiter plates (Greiner) and contained 1
μM PDK1, 10 mM HEPES pH 7.5, 150 mM NaCl, 1/1000 SYPRO
Orange, and 1 mM dithiotreitol. PIFtide or compounds (1% final
DMSO concentration) were added to this reaction mixture. The
temperature gradient was performed in steps of 0.3 °C in the range
25−70 °C. To calculate the midpoint temperature of transition (Tm)
of PDK1 for each compound concentration, the data were exported to
GraphPad Prism software, and the curves fitted to a Boltzmann
sigmoidal equation with all R2 > 0.998. Subtraction of the value
obtained for the DMSO control (T0) yielded ΔTm values. Each ΔTm
data point reflected the average of a duplicate experiment. The ΔTm
values from different compound concentrations were then used to
calculate EC50 values following to a dose−response curve fitting
procedure embedded in Origin 8 software.
(4) Garcia-Echeverria, C.; Sellers, W. R. Drug discovery approaches
targeting the PI3K/Akt pathway in cancer. Oncogene 2008, 27, 5511−
5526.
(5) Sato, S.; Fujita, N.; Tsuruo, T. Interference with PDK1-Akt
survival signaling pathway by UCN-01 (7-hydroxystaurosporine).
Oncogene 2002, 21, 1727−1738.
(6) Zhu, J.; Huang, J. W.; Tseng, P. H.; Yang, Y. T.; Fowble, J.; Shiau,
C. W.; Shaw, Y. J.; Kulp, S. K.; Chen, C. S. From the cyclooxygenase-2
inhibitor celecoxib to a novel class of 3-phosphoinositide-dependent
protein kinase-1 inhibitors. Cancer Res. 2004, 64, 4309−4318.
(7) Feldman, R. I.; Wu, J. M.; Polokoff, M. A.; Kochanny, M. J.;
Dinter, H.; Zhu, D.; Biroc, S. L.; Alicke, B.; Bryant, J.; Yuan, S.;
Buckman, B. O.; Lentz, D.; Ferrer, M.; Whitlow, M.; Adler, M.;
Finster, S.; Chang, Z.; Arnaiz, D. O. Novel small molecule inhibitors of
3-phosphoinositide-dependent kinase-1. J. Biol. Chem. 2005, 280,
19867−19874.
(8) Gopalsamy, A.; Shi, M.; Boschelli, D. H.; Williamson, R.; Olland,
A.; Hu, Y.; Krishnamurthy, G.; Han, X.; Arndt, K.; Guo, B. Discovery
of dibenzo[c,f ][2,7]naphthyridines as potent and selective 3-
phosphoinositide-dependent kinase-1 inhibitors. J. Med. Chem. 2007,
50, 5547−5549.
ASSOCIATED CONTENT
* Supporting Information
■
S
Synthetic procedures and analytical data of compounds 0m,
1a−g, 1i−n, 1p−v, 2a−f, 2i−n, 2p−v, 3m, 4a−g, 4i−n, 4p−v,
5a, 5q, 6a−d, 6f, 6g, 6i, 6k, 6n, 6q−s, and 6u; method for
separation of 4h into its enantiomers by HPLC including the
analytical chromatograph. This material is available free of
charge via the Internet at http://pubs.acs.org.
(9) Islam, I.; Brown, G.; Bryant, J.; Hrvatin, P.; Kochanny, M. J.;
Phillips, G. B.; Yuan, S.; Adler, M.; Whitlow, M.; Lentz, D.; Polokoff,
M. A.; Wu, J.; Shen, J.; Walters, J.; Ho, E.; Subramanyam, B.; Zhu, D.;
Feldman, R. I.; Arnaiz, D. O. Indolinone based phosphoinositide-
dependent kinase-1 (PDK1) inhibitors. Part 2: optimization of BX-
517. Bioorg. Med. Chem. Lett. 2007, 17, 3819−3825.
Accession Codes
The PDB code for 4h complexed with PDK1 is 4AW1.
(10) Peifer, C.; Alessi, D. R. Small-molecule inhibitors of PDK1.
ChemMedChem 2008, 3, 1810−1838.
AUTHOR INFORMATION
Corresponding Author
(11) Nagashima, K.; Shumway, S. D.; Sathyanarayanan, S.; Chen, A.
H.; Dolinski, B.; Xu, Y.; Keilhack, H.; Nguyen, T.; Wiznerowicz, M.;
Li, L.; Lutterbach, B. A.; Chi, A.; Paweletz, C.; Allison, T.; Yan, Y.;
Munshi, S. K.; Klippel, A.; Kraus, M.; Bobkova, E. V.; Deshmukh, S.;
Xu, Z.; Mueller, U.; Szewczak, A. A.; Pan, B. S.; Richon, V.; Pollock,
R.; Blume-Jensen, P.; Northrup, A.; Andersen, J. N. Genetic and
pharmacological inhibition of PDK1 in cancer cells: characterization of
a selective allosteric kinase inhibitor. J. Biol. Chem. 2011, 286, 6433−
6448.
■
*Phone: +49 681 302 70312; fax: +49 681 302 70308; e-mail:
ma.engel@mx.uni-saarland.de.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
(12) Copps, K. D.; White, M. F. Regulation of insulin sensitivity by
serine/threonine phosphorylation of insulin receptor substrate
proteins IRS1 and IRS2. Diabetologia 2012, 55, 2565−2582.
(13) Bayascas, J. R.; Sakamoto, K.; Armit, L.; Arthur, J. S.; Alessi, D.
R. Evaluation of approaches to generation of tissue-specific knock-in
mice. J. Biol. Chem. 2006, 281, 28772−28781.
(14) Um, S. H.; Frigerio, F.; Watanabe, M.; Picard, F.; Joaquin, M.;
Sticker, M.; Fumagalli, S.; Allegrini, P. R.; Kozma, S. C.; Auwerx, J.;
Thomas, G. Absence of S6K1 protects against age- and diet-induced
obesity while enhancing insulin sensitivity. Nature 2004, 431, 200−
205.
We thank Prof. Dr. R.W. Hartmann for generously providing
lab facilities to the M.E. group and Prof. Dr. S. Zeuzem for his
support of the R.M.B. lab. We acknowledge the support by the
Deutsche Krebshilfe (106388) to R.M.B., by the Deutsche
Forschungsgemeinschaft (DFG) (EN381/2-3) to M.E., and by
the GoBio grant from the German Federal Ministry of
Education and Science (0315102) to R.M.B. We are grateful
to Michael Zender for his assistance with separation of the
racemic compound 4h.
(15) Um, S. H.; D’Alessio, D.; Thomas, G. Nutrient overload, insulin
resistance, and ribosomal protein S6 kinase 1, S6K1. Cell Metab. 2006,
3, 393−402.
ABBREVIATIONS USED
■
PDK1, phosphoinositide-dependent kinase-1; PI3, phosphati-
dylinositol-3; S6K, p70 S6 kinase; RSK, p90 ribosomal S6
kinase; SGK, serum- and glucocorticoid-induced protein kinase;
PRK2, protein kinase C-related kinase 2; PIF, PRK2-interacting
fragment; HM, hydrophobic motif; PH, pleckstrin homology;
GST, glutathione-S-transferase; IRS, insulin receptor substrate;
IGF, insulin-like growth factor
(16) Sale, E. M.; Sale, G. J. Protein kinase B: signalling roles and
therapeutic targeting. Cell. Mol. Life Sci. 2008, 65, 113−127.
(17) Ellwood-Yen, K.; Keilhack, H.; Kunii, K.; Dolinski, B.; Connor,
Y.; Hu, K.; Nagashima, K.; O’Hare, E.; Erkul, Y.; Di Bacco, A.;
Gargano, D.; Shomer, N. H.; Angagaw, M.; Leccese, E.; Andrade, P.;
Hurd, M.; Shin, M. K.; Vogt, T. F.; Northrup, A.; Bobkova, E. V.;
Kasibhatla, S.; Bronson, R. T.; Scott, M. L.; Draetta, G.; Richon, V.;
Kohl, N.; Blume-Jensen, P.; Andersen, J. N.; Kraus, M. PDK1
attenuation fails to prevent tumor formation in PTEN-deficient
transgenic mouse models. Cancer Res. 2011, 71, 3052−3065.
(18) Westmoreland, J. J.; Wang, Q.; Bouzaffour, M.; Baker, S. J.;
Sosa-Pineda, B. Pdk1 activity controls proliferation, survival, and
growth of developing pancreatic cells. Dev. Biol. 2009, 334, 285−298.
(19) Biondi, R. M.; Cheung, P. C.; Casamayor, A.; Deak, M.; Currie,
R. A.; Alessi, D. R. Identification of a pocket in the PDK1 kinase
domain that interacts with PIF and the C-terminal residues of PKA.
EMBO J. 2000, 19, 979−988.
REFERENCES
■
(1) Pearce, L. R.; Komander, D.; Alessi, D. R. The nuts and bolts of
AGC protein kinases. Nat. Rev. Mol. Cell Biol. 2010, 11, 9−22.
(2) Mora, A.; Komander, D.; van Aalten, D. M.; Alessi, D. R. PDK1,
the master regulator of AGC kinase signal transduction. Semin. Cell
Dev. Biol. 2004, 15, 161−170.
(3) Bayascas, J. R.; Leslie, N. R.; Parsons, R.; Fleming, S.; Alessi, D. R.
Hypomorphic mutation of PDK1 suppresses tumorigenesis in
PTEN( ) mice. Curr. Biol. 2005, 15, 1839−1846.
9828
dx.doi.org/10.1021/jm3010477 | J. Med. Chem. 2012, 55, 9817−9830

Journal of Medicinal Chemistry
Article
(20) Biondi, R. M.; Kieloch, A.; Currie, R. A.; Deak, M.; Alessi, D. R.
The PIF-binding pocket in PDK1 is essential for activation of S6K and
SGK, but not PKB. EMBO J. 2001, 20, 4380−4390.
(36) Sadowsky, J. D.; Burlingame, M. A.; Wolan, D. W.; McClendon,
C. L.; Jacobson, M. P.; Wells, J. A. Turning a protein kinase on or off
from a single allosteric site via disulfide trapping. Proc. Natl. Acad. Sci.
U.S.A. 2011, 108, 6056−6061.
(37) Dettori, R.; Sonzogni, S.; Meyer, L.; Lopez-Garcia, L. A.;
Morrice, N. A.; Zeuzem, S.; Engel, M.; Piiper, A.; Neimanis, S.; Frodin,
M.; Biondi, R. M. Regulation of the interaction between protein kinase
C-related protein kinase 2 (PRK2) and its upstream kinase, 3-
phosphoinositide-dependent protein kinase 1 (PDK1). J. Biol. Chem.
2009, 284, 30318−30327.
(38) Xu, C.; Bartley, J. K.; Enache, D. I.; Knight, D. W.; Hutchings,
G. J. High Surface Area MgO as a Highly Effective Heterogeneous
Base Catalyst for Michael Addition and Knoevenagel Condensation
Reactions. Synthesis 2005, 19, 3468−3476.
(21) Bogan, A. A.; Thorn, K. S. Anatomy of hot spots in protein
interfaces. J. Mol. Biol. 1998, 280, 1−9.
(22) Stephens, L.; Anderson, K.; Stokoe, D.; Erdjument-Bromage,
H.; Painter, G. F.; Holmes, A. B.; Gaffney, P. R.; Reese, C. B.;
McCormick, F.; Tempst, P.; Coadwell, J.; Hawkins, P. T. Protein
kinase B kinases that mediate phosphatidylinositol 3,4,5-trisphosphate-
dependent activation of protein kinase B. Science 1998, 279, 710−714.
(23) Anderson, K. E.; Coadwell, J.; Stephens, L. R.; Hawkins, P. T.
Translocation of PDK-1 to the plasma membrane is important in
allowing PDK-1 to activate protein kinase B. Curr. Biol. 1998, 8, 684−
691.
(24) Beeson, M.; Sajan, M. P.; Dizon, M.; Grebenev, D.; Gomez-
Daspet, J.; Miura, A.; Kanoh, Y.; Powe, J.; Bandyopadhyay, G.;
Standaert, M. L.; Farese, R. V. Activation of protein kinase C-zeta by
insulin and phosphatidylinositol-3,4,5-(PO4)3 is defective in muscle in
type 2 diabetes and impaired glucose tolerance: amelioration by
rosiglitazone and exercise. Diabetes 2003, 52, 1926−1934.
(25) Farese, R. V.; Sajan, M. P.; Standaert, M. L. Insulin-sensitive
protein kinases (atypical protein kinase C and protein kinase B/Akt):
actions and defects in obesity and type II diabetes. Exp. Biol. Med.
(Maywood) 2005, 230, 593−605.
(39) Ma, S.; Yu, S.; Yin, S. Studies on K2CO3-catalyzed 1,4-addition
of 1,2-allenic ketones with diethyl malonate: controlled selective
synthesis of beta,gamma-unsaturated enones and α-pyrones. J. Org.
Chem. 2003, 68, 8996−9002.
(40) Ma, S.; Yin, S.; Li, L.; Tao, F. K2CO3-catalyzed Michael
addition−lactonization reaction of 1,2-allenyl ketones with electron-
withdrawing group substituted acetates. An efficient synthesis of α-
pyrone derivatives. Org. Lett. 2002, 4, 505−507.
(41) March, J. The decarboxylation of organic acids. J. Chem. Educ.
1963, 40, 212−213.
(42) Ni, J.; Auston, D. A.; Freilich, D. A.; Muralidharan, S.; Sobie, E.
A.; Kao, J. P. Photochemical gating of intracellular Ca2+ release
channels. J. Am. Chem. Soc. 2007, 129, 5316−5317.
(26) Frojdo, S.; Vidal, H.; Pirola, L. Alterations of insulin signaling in
type 2 diabetes: a review of the current evidence from humans.
Biochim. Biophys. Acta 2009, 1792, 83−92.
(27) Chou, M. M.; Hou, W.; Johnson, J.; Graham, L. K.; Lee, M. H.;
Chen, C. S.; Newton, A. C.; Schaffhausen, B. S.; Toker, A. Regulation
of protein kinase C ζ by PI 3-kinase and PDK-1. Curr. Biol. 1998, 8,
1069−1077.
(28) Balendran, A.; Biondi, R. M.; Cheung, P. C.; Casamayor, A.;
Deak, M.; Alessi, D. R. A 3-phosphoinositide-dependent protein
kinase-1 (PDK1) docking site is required for the phosphorylation of
protein kinase Cζ (PKCζ) and PKC-related kinase 2 by PDK1. J. Biol.
Chem. 2000, 275, 20806−20813.
(29) Newton, A. C. Protein kinase C: poised to signal. Am. J. Physiol.:
Endocrinol. Metab. 2010, 298, E395−402.
(43) Mai, A.; Cheng, D.; Bedford, M. T.; Valente, S.; Nebbioso, A.;
Perrone, A.; Brosch, G.; Sbardella, G.; De Bellis, F.; Miceli, M.; Altucci,
L. Epigenetic multiple ligands: mixed histone/protein methyltransfer-
ase, acetyltransferase, and class III deacetylase (sirtuin) inhibitors. J.
Med. Chem. 2008, 51, 2279−2290.
(44) Busschots, K.; Lopez-Garcia, L. A.; Lammi, C.; Stroba, A.;
Zeuzem, S.; Piiper, A.; Alzari, P. M.; Neimanis, S.; Arencibia, J. M.;
Engel, M.; Schulze, J. O.; Biondi, R. M. Substrate-selective inhibition of
protein kinase PDK1 by small compounds that bind to the PIF-pocket
allosteric docking site. Chem. Biol. 2012, in press.
(45) Lu, Y.; Shi, T.; Wang, Y.; Yang, H.; Yan, X.; Luo, X.; Jiang, H.;
Zhu, W. Halogen bonding–a novel interaction for rational drug
design? J. Med. Chem. 2009, 52, 2854−2862.
(30) Engel, M.; Hindie, V.; Lopez-Garcia, L. A.; Stroba, A.; Schaeffer,
F.; Adrian, I.; Imig, J.; Idrissova, L.; Nastainczyk, W.; Zeuzem, S.;
Alzari, P. M.; Hartmann, R. W.; Piiper, A.; Biondi, R. M. Allosteric
activation of the protein kinase PDK1 with low molecular weight
compounds. EMBO J. 2006, 25, 5469−5480.
(46) Frohner, W.; Lopez-Garcia, L. A.; Neimanis, S.; Weber, N.;
̈
Navratil, J.; Maurer, F.; Stroba, A.; Zhang, H.; Biondi, R. M.; Engel, M.
4-Benzimidazolyl-3-phenylbutanoic acids as novel PIF-pocket-target-
ing allosteric inhibitors of protein kinase PKCζ. J. Med. Chem. 2011,
54, 6714−6723.
(31) Stroba, A.; Schaeffer, F.; Hindie, V.; Lopez-Garcia, L. A.; Adrian,
I.; Frohner, W.; Hartmann, R. W.; Biondi, R. M.; Engel, M. 3,5-
̈
Diphenylpent-2-enoic acids as allosteric activators of the protein kinase
PDK1: Structure−activity relationships and thermodynamic character-
isation of binding as paradigms for PIF-binding pocket-targeting
compounds. J. Med. Chem. 2009, 52, 4683−4693.
(47) Lopez-Garcia, L. A.; Schulze, J. O.; Frohner, W.; Zhang, H.; Suß,
̈
̈
E.; Weber, N.; Navratil, J.; Amon, S.; Hindie, V.; Zeuzem, S.;
Jørgensen, T. J. D.; Alzari, P. M.; Neimanis, S.; Engel, M.; Biondi, R.
M. Allosteric regulation of protein kinase PKCζ by the N-terminal C1
domain and small compounds to the PIF-pocket. Chem. Biol. 2011, 18,
1463−1473.
(32) Hindie, V.; Stroba, A.; Zhang, H.; Lopez-Garcia, L. A.; Idrissova,
L.; Zeuzem, S.; Hirschberg, D.; Schaeffer, F.; Jorgensen, T. J.; Engel,
M.; Alzari, P. M.; Biondi, R. M. Structure and allosteric effects of low-
molecular-weight activators on the protein kinase PDK1. Nat. Chem.
Biol. 2009, 5, 758−764.
(48) Olsen, J. A.; Banner, D. W.; Seiler, P.; Obst Sander, U.; D’Arcy,
A.; Stihle, M.; Muller, K.; Diederich, F. A fluorine scan of thrombin
inhibitors to map the fluorophilicity/fluorophobicity of an enzyme
active site: evidence for C−F···CO interactions. Angew. Chem., Int.
Ed. 2003, 42, 2507−2511.
(49) Niimi, T.; Orita, M.; Okazawa-Igarashi, M.; Sakashita, H.;
Kikuchi, K.; Ball, E.; Ichikawa, A.; Yamagiwa, Y.; Sakamoto, S.; Tanaka,
A.; Tsukamoto, S.; Fujita, S.; Tatsuta, K.; Maeda, Y.; Chikauchi, K.
Design and synthesis of non-peptidic inhibitors for the Syk C-terminal
SH2 domain based on structure-based in-silico screening. J. Med.
Chem. 2001, 44, 4737−4740.
(50) Gao, Y.; Luo, J.; Yao, Z. J.; Guo, R.; Zou, H.; Kelley, J.; Voigt, J.
H.; Yang, D.; Burke, T. R., Jr. Inhibition of Grb2 SH2 domain binding
by non-phosphate-containing ligands. 2. 4-(2-Malonyl)phenylalanine
as a potent phosphotyrosyl mimetic. J. Med. Chem. 2000, 43, 911−920.
(51) Fedorov, O.; Marsden, B.; Pogacic, V.; Rellos, P.; Muller, S.;
Bullock, A. N.; Schwaller, J.; Sundstrom, M.; Knapp, S. A systematic
(33) Stockman, B. J.; Kothe, M.; Kohls, D.; Weibley, L.; Connolly, B.
J.; Sheils, A. L.; Cao, Q.; Cheng, A. C.; Yang, L.; Kamath, A. V.; Ding,
Y. H.; Charlton, M. E. Identification of allosteric PIF-pocket ligands
1
for PDK1 using NMR-based fragment screening and H-15N TROSY
experiments. Chem. Biol. Drug Des. 2009, 73, 179−188.
(34) Wei, L.; Gao, X.; Warne, R.; Hao, X.; Bussiere, D.; Gu, X. J.;
Uno, T.; Liu, Y. Design and synthesis of benzoazepin-2-one analogues
as allosteric binders targeting the PIF pocket of PDK1. Bioorg. Med.
Chem. Lett. 2010, 20, 3897−3902.
(35) Bobkova, E. V.; Weber, M. J.; Xu, Z.; Zhang, Y. L.; Jung, J.;
Blume-Jensen, P.; Northrup, A.; Kunapuli, P.; Andersen, J. N.; Kariv, I.
Discovery of PDK1 kinase inhibitors with a novel mechanism of action
by ultrahigh throughput screening. J. Biol. Chem. 2010, 285, 18838−
18846.
9829
dx.doi.org/10.1021/jm3010477 | J. Med. Chem. 2012, 55, 9817−9830

Journal of Medicinal Chemistry
Article
interaction map of validated kinase inhibitors with Ser/Thr kinases.
Proc. Natl. Acad. Sci. U.S.A. 2007, 104, 20523−20528.
(52) Kroe, R. R.; Regan, J.; Proto, A.; Peet, G. W.; Roy, T.; Landro,
L. D.; Fuschetto, N. G.; Pargellis, C. A.; Ingraham, R. H. Thermal
denaturation: a method to rank slow binding, high-affinity P38α MAP
kinase inhibitors. J. Med. Chem. 2003, 46, 4669−4675.
(53) Jansen, A. B.; Russell, T. J. Some novel penicillin derivatives. J.
Chem. Soc. 1965, 65, 2127−2132.
(54) Daehne, W.; Frederiksen, E.; Gundersen, E.; Lund, F.; Morch,
P.; Petersen, H. J.; Roholt, K.; Tybring, L.; Godtfredsen, W. O.
Acyloxymethyl esters of ampicillin. J. Med. Chem. 1970, 13, 607−612.
(55) Ferres, H. Prodrugs of beta-lactam antibiotics. Drugs Today
1983, 19, 499−538.
(56) Bundgaard, H.; Nielsen, N. M. Esters of N,N-disubstituted 2-
hydroxyacetamides as a novel highly biolabile prodrug type for
carboxylic acid agents. J. Med. Chem. 1987, 30, 451−454.
(57) D’Arezzo, S.; Incerpi, S.; Davis, F. B.; Acconcia, F.; Marino, M.;
Farias, R. N.; Davis, P. J. Rapid nongenomic effects of 3,5,3′-triiodo-L-
thyronine on the intracellular pH of L-6 myoblasts are mediated by
intracellular calcium mobilization and kinase pathways. Endocrinology
2004, 145, 5694−5703.
(58) James, A. M.; Wei, Y. H.; Pang, C. Y.; Murphy, M. P. Altered
mitochondrial function in fibroblasts containing MELAS or MERRF
mitochondrial DNA mutations. Biochem. J. 1996, 318 (Pt 2), 401−407.
(59) Zamecnik, P.; Aghajanian, J.; Zamecnik, M.; Goodchild, J.;
Witman, G. Electron micrographic studies of transport of oligodeox-
ynucleotides across eukaryotic cell membranes. Proc. Natl. Acad. Sci.
U.S.A. 1994, 91, 3156−3160.
(60) Niesen, F. H.; Berglund, H.; Vedadi, M. The use of differential
scanning fluorimetry to detect ligand interactions that promote protein
stability. Nat. Protoc. 2007, 2, 2212−2221.
9830
dx.doi.org/10.1021/jm3010477 | J. Med. Chem. 2012, 55, 9817−9830