Biaryl sulfonamides as: Cisoid azosteres for photopharmacology

Article abstract of DOI:10.1039/d1cc00950h

Biaryl sulfonamides are excellent candidates for the azologization approach that yields photoswitchable drugs more active in their metastable cis state, compared to the stable trans state. Here we present the scope and limitations of this strategy for rational design in photopharmacology. This journal is

Full text of DOI:10.1039/d1cc00950h

ChemComm
View Article Online
View Journal | View Issue
COMMUNICATION
Biaryl sulfonamides as cisoid azosteres
for photopharmacology†
Cite this: Chem. Commun., 2021,
57, 4126
a
c
f
Piermichele Kobauri,
Sebastian Thallmair,
Wiktor Szymanski, *b Fangyuan Cao,
de
d
Received 19th February 2021,
Accepted 17th March 2021
Siewert J. Marrink,
Martin D. Witte,
Frank J. Dekker *c and Ben L. Feringa
*
a
DOI: 10.1039/d1cc00950h
rsc.li/chemcomm
Biaryl sulfonamides are excellent candidates for the azologization bioactivities. Nevertheless, two rational design strategies emerged
approach that yields photoswitchable drugs more active in their in azobenzene-based photopharmacology: azo-extension and azolo-
metastable cis state, compared to the stable trans state. Here we gization. The former explores the structure–activity relationship
present the scope and limitations of this strategy for rational design (SAR) of existing drug scaffolds to identify suitable sites for the
in photopharmacology.
attachment of an azobenzene.4 In the latter approach, azobenzenes
are introduced in the place of aromatic moieties that are sterically
Photopharmacology employs light to control the bioactivity of and electronically similar. Those privileged moieties were termed
drugs with exceptional bio-orthogonality and spatiotemporal ‘‘azosteres’’.7 In most applications,8 photopharmacology aims to
precision, thus potentially increasing site- and organ-selectivity design photoswitchable drugs that are more active in the metastable
in pharmaceutical applications.1–4 The incorporation of mole- state, i.e. ‘‘cis-on’’9 in the case of classical azobenzenes. In particular,
cular photoswitches5 into drugs enables this non-invasive con- if this property is combined with a large difference in potency
trol, since their irradiation induces reversible changes in the between the photoisomers, it ensures effective control of biological
structure and properties of the drug.2 Because of favorable processes.1–3,8 However, most attempts at azologization resulted in
photochemical properties and ease of synthesis, azobenzene ‘‘trans-on’’ ligands, especially for planar azosteres that adopt a
is the most common switch in photopharmacology.6 Its photo- transoid conformation (such as benzyl ether,7 benzanilide,10 and
isomerization from the thermally stable trans-isomer to the diphenylacetylene11). An alternative strategy, named ‘‘sign inver-
metastable cis-isomer results in substantial changes in mole- sion’’, uses cyclic diazocines12 with thermally stable cis isomers,13
cular shape (from planar to globular), end-to-end distance but their limited synthetic accessibility might hinder its appli-
(from 9 Å to 6 Å) and dipole moment (from 0 D to B3 D).6
cability.14 Conversely, only a few examples of cisoid azosteres have
The design of photoswitchable bioactive compounds often been reported.9,15,16 Nevertheless, an explicit set of general criteria
relies on trial-and-error approaches and the assumption that for the design of cis-on photoswitchable drugs is still missing,
photoinduced changes will inherently result in different although they are expected to significantly expand the chemical
space available for photopharmacology. Here we present the identifi-
cation of the biaryl sulfonamide scaffold as a prevalent cisoid
azostere, because of its structural and electronic similarity with
cis-azobenzene.
a Stratingh Institute for Chemistry, University of Groningen, Nijenborgh 4,
Groningen, 9747 AG, The Netherlands. E-mail: b.l.feringa@rug.nl
b Medical Imaging Center, University of Groningen, University Medical Center
Groningen, Hanzeplein 1, Groningen 9713 GZ, The Netherlands.
E-mail: w.szymanski@umcg.nl
First, we queried the Cambridge Structural Database (CSD)
on the geometry of azobenzene and the most generic azostere,
i.e., two (hetero)aromatic rings linked by two atoms. Measure-
ment of dihedral and centroid angles, ring distances and ring
angles provided insights into the requirements for a cisoid
geometry (see ESI,† Fig. S18), in particular the centroid angle
being less than 1001 and the ring distance being less than 5 Å.
Biaryl sulfonamides (BAS) emerged as cisoid scaffolds (Fig. 1A
and B), in accordance with their known preference for a
staggered conformation.17 This geometrical bias18 has been
employed for conformational control in drug design.19 In
addition to structural similarity to cis-azobenzene, electronic
c Chemical and Pharmaceutical Biology, Groningen Research Institute of Pharmacy,
University of Groningen, A. Deusinglaan 1, Groningen, 9713 AV, The Netherlands.
E-mail: f.j.dekker@rug.nl
d Groningen Biomolecular Sciences and Biotechnology Institute & Zernike Institute
for Advanced Materials, University of Groningen, Nijenborgh 7,
Groningen 9747 AG, The Netherlands
e Frankfurt Institute for Advanced Studies, Ruth-Moufang-Straße 1,
Frankfurt am Main 60438, Germany
f Chemical Biology II, Stratingh Institute for Chemistry, University of Groningen,
Groningen 9747 AG, The Netherlands
† Electronic supplementary information (ESI) available. See DOI: 10.1039/
d1cc00950h
This journal is © The Royal Society of Chemistry 2021
4126 | Chem. Commun., 2021, 57, 4126–4129

View Article Online
Communication
ChemComm
ensures optimal contacts of the capping group with the enzyme
surface.24
The starting point for the azologization of Sulf-1 was the
crystal structure of its complex with Lp-PLA2.25 The sulfona-
mide is involved in hydrogen bonds with Leu153, Ser273 and a
water molecule occupying the oxyanion hole.26 Ligand binding
was first explored in silico, using induced fit docking27 (IFD) to
enable a certain degree of protein flexibility during the calcula-
tions. The docking poses suggested that cis-Azo-1 would be able
to mimic the binding mode of the parent compound much
better than its trans-isomer, albeit at the cost of the hydrogen
bonding interactions (Fig. 2A). The stability of the docking
poses over time was then evaluated through three replicas of
100 ns molecular dynamics (MD) simulations, which indicate
that only trans-Azo-1 would unbind quickly from the protein
Fig. 1 (A). Biaryl sulfonamides as cisoid azosteres. (B). Structural similarity:
distributions of centroid angles and ring distances in the CSD. (C). Electro-
nic similarity: dipole moments and electrostatic potential surfaces, where
red color represents max. negative potential and violet color represents
max. positive potential.
similarity was explored by density functional theory (DFT) during the simulation time, thus providing additional time-
calculations (see ESI,† section S4.6 for details). As compared domain support for a preferred binding of the cis over the trans
to trans-azobenzene, cis-azobenzene better approximates the conformation. Moreover, the electronic similarity in terms of
dipole moment of BAS, as well as its electrostatic potential ESP surfaces and dipole moments found for the fragments was
(ESP) surface (Fig. 1C). Finally, this moiety is widely used in also confirmed for the entire ligands (see ESI,† Fig. S50
medicinal chemistry20 (see ESI,† section S4.3 for details).
and S52).
Ligands Sulf-1 and Azo-1 were synthesized (see ESI,†
We hypothesized that the azologization of BAS would con-
stitute a general strategy for designing cis-on ligands. However, section S1) and the photochemical properties of Azo-1 were
the replacement of two hydrogen-bond acceptors and one investigated in Lp-PLA2 assay aqueous buffer (Fig. 3A and B).
donor in the –SO2NH–linker part with two acceptors of the Irradiation with l = 365 nm light resulted in trans–cis photo-
–NQN– azo bond poses risks of loss in affinity to the target. To isomerization (Fig. 3A), reaching a high photostationary state
test our hypothesis and its limitations, two bioactive com- distribution (PSD) of 92% cis, typical of alkyl- and
pounds containing the BAS moiety were identified and inves- aryl-oxy substituted azobenzenes.28 Successive irradiation
tigated by molecular modeling, photochemical evaluation and with l = 420 nm light promoted cis–trans back-isomerization,
bioactivity studies. They were chosen because of their low- reaching a PSD of 84% trans (see ESI,† Fig. S9). The photo-
nanomolar potency and the different roles of the sulfonamide switch showed no signs of fatigue upon multiple cycles of
moiety in binding to their biological targets. The first ligand photochemical isomerization in the presence of the reducing
(Sulf-1, Fig. 2A) is a nanomolar inhibitor of lipoprotein- agent glutathione (Fig. 3B). Thermal cis–trans isomerization
associated phospholipase 221 (Lp-PLA2), which is a potential occurred on a time scale of days (see ESI,† Fig. S6 and S7),
therapeutic target.22 Here, the BAS moiety is buried and inter- indicating high stability of cis isomer. Azo-1 showed good
acts deeply in the binding pocket.21 The second ligand (Sulf-2, solubility in water (see ESI,† Fig. S11) up to the highest
Fig. 2B) is belinostat,23 an FDA-approved histone deacetylase concentration used in the biological assay (100 mM).
(HDAC) inhibitor used in chemotherapy. In this case, the biaryl
To assess the inhibitory potency of Sulf-1 and Azo-1 on Lp-PLA2,
sulfonamide acts as a partially solvent-exposed linker that we performed fluorescence-based competition assays29 (Fig. 3C).
Fig. 2 (A). Azologization of Sulf-1. Docking poses of trans-(cyan) and cis-Azo-1 (magenta) superposed with the complex Sulf-1-Lp-PLA2 (green-gray,
PDB: 5YEA). Ligand RMSD from a representative replica of 100 ns MD simulation of the protein–ligand complexes. (B). Azologization of Sulf-2. Docking
poses of trans-(cyan), cis-Azo-2 (magenta) and Sulf-2 (green) into HDAC2 (gray, PDB: 4LXZ). Ligand RMSD from a representative replica of 100 ns MD
simulation of the protein–ligand complexes. For additional MD simulations, see ESI† (section S4.5).
This journal is © The Royal Society of Chemistry 2021
Chem. Commun., 2021, 57, 4126–4129 | 4127

View Article Online
ChemComm
Communication
Fig. 3 Photochemical and pharmacological evaluation of Azo-1. (A). UV-Vis
spectra of Azo-1 (20 mM, 1% DMSO in Lp-PLA2 assay buffer) at the thermal
equilibrium (light blue), PSS365 nm (red) and PSS420 nm (dark blue). (B). Repeated
photoisomerization of Azo-1 with l = 365 nm and 420 nm light (20 mM, 1%
DMSO in Lp-PLA2 assay buffer, 10 mM glutathione). (C). Dose–response
curves for inhibitors Sulf-1/Azo-1 against Lp-PLA2 and obtained IC50 values.
(D). Reversible control of Lp-PLA2 activity with Azo-1 (25 mM).
Fig. 4 Photochemical and pharmacological evaluation of Azo-2. (A).
UV-Vis spectra of Azo-2 (20 mM, 1% DMSO in HDAC2 assay buffer) at
the thermal equilibrium (light blue), PSS365 nm (red) and PSS420 nm (dark
blue). (B). Photoisomerization of Azo-2 with l = 365 nm and 420 nm light
(20 mM, 1% DMSO in HDAC2 assay buffer, 10 mM glutathione). (C). Dose–
response curves for inhibitors Sulf-2/Azo-2 against HDAC2 and obtained
IC50 values. (D). Solvent exposure of Sulf-2 and cis-Azo-2 docked into
HDAC2 (gray surface).
The parent inhibitor Sulf-1 maintained its high reported activity in
the nanomolar range, whereas the thermally adapted sample of revealed that all the ligands were more flexible and explored
Azo-1 showed very low activity. On the other hand, UV irradiation larger portions of the protein surface, as the active site is
resulted in an 410-fold activation of Azo-1. Cycles of irradiation exposed to the solvent (see Fig. 4D and ESI,† Fig. S36, S41,
(Fig. 3D), starting from the thermally adapted sample and then S46, S47). This predicted behavior suggests that the differences
alternating between the solutions that represent PSD at l = 365 nm in activity between the ligands could be less pronounced than
and 420 nm light, showed reversible behavior, indicating that the in the case of the Azo-1/Lp-PLA2 system.
change in potency is caused by photoisomerization of Azo-1 and
Photoisomerization of Azo-2 was achieved in HDAC2 assay
not by a light-induced irreversible process. Nevertheless, we buffer (Fig. 4A), reaching a high PSD of 95% cis. Furthermore,
observed an overall 100-fold decrease in activity of Azo-1 as the compound showed no fatigue after multiple cycles of
compared to the parent Sulf-1, which might be due to the loss of photoisomerization under reducing conditions (Fig. 4B).
crucial hydrogen bonds with the oxyanion hole of Lp-PLA2.21
HDAC2 inhibition assay confirmed the high potency of Sulf-
As a second test for our proposed azologization strategy, we 2,32 and much to our delight Azo-2 also showed activity in the
chose belinostat24 (Sulf-2), an inhibitor of HDACs used in sub-micromolar range (Fig. 4C). We found that irradiation with
clinical practice. In this case, a p-alkoxyazobenzene was l = 365 nm light resulted in a 2-fold activation of Azo-2 (Fig. 4C),
employed, as it is known to show very high PSDs28 and thereby showing the desired cis-on behavior of the photoswitch-
appeared to be tolerated during previous SAR studies.23 able inhibitor. Moreover, in this case we observed a smaller,
Because of the available biological assay, we selected HDAC2 12-fold decrease in activity compared to the parent ligand. In
as protein target. Therefore, we needed to dock Sulf-2 into contrast to the previous example, the azologized BAS moiety
HDAC2,30 as the only crystal structure of the compound in the has a less essential role in binding, mainly providing the bent
PDB is a complex with HDAC6.31 Although the two enzymes geometry for conformational control and does not contribute
belong to different classes of HDACs,24 the inhibitor binding specific interactions.19,24
site is highly conserved (see ESI,† Fig. S27). The obtained
The inhibition data for Azo-1 and Azo-2 indicate that the
docking pose revealed that the hydroxamic acid chelates the pharmacophoric role of the BAS substructure influences the
zinc cation and engages in hydrogen bonds with His145 and outcome of the azologization. When this moiety formed key
Tyr308, while the sulfonamide moiety is not involved in specific interactions with buried residues of the oxyanion hole,25 its
interactions with the enzyme (Fig. 2B). These contacts pre- azostere Azo-1 suffered a larger loss of activity. Concurrently, a
dicted for HDAC2 are in agreement with the binding mode larger difference in activity between the photoisomers was
observed experimentally with HDAC6 (see ESI,† Fig. S32 and achieved, suggesting that a cisoid geometry was strictly required
S33). IFD poses of Azo-2 indicated that the cis-isomer of the for biological activity. However, when the sulfonamide moiety
ligand would adopt a similar binding mode and keep the served only as a solvent-exposed linker to guarantee optimal
original bent geometry. However, subsequent MD simulations contacts between the capping group and the target,24 the azolog
This journal is © The Royal Society of Chemistry 2021
4128 | Chem. Commun., 2021, 57, 4126–4129

View Article Online
Communication
ChemComm
´
10 S. Pittolo, X. Gomez-Santacana, K. Eckelt, X. Rovira, J. Dalton,
C. Goudet, J. P. Pin, A. Llobet, J. Giraldo, A. Llebaria and
P. Gorostiza, Nat. Chem. Biol., 2014, 10, 813–815.
showed a smaller loss of original activity. For Azo-2, the
azologization involved a more peripheral region of the pharma-
cophore, while it did not affect the hydroxamic acid (i.e., the
zinc binding domain). This substitution resulted in a less
distinct difference between photoisomers.
11 N. J. Hauwert, T. A. M. Mocking, D. Da Costa Pereira, K. Lion,
Y. Huppelschoten, H. F. Vischer, I. J. P. De Esch, M. Wijtmans and
R. Leurs, Angew. Chem., Int. Ed., 2019, 58, 4531–4535.
12 R. Siewertsen, H. Neumann, B. Buchheim-Stehn, R. Herges,
In conclusion, we show here that BAS are promising cisoid
azosteres for photopharmacology. While guidelines for transoid
azosteres have been described and applied,4,33 the definition of
criteria for the rational design of cis-on photoswitchable
ligands is still lacking. To explore the requirements for mole-
cular similarity with cis-azobenzene, we have analyzed the
geometrical and electrostatic properties of two-atom-linked
biaryl systems and selected the BAS motif based on its bent
geometry and favorable dipole moment. We have demonstrated
that biaryl sulfonamides are promising azosteres if they
are considered for their preferred cisoid geometry, rather than
being discarded for their poor similarity to trans-azobenzene.33
Since biaryl sulfonamides are common motifs in numerous
ligands (see ESI,† Table S3) for various biological targets (e.g.,
anti-apoptotic protein MCL-134 and bromodomain-containing
protein 435), we believe that they may provide a rich source of
inspiration for photopharmacology. Azologization of cisoid
substructures has the potential to guide the rational design of
cis-on photoswitchable ligands.
¨
C. Nather, F. Renth and F. Temps, J. Am. Chem. Soc., 2009, 131,
15594–15595.
¨
13 J. B. Trads, K. Hu¨ll, B. S. Matsuura, L. Laprell, T. Fehrentz, N. Gorldt,
¨
K. A. Kozek, C. D. Weaver, N. Klocker, D. M. Barber and D. Trauner,
Angew. Chem., Int. Ed., 2019, 58, 15421–15428.
14 S. Li, N. Eleya and A. Staubitz, Org. Lett., 2020, 22, 1624–1627.
15 M. Borowiak, W. Nahaboo, M. Reynders, K. Nekolla, P. Jalinot,
J. Hasserodt, M. Rehberg, M. Delattre, S. Zahler, A. Vollmar,
D. Trauner and O. Thorn-Seshold, Cell, 2015, 162, 403–411.
16 M. W. H. Hoorens, M. E. Ourailidou, T. Rodat, P. E. van der
Wouden, P. Kobauri, M. Kriegs, C. Peifer, B. L. Feringa,
F. J. Dekker and W. Szymanski, Eur. J. Med. Chem., 2019, 179,
133–146.
17 K. A. Brameld, B. Kuhn, D. C. Reuter and M. Stahl, J. Chem. Inf.
Model., 2008, 48, 1–24.
18 G. Schwertz, M. S. Frei, M. C. Witschel, M. Rottmann,
U. Leartsakulpanich, P. Chitnumsub, A. Jaruwat, W. Ittarat,
¨
A. Schafer, R. A. Aponte, N. Trapp, K. Mark, P. Chaiyen and
F. Diederich, Chem. – Eur. J., 2017, 23, 14345–14357.
19 Y. Zheng, C. M. Tice and S. B. Singh, Bioorg. Med. Chem. Lett., 2017,
27, 2825–2837.
20 E. A. Ilardi, E. Vitaku and J. T. Njardarson, J. Med. Chem., 2014, 57,
2832–2842.
21 Q. Liu, F. Huang, X. Yuan, K. Wang, Y. Zou, J. Shen and Y. Xu, J. Med.
Chem., 2017, 60, 10231–10244.
Financial support from the EU Horizon 2020 program
(ALERT co-fund No. 713482 for B. L. F.) and the Dutch Scientific
Organization (VIDI grant nr. 723.014.001 for W. S., VIDI grant
nr. 723.015.004 for M. D. W.) is kindly acknowledged. P. K.
acknowledges dr. Romain Costil and Isabel Sieders (University
of Groningen) for fruitful discussions, and the Center for
Information Technology of the University of Groningen for
their support and for providing access to the Peregrine HPC
cluster.
22 F. Huang, K. Wang and J. Shen, Med. Res. Rev., 2020, 40, 79–134.
23 P. W. Finn, M. Bandara, C. Butcher, A. Finn, R. Hollinshead,
N. Khan, N. Law, S. Murthy, R. Romero, C. Watkins, V. Andrianov,
R. M. Bokaldere, K. Dikovska, V. Gailite, E. Loza, I. Piskunova,
I. Starchenkov, M. Vorona and I. Kalvinsh, Helv. Chim. Acta, 2005,
88, 1630–1657.
24 M. Mottamal, S. Zheng, T. L. Huang and G. Wang, Molecules, 2015,
20, 3898–3941.
25 P. Canning, B. A. Kenny, V. Prise, J. Glenn, M. H. Sarker, N. Hudson,
M. Brandt, F. J. Lopez, D. Gale, P. J. Luthert, P. Adamson,
P. Turowski and A. W. Stitt, Proc. Natl. Acad. Sci. U. S. A., 2016,
113, 7213–7218.
26 U. Samanta and B. J. Bahnson, J. Biol. Chem., 2008, 283,
31617–31624.
27 W. Sherman, T. Day, M. P. Jacobson, R. A. Friesner and R. Farid,
J. Med. Chem., 2006, 49, 534–553.
28 W. Szymanski, M. E. Ourailidou, W. A. Velema, F. J. Dekker and
B. L. Feringa, Chem. – Eur. J., 2015, 21, 16517–16524.
29 F. Huang, H. Hu, K. Wang, C. Peng, W. Xu, Y. Zhang, J. Gao, Y. Liu,
H. Zhou, R. Huang, M. Li, J. Shen and Y. Xu, J. Med. Chem., 2020, 63,
7052–7065.
30 B. E. L. Lauffer, R. Mintzer, R. Fong, S. Mukund, C. Tam,
I. Zilberleyb, B. Flicke, A. Ritscher, G. Fedorowicz, R. Vallero,
D. F. Ortwine, J. Gunzner, Z. Modrusan, L. Neumann, C. M. Koth,
J. S. Kaminker, C. E. Heise and P. Steiner, J. Biol. Chem., 2013, 288,
26926–26943.
Conflicts of interest
There are no conflicts to declare.
Notes and references
1 W. A. Velema, W. Szymanski and B. L. Feringa, J. Am. Chem. Soc.,
2014, 136, 2178–2191.
2 M. M. Lerch, M. J. Hansen, G. M. van Dam, W. Szymanski and
B. L. Feringa, Angew. Chem., Int. Ed., 2016, 55, 10978–10999.
3 M. W. H. Hoorens and W. Szymanski, Trends Biochem. Sci., 2018, 43,
567–575.
31 Y. Hai and D. W. Christianson, Nat. Chem. Biol., 2016, 12, 741–747.
32 H. Y. Lee, C. Y. Chang, C. J. Su, H. L. Huang, S. Mehndiratta,
Y. H. Chao, C. M. Hsu, S. Kumar, T. Y. Sung, Y. Z. Huang, Y. H. Li,
C. R. Yang and J. P. Liou, Eur. J. Med. Chem., 2016, 122, 92–101.
33 J. Morstein, M. Awale, J. L. Reymond and D. Trauner, ACS Cent. Sci.,
2019, 5, 607–618.
4 J. Broichhagen, J. A. Frank and D. Trauner, Acc. Chem. Res., 2015, 48,
1947–1960.
5 B. L. Feringa and W. R. Browne, Molecular Switches, Wiley-VCH,
Weinheim, Germany, 2011.
6 A. A. Beharry and G. A. Woolley, Chem. Soc. Rev., 2011, 40,
4422–4437.
34 B. Follows, S. Fessler, T. Baumeister, A. M. Campbell,
M. M. Zablocki, H. Li, D. Gotur, Z. Wang, X. Zheng, L. Molz,
C. Nguyen, T. Herbertz, L. Wang and K. Bair, Bioorg. Med. Chem.
Lett., 2019, 29, 2375–2382.
7 M. Schoenberger, A. Damijonaitis, Z. Zhang, D. Nagel and
D. Trauner, ACS Chem. Neurosci., 2014, 5, 514–518.
8 I. M. Welleman, M. W. H. Hoorens, B. L. Feringa, H. H. Boersma and
´
W. Szymanski, Chem. Sci., 2020, 11, 11672–11691.
¨
9 C. Matera, A. M. J. Gomila, N. Camarero, M. Libergoli, C. Soler and 35 B. K. Allen, S. Mehta, S. W. J. Ember, J. Y. Zhu, E. Schonbrunn,
P. Gorostiza, J. Am. Chem. Soc., 2018, 140, 15764–15773.
N. G. Ayad and S. C. Schu¨rer, ACS Omega, 2017, 2, 4760–4771.
This journal is © The Royal Society of Chemistry 2021
Chem. Commun., 2021, 57, 4126–4129 | 4129