From Complex Natural Products to Simple Synthetic Mimetics by Computational de Novo Design

Article abstract of DOI:10.1002/anie.201601941

We present the computational de novo design of synthetically accessible chemical entities that mimic the complex sesquiterpene natural product (-)-Englerin A. We synthesized lead-like probes from commercially available building blocks and profiled them for activity against a computationally predicted panel of macromolecular targets. Both the design template (-)-Englerin A and its low-molecular weight mimetics presented nanomolar binding affinities and antagonized the transient receptor potential calcium channel TRPM8 in a cell-based assay, without showing target promiscuity or frequent-hitter properties. This proof-of-concept study outlines an expeditious solution to obtaining natural-product-inspired chemical matter with desirable properties.

Full text of DOI:10.1002/anie.201601941

Angewandte
Communications
Chemie
International Edition: DOI: 10.1002/anie.201601941
German Edition: DOI: 10.1002/ange.201601941
Natural Product Mimetics
From Complex Natural Products to Simple Synthetic Mimetics by
Computational De Novo Design
Lukas Friedrich, Tiago Rodrigues, Claudia S. Neuhaus, Petra Schneider, and Gisbert Schneider*
Abstract: We present the computational de novo design of
synthetically accessible chemical entities that mimic the com-
plex sesquiterpene natural product (À)-Englerin A. We syn-
thesized lead-like probes from commercially available build-
ing blocks and profiled them for activity against a computa-
tionally predicted panel of macromolecular targets. Both the
design template (À)-Englerin A and its low-molecular weight
mimetics presented nanomolar binding affinities and antago-
nized the transient receptor potential calcium channel TRPM8
in a cell-based assay, without showing target promiscuity or
frequent-hitter properties. This proof-of-concept study outlines
an expeditious solution to obtaining natural-product-inspired
chemical matter with desirable properties.
in A mimetics has failed to deliver chemical entities with
potent anticancer activity.[9]
Bioactivity-guided scaffold trees have proven useful for
the stepwise exploration of natural-product-derived molec-
ular frameworks.[10] We therefore envisaged that ligand-
based, computer-assisted de novo design might be able to
translate the complex pharmacophore patterns of natural
product templates into simpler and synthetically more
accessible entities.
We started off by using the ligand-based software tool
DOGS (Design of Genuine Structures)[11] to computationally
generate mimetics of (À)-Englerin A (Figure 1). This algo-
rithm performs a reaction-based molecular design process
using 83 organic synthesis schemes and 25214 molecular
building blocks, and has been successfully used for the
de novo design of lead- and drug-like new chemical entities
(NCEs) before.[12] The automated design process resulted in
903 in silico structures (Figure 2), which were then compared
to (À)-Englerin A in terms of their topological pharmaco-
phore feature similarity (CATS method).[13] This re-scoring
step was done to increase the chance of finding isofunctional
NCEs as a consensus of two similarity metrics, namely the
DOGS scoring function[14] and the CATS similarity (Support-
ing Information).[15] The resulting top-scoring compounds
were further analyzed for their potential macromolecular
targets. For this purpose, we used software that has been
shown to be able to predict the targets of drug- and fragment-
like compounds and complex natural products (SPiDER
method).[16] SPiDER infers potential drug targets from
pharmacophore and property similarities between the query
compound and known pharmacologically active ligands with
known targets.
Based on the similarity analysis and the target predictions,
we selected designs 1 and 2 on ranks 4 and 18 of the result list
for further consideration, taking into account the availability
of starting materials, absence of reactive moieties, and
predicted solubility (Figure 1; Supporting Information). For
ease and reduced cost of synthesis, we converted the aliphatic
rings to arene systems. Compound 2 was obtained from the
reaction of the required oxazolidinone and acyl chloride,
followed by installation of the furanyl moiety. Suzuki coupling
of N- and C-protected 3-bromophenylalanine afforded com-
pound rac-3 (Scheme 1).
N
atural products are of fundamental interest as chemical
matter for interrogating biological systems.[1,2] They contain
biologically pre-validated architectures that allow the explo-
ration of drug-relevant chemical space.[3] However, drug
discovery still awaits the full exploitation of pharmacologi-
cally active natural products because, among other reasons,
their supply is limited and they often possess complex
chemical structures rendering total syntheses difficult.[4]
Herein, we present the computational de novo design of
synthetically accessible, isofunctional mimetics of the struc-
turally intricate natural product (À)-Englerin A (1), a sesqui-
terpene from Phyllanthus engleri. Nanomolar binding affin-
ities and high ligand efficiencies to the transient receptor
potential melastatin 8 (TRPM8) ion channel designate these
designer compounds.
The molecular basis for the potent antiproliferative
activity of (À)-Englerin A was discovered by Waldmann and
co-workers.[5] The natural product selectively kills renal
cancer cells (GI50 = 1–87 nm) through activation of TRP
canonical 4/5 (TRPC4/5) calcium channels, but its develop-
ment into a clinical candidate may be precluded by acute
toxicity.[6] Its total synthesis was first achieved by Willot
et al.[7] and has recently been simplified to 14 steps.[8]
Recently, the rational design of terpene-based (À)-Engler-
[*] L. Friedrich, Dr. T. Rodrigues, C. S. Neuhaus, Dr. P. Schneider,
Prof. Dr. G. Schneider
Department of Chemistry and Applied Biosciences
Swiss Federal Institute of Technology (ETH)
Vladimir-Prelog-Weg 4, 8093 Zurich (Switzerland)
E-mail: gisbert.schneider@pharma.ethz.ch
With both compounds in hand, we tested them in func-
tional cell-based assays for TRPM8 and TRPV1 modulation
(Table 1). Compounds 2 and 3 potently blocked the TRPM8
subtype (KB = 1.4 and 0.2 mm, respectively), but did not inhibit
the V-type TRP (VR1) calcium channel at a concentration of
10 mm. (À)-Englerin A equipotently blocked TRPM8 (KB =
0.4 mm), thereby revealing an apparently broader activity
Dr. P. Schneider
inSili.com GmbH
Segantinisteig 3, 8049 Zurich (Switzerland)
Supporting information for this article can be found under:
http://dx.doi.org/10.1002/anie.201601941.
Angew. Chem. Int. Ed. 2016, 55, 6789 –6792
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
6789

Angewandte
Communications
Chemie
designer compounds are isofunctional antago-
nists of the TRPM8 calcium channel. While
(À)-Englerin A shows acute toxicity,[6] the two
de novo designed compounds 2 and 3 were
non-cytotoxic in a preliminary study with
human primary glioblastoma cells (U-87 MG
cells), which do not express TRPM8 (Support-
ing Information). Furthermore, these com-
pounds display high ligand efficiencies (LE >
0.30), rendering them suitable for further
exploration as TRPM8 antagonists.[17]
To exclude artifact measurements, we
measured the solubility of compound 3 and
performed in silico frequent hitter assessment.
We did not observe colloidal aggregation in
water up to a concentration of 60 mm. Its
molecular structure is free from apparent
reactive and otherwise undesired substructure
flags, and possesses a predicted low target
promiscuity score (< 2%).[18] We therefore
Figure 1. The molecular design strategy. In the first step (i), the design template 1 was
computationally converted into 903 de novo designs by reaction-based building block
fusion. Re-scoring of the computer-generated structures with the topological pharmaco-
phore metric CATS[13] (lower values suggest greater similarity to (À)-Englerin A) and the
assessment of synthetic feasibility led to the selection of the original designs 1 and 2.
Simplification of their chemical structure for ease of synthesis resulted in compounds 2
and 3. The substructures of designs 1 and 2 that are not contained in the final
compounds 2 and 3 are highlighted in bold. The chiral centers of the designed
compounds are not assigned, because the software did not consider stereochemistry.
concluded that compound
3 blocks the
TRPM8-mediated Ca2+ flux in a specific con-
centration-dependent fashion and has a low
off-target potential. This conclusion was fur-
ther substantiated by activity testing against
a panel of 12 potential macromolecular targets predicted by
the SPiDER software (Supporting Information). Of these
targets, only PXR was mildly antagonized by compound 3 in
a concentration of 10 mm (52% replacement of radiolabeled
agonist T0901317, Ki = 100 nm). This outcome indicates
natural products as suitable templates for the de novo
design of synthetic mimetics. It equally corroborates the
target promiscuity prediction model (inSili.com GmbH,
Zurich, Switzerland), which correctly designated compound
3 as a target-family selective ligand.[18]
To obtain preliminary structure–activity relationship data,
we synthesized derivatives of compound 3. We chose the site
of structural variation by flexibly aligning compound 3 in the
R and S configurations, respectively, on a low-energy con-
formation of (À)-Englerin A and
Figure 2. The most frequent molecular scaffolds of 903 de novo
designed (À)-Englerin A mimetics. The numbers are absolute frequen-
cies.
analyzing their superimposed phar-
macophores (Figure 3). The S enan-
tiomer led to a better overall match
and we decided to explore the least
well aligned region in more detail.
The alignment suggested (S)-3 as
the bioactive enantiomer. Deriva-
tives 4–7 were obtained in modular
fashion to interrogate the effect of
substituting the furanyl moiety on
Scheme 1. Synthesis of the (À)-Englerin A mimetics 2 and 3. Reagents and conditions: i) n-BuLi, 2-
(4-bromophenyl)acetyl chloride, THF, À788C!RT, 4 h; ii) furan-2-ylboronic acid, Pd(PPh3)2Cl2,
Na2CO3, DMF/H2O, 1008C, 5 h; iii) Cbz-Cl, Na2CO3, dioxane, 08C!RT, 24 h; iv) H2SO4, MgSO4,
MeOH, CH2Cl2, RT 24 h; v) furan-2-ylboronic acid, Pd(PPh3)4, Cs2CO3, DMF/H2O, 1008C, 5 h.
activity against TRPM8 (Table 1).
With the exception of compound 7,
all derivatives potently blocked the
ion channel with sub-micromolar
activity.
spectrum than originally assumed by Akbulut et al.[5] Signifi-
cantly, the screening results showed that the de novo designs
inherited the binding potential to the TRP target family from
their mother natural product. Both (À)-Englerin A and the
The results of this study demonstrate that computational
de novo design can be productively applied to finding
synthetically accessible NCEs from structurally complex
natural products. Although we used a two-dimensional
6790 www.angewandte.org
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2016, 55, 6789 –6792

Angewandte
Communications
Chemie
Table 1: Inhibitory activities of (À)-Englerin A, 1, and the designed
compounds 2–7 (see Scheme 1) against the target ion channel TRPM8.
aimed at finding TRPM8 ligands, the main advantage lies in
the discovery of new chemotypes without the need for high-
throughput screening.[20] It represents an automated and
largely unbiased design approach. However, de novo design
does not eliminate the need for subsequent hit-to-lead
optimization.
No.
R1
R2
IC50/ mm
KB [mm]
1[a]
2[b]
3.0Æ0.08 log units
10Æ0.08 log units
0.4
1.4
Despite providing an attractive solution to drive the
design of biologically-relevant chemical matter, a single
proof-of-concept application should not be over-interpreted
and oversold. A thorough assessment of the potential and the
limitations of this de novo design approach with regard to
diverse natural product templates will be mandatory. Keeping
a healthy skepticism, the molecular design concept outlined
here might open a new path for natural-product-inspired
chemical biology and medicinal chemistry.[2,21]
3
4
Me
1.7Æ0.12 log units
0.2
0.3
t-Bu
2.0Æ0.13 log units
5
Me
4.5Æ0.07 log units
0.6
0.5
6
7
Me
Me
3.7Æ0.50 log units
>10
Acknowledgements
We thank A. Finkelmann, D. Reker, and S. Haller for
technical support. This research was financially supported by
ETH Zurich and a grant from the OPO Foundation, Zurich,
Switzerland. P. S. and G.S. are co-founders of inSili.com
GmbH, Zurich.
[a] (À)-EnglerinA was purchased from Axon Lab AG (Baden, Switzer-
land). [b] KB =IC50[1+(C/EC50.C)]À1, where C is the concentration of
reference agonist in the assay and EC50.C its EC50 value.
Keywords: chemical biology · computer-assisted drug design ·
drug discovery · organic synthesis · polypharmacology
How to cite: Angew. Chem. Int. Ed. 2016, 55, 6789–6792
Angew. Chem. 2016, 128, 6901–6904
[1] a) A. L. Harvey, R. Edrada-Ebel, R. J. Quinn, Nat. Rev. Drug
Discovery 2015, 14, 111 – 129; b) A. Bauer, M. Bronstrup, Nat.
Prod. Rep. 2014, 31, 35 – 60.
[2] a) T. Rodrigues, D. Reker, P. Schneider, G. Schneider, Nat.
Chem. 2016, 8, DOI: 10.1038/nchem.2479; b) E. A. Crane, K.
Gademann, Angew. Chem. Int. Ed. 2016, 55, 3882 – 3902; Angew.
Chem. 2016, 128, 3948 – 3970; c) P. Schneider, G. Schneider, J.
Med. Chem. 2016, 59, DOI: 10.1021/acs.jmedchem.5b01849.
[3] a) R. M. Wilson, S. J. Danishefsky, J. Org. Chem. 2006, 71, 8329 –
8351; b) M. L. Lee, G. Schneider, J. Comb. Chem. 2001, 3, 284 –
289; c) L. Du, A. J. Robles, J. B. King, D. R. Powell, A. N. Miller,
S. L. Mooberry, R. H. Cichewicz, Angew. Chem. Int. Ed. 2014,
53, 804 – 809; Angew. Chem. 2014, 126, 823 – 828; d) H. van Hat-
tum, H. Waldmann, J. Am. Chem. Soc. 2014, 136, 11853 – 11859;
e) V. Eschenbrenner-Lux, P. Kuchler, S. Ziegler, K. Kumar, H.
Waldmann, Angew. Chem. Int. Ed. 2014, 53, 2134 – 2137; Angew.
Chem. 2014, 126, 2166 – 2169; f) R. Narayan, J. O. Bauer, C.
Strohmann, A. P. Antonchick, H. Waldmann, Angew. Chem. Int.
Ed. 2013, 52, 12892 – 12896; Angew. Chem. 2013, 125, 13130 –
13134; g) M. C. McLeod, G. Singh, J. N. Plampin III, D. Rane,
J. L. Wang, V. W. Day, J. Aube, Nat. Chem. 2014, 6, 133 – 140;
h) T. Miyao, D. Reker, P. Schneider, K. Funatsu, G. Schneider,
Planta Med. 2015, 81, 429 – 435; i) Y. L. Bennani, Drug Discovery
Today 2011, 16, 779 – 792.
Figure 3. Alignment of low-energy conformations of (À)-Englerin A
(green) and compound (S)-3 (magenta). Two perfectly superimposed
pharmacophore features are highlighted (F1: aromatic, F2: H-bridge
acceptor).
ligand-based approach, one can speculate that the intricate
stereochemistry of the template (À)-Englerin A was implic-
itly captured by the similarity metrics that guided the in silico
design process. Evidently, the simplified structural and
pharmacophore features of the designed compounds were
sufficient for blocking the TRPM8 channel. This observation
may not be transferable to structurally more demanding
natural product templates (for example, flexible macrocycles)
and other targets. For such cases, the design approach could
be extended by conformation-sensitive compound re-scoring,
as a first next step. By substituting some of the “three-
dimensionality” of the natural product with planar arene
systems in the molecule simplification step (Figure 1), we
eliminated a perceived advantage of natural products.[19] It
remains to be shown if this idea is generally applicable.
Comparing our design strategy to alternative approaches that
[4] A. G. Atanasov, B. Waltenberger, E. M. Pferschy-Wenzig, T.
Linder, C. Wawrosch, P. Uhrin, V. Temml, L. Wang, S. Schwaiger,
E. H. Heiss, J. M. Rollinger, D. Schuster, J. M. Breuss, V.
Bochkov, M. D. Mihovilovic, B. Kopp, R. Bauer, V. M. Dirsch,
H. Stuppner, Biotechnol. Adv. 2015, 33, 1582 – 1614.
[5] Y. Akbulut, H. J. Gaunt, K. Muraki, M. J. Ludlow, M. S. Amer,
A. Bruns, N. S. Vasudev, L. Radtke, M. Willot, S. Hahn, T. Seitz,
S. Ziegler, M. Christmann, D. J. Beech, H. Waldmann, Angew.
Angew. Chem. Int. Ed. 2016, 55, 6789 –6792
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
6791

Angewandte
Communications
Chemie
Chem. Int. Ed. 2015, 54, 3787 – 3791; Angew. Chem. 2015, 127,
3858 – 3862.
[14] a) M. Rupp, E. Proschak, G. Schneider, J. Chem. Inf. Model.
2007, 47, 2280 – 2286; b) M. Rupp, G. Schneider, Mol. Inf. 2010,
29, 266 – 273.
[6] C. Carson, P. Raman, J. Tullai, L. Xu, M. Henault, E. Thomas, S.
Yeola, J. Lao, M. McPate, J. M. Verkuyl, G. Marsh, J. Sarber, A.
Amaral, S. Bailey, D. Lubicka, H. Pham, N. Miranda, J. Ding, H.-
M. Tang, H. Ju, P. Tranter, N. Ji, P. Krastel, R. K. Jain, A. M.
Schumacher, J. J. Loureiro, E. George, G. Berellini, N. T. Ross,
S. M. Bushell, G. Erdemli, J. M. Solomon, PLoS ONE 2015, 10,
e0127498.
[7] M. Willot, L. Radtke, D. Kçnning, R. Frçhlich, V. H. Gessner, C.
Strohmann, M. Christmann, Angew. Chem. Int. Ed. 2009, 48,
9105 – 9108; Angew. Chem. 2009, 121, 9269 – 9272.
[8] H. Kusama, A. Tazawa, K. Ishida, N. Iwasawa, Chem. Asian J.
2016, 11, 64 – 67.
[9] M. J. Acerson, B. S. Bingham, C. A. Allred, M. B. Andrus,
Tetrahedron Lett. 2015, 56, 3277 – 3280.
[10] S. Renner, W. A. van Otterlo, M. Dominguez Seoane, S. Mçck-
linghoff, B. Hofmann, S. Wetzel, A. Schuffenhauer, P. Ertl, T.
Oprea, D. Steinhilber, L. Brunsveld, D. Rauh, H. Waldmann,
Nat. Chem. Biol. 2009, 5, 585 – 592.
[15] P. Willett, Mol. Inf. 2014, 33, 403 – 413.
[16] a) T. Rodrigues, D. Reker, J. Kunze, P. Schneider, G. Schneider,
Angew. Chem. Int. Ed. 2015, 54, 10516 – 10520; Angew. Chem.
2015, 127, 10662 – 10666; b) D. Reker, T. Rodrigues, P.
Schneider, G. Schneider, Proc. Natl. Acad. Sci. USA 2014, 111,
4067 – 4072; c) D. Reker, A. M. Perna, T. Rodrigues, P.
Schneider, M. Reutlinger, B. Monch, A. Koeberle, C. Lamers,
M. Gabler, H. Steinmetz, R. Muller, M. Schubert-Zsilavecz, O.
Werz, G. Schneider, Nat. Chem. 2014, 6, 1072 – 1078.
[17] a) P. PØrez-Faginas, M. Teresa Aranda, R. de la Torre-Martínez,
S. Quirce, A. Fernµndez-Carvajal, A. Ferrer-Montiel, R. Gon-
zµlez-MuÇiz, RSC Adv. 2016, 6, 6868 – 6877; b) A. Bertamino, C.
Ostacolo, P. Ambrosino, S. Musella, V. Di Sarno, T. Ciaglia,
M. V. Soldovieri, N. Iraci, A. Fernandez Carvajal, R. de la Torre-
Martinez, A. Ferrer-Montiel, R. Gonzalez Muniz, E. Novellino,
M. Taglialatela, P. Campiglia, I. Gomez-Monterrey, J. Med.
Chem. 2016, 59, 2179-2191.
[11] a) M. Hartenfeller, H. Zettl, M. Walter, M. Rupp, F. Reisen, E.
Proschak, S. Weggen, H. Stark, G. Schneider, PLoS Comput.
Biol. 2012, 8, e1002380; b) M. Hartenfeller, M. Eberle, P. Meier,
C. Nieto-Oberhuber, K.-H. Altmann, G. Schneider, E. Jacoby, S.
Renner, J. Chem. Inf. Model. 2012, 52, 1167 – 1178; c) M.
Hartenfeller, G. Schneider, Methods Mol. Biol. 2011, 672, 299 –
323.
[12] a) T. Rodrigues, G. Schneider, Synlett 2014, 170 – 178; b) T.
Rodrigues, D. Reker, M. Welin, M. Caldera, C. Brunner, G.
Gabernet, P. Schneider, B. Walse, G. Schneider, Angew. Chem.
Int. Ed. 2015, 54, 15079 – 15083; Angew. Chem. 2015, 127, 15294 –
15298.
[18] P. Schneider, M. Rçthlisberger, D. Reker, G. Schneider, Chem.
Commun. 2016, 52, 1135 – 1138.
[19] P. A. Clemons, N. E. Bodycombe, H. A. Carrinski, J. A. Wilson,
A. F. Shamji, B. K. Wagner, A. N. Koehler, S. L. Schreiber, Proc.
Natl. Acad. Sci. USA 2010, 107, 18787 – 18792.
[20] a) M. A. Sherkheli, A. K. Vogt-Eisele, D. Bura, L. R.
Beltrµn Mµrques, G. Gisselmann, H. Hatt, J. Pharm. Pharm.
Sci. 2010, 13, 242 – 253; b) J. DeFalco, M. A. Duncton, D.
Emerling, Curr. Top. Med. Chem. 2011, 11, 2237 – 2252.
[21] a) J. M. Rollinger, Curr. Pharm. Des. 2010, 16, 1654 – 1655; b) H.
Lachance, S. Wetzel, K. Kumar, H. Waldmann, J. Med. Chem.
2012, 55, 5989 – 6001.
[13] a) M. Reutlinger, C. P. Koch, D. Reker, N. Todoroff, P.
Schneider, T. Rodrigues, G. Schneider, Mol. Inf. 2013, 32, 133 –
138; b) G. Schneider, W. Neidhart, T. Giller, G. Schmid, Angew.
Chem. Int. Ed. 1999, 38, 2894 – 2896; Angew. Chem. 1999, 111,
3068 – 3070.
Received: February 24, 2016
Revised: April 3, 2016
Published online: April 25, 2016
6792 www.angewandte.org
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2016, 55, 6789 –6792