Design, synthesis and 3D-QSAR studies of novel 1,4-dihydropyridines as TGFβ2/Smad inhibitors

Article abstract of DOI:10.1016/j.ejmech.2015.03.027

Targeting TGFβ2/Smad signaling is an attractive strategy for several therapeutic applications given its role as a key player in many pathologies, including cancer, autoimmune diseases and fibrosis. The class of b-annelated 1,4-dihydropyridines (DHPs) repr

Full text of DOI:10.1016/j.ejmech.2015.03.027

Accepted Manuscript
Design, synthesis and 3D-QSAR studies of novel 1,4-dihydropyridines as TGFβ/
Smad inhibitors
Daniel Längle, Viktoria Marquardt, Elena Heider, Brigita Vigante, Gunars Duburs,
Iveta Luntena, Dirk Flötgen, Christopher Golz, Carsten Strohmann, Oliver Koch,
Dennis Schade
PII:
S0223-5234(15)00194-4
10.1016/j.ejmech.2015.03.027
EJMECH 7772
DOI:
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 20 January 2015
Revised Date: 11 March 2015
Accepted Date: 13 March 2015
Please cite this article as: D. Längle, V. Marquardt, E. Heider, B. Vigante, G. Duburs, I. Luntena, D.
Flötgen, C. Golz, C. Strohmann, O. Koch, D. Schade, Design, synthesis and 3D-QSAR studies of novel
1,4-dihydropyridines as TGFβ/Smad inhibitors, European Journal of Medicinal Chemistry (2015), doi:
10.1016/j.ejmech.2015.03.027.
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT
Design, synthesis and 3D-QSAR studies of novel
1
,4-dihydropyridines as TGFβ/Smad inhibitors
1
1
1
2
2
Daniel Längle, Viktoria Marquardt, Elena Heider, Brigita Vigante, Gunars Duburs, Iveta
2
1
1
1
1
Luntena, Dirk Flötgen, Christopher Golz, Carsten Strohmann, Oliver Koch, Dennis
1
Schade *
1
TU Dortmund University, Department of Chemistry & Chemical Biology, Otto-Hahn-Str. 6,
4
4227 Dortmund, Germany
2
Latvian Institute of Organic Synthesis, Aizkraukles 21, Riga LV-1006, Latvia
Corresponding author: dennis.schade@tu-dortmund.de, +49-231-7557083
*
KEYWORDS
Transforming Growth Factor β (TGFβ), structure-activity relationships (SARs), b-anellated
1
,4-dihydropyridines (1,4-DHPs), absolute configuration, crystal structure, molecular
electrostatic potential (MEP), 3D-QSAR model
1

ACCEPTED MANUSCRIPT
TOC Graphic
HIGHLIGHTS
•
•
•
•
Synthesis of novel DHPs and SAR studies reveal possibility of incorporating polar groups.
Structural examination of pathway selectivity against TGFβ over Activin A.
Absolute configuration of active (+)-enantiomers by X-ray crystallography.
3D-QSAR model reinforces a specific mode of action for DHPs as new TGFβ inhibitors,
which is largely independent from intrinsic compound properties (e.g., clogP)
2

ACCEPTED MANUSCRIPT
ABSTRACT
Targeting TGFβ/Smad signaling is an attractive strategy for several therapeutic applications
given its role as a key player in many pathologies, including cancer, autoimmune diseases and
fibrosis. The class of b-annelated 1,4-dihydropyridines (DHPs) represents promising novel
pharmacological tools as they interfere with this pathway in a novel fashion, i.e. through
induction of TGFβ receptor type II degradation. In the present work, >40 rationally designed,
novel DHPs were synthesized and evaluated for TGFβ inhibition, substantially expanding the
current understanding of the SAR profile. Key findings include that the 2-position tolerates a
wide variety of polar functionalities, suggesting that this region could possibly be solvent-
exposed within the (thus far) unknown cellular target. A structural explanation for pathway
selectivity is provided based on a diverse series of 4’’-substituted DHPs, including molecular
electrostatic potential (MEP) calculations. Moreover, the absolute configuration for the chiral
4
-position was determined by x-ray crystal analysis and revealed that the bioactive
(+)-enantiomers are (R)-configured. Another key objective was to establish a 3D-QSAR model
2
2
which turned out to be robust (r = 0.93) with a good predictive power (r
= 0.69). This data
pred
further reinforces the hypothesis that this type of DHPs exerts its novel TGFβ inhibitory mode of
action through binding a distinct target and that unspecific activities that would derive from
intrinsic properties of the ligands (e.g., lipophilicity) play a negligible role. Therefore, the present
study provides a solid basis for further ligand-based design of additional analogs or DHP
scaffold-derived compounds for hit-to-lead optimization, required for more comprehensive
pharmacological studies in vivo.
3

ACCEPTED MANUSCRIPT
1
. INTRODUCTION
The superfamily of transforming growth factor-β (TGFβ) ligands comprises of a series of
structurally closely related cytokines that modulate a plethora of cellular processes such as cell
division, proliferation, differentiation, migration, adhesion and apoptosis in a tissue
context-specific manner.(1) They play a key role during embryogenesis but also for homeostasis
in adult tissues. Disruption or pathologic alterations of these pathways are associated with a
number of diseases such as cancer, autoimmune diseases and fibrosis.(2) Therefore, the
pharmacologic modulation of TGFβ-dependent processes is of great therapeutic relevance,
underlined by the many ‘biologicals’ and small molecules currently in clinical trials for distinct
applications.(3, 4) In general, three strategies are being investigated for therapeutic intervention
with TGFβ signaling: 1) inhibition at the translational level by anti-sense oligonucleotides, 2)
blockade of ligand-receptor interaction using antibodies and 3) inhibition of TGFβ receptor
kinases (TGFβRI+II) by small molecules.(4, 5) The latter approach provides several advantages
over biological agents, because small molecules can be administered orally and may reach and
infiltrate target tissues more efficiently than biomacromolecules. In addition, target and pathway
selectivity can often be greater by using defined and pharmacologically optimized small
molecule agents.(6) In this regard, the herein presented compound class of dihydropyridines
might be of particular therapeutic value in a regenerative medicine setting.
We previously reported on a novel class of TGFβ inhibitors, b-annelated 1,4-dihydropyridines
(DHPs), which were identified in a phenotypic screen for small molecules that stimulate cardiac
differentiation of stem cells, with ITD-1 (1) as the hit compound (Figure 1).(7) It was shown that
these DHPs inhibit TGFβ signaling via a novel mode of action, which is the downregulation of
4

ACCEPTED MANUSCRIPT
the TGFβ type II receptor (TGFβRII), giving these 1,4-DHPs the name ITDs (i.e., inducers of
TGFβ receptor degradation). This novel mechanism sets these compounds apart from the
“classic” TGFβ inhibitors that typically target the receptor kinase domains, with the consequence
of redundant activity on closely related signaling pathways or kinases, respectively. An initial
focused medicinal chemistry study was done which revealed a rather steep SAR profile for the
observed biological activity.(8)
Here, we report on the extension of these SARs with additional, more versatile substituted and
functionalized derivatives addressing distinct shortcomings of our current SAR picture of these
compounds (Figure 1). Along these lines, it was key to determine the - up to now - unknown
absolute configuration of the active (+)-enantiomers. In addition, the entire dataset known until
today was used to establish a quantitative correlation between structure and activity via a 3D-
QSAR analysis. The primary aim was here to further reinforce the selective nature of observed
TGFβ inhibiting activity, possibly addressing one relevant target and ruling out unspecific
effects in the cellular assays. This work eventually sets the basis for the design of novel DHPs or
DHP scaffold-derived compounds for rational hit-to-lead optimization that will allow conducting
more comprehensive pharmacological studies in an in vivo setting.
5

ACCEPTED MANUSCRIPT
Ph
R3
pathway selectivity?
absolute configuration?
O
'exploring'
O
COOEt
R2
R1
size & functional groups?
polar / hydrophilic groups?
*
N
H
N
H
ITD-1 (1)
Figure 1. Chemical structure of cardiogenic b-annelated 1,4-DHPs as TGFβ/Smad inhibitors,
with ITD-1 (1) as the “hit compound” from preliminary studies,(8) highlighting regions of the
molecule to be explored for expanding key SAR features.
6

ACCEPTED MANUSCRIPT
2
. RESULTS AND DISCUSSION
2
.1. Chemistry
As indicated in the introduction, our preliminary understanding of the structure activity
relationships for this novel class of TGFβ/Smad signaling inhibitors has been limited and
generated rather lipophilic derivatives which, accordingly, suffer from the respective
pharmaceutical and pharmacokinetic shortcomings (e.g., solubility, ADME aspects, formulation
technologies). Therefore, several DHP derivatives were designed to incorporate heteroatom-
containing, polar functional groups but also to explore steric aspects of certain regions of the
molecules.
Scheme 1 summarizes the synthetic routes to novel DHPs with various substituents in 2-, 3-,
4
-, 6- and 7-position. Using such classic Hantzsch conditions, it was possible to access many of
the desired compounds 1, 6-21 with yields ranging from 17-68%. Through replacement of the
β-dicarbonyl component 2 (mostly dimedone) with appropriate acetylenedicarboxylates 3, 2,3-
diester substituted DHPs were readily available. Notably, these 2-carboxylic ester groups could
be selectively hydrolyzed under basic conditions (i.e., 3 equiv of NaOH in EtOH) to give the free
carboxylic acids 20a and 21a in good yields (68-87%) without compromising the 3-esters. This
underlines the more stable carbamate nature of the 3-esters through conjugation with the
1
,4-DHP core. For the introduction of a nitrogen atom in position 6 (X = NH, Scheme 1),
piperidine-2,4-dione instead of dimedone (2) was used and gave compound 12 in 36% yield.
If the desired aldehyde was not commercially available, 4’-bromophenyl-derivative 7 was
prepared as the starting material and subsequently coupled under Suzuki conditions to afford
compounds 22-32 in yields between 15-77% under non-optimized conditions (Scheme 2). For
7

ACCEPTED MANUSCRIPT
compound 32 the O-TBDMS-protected 4-hydroxyphenylboronic acid was used in the Suzuki
reaction, followed by deprotection with TBAF in THF.
1
2
3
cmpd
X
Y
R
R
R
yield route
1
6
7
8
9
CH
CH
CH
CH
CH
CH
CH
2
2
2
2
2
2
2
CMe
CMe
CMe
CMe
CMe
CMe
CMe
2
2
2
2
2
2
2
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Et
COOEt
Ph
38%
40%
38%
37%
17%
33%
23%
36%
71%
55%
21%
31%
68%
67%
33%
42%
A
A
A
A
A
A
A
A
B
B
B
B
B
B
B
B
COOEt
Br
COOEt
4-BrPh
Ethinyl
Ph
COOEt
CONHEt
COOtBu
COOAllyl
COOEt
1
1
1
1
1
1
1
1
1
0
1
2
3
4
5
6
7
8
Ph
Ph
NH
CH
Ph
2
CH
CH
CH
CH
CH
CH
CH
CH
CMe
CMe
CMe
CMe
CMe
CMe
CMe
CMe
COOC12
COnPr
COnBu
COEt
H
25
Ph
2
2
2
2
2
2
2
2
2
Ph
2
2
2
2
2
2
2
Ph
Ph
Me
Me
Me
CONHBn
Ph
COO(CH
) OnPr Ph
2 2
19*
CN
Ph
Ph
2
0
COOMe COOMe
8

ACCEPTED MANUSCRIPT
2
1
CH
2
CMe
2
COOEt COOEt
Ph
53%
B
Scheme 1. Synthetic routes to novel TGFβ inhibiting b-anellated 1,4-DHPs with various
substituents in 2-, 3-, 4-, 6- and 7-position. * β-Aminocrotononitrile instead of alkane-1,3-dione
component and NH OAc was used.
4
cmpd R
yield
2
2
2
2
2
2
2
2
3
3
3
2
3
4
5
6
7
8
9
0
1
2
4’’-CF
4’’-SO
3
-Ph
77%
Me-Ph 77%
2
4’’-Cl-Ph
4’’-F-Ph
76%
72%
52%
50%
46%
39%
4’’-OCF -Ph
3
4’’-CN-Ph
4’’-Pyridinyl
4’’-CH
3
-Ph
4’’-NMe
-Ph 36%
2
4’’-OMe-Ph
4’’-OH-Ph
25%
15% (2 Steps)
Scheme 2. Synthesis of novel 4’-substituted b-annelated 1,4-DHPs.
9

ACCEPTED MANUSCRIPT
Scheme 3. Synthetic routes to novel polycyclic 1,4-DHPs. Reagents and conditions: (i)
Dimedone, piperidine, EtOH, H O; (ii) NH OAc, AcOH, 33% (2 steps); (iii) Indane-1,3-dione,
2
4
AcOH, H
2
SO
4
; (iv) AcOH, 34% (2 steps).
Moreover, two polycyclic 1,4-dihydropyridines were synthesized according to the general
Hantzsch concept, i.e. decahydroacridine 34 and hexahydroindeno[1,2-b]quinoline 36. For these
compounds, however, it was necessary to first isolate the initial condensation products under
distinct cyclization conditions.
2-Amino-substituted 1,4-dihydropyridines were accessible by refluxing amidinoacetic acid
ethyl ester 38, dimedone 37 and aldehyde 39 in a sodium ethanolic solution as described in the
literature (Scheme 4).(9)
10

ACCEPTED MANUSCRIPT
Scheme 4. Synthesis of novel 2-amino-substituted b-annelated 1,4-DHPs.
However, in order to access more basic amines in 2-position (i.e., not conjugated with the DHP
core, vinylogous carbamate), the use of a terminal amino group in the alkane-1,3-dione
derivative for the Hantzsch DHP synthesis proved challenging. For this reason, the 2-
bromomethyl compound 42 was prepared in moderate yields (51%) according to a procedure
described by Petrova et al., treating ITD-1 (1) with NBS for 3 days (Scheme 5).(10) In fact, this
2
-bromomethyl intermediate proved to be a key intermediate to install several functional groups
in 2-position. 42 was then reacted with potassium phthalimide as a protected precursor of the
primary amino derivative. Unfortunately, after removing the protecting group, the desired free
amine could not be isolated due to cyclization to lactame 43. Although a primary amine was not
accessible via the routes described above, secondary amines 44 and 45 could be successfully
prepared and isolated as hydrochlorides in 27% and 58% yields after reacting 2-bromomethyl
intermediate 42 with ethylamine or n-propylamine, respectively.
Furthermore, the above described condensation reaction causing cyclization to a 5-membered
ring (i.e., pyrrolo[3,4-b]quinolone) was used to explore a series of different tricyclic derivatives
(Scheme 5). Aside from this N-unsubstituted secondary lactam 43, the corresponding benzyl- and
ethyl analogs were prepared. N-ethyl lactam 46 was accessible from the N-ethylaminomethyl-
11

ACCEPTED MANUSCRIPT
substituted DHP 44 in good yields (59%) under basic conditions. An attractive alternative to this
reaction was discovered when incorporating the desired amine in the form of the 2-carboxylic
amide 17 (see Scheme 1)(8) followed by treatment with NBS for 24 hours, furnishing the desired
N-benzylated pyrrolo[3,4-b]quinolone 55 in 81% yield (not shown in any Scheme, see Exp.
Section). In a similar approach, reacting 2-bromomethyl intermediate 42 with KOH (in
chloroform) did not provide access to the primary alcohol since this intermediate rapidly
cyclized to furo[3,4-b]quinolone 47. Still, a non-cyclized analog (i.e., 2-methoxymethyl DHP,
4
8) was successfully obtained in high yields (91%) after treating 42 with KOH in methanol.
For the preparation of 1,4-dihydropyridines carrying distinct ester groups in 3-position, and
which were not accessible by the general procedure outlined in Scheme 1, a protocol was used
that proved useful for similar derivatives in the past (Scheme 4).(8) The already above described
possibility for orthogonal deprotection of the 2-carboxylic esters under basic conditions, while
leaving the 3-esters intact, implies a high chemical stability of the latter. In fact, hydrolysis of
ethyl ester 1 to carboxylic acid 49 could only be accomplished under strongly Lewis acidic
conditions with BCl . Subsequent alkylation with different alkyl bromides and potassium
3
carbonate in DMF delivered the desired esters 50-53.
12

ACCEPTED MANUSCRIPT
Scheme 5. Synthetic routes to novel 2- and 3-substituted b-annelated DHPs. Reagents and
1
conditions: i): NBS, MeOH; ii): KOH, MeOH; iii): H
2
NR , DCM; iv): KOH, EtOH: v): KOH,
3 2 3 3
CHCl ; vi): Potassium phthalimide, DMF; H NCH , 22% (2 Steps); vii): BCl , DCM; viii):
2
R Br, DMF.
13

ACCEPTED MANUSCRIPT
2
.2. X-Ray crystal structure analysis
In our earlier reports, we elaborated on the fact that the observed activity on cardiomyocyte
differentiation from murine embryonic stem cells and inhibition of TGFβ signaling is linked to
the (+)-enantiomers.(7, 8) Interestingly, these mechanistic studies also revealed that the
stereoselective mode of action was reversed for calcium channel antagonism since only the (‒)-
enantiomers strongly inhibited calcium transients in rat cardiomyocytes whereas the TGFβ
inhibiting (+)-enantiomers were weakly active.(8)
One of the designated goals of our ongoing work is to define a detailed picture of the
b-annelated DHPs’ pharmacophore for novel ligand-based approaches, especially since the
cellular target has not yet been identified. A key requirement was thus to determine the absolute
configuration of the active (+)-enantiomer of ITD-1 (54b) by x-ray crystallography. For this, the
synthesis of single enantiomers was done as reported before, i.e. via diastereomeric threonine-
based 3-esters.(8) However, it was crucial to use enantiomers of highest possible optical purity
for crystallization. High enantiomeric purity was achieved after separating diastereomeric
intermediates by preparative HPLC (eluent: H O/ACN/MeOH, 30/27/43) instead of the
2
previously reported tedious procedure that used gravity flow chromatography with a
toluene/EtOAc eluent system. Using this method, the ITD-1 (+)-enantiomer 54b was prepared in
ꢁꢂ
2
3% yields over three steps and an optical purity of >98%ee (ꢀ ꢃ = +74) (lit. +68)(8), and could
be crystallized for x-ray analysis by slow evaporation of an ethanolic solution.
The crystal structure of 54b revealed that the cardiomyogenic, TGFβ inhibiting (+)-enantiomer
is (R)-configured as depicted in Figure 2a. In accordance with recently reported x-ray crystal
data on differently 4-substituted, b-annelated DHPs,(11) the typical structural features of classic
1
,4-dihydropyridines are conserved for these DHPs, such as the flattened boat conformation and
14

ACCEPTED MANUSCRIPT
the pseudoaxial orientation of the 4-substituent. Notably, however, the dihedral (torsion) angle of
1
07 ° for the C4-C5-C18-C19 plane relative to the C6-C7 plane is rather shallow (54b),
compared to similar DHPs, e.g. a ring-open (non-fused) 4-biphenyl analog (91-92 °, torsion
angle)(11) or a b-annelated 4-phenyl analog (97-99 °, torsion angle).(12) This becomes also
obvious comparing the angles of 4-biphenyl substituents above the DHP core, with 118.9 ° for
5
4b and 104.3 ° and 116.3° for the ring-open 4-biphenyl analog or annelated 4-phenyl analog,
respectively (see Supporting Information, Figures S1-3). To which extent these conformational
aspects play a role in the context of TGFβ inhibition remains to be clarified and is subject of our
current research.
Figure 2. X-ray crystal structure analysis of the cardiomyogenic, TGFβ inhibiting
(R)-configured ITD-1 (+)-enantiomer 54b. a) ORTEP plot, chemical structure and optical
rotation data; b) Packing diagram of the title compound showing intermolecular hydrogen bonds
between N1 and the carbonyl oxygen of the b-annelant (symmetry codes: A = 2-y, 1+x-y, 1/3+z;
B = 1-x+y, 2-x, -1/3+z).
15

ACCEPTED MANUSCRIPT
2
.3. TGFβ/Smad inhibition
As outlined in the introduction, we aimed at expanding on our preliminary SAR info for this
subclass of b-annelated 1,4-DHPs as cardiomyogenic TGFβ inhibitors. This SAR has been steep
and generated rather lipophilic derivatives aside from the fact the many regions of the molecule
were not explored at all. Therefore, the newly synthesized compounds were designed to
incorporate heteroatom-containing, polar functional groups but also to explore steric aspects of
certain regions of the molecules. Compounds were evaluated in a TGFβ/Smad reporter gene
assay in HEK293T as described before,(8) and the data is summarized in Table 1.
Thus far, the 2-position of these DHPs as TGFβ inhibitors has not been investigated at all. We
tested several carboxylate, ether and amino functionalities for activity and found that quite a few
groups are indeed well-tolerated. For example, replacing the lipophilic 2-methyl group by a
2
1
-amino group decreased potency only to a low extent. 2-Amino DHP 40 had an IC50 value of
.3 µM compared to 0.8 µM of its methyl analog ITD-1 (1). The respective 4’-t-Bu analog 41
was ranging even in the submicromolar range (i.e., IC50 = 0.5 µM) implicating that the
introduction of a 2-amino group still provides access to the most potent DHP-based TGFβ
inhibitors.
Since the direct attachment of an amino group to the DHP core does not generate a ‘true
amine’ but a vinylogous carbamate instead, we reasoned to keep a methylene group spacer and
introduced the ethyl (44, IC = 3.4 µM) and n-propyl (45, IC = 1.6 µM) amines. Both of the
5
0
50
amines revealed loss in potency which was moderate compared to the 2-amino analog 40 (IC =
5
0
1
.3 µM) but at least 2-fold when compared to the parent DHP 1 (IC = 0.8 µM). However, this
50
little loss in potency may well be compensated when decorating other regions of the molecule
16

ACCEPTED MANUSCRIPT
with known affinity-increasing substituents, while at the same time buying-in an improved
aqueous solubility.
Interestingly, several O-functionalized analogs were also still active. 2-Methyl ether derivative
4
8 showed a TGFβ inhibitory activity of 78% at 2.5 µM (IC = 1.8 µM). Even replacement of
50
the 2-methyl group with carboxylic esters rendered the molecules active, although losing potency
by ca. 2-fold (20 and 21, IC50’s = 2.0 and 1.8 µM) compared to DHP-1. As exemplified with two
derivatives (20a and 21a), a carboxylic acid group pointed out the limitation of introducing a
negatively charged moiety in the 2-position since both compounds were inactive. However, it
cannot be excluded that this phenomenon is solely due to the limited cell permeability of such
free acids. Taken together, these findings support the hypothesis that the region surrounding the
2
-position could possibly be solvent-exposed, and is rather not key for high-affinity interaction
with its unknown cellular target.
Next, further building on the 3-carboxylic acid (ester) group, several strategic SAR extensions
were realized. Starting off with esters of specific size and branching we could support the
hypothesis that particularly C -C containing esters were most potent (i.e., IC = 0.5-0.7 µM),
3
4
50
while groups larger than C were detrimental to biological activity. Similar hints were already
4
obtained in our previous study.(8) To address the question whether a simple increase in
lipophilicity and/or enrichment in lipid membranes could contribute to potent TGFβ inhibition in
a cellular assay setup, long chain ethylene glycol- (18) and fatty (lauryl) alcohol (13) linker
moieties were evaluated but turned out inactive.
Importantly, the electronic and H-bond donor/acceptor qualities of the direct 3-substituent
were evaluated. Switching from the ethyl ester 1 to the respective amide 9 (inactive) revealed
that an H-bond donor diminishes TGFβ inhibitory activity. This finding is in accordance with the
17

ACCEPTED MANUSCRIPT
fact that the free 3-carboxylic acid is also inactive.(8) An interesting observation was the
moderate but still surprisingly high potency of a simple 3-cyano derivative (19, IC = 2.9 µM),
5
0
considering that a clear SAR and increase in potency was typically obtained for proper-sized
alkyl residues which were completely lacking here. This raised the question of the significance
of the H-bond acceptor properties close to the DHP core. Therefore, a series of ketones (14-16)
were introduced that indeed further underlined the relevance of this H-bond acceptor. Based on
the 3-substituent chain length, the direct ketone analog to ITD-1 (1, IC50 = 0.8 µM) turned out to
have high TGFβ inhibitory potency (14, IC = 0.5 µM).
5
0
Together, we have now compelling evidence that a specific and important molecular
interaction originates from the 3-position. In particular H-bond acceptor properties, exemplified
with a simple ketone, together with a C -C aliphatic, low-branched side chain defines the spatial
3
4
and electronic requirements for this TGFβ inhibitor pharmacophore. In a similar approach, via
incorporation of a nitrogen at the 6-position, i.e. the b-annelation site of the DHPs, the same
question was addressed as to whether an amide (lactam) abolished activity (12, 20% inhibition at
2
.5 µM).
From our previous endeavors with this compound class, the significance of the 4-substituent
was evident and has been further explored ever since. Although this region of the molecule does
not permit modifications that would alter the linear orientation along the 4-phenyl-4’ axis, a
distinct 4’’-substitution pattern of the biphenyl turned out to be key for the desired signaling
pathway selectivity, i.e. TGFβ versus Activin A.(8) In fact, a 4’’-CF -substituted DHP 22 did
3
exert a higher inhibitory efficiency against TGFβ versus Activin A compared to its 4’’-CH
3
analog 29. In order to provide a more comprehensive, SAR-based explanation to verify this
inhibitory profile, a larger series of 4’’-substituted DHPs were designed, synthesized and tested
18

ACCEPTED MANUSCRIPT
against both of these TGFβ superfamily ligands. The assay data results are summarized in Table
1
(TGFβ inhibition) and further illustrated in Figure 3 (TGFβ versus Activin A inhibition),
which reveals a clear SAR trend based on the Hammett constant of the 4’’-substituents.
Although overall inhibitory potency against TGFβ was comparably high for compounds bearing
either electron-withdrawing (= 7, 22, 24, 25, 26) or -donating groups (= 29, 31, 32, 30) of certain
size (74-95% inhibition at 2.5 µM), this activity did not parallel the determined anti-Activin A
activity. For the latter, the electronic nature of this 4’’-substituent indeed made a difference:
DHPs containing electron-withdrawing groups consistently exhibited a low interference with
Activin A signaling, e.g. <50% inhibition for 22 (4’’-CF ), 26 (4’’-OCF ) and 24 (4’’-Cl). In
3
3
accordance with this data, it was tempting to speculate that a further increase of pathway
selectivity could be realized through incorporation of 4’’-substituents with even higher Hammett
constant values than the OCF or CF groups. Therefore, sulfone DHP 23 and nitrile 27 were
3
3
synthesized and tested. However, the outcome was disappointing since not only pathway
selectivity was completely abrogated but also overall inhibitory potency was significantly
diminished. Concluding from these results, a small-sized electron withdrawing 4’’-substituent,
devoid of π-bonds, ensures the highest degree of selectivity against TGFβ over Activin A
inhibition.
In an attempt to explain the distinct pathway inhibitory profiles of 4’’-electron-withdrawing
and -donating group substituted DHPs, molecular electrostatic potentials (MEPs, Figure 3b-d)
were calculated for 22 (4’’-CF ), 29 (4’’-CH ) and the 4’’-unsubstituted ITD-1 (1). Interestingly,
3
3
against our intuitive expectation that electron-withdrawing substituents in 4’’-position would
generate an electron-deficient distal aromatic ring, the connolly surface of 22 (Fig. 3b) revealed
19

ACCEPTED MANUSCRIPT
the opposite: The 4’’-CF
3
group creates an electron-rich distal aromatic by pulling electron
density from the proximal 4-phenyl into the distal ring (solid arrow, Fig. 3b).
Figure 3. Structural basis for pathway selective inhibition of b-annelated DHPs: a) TGFβ versus
Activin A. Inhibition values were determined in a Smad4-binding element (SBE4) reporter gene
assay in HEK293T cells using either TGFβ-2 or Activin A for pathway induction.(8) Data
represent means ± SD (n = 2-4), normalized to DMSO controls (= 0% inhibition). Statistical
analysis: Two-paired student t-test p values: *** (p ≤ 0.0001), ** (p ≤ 0.001), * (p ≤ 0.05), ns
(
not significant). b-c) Connolly surface (probe radius = 1 Å) mapped with the molecular
electrostatic potential (MEP), quantum chemical calculations performed with Gaussian 03
revision E.01); Blue/red areas represent negative/positive electrostatic potentials.
(
20

ACCEPTED MANUSCRIPT
This finding supports the idea that this distal region of the molecule is important for potent
selective) interaction with the unknown cellular target and possibly several of the examined 4’’-
(
substituents exert additional beneficial effects due to specific molecular interactions within the
pocket of the unknown target. Obviously, the effect is less pronounced for a 4’’-CH -substituted
3
(
29, Fig. 3c) or 4’’-unsubstituted DHP (1, Fig. 3d). Therefore, illustration of MEP connolly
surfaces did not only extend the Hammett constant-based explanation of pathway selectivity
Fig. 3a), it also provided an unexpected insight into electron density distribution of this series of
-biphenyl DHPs. In this regard, yet another surprising observation was a high electron density
(
4
above the C8-methylene group in the b-annelant (dashed arrows, Fig. 3b-d) which was expected
to be rather around the N1 atom of the DHP core. The generation of an electron-rich center
above the b-annelant may represent one explanation for the fact that annelation is required for
TGFβ inhibiting activity.
In addition to a simple SAR extension of the 2-, 3- and 4-substituents, it was vital for our
understanding of the pharmacophore to increase the chemical space of these DHPs. A series of
compounds was synthesized and tested which contained an additional 5-membered d-annelant.
Lactams 43 and 46 as well as lactone 47 were weakly active underlining that the increase in
rigidity negatively affects the desired activity. An exception was the N-benzyl derivative 55 that
was comparably potent to the original ITD-1 with an IC50 of 0.8 µM, and therefore, represents an
‘outlier’ of the thus far devised SAR model for which we do not have an explanation yet.
However, this finding prompted us to explore other, even larger tri- and tetracyclic structures.
With the indenone 36 an aromatic residue was incorporated in analogy to the active N-benzyl
21

ACCEPTED MANUSCRIPT
lactam 55, but obviously much less floppy. Interestingly, 36 was also still quite active (IC50
=
1
.26 µM), similar to the symmetric tricyclic aliphatic derivative 34 (IC = 1.88 µM).
50
22

ACCEPTED MANUSCRIPT
a
Table 1. TGFβ inhibition of novel b-annelated 1,4-dihydropyridines.
%
inhibition
1
2
3
compd R
R
R
Y
IC50 (µM)
(
2.5 µM)
1
4
4
4
4
4
2
2
2
2
5
5
1
1
5
5
1
1
Me
NH
NH
COOEt
COOEt
COOEt
COOEt
Ph
Ph
tBu
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
0.85 ± 0.30
1.33 ± 0.21
0.51 ± 0.03
3.44 ± 0.33
1.60 ± 0.38
1.89 ± 0.33
2.01 ± 0.60
n.d.
83 ± 8
61 ± 2
89 ± 5
64 ± 27
57 ± 6
78 ± 14
46 ± 16
4 ± 5
0
2
2
1
4
CH
CH
CH
2
2
2
NHEt
5
NHnPr COOEt
8
OMe
COOEt
0
COOMe
COOH
COOEt
COOH
Me
COOMe
COOMe
COOEt
0a
1
1.80 ± 0.20
n.d.
56 ± 6
0 ± 2
1a
0
COOEt
COOnPr
COOiPr
COOAllyl
COOtBu
COOiBu
COOcPn
0.49 ± 0.24
0.66 ± 0.13
0.62 ± 0.02
0.67 ± 0.17
0.42 ± 0.27
1.11 ± 0.52
n.d.
93 ± 3
85 ± 3
91 ± 5
84 ± 1
91 ± 3
77 ± 12
0 ± 3
1
Me
1
Me
0
Me
3
Me
2
Me
8
Me
COO(CH ) OnPr
2 2
3
Me
COOC12
H
25
n.d.
0 ± 18
23

ACCEPTED MANUSCRIPT
9
1
1
1
1
1
2
2
2
3
3
2
7
2
3
2
2
2
8
4
4
4
5
1
3
3
Me
Me
Me
Et
CONHEt
CONHBn
CN
Ph
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
O
n.d.
31 ± 1
32 ± 13
44 ± 6
62 ± 8
84 ± 11
80 ± 2
88 ± 6
94 ± 2
95 ± 1
74 ± 6
93 ± 4
91 ± 2
94 ± 3
74 ± 16
95 ± 2
44 ± 1
52 ± 2
27 ± 11
59 ± 5
37 ± 4
28 ± 8
25 ± 15
82 ± 5
21 ± 2
70 ± 2
55 ± 6
7
9
6
4
5
2
9
6
2
1
4
Ph
5.19 ± 3.45
2.95 ± 0.68
1.63 ± 0.56
0.50 ± 0.13
0.56 ± 0.11
0.97 ± 0.29
0.35 ± 0.03
0.48 ± 0.06
0.73 ± 0.26
0.42 ± 0.21
0.76 ± 0.22
0.46 ± 0.25
1.23 ± 0.02
0.33 ± 0.15
2.76 ± 0.18
1.79 ± 0.42
6.14 ± 1.55
1.71 ± 0.60
4.30 ± 1.71
Ph
COEt
COnPr
COnBu
COOEt
COOEt
COOEt
COOEt
COOEt
COOEt
COOEt
COOEt
COOEt
COOEt
COOEt
COOEt
COOEt
-
Ph
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
-
Ph
Ph
4’’-CF
Ph
3
4’’-CH
4’’-OCF
4’’-OHPh
3
Ph
Ph
3
4’’-OCH
4’’-ClPh
4’’-BrPh
4’’-FPh
3
Ph
5
0
3
7
8
4’’-NMe
2
Ph
4’’-SO
MePh
2
4’’-CNPh
4’’-Pyridinyl
Ethinyl
7
3
6
5
2
6
4
-
-
-
-
-
-
-
-
-
NH n.d.
-
-
NEt 5.03 ± 1.31
NBn 0.87 ± 0.20
-
-
-
-
-
-
-
n.d.
-
-
1.26 ± 0.01
1.88 ± 0.27
-
-
24

ACCEPTED MANUSCRIPT
a
Data represent means ± SD of n = 2-4 independent experiments in a TGFβ/Smad reporter gene
b
assay. For this compound a single experiment was done. n.d., not determined.
2
.4. 3D-QSAR studies
Quantitative structure activity relationships (QSAR) describe the correlation of biological
activity with the chemical structure of a set of molecules. QSAR analyses have become a widely
used technique in drug design (13) since their first description in 1964 by Hansch and Free-
Wilson.(14, 15) They are commonly used to validate structure activity relationships as well as to
predict the biological activity of newly designed compounds. This approach is intrinsically most
powerful when building on data from biochemical assays with compound activities covering a
broad dose-response range. Nevertheless, also for cell-based data, 3D-QSAR information can be
of high value. In fact, such in vitro cellular studies have been applied for the validation and
QSAR analysis of biological activities of several 1,4-dihydropyridine-type of compounds, for
example regarding antitumor activity in HL-60 cells,(16) inhibition of amyloid beta peptide
accumulation in CHO cells,(17) antitubercular activity in a BACTEC 460 assay system,(18)
aside from classic calcium channel antagonist activity in guinea pig ileum smooth muscle
assays.(19) In all cases, the requirement for calculating a QSAR model is the assumption that all
evaluated derivatives share the same target or target binding site, and unspecific effects, such as
permeability, have no significance influence or do not decisively determine the desired activity,
respectively.
In the context of the TGFβ inhibitors described herein, the cellular target has not yet been
identified. Thus, the aim was in the first place to support the hypothesis that these type of
b-annelated 1,4-DHPs exert their novel TGFβ inhibitory mode of action through binding a
25

ACCEPTED MANUSCRIPT
distinct target in case the observed SAR can be quantitatively captured. Unspecific activities that
would derive from intrinsic properties of the ligands (e.g., lipophilicity) rather than from direct
interaction with the target binding site were supposed to be ruled out. At the same time, of
course, unspecific ligand features or “off-target” effects always play a role in cell- and organism-
based assay systems.
A 3D-QSAR analysis was performed using TGFβ inhibitors from this study and our previous
report that showed a “measurable activity” which led to a training dataset of 58 DHPs and a final
test dataset of 19 DHPs. The 3D conformations of all ligands were prepared based on the crystal
structure of the active, (R)-configured ITD-1 (+)-enantiomer 54b that we disclosed here for the
first time. Since the absolute configuration at the 4-position is vital for the compounds’ TGFβ
inhibitory activity, this information was likewise important for a reliable 3D-QSAR model.
2
Although dealing with cell-based data from a reporter gene assay, the model had a high r of
2
2
0
.93 and significant, good values for the internal cross-validation (q
= 0.74, q
= 0.70),
LMO
LOO
which indicates a robust model (Figure 4b). However, the predictive power of this model on a
test dataset (i.e., that was not involved in the training dataset) is the most challenging and most
2
important part of a QSAR analysis. The final test dataset (19 compounds) revealed an r pred of
0
.69 which confirmed the predictive power of the obtained 3D-QSAR model. Overall, a
significant and robust model with predictive power could be devised for the DHP dataset which
supports the assumption that a defined target (binding site) or mechanism is addressed.
26

ACCEPTED MANUSCRIPT
a
c
b
d
2
2
2
2
r
q
LOO
q
LMO
r
pred
0
.93
0.74
0.70
0.69
Figure 4. 3D-QSAR analysis results of TGFβ inhibitory activities of DHPs. a) Predicted versus
experimental inhibition of the trainings dataset (black circles, black line) and of the test dataset
(red triangles). b) Statistical parameters of the final 3D-QSAR model. c) Alignment of trainings
data set and Van-der-Waals and electrostatic contour maps for 3D-QSAR model (green/yellow:
steric bulky regions are positively/negatively correlated with inhibitory activity; red/blue:
positively/negatively charged substituents are positively correlated with inhibitory activity). d)
Correlation plot of calculated logP (clogP) versus TGFβ inhibition (% at 2.5 µM).
27

ACCEPTED MANUSCRIPT
One advantage of 3D-QSAR modelling is the visualization of molecular interaction field
coefficients as contour maps. The alignment of the 58 compound-comprising training data set is
shown in combination with the contour map of the model in Figure 4c. Yellow contours
highlight regions where steric bulky substituents correlate negatively with inhibitory activity and
whereas bulky substituents within the green ones correlate positively. Moreover, electrostatic
interactions that correlate positively with the biological activity are shown in red (negative
partial charges) and blue (positive partial charges). Of course, the detailed influence of
substituents on electrostatic potential as shown in Figure 3b-d cannot be seen in this kind of
QSAR studies, since basic force-field based molecular interaction fields are used. However, for
structure-activity relationships these interaction fields have proven their usefulness.
A combination of a blue and red surface region near the 3-carbonyl-like groups can be
explained by ester functionalities of highly potent DHPs where the alkoxy oxygen is close to the
red surface. In contrast, molecules containing a lactam or lactone (e.g., 46 and 47) show low
inhibition activity and are surrounded by a blue surface. The electronegative charge of the
carbonyl group negatively correlates with inhibition activity which is represented by the
electropositive surface nearby. Similarly, the nitrile group of 19 lies also within this area.
The steric contour map in Figure 4c shows a green surface around the distal 4’-substituent
which implies that sterically demanding groups are well-tolerated and correlate positively with
activity. This is supported by DHPs with either a lack of 4’-substituents (e.g.
S8) or small substituents in 4’-position (e.g., 7*, S3 S4) that show a low inhibition. In contrast,
the yellow surface above the DHP core indicates areas which should not be covered with
substituents. Such compounds were for example 3‘-aryl-substituted DHPs (39* 40*-43*) that
were inactive or had a low potency. Although, among the most potent molecules, the contour
, S1, S2, S5-S6, 8*,
,
,
28