3,5-Diphenylpent-2-enoic acids as allosteric activators of the protein kinase PDK1: Structure-activity relationships and thermodynamic characterization of binding as paradigms for PIF-binding pocket-targeting compounds

Article abstract of DOI:10.1021/jm9001499

The modulation of protein kinase activities by low molecular weight compounds is a major goal of current pharmaceutical developments. In this line, important efforts are directed to the development of drugs targeting the conserved ATP binding site. Howeve

Full text of DOI:10.1021/jm9001499

J. Med. Chem. 2009, 52, 4683–4693 4683
DOI: 10.1021/jm9001499
3,5-Diphenylpent-2-enoic Acids as Allosteric Activators of the Protein Kinase PDK1:
Structure-Activity Relationships and Thermodynamic Characterization of Binding as Paradigms for
PIF-Binding Pocket-Targeting Compounds†
Adriana Stroba,‡ Francis Schaeffer,§ Valerie Hindie, Laura Lopez-Garcia, Iris Adrian, Wolfgang Frohner,‡
¨
Rolf W. Hartmann,‡ Ricardo M. Biondi, and Matthias Engel*,‡
‡
§
Pharmaceutical and Medicinal Chemistry, Saarland University, P.O. Box 151150, D-66041 Saarbrucken, Germany, Unite de Biochimie
ꢀ
¨
Structurale (CNRS-URA 2185), Institut Pasteur, F-75724 Paris, France, and Department of Internal Medicine I, University of Frankfurt,
Theodor-Stern-Kai 7, D-60590 Frankfurt a.M., Germany
Received February 5, 2009
The modulation of protein kinase activities by low molecular weight compounds is a major goal of
current pharmaceutical developments. In this line, important efforts are directed to the development of
drugs targeting the conserved ATP binding site. However, there is very little experience on targeting
allosteric, regulatory sites, different from the ATP binding site, in protein kinases. Here we describe the
synthesis, cell-free activation potency, and calorimetric binding analysis of 3,5-diphenylpent-2-enoic
acids and derivatives as allosteric modulators of the phosphoinositide-dependent kinase-1 (PDK1)
catalytic activity. Our SAR results combined with thermodynamic binding analyses revealed both
favorable binding enthalpy and entropy and confirmed the PIF-binding pocket of PDK1 as a druggable
site. In conclusion, we defined the minimal structural requirements for compounds to bind to the PIF-
binding pocket and to act as allosteric modulators and identified two new lead structures (12Z and 13Z)
with predominating binding enthalpy.
Introduction
The PIF-binding pocket of PDK1 is a hydrophobic surface
pocket neighboring a putative phosphate coordination
site and is located on the small lobe of the PDK1 catalytic
domain.4,7 Its physiological ligands are C-terminal regions of
PDK1 substrates that utilize a considerably larger area for
interaction than just the hydrophobic pocket. The PDK1 PIF-
binding pocket serves two functions: first, it is a docking site
for transient interactions with the substrate proteins via their
C-terminal hydrophobic motif (HM), comprising the se-
quence Phe-Xaa-Xaa-Phe. Second, binding of the peptides
increases the intrinsic catalytic activity of PDK1. Almost all
substrates of PDK1, including S6K, SGK, RSK, and the
atypical PKCs, require this docking interaction with the
PDK1 PIF-binding pocket in order to become phosphory-
lated at the activation loop, which results in full activation of
the substrate kinase.5,6,8 As the only known exception, PKB
does not require interaction with the PIF-binding pocket of
PDK1.5,9
Protein kinases are actively involved in the control and
regulation of cellular processes. In addition, overexpression of
protein kinases or loss of kinase regulatory mechanisms are
observed in many human diseases such as cancer, inflammation,
neurological (e.g., Alzheimer0s disease), and metabolic disorders
(e.g., type 2 diabetes). Thus, protein kinases are an important
drug target class for the treatment of diverse human diseases.1
The phosphoinositide-dependent protein kinase 1 (PDK1a)
is in the center of growth factor and insulin signaling and is a
master kinase that phosphorylates the activation loop of
several protein kinases from the AGC group, including all
PKC, PKB (also termed AKT), S6K, RSK, and SGK iso-
forms.2 PDK1 has gained importance as a potential thera-
peutic target for cancer because several of the substrate
protein kinases, comprising S6K, RSK, and PKB, regulate
processes that are essential to the tumor cell such as cell
growth and survival.2,3
Several reports have described small molecule inhibitors of
PDK1 that target the ATP binding site.10-14 One such non-
specific inhibitor, UCN-01 (7-hydroxystaurosporine), is in
clinical trials.15-18 While these compounds have shown anti-
cancer activities in vivo, coinhibition of additional protein
kinases increases the risk of dose-limiting off-target effects. In
fact, most ATP competitive inhibitors show limited selecti-
vity,19,20 implying that certain protein kinases have not been
considered as suitable for drug development, in particular
protein kinases for which closely related isoforms exist.
As an alternative and as a complement to ATP binding site-
directed compounds, there is increasing interest in the deve-
lopment of non-ATP competitive strategies for protein kinase
drug developments.21 Most interestingly, the characterization
We have discovered an allosteric site on PDK1 (termed the
PIF-binding pocket), distant from the ATP binding site, and
suggested the site asa possible target for drugs.4,5 Interestingly,
an equivalent hydrophobic motif (HM) binding pocket reg-
ulatory site exists on other AGC kinases. Therefore, our
research on PDK1 may open the field to drug developments
on other kinases sharing a similar mechanism of regulation.6
†PDB code of 2Z with PDK1: 3HRF.
*To whom correspondence should be addressed. Phone: þ49 681 302
70312. Fax: þ49 681 302 70308. E-mail: ma.engel@mx.uni-saarland.de.
Web site: http://www.pharmmedchem.de.
a Abbreviations: HM, hydrophobic motif; HWE, Horner-Wads-
worth-Emmons; wt, wild type; ITC, isothermal titration calorimetry;
PDK1, phosphoinositide-dependent kinase-1.
r
2009 American Chemical Society
Published on Web 07/17/2009
pubs.acs.org/jmc

4684 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 15
Stroba et al.
Scheme 1. Synthesis of Compounds 2E-13E, 2Z-13Z, and
21Ea
Figure 1. Hit compound (1) and new lead compound (2Z).
of novel allosteric sites as druggable sites might enable the
development of drugs with more subtle modulation of the
protein kinase activity as compared to the complete inhibition
of the enzymatic activity. For example, the targeting of
allosteric-regulatory sites could potentiallygenerate inhibitors
but also protein kinase activating drugs. Recently, we have
provided first evidence that a small molecule can trigger
activation of PDK1.22 Similarly, anacardic acid (6-pentade-
catrienyl salicylic acid) was reported recently as an allosteric
activator of Aurora A kinase.23
In the case of GPCRs, modulation of the receptor function
by allosteric ligands has provided high selectivity and a far
greater diversity in the repertoire of pharmacological effects
can be achieved.24 In the present report, we describe the design
and synthesis of 3,5-diarylpent-2-enoic acids as novel PIF-
binding pocket directed compounds and present SAR data on
their allosteric activation potency. We provide the binding
energetics of allostericactivators based onisothermal titration
calorimetry, shedding light on relationships between structur-
al properties, binding affinity, and allosteric activation. The
general requirements for HM/PIF-binding pocket com-
pounds are presented and discussed based on our recent
cocrystallization of compound 2Z with PDK1.25
a Reagents and conditions: (i) method A: NaOH, EtOH, 1 h, rt;
(ii) piperidine, EtOH, reflux, 16 h; (iii) method B: 3,5-bis(ethoxycarbo-
nyl)-1,4-dihydro-2,6-dimethylpyridine, toluene, silica gel, 70 °C, 16 h;
(iv) method C: triethyl phosphonoacetate, NaH, DME, 80 °C, 4 h;
(v) method D: NaOH, EtOH, rt, 3 h. For substituents R, see Table 1.
Scheme 2. Synthesis of Compounds 14E, 14Z, and 15Ea
Results and Discussion
1. Chemistry. In a previous report, we had published hit
compound 1 as an activator of PDK1 (Figure 1). Drawbacks
of this compound were the existence of a chiral center,
yielding a racemic mixture with unknown contribution of
each enantiomer to the total activity, and the sulfanyl
moiety, which is prone to oxidations and potentially retro-
Michael reactions. The design of the new structural analo-
gues aimed at replacement of the chiral center by a double
bond while retaining the combination of two sp3 and one sp2
hybridized C-atoms in the chain connecting the benzene
rings, thus leading to compound 2Z (Figure 1).
Syntheses of 2 and analogues 3-13 started with a
Claisen-Schmitt condensation (method A) between acetophe-
none and a series of benzaldehydes 2d-13d to obtain the
chalcones 2c-13c (Scheme 1). To selectively reduce the
conjugated double bond of the chalcones, we utilized a
convenient hydride transfer reaction (method B) from 3,5-
bis(ethoxycarbonyl)- 1,4-dihydro-2,6-dimethylpyridine
(HEH) catalyzed by silica gel,26 which quantitatively af-
forded the saturated ketones 2b-13b. Olefination of the
carbonyl group in 2b-13b was achieved by Horner-Wads-
worth-Emmons reaction (HWE) (method C) with triethyl
phosphonoacetate, yielding E/Z-mixtures of ethyl 3,5-diphe-
nylpent-2-enoates (2a-13a, Scheme 1), which could be
efficiently separated by flash column chromatography in
all cases. We intentionally chose the reaction conditions to
favor formation of both stereoisomers in order to isolate
both compounds and test them separately. In the last step,
the ethyl ester was hydrolyzed (method D) under basic
conditions to yield the free 3,5-diphenylpent-2-enoic acids
(2E-13E, 2Z-13Z). For 2E and 2Z, we determined the E/Z
configuration exemplarily using 2D-NOESY-1H NMR
aReagents and conditions: (i) K2CO3, EtOH, 80 °C, 1 h; (ii) 5%
BnNEt3Cl, 30% NaOH, CH2Cl2, rt, 16 h; (iii) method C: triethyl
phosphonoacetate, NaH, DME, 80 °C, 4 h; (iv) method D: NaOH,
EtOH, rt, 3 h.
(see Supporting Information). The E/Z assignment of the
other compound pairs was done by comparison of the
corresponding NMR spectra with those of 2E and 2Z. Thus,
by introducing a shorter olefinic carboxyl side chain, we
reduced the total number of rotatable bonds from seven (1)
to five (2Z). An even more rigid, cyclized analogue 21E was
prepared by condensation of 4-chlorobenzaldehyde 2d with
1-indanone, reduction of the resulting cyclized chalcone,
HWE reaction, and subsequent ester hydrolysis (Scheme 1).
21E was isolated as enantiomeric mixture exclusively in the
(E)-form.
In another subset of compounds, we replaced the ring A
benzylic methylene by oxygen or sulfur. The strategy for the
synthesis of these heteroanalogues of 2Z is illustrated in
Scheme 2. Bromoacetophenone was subjected to an SN
reaction with 4-chlorophenol (14c) and 4-chlorophenyl mer-
captane 15c, respectively, to yield 2-(4-chlorophenoxy)acet-
ophenone (14b) and 2-(4-chlorophenylthio)acetophenone
(15b). The final compounds, (E)- and (Z)-4-(4-chlorophe-
noxy)-(14E,14Z) and (E)-4-(4-chlorophenylthio)-3-phenyl-
but-2-enoic acid (15E), were obtained via HWE reaction
(method C) and hydrolysis (method D) analogously to the
3,5-diphenylpentenoic acids.

Article
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 15 4685
Scheme 3. Synthesis of Compounds 16-17, 18E, and 18Za
Figure 2. Biochemical evidence from site-directed mutagenesis that
the PIF-binding pocket of PDK1 is the target site of 2Z. The effect
of 2Z on PDK1 specific activity was measured with recombinant
wild type enzyme and PIF pocket-binding mutants as indicated.
Mutation of valine-127 to threonine abrogates the activation of
PDK1 by 2Z but not by the 22 amino acid residue peptide PIFtide.
For comparison, results obtained with the original compound 1 are
also given. Concentrations of the compounds used were 2 μM for
PIFtide and 20 μM for 1 and 2Z.
In addition, we prepared heterocyclic analogues via two
different synthetic pathways (Scheme 3). Rap-Stoermer
reaction of salicylaldehydes 16c and 17c with phenacyl
bromide (method E) provided the substituted 2-benzoylben-
zofurans 16b and 17b,27 while synthesis of 2-(4-chloroben-
zoyl)benzothiophene (18b) was accomplished by a one-step
nucleophilic aromatic displacement of an activated nitro
function by mercaptoacetophenone followed by an intramo-
lecular aldol condensation.28 The heterocyclic precursors
were further processed to the corresponding acrylic acids
(16-17, 18E, 18Z) analogously to compounds 2b-13b.
Finally, we also reduced the acrylic acids (16E/Za-17E/Za)
to the corresponding propionic acids (19a-20a) by means of a
catalytic transfer hydrogenation (method F) using sodium
hypophosphite in combination with Pd/C.
2. Biological Activity. 2.1. Biochemical Evidence that
Compound 2Z Activates PDK1 by Binding to the PIF-Binding
Pocket. To measure the effect of compound 2Z on the PDK1
catalytic activity, we performed a radioactive kinase activity
assay employing T308tide, a peptide derived from the activa-
tion loop of PKB, as a substrate. Similarly to the 24 amino
acid polypeptide PIFtide, characterized to bind to the PIF-
binding pocket and activate PDK1,4 compound 2Z in-
creased the activity of wild type (wt) PDK1 with the same
activation efficacy (data for PIFtide not shown) (Figure 2).
In addition, 2Z displayed a 4-fold lower AC50 than the hit
compound 1 (Table 1).22 To biochemically characterize the
binding site of 2Z on PDK1, we compared the ability of 2Z to
activate wt PDK1 vs PDK1 proteins mutated within the PIF-
binding pocket. The mutant PDK1L155S, in which the char-
acter of the PIF-binding pocket is completely changed, could
not be activated by PIFtide nor by 2Z, suggesting that the
PIF-binding pocket site was required by both effectors. In
contrast, the ability of 2Z to activate PDK1 was lost toward
the PDK1V127T mutant while this mutant was still activated
by PIFtide. This was not surprising because previous work
had shown that when V127 was mutated to a larger hydro-
phobic residue (Leu), the mutant PDK1 protein also retained
aReagents and conditions: (i) method E: K2CO3, DMF, reflux, 2 h;
(ii) K2CO3, DMF, 0 °C to rt, 3 h; (iii) method C: triethyl phosphonoa-
cetate, NaH, DME, 80 °C, 4 h; (iv) method F: NaH2PO2 H2O, 10% Pd/C,
3
EtOH/H2O, 60 °C, 2.5 h; (v) method D: NaOH, EtOH, rt, 3 h.
the ability to bind and be activated by PIFtide, and to
efficiently interact and phosphorylate SGK, a substrate
that relies on the binding of its hydrophobic motif to the
PIF-binding pocket on PDK1.22 Thus, it appears that
HM polypeptides do not rely so heavily on the identity of
the residue at position 127, while the much smaller
compound 2Z required a valine residue at position 127
within the hydrophobic PIF-binding pocket. Interestingly,
the PDK1I119A mutant was well activated by 2Z, whereas
the activatory potencies of both compound 1 and PIFtide
were affected by this mutation. Altogether, the mutagenesis
results suggested that the binding of 2Z was indeed within
the hydrophobic PIF-binding pocket. However, there were
significant differences on the requirements between 2Z, 1,
and PIFtide, which had also been suggested to bind to the
same site.
The cocrystallization of 2Z with PDK1 confirmed that
binding occurs exclusively in the PIF-binding pocket (Fig-
ure 3; crystallographic details together with studies on the
allosteric mechanism of PDK1 activation by low molecular
weight compounds will be published elsewhere25). The com-
plex structure showed that the two phenyl rings bound to the
two hydrophobic subpockets separated by L155 and bor-
dered by V127, explaining the lack of activity of compound
2Z on PDK1 proteins mutated at these sites (Figure 2). In
contrast, the ethyl branch of I119 borders one side of the
pocket and interacts only marginally with the chlorine of 2Z,
therefore explaining why the substitution for a smaller hydro-
phobic residue did not affect the ability of 2Z to activate
PDK1. Altogether, the crystal structure of the catalytic do-
main of PDK1 was in agreement with the biochemical studies
performed here in solution with PIF-binding pocket mutants

4686 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 15
Stroba et al.
Table 1. Effect of Compounds on Catalytic Activity of PDK1 and Thermodynamic Characterization of Binding
aMean value of at least two independent experiments, standard deviation <20%. bAs a particularity for compounds with activatory properties, it was
necessary to indicate the maximum activation that was achievable with a compound, as compared to the basal activity of PDK1 (set to 100%). Error bars
were calculated based on two independent titrations of PDK1 with 12Z as follows: cStandard deviations for Ka and Kd are of 11%. dError bar for ΔH=
þ 0.30 kcal/mol. eError bar for TΔS=þ 0.37 kcal/mol. fError bar for ΔG = þ0.07 kcal/mol. gValues were taken from ref 25. h Mixture of isomers, ca.
75% Z/25% E for 16, and 70% Z/30% E for 17. n.e.: no effect; n.b.: no binding; n.d.: not determined.

Article
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 15 4687
Figure 4. Characterization of 2Z and 2E isomers interactions with
PDK150-359 by ITC. The top panel shows the raw heat signal for
successive injections of dissolved compounds 12Z and 12E into a
PDK150-359 solution at 20 °C. The bottom panel shows the inte-
grated heats of injections corrected for heats of dilution for 12Z
(filled squares) and 12E (open squares), with solid lines correspond-
ing to the best fit of the data to a bimolecular binding model. No
binding of compound 12E to PDK150-359 is detected in conditions
where 12Z shows clear binding. Thermodynamic parameter values
are given in Table 1.
Figure 3. Compound 2Z bound in the PIF-binding pocket of
PDK1. The pocket is bordered by the RC helix and the short RB
helix as indicated. Dashed yellow lines display the interactions of the
carboxylate group with R131, T148, and K76 and with Q150 via a
water molecule (red ball). L155 divides the hydrophobic groove into
two subpockets that are occupied by the benzene rings. Phenyl ring
B (3-phenyl, see Figure 1) additionally contributes to the binding
energy by forming edge-to-face CH-π interactions with F157. As
shown in Figure 2, mutation of V127 to threonine abolished the
activation by 2Z. The structure was generated using PyMOL
(http://pymol.sourceforge.net/).
phosphate binding site residues were required for prompting
the activation of PDK1. Furthermore, the complete lack of
activity of the 2Z ethyl ester derivative (Z)-2a also corrobo-
rated that in particular the strong ionic interactions with
phosphate binding site residues neighboring the hydropho-
bic PIF-binding pocket provided an important contribution
to the ability to activate PDK1 (data not shown).
of full length PDK1, further confirming the PIF-binding
pocket as the binding site for 2Z.
2.2. Dependency of Allosteric Potency and Binding of the
Compounds on Electrostatic Interactions Involving the Car-
boxyl Group. The 2Z/PDK1 cocrystal structure had revealed
that the carboxylate of 2Z interacted with the receptor site in
a network of ionic and H-bonds, comprising interactions
with K76 (ε-NH2), T148 (OH), R131, and water-mediated
with the amide of Q150 (Figure 3).25 Thus the carboxylate
mimicked the phosphate group of phosphoserine/threonine
residues from natural ligands, which is expected to bind to
the equivalent site. In good agreement with this cocrystal
structure result, we found that the potency to activate PDK1
was associated only with the geometric isomers carrying the
carboxy group cis relative to the phenyl ring B (generally
termed “cis isomers” for easier readability in the following=
Z isomer for 2Z/2E-12Z/E, E isomer for 14E/14Z and 18E/
18Z), whereas the isomers with the opposite configuration
(termed “trans isomers” in the following) were mostly in-
active (Table 1). The only exception was 13E, which dis-
played only a slight increase of the AC50 compared with the
Z form but, however, was activating with only low efficacy
(2.2 fold, Table 1).
If the ring systems of the trans isomers would bind to the
PIF-binding pocket similar to the two phenyls from 2Z, the
respective carboxy group would point to the opposite direc-
tion (see Figure 3). Therefore, on the basis of the crystal
structure of PDK1 bound to 2Z, it becomes clear that the
carboxy moiety of the trans isomers cannot interact with the
same type and quantity of bonding as seen with the cis isomer
2Z. The trans orientated carboxy group could potentially
form H-bonds directly with Q150; however, the lack of
activity of the respective trans isomers 2E, 9E, 12E, 14Z,
and 18Z indicated that a broader range of interactions with
The above data indicated that all the cis isomers 2Z-12Z,
14E, and 18E activated PDK1, whereas activation potency
of the corresponding trans isomers was completely absent or
strongly reduced in the case of 13E. Next we performed
calorimetry experiments to investigate whether the trans
isomers show a complete lack of binding or if there is binding
that is uncoupled from allosteric activation. We found
that under identical experimental conditions, all the trans
isomers analyzed, 2E, 9E, 12E, and 13E, did not interact with
PDK1 (estimated detection limit: Kd>100 μM) (Table 1), in
contrast to the corresponding cis isomers, which all bound in
a 1:1 stoichiometry. For a direct comparison, example
calorimetric data are shown for the stereoisomers 12Z and
12E in Figure 4.
On the basis of the current data, we cannot explain why
13E still displayed a weak activation of PDK1 while there
was no detectable binding in the ITC experiments. Because
the ITC measurements were performed in the absence of
ATP, it could be speculated that in the activity assays,
allosteric cooperative effects triggered by ATP binding might
enable interaction of 13E with the PIF-binding pocket, thus
stabilizing a slightly more active conformation of PDK1;
thereby the indole ring might promote the binding by
H-bond interaction with Q150, thus distinguishing 13E from
the other trans configured compounds that were analyzed by
ITC.
In general, our ITC findings confirmed that strong binding
was only possible when the carboxy group was in the cis
configuration. Furthermore, the similar results obtained
with the different compound analogues suggested that all
cis compounds may be docked to the phosphate binding site
in a similar way, further positioning the phenyl ring B into

4688 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 15
Stroba et al.
Figure 5. Scheme summarizing the specific interactions of lead
compound 12Z with the PDK1 PIF-binding pocket. Blue dotted
lines denote interactions that have been identified for the analogue
2Z in the co-crystal structure with PDK1; red dotted lines indicate
the assumed interactions as inferred from the ITC experimental
data.
the same subpocket on the PDK1 PIF-binding pocket as
experimentally described for 2Z.
2.3. Interrelationships between Structure, Activity, and
Thermodynamic Binding Signature. We found that phenyl
ring B from 2Z (as defined in Figure 1) participated in an
edge-to-face CH-π interaction with F157 (Figure 3). Such
T-shaped interactions between benzene dimers are stabli-
lized by 2.4 kcal/mol in the gas phase29 and were found to be
the most favored formation in solution.30 The distance
between the two phenyl components of the T-shape interac-
Figure 6. General correlations based on calorimetric measure-
ments. (A) Enthalpy-entropy compensation phenomenon ob-
served upon binding of Z isomers to PDK1. Large changes in the
binding enthalpies (ΔH) and binding entropies (TΔS) were mea-
sured. However, the observed linear relationship between the ΔH
and TΔS values with a slope of 1 shows that a favorable increase in
binding enthalpy was compensated by an equal unfavorable loss of
entropy in the binding free energy (ΔG = ΔH - TΔS), which left
nearly unchanged the binding affinity (Kd = exp[ΔG/RT]). (B) A
gross linear correlation was observed between the AC50 values for
PDK1 activation and the dissociation constants as measured by
ITC. This correlation suggested that, for most compounds, the
binding affinity to the intermediate active conformation of the
kinase, which might be represented rather by the AC50, does not
differ strongly from the affinity to the activated conformation of the
kinase, which might be described by the Kd. Only for compound 7Z
(not used to calculate the trend line), a low AC50 was associated with
an overproportionally high Kd.
˚
tion (3.22 A between the 2Z C4 and the ring center of F157 in
the cocrystal) is significantly shorter than the sum of the
C-H C van der Waals radii and thus indicative of a rather
3 3 3
strong interaction. Therefore, to investigate the effect of
different ring substitutions on the potency of the com-
pounds, we did not modify ring B but rather focused on
substitutions on ring A. Indeed, our working model of the
compound 2Z binding suggested larger unfilled space in the
ring A subpocket binding site (cf. Figure 3).
To optimize ring A, the halogen substituents were system-
atically varied in type and position. We observed no major
differences in activation potency (AC50) among the com-
pounds carrying p-Cl, p-Br, or p-CF3 substituents (2Z, 5Z,
6Z, Table 1), whereas fluorine in 4Z was not tolerated and
caused a largeincrease in AC50. The highestactivation efficacy
(Amax) of the compound series was achieved by the p-CF3
substituent. Shifting the p-Cl to the m-position (3Z) adversely
affected the potency to activate (2.2 fold) because it remained
also low in the 3,4-dichloro substituted compound 7Z.
Both dichloro substituted compounds, 7Z and 8Z, showed
remarkable properties that were distinct among each other
and from the monosubstituted compounds. This indicated
that it was possible to control the binding and allosteric
properties of PIF-binding pocket directed compounds by
selecting different substitution positions. The 3,4-dichloro
substitution (7Z) was breaking the general correlation be-
tween AC50 and Kd as discussed below (Figure 6B). Our
thermodynamic data indicated that part of the binding
entropy seen with 2Z was lost, suggesting that the 3-Cl was
impeding the hydrophobic interaction of the substituted
phenyl ring with the pocket, although a precise explanation
was lacking.
The 2,4-dichloro substitution pattern in 8Z, however,
had a very distinct and interesting effect on the allosteric
potency, as the o-chlorine induced a rather “antagonist”
like behavior of the compound: although 8Z displayed one
of the lowest AC50 values and even the lowest Kd, the
potency to allosterically activate PDK1 decreased to about
half of the Amax of 2Z. In contrast to the 3,4-dichloro
substituted analogue 7Z, 8Z displayed a predominantly
entropy-driven binding (Table 1). Thus, although we can-
not provide a final answer based on our current data, the
more entropy-driven binding of ring A in 8Z might allow a
greater flexibility of positioning within the hydrophobic
pocket in order to avoid a steric clash of the 2-chlorine with
either T128 or Q150 (compare with ring A position of 2Z,
Figure 3), whereas the binding positions might be more
restricted in the case of highly directed H-bonds and
dispersion forces that usually contribute to the enthalpy
term.

Article
Such an entropy-enabled binding shift might propagate
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 15 4689
substituted 10Z, which is incapable of forming additional
CH-π bonds, was nearly inactive.
into suboptimal interaction of the 8Z carboxyl side chain
with R131, resulting in a reduced allosteric activation with-
out compromising binding affinity. A comparison of 8Z
with 9Z showed that entropy-driven binding alone was not
sufficient to confer “antagonist”-like properties, rather a
steric impact, like from the o-chlorine, was also required.
This was corroborated by the fact that the maximum
activation potency of the 4-bromo derivative 5Z was not
affected when a smaller o-fluorine was introduced (cf. 9Z)
despite the significant increase in the entropy/enthalpy
ratio of binding. It will be interesting to see whether 8Z
binds to and stabilizes the intermediate active, basal con-
formation of the PIF pocket without causing major con-
formational changes.
For the phenyloxy- and phenylthio-substituted series,
both a slight reduction of the maximum activation of
PDK1 and an increase of the AC50 values was noted com-
pared with 2Z (14E and 15E). The trans-4-chloro-phenoxy
compound 14Z was completely inactive. This suggested that
the compounds bound to the PDK1 PIF-binding pocket in a
similar way but interacted with the key allosteric residues in a
slightly different manner so that full activation of the cata-
lytic activity could not occur. In addition, we concluded that
the oxygen of 14E was probably too distanced from the Q150
amide or from the T128 hydroxyl for H-bond interactions
(cf. Figure 3). Rather, the increase in polarity adversely
affected hydrophobic interactions with the pocket.
Because the 3,4-dichloro substitution (7Z) decreased the
AC50 almost 3-fold compared to the 4-chloro compound 2Z,
we wanted to explore further enlargement of the ring system
at 3,4-positions, leading to the synthesis of 12Z and 13Z. In
contrast to 7Z, the ring extension by a second phenyl in 12Z
did not abolish the activation potency. Again, the applica-
tion of isothermal calorimetry allowed us to gain insight in
the quality of interactions between the bicyclic rings of 12Z/
13Z and the PIF binding pocket. Both compounds showed
the highest ΔH/ΔG ratios in the analyzed series (12Z: 68.3%,
13Z: 60.8%) due to a marked increase of the enthalpy
relative the entropy of binding (Table 1). This was in sharp
contrast to all halogen-substituted phenyl analogues (with
the exception of 7Z). Considering that the unsubstituted
bicyclic rings have a higher π-electron density combined
with a larger surface, the increase of the binding enthalpy
argued for a stronger contribution of CH-π interactions
than present with the halogen-substituted phenyl rings.
Although the enthalpy for one unit CH-π interaction is
small, around 1 kcal/mol,31 we hypothesized that multiple
CH groups participated simultaneously in interactions with
π electrons, thus adding up to a significant increase in
enthalpy, consistent with our experimental data. According
to the cocrystal structure, the following residues of the PIF-
binding pocket were possible candidates for a putative
interaction with the extended π-electron system of 12Z or
13Z: most certainly the geminal dimethyl of L155, and,
depending on which rotamer of the naphthyl or indolyl
linking bond is preferred, either the terminal methyl of
I118 or γ-methylene of the K115 side chain (cf. Figure 3).
Both of the latter residues are within reach of the bicyclic
rings of 12Z and 13Z, and the two possible corresponding
rotamers could dock into the subpocket provided that the
available space is slightly increased by a movement of
the I119 chain. Because terminal methyls of Leu, Ile, or
Val are more commonly found in CH-π interactions,31 we
included a hypothetical interaction between the naphthyl of
12Z and the methyl of I118 in the interaction scheme for 12Z
(Figure 5).
Next, the rather flexible structure of 2Z and its derivatives
was rigidified in order to increase binding affinity but also to
probe the potential overall conformation required for a
compound to bind to the PIF-binding pocket. Our initial
working model suggested that a benzoannelated heterocycle
linking carbon 4 with the ortho position of ring A in 2Z might
be compatible with the presumed V-shaped binding mode of
an active compound. In fact, the conformationally con-
straint versions of 14E and 15E could significantly activate
purified PDK1 (compounds 16, 17, and 18E), providing
clues on the conformation required for a HM/PIF-binding
pocket directed compound to trigger activation. With the
benzofuran compounds 16 and 17, chromatographic separ-
aration of the E and Z isomers was not possible, thus
precluding separate determination of the biological activity.
However, from the benzothiophene compounds, the E iso-
mer was a considerably better activator of PDK1 then the Z
form, thus suggesting that cyclized and open chain analogues
have similar binding modes in the PIF-binding pocket.
Moreover, this result suggests that the pure E isomer from
16 would have a lower AC50 in the same range as 18E because
the Z isomer in the mixture very probably did not contribute
to the activity. In summary, cyclizations involving ring A did
not increase potency, probably because they abolished part
of the flexibility required to optimally adapt to the pocket
shape. Nevertheless, the remaining activity of 16 and 18E
proved that a broad range of angles is tolerated for docking
of the ring system equivalent to ring A (Figure 1) to the
subpocket. The narrow subpocket for the unsubstituted
phenyl ring (ring B) forces a steeper docking, thus the 120°
angled benzothiophene in 18E must bind in a considerably
more flat position in the other subpocket than ring A from
2Z, where the prolonged axes of the phenyl rings enclose
an angle of approximately 46° (measured in the cocrystal
structure).
An analogous cyclization involving ring B led to 21 with a
strongly reduced maximum activation of PDK1 as compared
to 2Z. This result indicated another essential property for
PDK1 activator compounds, which is the ability of ring B to
rotate out of the double bond plane. In accordance with this
experimental finding, the corresponding dihedral angle in
the cocrystal conformation of 2Z was 118°.
2.4. Overall Correlations between Enthalpy and Entropy of
Binding and the Biological Activity Parameters of Allosteric
PDK1 Activators. Regardless of the pivotal ionic interaction
which contributed essentially to the binding of all active
compounds, we found a surprisingly broad thermodynamic
binding spectrum from mainly entropy-driven to predomi-
nantly enthalpy-driven binding. A correlation analysis
It is conceivable that the enthalpy gain due to the assumed
reinforcement of CH-π interactions could compensate for a
less favorable entropy associated with potential movements
of side chains. This hypothesis was supported by the differ-
ential activation potencies of 10Z and 11Z; both compounds
can only bind to the PIF-subpocket in the same way as the
chlorophenyl ring of 2Z when I119 moves away in order to
avoid a steric clash (cf. Figure 3). While the active compound
11Z offers potential for additional CH-π interactions,
which could compensate for an entropic penalty, the p-ethyl

4690 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 15
Stroba et al.
involving the thermodynamic binding signature and the
activity data revealed two strong relationships; as the most
obvious correlation, we identified a sharp enthalpy-entropy
compensation effect, which has been described for many
ligand-receptor systems such as agonists and antagonists
interacting with membrane receptors32-34 and inhibitors
binding to HIV protease35 (see ΔH/TΔS correlation,
Figure 6A). This effect describes the fact that among a series
of compound analogues with comparable size, any increase
in binding enthalpy is accompanied by a loss of entropy to
about the same degree, leading to little or no change in the
binding affinity.36,37 The compensation effect was most
obvious when comparing the group of halogenated com-
pounds (2Z, 6Z, 8Z, and 9Z) with 12Z and 13Z (Figure 6A).
The bicyclic aromatic rings caused binding with highly
favorable enthalpy, probably due to additional CH-π inter-
actions as described above (Figure 5). Our ITC data indi-
cated that the additional binding enthalpy was gained at the
cost of entropy, accounting for the enthalpy-entropy com-
pensation in the concrete case. A possible explanation for the
concomitant loss of entropy with 12Z and 13Z could be that
due to the higher polarity of the π-electron-rich rings, the
contacting water molecules are less ordered and less entropy
is gained upon water release from the compound. In con-
trast, fewer enthalpically favorable van der Waals or CH-π
interactions are possible with one phenyl ring where the
density of the π-electrons is further decreased by halogen
substituents. Apparently, this loss of H-bond capacity in the
halogenated compounds was exactly balanced out by an
opposite increase in hydrophobic interaction, giving raise to
the entropy term. These mutual compensation processes
prevented significant changes in the total binding free en-
ergies of compounds (compare e.g. 2Z, ΔG=-6.73 kcal/mol
with 12Z, ΔG=-6.67 kcal/mol), which otherwise differed
substantially in their ΔH/ΔG ratio (Table 1). This phenome-
non is not uncommon in the lead optimization process and
has to be overcome in order to increase binding affinity. (see
Section 2.5 below).
engineer because they arise primarily from specific van der
Waals and H-bonding interactions, for which a good geo-
metric complementarity between the involved functional
groups of drug and receptor is a prerequisite. To optimize
the entropy contribution on the other hand has proven
much easier, as this is almost inevitably achieved by intro-
ducing nonpolar groups that increase the hydrophobic
effect.39 From the perspective of allosteric modulators, it
might be worth discussing whether 8Z, which displayed
stronger binding in connection with the opposite trend in
activation potency, should not be pursued in parallel as lead
compounds toward the development of nonactivators or
even allosteric inhibitors. However, because very high
affinity can only be achieved if both enthalpy and entropy
contribute favorably and in a balanced manner to the
binding free energy,38,40 enthalpy-dominated binders are
doubtlessly the preferred starting point for lead optimiza-
tion. Moreover, it was reported by Nezami et al. that in
order to overcome the enthalpy-entropy compensation
effect, an optimal balance with the free energy partitioned
approximately as one-third enthalpy and two-thirds entro-
py (TΔS) was required.40 Only compounds with this com-
bination reached nanomolar affinity to the target
plasmepsin II in the reported case; the same conclusion
could also be drawn for allosteric glycogen phosporylase
inhibitors in a recent study.38 In that sense, calorimetric
analysis has disclosed 8Z rather as a “dead end” because the
balance has been reached already (ΔH/ΔG = 29.5%) and
any addition of another chemical moiety would provoque
enthalpy/entropy compensation effects without substan-
tially increasing ΔG. Similarly, the thermodynamic binding
data for 9Z revealed the compound as a less appropriate
starting point to escape enthalpy-entropy compensation,
even though the AC50 was two times lower than that of 2Z
(Table 1).
In summary, 12Z and 13Z were found to possess the most
ideal thermodynamic lead profile among the compounds
analyzed by ITC because the enthalpic term strongly pre-
dominates the binding free energy.
Another relationship became apparent when the thermo-
dynamic binding profiles were correlated with the activatory
potency of the compounds. We observed a rough correlation
between the AC50 values for PDK1 activation and the
dissociation constants as measured by ITC (Figure 6B). This
correlation suggested that for most compounds, the binding
affinity to the intermediate active conformation of the
kinase, which might be represented rather by the AC50, does
not differ strongly from the affinity to the activated con-
formation of the kinase, which might be described by the Kd.
The variations in activation efficacy (Amax) did not affect this
correlation. The relationship between AC50 and Kd was not
self-evident because binding to the intermediate active state
and dissociation of the complex after the conformational
change could be perceived as two separate events character-
ized by divergent affinity constants. Our finding rather
suggests that during the allosteric activation by the majority
of compounds, only subtle and rapid changes occur within
the PIF-binding pocket itself, while the overall geometry of
the PIF-binding pocket remains unaltered.
4. Conclusions
The compounds presented herein represent a rare case of
truly allosteric compounds targeting a regulatory binding site
on a protein kinase catalytic domain that is not adjacent to or
overlappingwiththe ATP bindingsite. Itshouldbementioned
that many inhibitors reported in the literature to be “allos-
teric” still bind to the ATP binding site at least partially. As in
the case of imatinib mesylate (gleevec), the term is used to
signify that an inactive, open conformation of the kinase is
stabilized.21 This study is, to the best of our knowledge, the
first one correlating the structural motifs of allosteric activa-
tors with both the effects on catalytic activity (Amax and AC50)
and the energetics of the interaction (quantified by isothermal
titration calorimetry).
Altogether, our work might pave the way to the develop-
ment of allosteric modulators for PDK1 based on the follow-
ing findings: (i) We have defined the minimum requirements
for a small compound to function as a PDK1 activator, two
aromatic moieties connected by an aliphatic chain, bearing a
two atom membered side chain with a free carboxylic group; a
V-shaped overall conformation of the aryl rings toward each
other is also required to achieve complementarity to the
binding pocket.
2.5. Implications of the Calorimetric Characterization for
Lead Selection and Optimization. In a scenario where the
goal is to increase binding affinity of small molecules, a
proven strategy is to choose as lead candidates compounds
that show enthalpy-dominated binding.35,38 The reason
is that enthalpy-driven interactions are more difficult to

Article
These prototype compounds, despite the low molecular
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 15 4691
TSQ quantum (Thermo Electron Corporation) instrument
equipped with an 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 4.0ꢀ105 Pa, an auxiliary gas pressure of 1.0ꢀ105
Pa, a capillary temperature of 400 °C, 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 mmꢀ
3 mm) column (Macherey-Nagel) was used as stationary phase.
The solvent system consisted of water containing 0.1% TFA (A)
and 0.1% TFA in acetonitrile (B). HPLC-method: flow rate
400 μL/min. The percentage of B started at an initial of 5%, was
increased up to 100% during 16 min, kept at 100% for 2 min,
and flushed back to the 5% in 2 min. All masses were reported as
those of the protonated parent ions. Melting points were
determined on a Mettler FP1 melting point apparatus and are
uncorrected.
masses, already displayed activities in the low μM range,
suggesting that mass and ligand efficiency should remain in
a reasonable range after lead optimization. (ii) Some im-
portant CH-π interactions between ligand and receptor
that can essentially contribute to increased selectivity and
potency of PIF-pocket directed compounds have been
identified by virtue of their favorable binding enthalpy.
(iii) Calorimetric analysis revealed compounds binding
with variable proportions of entropy and enthalpy, allow-
ing for selection of appropriate structures as starting points
to further improve potency. Thus we were able to establish
that the HM/PIF-binding pocket of PDK1 as a druggable
site despite being located at the protein surface and phy-
siologically participating in extended protein-peptide in-
teractions.
Because hydrophobic pockets homologous to our target
binding site are conserved in many protein kinases comprising
the whole AGC family, our approach may be suitable for
similar developments of allosteric modulators for a broad
range of kinase targets. Thereby the mechanisms of action
would consist of allosteric effects on the kinase catalytic
activity, blocking of inter- or intramolecular protein-protein
interactions, or a combination of both, allowing a more subtle
manipulation of intracellular signaling pathways.
In contrast to the more established kinase inhibition stra-
tegies, direct allosteric activation of a protein kinase is a rather
new scientific concept. In the case of PDK1, activation of the
catalytic activity is achieved by binding to the HM/PIF-
binding pocket, which at the same time blocks the activation
of all substrates that require transient interaction with the
pocket. This is the case for all known PDK1 substrates,
including S6K, RSK, and some PKC isoforms but not for
PKB. Thus the compounds might act differentially on the two
subsets of PDK1 substrates, which either require docking
interaction with the PIF-binding pocket or not. Hence, in-
depth investigation of the compounds’ effects on the cellular
signaling pathways, using genetic tools and detailed phos-
phorylation analysis, are required and will be subject of our
future studies. In particular, detailed analyses in a cellular
setting will be required to study the biological effects of PIF-
binding pocket targeting compounds with high (e.g., 6Z) vs
low (e.g., 8Z) concomitant allosteric activation of PDK1
catalytic activity.
5.1.2. Method D: Hydrolysis. A solution of the corresponding
ester (2-13E/Za) (1 equiv) and NaOHaq (3 equiv) in EtOH was
refluxed for 4 h. After the completion of 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
organic layers were collected, washed with brine (20 mL), dried
over MgSO4, and evaporated to afford a residue, which was
purified by crystallization to afford the acids.
5.1.2.1. (E)-5-(4-Chlorophenyl)-3-phenylpent-2-enoic Acid (2E).
Synthesized according to method D using compound 2Ea (0.16 g,
0.51 mmol) and NaOHaq (1.1 mL, 5.1 mmol); white solid; yield:
1
0.105 g (72%); mp 117-118 °C. H NMR (CDCl3, 500 MHz):
δ 2.70-2.74 (m, 2H), 3.37-3.41 (m, 2H), 6,11 (s, 1H), 7.12 (d, 3J=
8.5 Hz, 2H), 7.22 (d, 3J=8.2 Hz, 2H), 7.39-7.42 (m, 3H), 7.44-
7.48 (m, 2H). 13C NMR (CDCl3, 125 MHz): δ 17.2, 33.5, 60.2,
126.9, 127.6, 128.0, 128.1, 129.5, 131.2, 141.3, 144.8, 148.8, 180.5.
NOESY experiment (1H NMR (CDCl3, 500 MHz) showed a
cross peak between the signal at 7.44-7.48 (m, 2H) and the signal
at 6.11 (s, 1H). LC/MS (þESI): m/z 287.6 [MHþ], Rt = 14.13
(g99%).
5.1.2.2. (Z)-5-(4-Chlorophenyl)-3-phenylpent-2-enoic Acid (2Z).
Synthesized according to method D using compound 2Za (0.139
g, 0.44 mmol) and NaOHaq (1.1 mL, 4.4 mmol); white solid; yield:
1
0.093 g (74%); mp 114-115 °C. H NMR (CDCl3, 500 MHz):
δ 2.64-2.67 (m, 2H), 2.71-2.77 (m, 2H), 5.87 (s, 1H), 7.04 (d, 3J=
8.5 Hz, 2H), 7.18 (m, 2H), 7.24 (d, 3J=8.5 Hz, 2H), 7.34 (m, 3H).
13C NMR (CDCl3, 125 MHz): δ 15.9, 32.8, 60.2, 127.1, 127.6,
128.0, 129.7, 131.8, 139.8, 142.4, 145.3, 170.5. NOESY experiment
(1H NMR (CDCl3, 500 MHz) showed a cross peak between the
signals at 2.64-2.67 (m, 2H) and at 2.71-2.77 (m, 2H), and the
signal at 5.87 (s, 1H). LC/MS (þESI):m/z 287.5[MHþ],Rt=13.56
(g99%).
5. Experimental Section
5.1.2.3. (Z)-3-Phenyl-5-(4-(trifluoromethyl)phenyl)pent-2-enoic
Acid (6Z). Synthesized according to method D using compound
6Za (0.15 g, 0.43 mmol) and NaOHaq (1.4 mL, 4.31 mmol); white
5.1. Chemistry. 5.1.1. General. 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 TLC analysis on Alugram SIL G/UV254 (Ma-
cherey-Nagel). Visualization was accomplished with UV light
and KMnO4 solution. 1H NMR, 13C, and 2D-NOESY spectra
were recorded at 500 MHz using a Bruker DRX-500 spectro-
meter. 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 Hz. The purities of the tested compounds 2Z-21 were
determined by HPLC coupled with mass spectrometry and were
higher than 95% for all compounds except 8E and 13Z, which
exhibited at least 93% and 94% purity, respectively. Mass
spectrometric analysis (HPLC-ESI-MS) was performed on a
1
solid; yield: 0.113 g (82%); mp 125-127 °C. H NMR (CDCl3,
500 MHz): δ 2.70-2.78 (m, 4H), 5.85 (s, 1H), 7.14-7.16 (m, 2H),
7.19-7.29 (m, 2H), 7.30-7.37 (m, 3H), 7.49-7.54 (m, 2H). 13C
NMR (CDCl3, 125 MHz): δ 33.5, 34.9, 125.3 (d, J (C, F)=3.7 Hz,
CH), 125.9, 127.2, 127.5, 128.1, 128.5, 128.6, 128.8, 130.0, 133.5,
144.6, 160.6, 170.0. LC/MS (þESI): m/z 321.5 [MHþ]; Rt=13.74
(g95%).
5.1.2.4. (Z)-5-(3,4-Dichlorophenyl)-3-phenylpent-2-enoic Acid
(7Z). Synthesized according to method D using compound 7Za
(0.10 g, 0.29 mmol) and NaOHaq (0.96 mL, 2.86 mmol); white
solid; yield: 0.085 g (91%); mp 130-134 °C. 1H NMR (CDCl3,
500 MHz): δ 2.62-2.65 (m, 2H), 2.74-2.77 (m, 2H), 5.86 (s,
4
3
1H), 6.93 (dd, J=2.2, J=8.2 Hz, 2H), 7.16-7.21 (m, 2H),
7.31-7.39 (m, 4H). 13C NMR (CDCl3, 125 MHz): δ 32.5, 35.5,
116.8, 125.8, 126.9, 127.5, 127.9, 128.0, 128.2, 130.2, 133.3,
139.9, 140.4, 160.3, 169.9. LC/MS (þESI): m/z 322.7 [MHþ];
Rt=14.32 (g99%).

4692 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 15
Stroba et al.
5.1.2.5. (Z)-5-(2,4-Dichlorophenyl)-3-phenylpent-2-enoic Acid
(8Z). Synthesized according to method D using compound 8Za
(0.10 g, 0.31 mmol) and NaOHaq (0.32 mL, 0.93 mmol); white
Acknowledgment. We thank Prof. Dr. S. Zeuzem for his
generous support of the R.M.B. lab. We acknowledge
the support by the Deutsche Forschungsgemeinschaft (BI
1
solid; yield: 0.09 g (87%); mp 122-124 °C. H NMR (CDCl3,
1044/2-2), by the Europrofession Foundation (Saarbrucken,
ꢀ
¨
Germany) to R.M.B. and M.E., by the Deutsche Jose
500 MHz): δ 2.67-2.74 (m, 4H), 5.88 (s, 1H), 7.03 (d, 3J =
8.2 Hz, 1H), 7.14 (dd, 3J=2.2, 4J=8.2 Hz, 1H), 7.19-7.21 (m,
2H), 7.33-7.39 (m, 4H). 13C NMR (CDCl3, 125 MHz): δ 32.5,
40.2, 116.9, 127.1, 127.3, 128.2, 128.4, 129.4, 131.1, 132.7, 134.5,
138.7, 141.4, 160.5, 176.5. LC/MS (þESI): m/z 322.7 [MHþ];
Rt=14.63 (g98%).
Carreras Leukamie-Stiftung (DJCLS R 06/07) to M.E., and
by the GoBio grant from the German Federal Ministry of
Education and Science to R.M.B. and M.E.
5.1.2.6. (E)-5-(Naphthalen-2-yl)-3-phenylpent-2-enoic Acid
(12E). Synthesized according to method D using compound
12Ea (0.17 g, 0.51 mmol) and NaOHaq (1.70 mL, 5.14 mmol);
white solid; yield: 0.143 g (93%); mp 147-150 °C. H NMR
Supporting Information Available: Synthetic procedures and
NMR spectroscopic data of compounds 2c-14c, 21c, 2b-13b,
15b-18b, 21b, 2a-20a, 21Ea, 3-5Z, 9-11Z, 14-15Z, 18Z, 3-
11E, 13-14E, 18E, 16-17E/Z, 19, 20, 21; 2D-NOESY-1H
NMR spectroscopic data on 2Z/2E. This material is available
free of charge via the Internet at http://pubs.acs.org.
1
(CDCl3, 500 MHz): δ 2.91-2.94 (m, 2H), 3.49-3.53 (m, 2H),
6.14 (s, 1H), 7.38-7.44 (m, 6H), 7.51-7.53 (m, 2H), 7.62 (s, 1H),
7.74-7.79 (m, 3H). 13C NMR (CDCl3, 125 MHz): δ 33.3, 35.4,
116.5, 125.2, 126.5, 126.8, 127.2, 127.4, 127.9, 128.4, 128.7,
128.7, 129.4, 130.9, 132.1, 138.6, 138.7, 162.4, 170.9. LC/MS
(þESI): m/z 303.6 [MHþ]; Rt=14.54 (g99%).
References
(1) Cohen, P. Protein kinases;the major drug targets of the twenty-
first century? Nat Rev. Drug Discovery. 2002, 1, 309–315.
(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.
(4) 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.
(5) 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.
(6) Frodin, M.; Antal, T. L.; Dummler, B. A.; Jensen, C. J.; Deak, M.;
Gammeltoft, S.; Biondi, R. M. A phosphoserine/threonine-bind-
ing pocket in AGC kinases and PDK1 mediates activation by
hydrophobic motif phosphorylation. EMBO J. 2002, 21, 5396–
5407.
(7) Biondi, R. M.; Komander, D.; Thomas, C. C.; Lizcano, J. M.; Deak,
M.; Alessi, D. R.; van Aalten, D. M. High resolution crystal
structure of the human PDK1 catalytic domain defines the regula-
tory phosphopeptide docking site. EMBO J. 2002, 21, 4219–4228.
(8) 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 Czeta (PKCzeta) and PKC-related kinase 2 by
PDK1. J. Biol. Chem. 2000, 275, 20806–20813.
(9) Collins, B. J.; Deak, M.; Arthur, J. S.; Armit, L. J.; Alessi, D. R. In
vivo role of the PIF-binding docking site of PDK1 defined by
knock-in mutation. EMBO J. 2003, 22, 4202–4211.
(10) 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.
5.1.2.7. (Z)-5-(Naphthalen-2-yl)-3-phenylpent-2-enoic Acid
(12Z). Synthesized according to method D using compound
12Za (0.180 g, 0.54 mmol) and NaOHaq (1.80 mL, 5.45 mmol);
1
white solid; yield: 0.150 g (92%); mp 117-120 °C. H NMR
(CDCl3, 500 MHz): δ 2.84-2.89 (m, 4H), 5.92 (s, 1H), 7.21 (dd,
4J=1.7, 3J=8.2 Hz, 2H), 7.30-7.47 (m, 6H), 7.47-7.52 (m, 1H),
7.55 (s, 1H), 7.74-7.82 (m, 2H). 13C NMR (CDCl3, 125 MHz):
δ 34.1, 42.5, 116.9, 125.6, 126.3, 126.7, 127.2, 127.5, 127.7, 127.9,
128.3, 128.4, 128.5, 128.7, 133.8, 138.3, 139.4, 159.8, 173.3. LC/
MS (þESI): m/z 303.5 [MHþ]; Rt=14.02 (g98%).
5.1.2.8. (Z)-5-(1H-Indol-3-yl)-3-phenylpent-2-enoic Acid
(13Z). Synthesized according to method D using compound
13Za (0.20 g, 0.63 mmol) and NaOHaq (0.65 mL, 1.89 mmol);
1
white solid; yield: 0.135 g (73%); mp 160-162 °C. H NMR
(CD3OD, 500 MHz): δ 2.68-2.71 (m, 2H), 2.79-2.82 (m, 2H),
5.89 (s, 1H), 7.05 (t, 3J=7.9 Hz, 1H), 6.95 (dt, 4J=0.9, 3J=7.6 Hz,
1H), 7.05 (dt, 4J=0.9, 3J=7.6 Hz, 1H), 7.10-7.11 (d, 4J=2.2 Hz,
1H), 7.23-7.25 (m, 2H), 7.29-7.38 (m, 4H), 7.41 (d, 3J=7.9 Hz,
1H), 10.78 (s, NH), 11.85 (s, OH). 13C NMR (CD3OD, 125
MHz): δ 24.6, 42.3, 112.2, 115.1, 118.7, 119.2, 119.5, 122.3, 123.0,
128.5, 128.6, 128.7, 128.9, 138.2, 141.4, 160.7, 169.8. LC/MS
(þESI): m/z 292.5 [MHþ]; Rt=11.93 (g94%).
5.2. Biology. 5.2.1. Protein Kinases and Kinase Assays. All
procedures were performed exactly as described previously.22 In
brief, PDK1 mutants were generated using site directed mutagen-
esis. Wild type and mutant protein kinases were expressed in
HEK293 cells after transient transfection as GST fusion proteins
and purified using glutathione sepharose. Cell-free protein kinase
activity assays were performed using T308tide as a substrate
peptide, and started using γ32P-ATP. Phosphorylated peptides
were spotted on P81 phosphocellulose paper (Whatman), washed
by diluted phosphoric acid and incorporated 32P quantified in a
PhosphoImager.
(11) Islam, I.; Bryant, J.; Chou, Y. L.; Kochanny, M. J.; Lee, W.;
Phillips, G. B.; Yu, H.; Adler, M.; Whitlow, M.; Ho, E.; Lentz, D.;
Polokoff, M. A.; Subramanyam, B.; Wu, J. M.; Zhu, D.; Feldman,
R. I.; Arnaiz, D. O. Indolinone based phosphoinositide-dependent
kinase-1 (PDK1) inhibitors. Part 1: design, synthesis and biological
activity. Bioorg. Med. Chem. Lett. 2007, 17, 3814–3818.
(12) Sato, S.; Fujita, N.; Tsuruo, T. Interference with PDK1-Akt
survival signaling pathway by UCN-01 (7-hydroxystaurosporine).
Oncogene 2002, 21, 1727–1738.
5.2.2. Isothermal Titration Calorimetry. The association reac-
tions of PDK1 with the Z isomers were quantified by isothermal
titration calorimetry using the high precision VP-ITC titration
calorimetric system (MicroCal Inc., MA) and protocols pre-
viously described.41 Briefly, PDK150-359 and the compounds
were dissolved in the same buffer (50 mM TrisHCl, pH 7.5,
200 mM NaCl, 1 mM DTT). The binding enthalpies were
obtained by injecting the activators (450 μM) into the calori-
metric cell containing the enzyme (20 μM). The solutions
were thoroughly degassed under vacuum, and each experiment
was performed at 20 °C by one injection of 2 μL followed by
29 injections of 10 μL with 210 s between injections using a
290 rpm rotating syringe. Heat signals were corrected for the
heats of dilution and normalized to the amount of compound
injected. Normalization and deconvolution of the binding
isotherms was carried out using Origin742 provided by the
manufacturer.
(13) Stauffer, F.; Maira, S. M.; Furet, P.; Garcia-Echeverria, C. Imi-
dazo[4,5-c]quinolines as inhibitors of the PI3K/PKB-pathway.
Bioorg. Med. Chem. Lett. 2008, 18, 1027–1030.
(14) Peifer, C.; Alessi, D. R. Small-Molecule Inhibitors of PDK1.
ChemMedChem 2008, 3, 1810–1838.
(15) Sausville, E. A.; Arbuck, S. G.; Messmann, R.; Headlee, D.; Bauer,
K. S.; Lush, R. M.; Murgo, A.; Figg, W. D.; Lahusen, T.; Jaken, S.;
Jing, X.; Roberge, M.; Fuse, E.; Kuwabara, T.; Senderowicz, A. M.
Phase I trial of 72-h continuous infusion UCN-01 in patients with
refractory neoplasms. J. Clin. Oncol. 2001, 19, 2319–2333.
(16) Welch, S.; Hirte, H. W.; Carey, M. S.; Hotte, S. J.; Tsao, M. S.;
Brown, S.; Pond, G. R.; Dancey, J. E.; Oza, A. M. UCN-01 in
combination with topotecan in patients with advanced recurrent

Article
ovarian cancer: a study of the Princess Margaret Hospital Phase II
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 15 4693
(29) Grover, J. R.; Walters, E. A.; Hui, E. T. Dissociation Energies of
the Benzene Dimer and Dimer Cation. J. Phys. Chem. 1987, 91,
3233–3237.
(30) Laatikainen, R.; Ratilainen, J.; Sebastian, R.; Santa, H. NMR
study of aromatic-aromatic interactions for benzene and
some other fundamental aromatic systems using alignment of
aromatics in strong magnetic field. J. Am. Chem. Soc. 1995, 117,
11006–11010.
(31) Nishio, M.; Umezawa, Y.; Hirota, M.; Takeuchi, Y. The CH/π
interaction: significance in molecular recognition. Tetrahedron
1995, 51, 8665–8701.
(32) Gilli, P.; Ferretti, V.; Gilli, G.; Borea, P. A. Enthalpy-Entropy
Compensation in Drug Receptor Binding. J. Phys. Chem. 1994, 98,
1515–1518.
(33) Grunwald, E.; Steel, C. Solvent Reorganization and Thermody-
namic Enthalpy-Entropy Compensation. J. Am. Chem. Soc. 1995,
117, 5687–5692.
(34) Grunwald, E. Thermodynamics of Molecular Species; Wiley &
Sons: New York, 1997.
(35) Velazquez-Campoy, A.; Luque, I.; Todd, M. J.; Milutinovich, M.;
Kiso, Y.; Freire, E. Thermodynamic dissection of the binding
energetics of KNI-272, a potent HIV-1 protease inhibitor. Protein
Sci. 2000, 9, 1801–1809.
(36) Dunitz, J. D. Win some, lose some: enthalpy-entropy compensa-
tion in weak intermolecular interactions. Chem. Biol. 1995, 2, 709–
712.
consortium. Gynecol. Oncol. 2007, 106, 305–310.
(17) Edelman, M. J.; Bauer, K. S.Jr.; Wu, S.; Smith, R.; Bisacia, S.;
Dancey, J. Phase I and pharmacokinetic study of 7-hydroxystaur-
osporine and carboplatin in advanced solid tumors. Clin. Cancer.
Res. 2007, 13, 2667–2674.
(18) Kortmansky, J.; Shah, M. A.; Kaubisch, A.; Weyerbacher, A.; Yi,
S.; Tong, W.; Sowers, R.; Gonen, M.; O’Reilly, E.; Kemeny, N.;
Ilson, D. I.; Saltz, L. B.; Maki, R. G.; Kelsen, D. P.; Schwartz, G.
K. Phase I trial of the cyclin-dependent kinase inhibitor and protein
kinase C inhibitor 7-hydroxystaurosporine in combination with
fluorouracil in patients with advanced solid tumors. J. Clin. Oncol.
2005, 23, 1875–1884.
(19) Davies, S. P.; Reddy, H.; Caivano, M.; Cohen, P. Specificity and
mechanism of action of some commonly used protein kinase
inhibitors. Biochem. J. 2000, 351, 95–105.
(20) Bain, J.; McLauchlan, H.; Elliott, M.; Cohen, P. The specificities of
protein kinase inhibitors: an update. Biochem. J. 2003, 371, 199–204.
(21) Kiselyov, A.; Balakin, K. V.; Tkachenko, S. E.; Savchuk, N. P.
Recent progress in development of non-ATP competitive small-
molecule inhibitors of protein kinases. Mini Rev. Med. Chem. 2006,
6, 711–717.
(22) 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 molec-
ular weight compounds. EMBO J. 2006, 25, 5469–5480.
(23) Kishore, A. H.; Vedamurthy, B. M.; Mantelingu, K.; Agrawal, S.;
Reddy, B. A.; Roy, S.; Rangappa, K. S.; Kundu, T. K. Specific
small-molecule activator of Aurora kinase A induces autopho-
sphorylation in a cell-free system. J. Med. Chem. 2008, 51, 792–797.
(24) Conn, P. J.; Christopoulos, A.; Lindsley, C. W. Allosteric mod-
ulators of GPCRs: a novel approach for the treatment of CNS
disorders. Nat. Rev. Drug Discovery 2009, 8, 41–54.
(37) Lafont, V.; Armstrong, A. A.; Ohtaka, H.; Kiso, Y.; Mario Amzel,
L.; Freire, E. Compensating enthalpic and entropic changes hinder
binding affinity optimization. Chem. Biol. Drug Des. 2007, 69, 413–
422.
(38) Anderka, O.; Loenze, P.; Klabunde, T.; Dreyer, M. K.; Defossa, E.;
Wendt, K. U.; Schmoll, D. Thermodynamic characterization of
allosteric glycogen phosphorylase inhibitors. Biochemistry 2008,
47, 4683–4691.
(25) 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. High resolution complex
structure and allosteric effects of low molecular weight activators
on the protein kinase PDK1. Nat. Chem. Biol. 2009, DOI: 10.1038/
nchembio.208.
(39) Ruben, A. J.; Kiso, Y.; Freire, E. Overcoming roadblocks in lead
optimization: a thermodynamic perspective. Chem. Biol. Drug Des.
2006, 67, 2–4.
(40) Nezami, A.; Luque, I.; Kimura, T.; Kiso, Y.; Freire, E. Identifica-
tion and characterization of allophenylnorstatine-based inhibitors
of plasmepsin II, an antimalarial target. Biochemistry 2002, 41,
2273–2280.
(41) Schaeffer, F.; Matuschek, M.; Guglielmi, G.; Miras, I.; Alzari, P.
M.; Beguin, P. Duplicated dockerin subdomains of Clostridium
thermocellum endoglucanase CelD bind to a cohesin domain of
the scaffolding protein CipA with distinct thermodynamic para-
meters and a negative cooperativity. Biochemistry 2002, 41, 2106–
2114.
(26) Yasui, S.; Fujii, M.; Ohno, A. NAD(P)þ-NAD(P)H Models. 64.
The Quantitative Elucidation of the Mechanistic Aspects in the
Silica Gel-Catalyzed Reduction of R,β-Unsaturated Carbonyl
Compounds by a Model of NAD(P)H. Bull. Chem. Soc. Jpn.
1987, 60, 4019–4026.
(27) Yoshizawa, K.; Toyota, S.; Toda, F.; Csoregh, I. Preparative and
¨
mechanistic studies of solvent free Ra-Stoermer reactions. Green
Chem. 2003, 5, 353–356.
(28) Beck, J. A. DirectSynthesisofBenzo[b]thiophene-2-carboxylate Esters
involving Nitro Displacement. J. Org. Chem. 1972, 37, 3224–3226.
(42) Wiseman, T.; Williston, S.; Brandts, J. F.; Lin, L. N. Rapid
measurement of binding constants and heats of binding using a
new titration calorimeter. Anal. Biochem. 1989, 179, 131–137.