Reversible optical control of F1Fo-ATP synthase using photoswitchable inhibitors

Article abstract of DOI:10.1002/1873-3468.12958

F₁F_o-ATP synthase is one of the best studied macromolecular machines in nature. It can be inhibited by a range of small molecules, which include the polyphenols, resveratrol and piceatannol. Here, we introduce Photoswitchable Inhibitors of ATP Synthase, termed PIAS, which were synthetically derived from these polyphenols. They can be used to reversibly control the enzymatic activity of purified yeast Yarrowia?lipolyticaATP synthase by light. Our experiments indicate that the PIAS bind to the same site in the ATP synthase F₁ complex as the polyphenols in their trans form, but they do not bind in their cis form. The PIAS could be useful tools for the optical precision control of ATP synthase in a variety of biochemical and biotechnological applications.

Full text of DOI:10.1002/1873-3468.12958

DR. DIRK TRAUNER (Orcid ID : 0000-0002-6782-6056)
Article type : Research Letter
Reversible Optical Control of F1Fo-ATP Synthase
Using Photoswitchable Inhibitors
Bianca Eisel1,2,#, Felix Hartrampf3,#, Thomas Meier1,2,*, Dirk Trauner3,4*
1Department of Structural Biology, Max Planck Institute of Biophysics, Max-von-
Laue-Straße 3, 60438, Frankfurt am Main, Germany
2Department of Life Sciences, Imperial College London, Exhibition Road, London
SW7 2AZ, United Kingdom
3Department of Chemistry, University of Munich, Butenandtstraße 5-13, D-81377
Munich, Germany
4Department of Chemistry, New York University, 100 Washington Square East,
Room 712, New York, NY 10003
#contributed equally to this work
*corresponding authors:
Dirk Trauner: dirktrauner@nyu.edu
Thomas Meier: t.meier@imperial.ac.uk
Keywords: Yarrowia lipolytica F1Fo-ATP synthase, photopharmacology
Running title: Photoswitchable control of ATP synthase
Abbreviations: Adenosine-5’-triphosphate (ATP)
This article has been accepted for publication and undergone full peer review but has
not been through the copyediting, typesetting, pagination and proofreading process,
which may lead to differences between this version and the Version of Record. Please
cite this article as doi: 10.1002/1873-3468.12958
This article is protected by copyright. All rights reserved.

Abstract
F1Fo-ATP synthase is one of the best studied macromolecular machines in
nature. It can be inhibited by a range of small molecules, which include the
polyphenols resveratrol and piceatannol. Here, we introduce Photoswitchable
Inhibitors of ATP Synthase, termed PIAS, which were synthetically derived
from these polyphenols. They can be used to reversibly control the enzymatic
activity of purified yeast Yarrowia lipolytica ATP synthase by light. Our
experiments indicate that the PIAS bind to the same site in the ATP synthase
F1 complex as the polyphenols in their trans form but they do not bind in their
cis form. The PIAS could be useful tools for the optical precision control of
ATP synthase in a variety of biochemical and biotechnological applications.
Introduction
F1Fo-type ATP synthase is a membrane-embedded, macromolecular rotary
machine that discharges the transmembrane electrochemical ion gradient to
synthesize adenosine triphosphate (ATP) from adenosine diphosphate (ADP)
and inorganic phosphate (Pi). This key metabolic enzyme uses a unique
mechanochemical rotary mechanism to produce the bulk amount of universal
energy currency ATP in all living cells, but it is also able to operate in reverse,
hydrolyzing ATP, to establish ion gradients by exploiting the energy released
from hydrolysis of ATP [1].
In eukaryotes, the ATP synthase is embedded in the inner membrane of
mitochondria or in the thylakoid membranes of chloroplasts, while in Bacteria
and Archaea it is located in the cytoplasmic membrane. In all organisms, the
ATP synthase shares an overall highly conserved architecture consisting of a
water soluble F1 complex (subunits 33) and a membrane-intrinsic Fo
complex (ab2c8-17)[2-4] joined together by a central stalk (subunits and )
and a peripheral stalk (subunits b2 and ). The 33 subunits envelope the
central stalk  subunit, which by itself introduces an inherent asymmetry into
the F1 headpiece. The lower part of the  and  subunit is in contact with the
This article is protected by copyright. All rights reserved.

membrane-embedded Fo rotor, formed by a number of identical copies of c-
subunit, called the c-ring. Recent advances in structural biology have provided
new insights into the structure and dynamics of completely assembled
complexes of ATP synthase. In particular it includes also valuable structural
information about the previously less well characterized Fo stator complex in
the membrane, its outer stalk region as well as the structural basis of
dimerization of mitochondrial ATP synthases [5-8]. For example, the yeast
Yarrowia lipolytica F1Fo-ATP synthase dimer consists of a total of more than
60 different proteins, which in mitochondria form a dimeric ATP synthase of
about 1.25 MDa in size and play an important role in the determination of
cristae morphology of the inner mitochondrial membrane [8].
From an enzymatic functional point of view, the F1 complex is the catalytic,
ATP producing or consuming mechanochemical motor, while the Fo complex
represents the electrical motor that generates torque by dissipating the ion
gradient by ion translocation. ATP synthesis is driven by the flow of ions
through Fo, leading to a rotation of the c-(rotor) ring, which transmits rotation
into F1 via the subunit. It is the intrinsically asymmetric subunit that finally
elicits sequential conformational changes in the three catalytic  subunits,
leading to ATP synthesis [9,10].
Inhibitors of ATP synthase have played an important role in the discovery and
biochemical characterization of ATP synthases over many decades (for a
review, see [11]). The ATP synthesis or hydrolysis can be inhibited by a range
of compounds that bind, for example, to the rotor-stator interface region within
the F1 headpiece thereby interfering either with the rotational ATP
synthesizing or ATP hydrolyzing mechanism, or both [11,12]. Among them is
one particular class of natural products, known as polyphenols, which
includes stilbene derivatives, such as resveratrol and piceatannol, and
flavonoids, such as quercetin (Fig. 1A). Natural polyphenols are found in
grapes, peanuts, berries, and red wine. Due to their pharmacokinetic
properties and relatively low affinities to human ATP synthases, they are non-
toxic at concentrations found in their natural sources. They have been shown
This article is protected by copyright. All rights reserved.

to extend the life span of simple organisms [13] but their value in human
medicine remains to be determined.
Stilbenes closely resemble the azobenzenes, a very well-established class of
synthetic photoswitches [14]. This suggested that replacement of the stilbene
moiety with an isosteric azobenzene (“azologization”) could convert
resveratrol and piceatannol into photoswitchable inhibitors whose activity
could be turned ON and OFF by light ([15], Fig. 1B). Azobenzenes undergo
fast photoisomerization from the thermodynamically more stable, linear trans
to the more unstable, bent cis form upon irradiation with UV-A or visible light
[16]. The cis form reverts thermally or can be switched back actively by a
different wavelength of light. The wavelength needed for photoisomerization
and the rate of thermal reversion can be tuned by modification of the
azobenzene chromophore [17,18]. Unlike caged compounds, whose uncaging
is irreversible, azobenzenes can be switched ON and OFF repeatedly and
over thousands of cycles [16]. They have played a central role in the
development of photopharmacology, which is an attempt to control biological
function with artificial photoswitches [19-21]. This concept has been applied to
biological targets as diverse as ion channels, G-protein coupled receptors,
enzymes, and microtubules [22-28].
We now report on the extension of photopharmacology to F-type ATP
synthases. To this end, we introduce Photoswitchable Inhibitors of ATP
Synthase, termed PIAS 1-4 (Fig. 1). We demonstrate that ATP hydrolysis of
our test system, the purified yeast Yarrowia lipolytica F1Fo-ATP synthase (Fig.
1D, E and F), can be optically switched ON and OFF in vitro by using these
molecules. Our results provide a blueprint for the development of precision
tools to spatiotemporally control ATP levels and pH gradients in biological
systems.
This article is protected by copyright. All rights reserved.

Results
Design, synthesis and photophysical characterization of the PIAS. The
design of the PIAS was based on an X-ray crystal structure of the bovine
mitochondrial ATP synthase in complex with resveratrol, piceatannol and
quercetin (Fig. 1). According to this structure, the polyphenols bind in a
hydrophobic pocket between the rotor  subunit C-terminal end and the
surrounding region formed by the stator  and  subunits (Fig. 1A, [29]). This
interaction blocks the rotation of the rotor against the stator and thereby
catalysis. Accordingly, we reasoned that the PIAS inhibit the ATPase in their
trans form, which closely resembles the polyphenols, and remain inactive in
their bent cis form (Fig. 1B), which could not be accommodated in the sleeve-
like binding site defined by the X-ray crystal structure.
The synthesis of PIAS 1-4 by azo-coupling is shown in Fig. 1C. Diazotization
of 3,5-dimethoxyaniline (1) gave the diazonium salt 2, which was treated in
situ with the sodium salt of either phenol or guaiacol to give PIAS-2 and PIAS-
3 in 75 and 70% yield, respectively. Global demethylation using excess boron
tribromide in dichloromethane yielded PIAS-1, an azolog of resveratrol, and
PIAS-4, an azolog of piceatannol, in 64% and 32% yield, respectively. PIAS
1-3 had previously been prepared by a similar sequence [30].
1
All four compounds (≥95% pure by H NMR) showed comparable absorption
maxima between 340 and 360 nm in acetonitrile/water solution. No
isomerization to the cis isomer was apparent upon irradiation with 365 nm
even with high-power LED light. This is due to the known very fast thermal
relaxation of azobenzenes that bear a para-hydroxy group [31,32]. In the dark
as well as under ambient light conditions, we could only observe the trans
1
isomer using H NMR spectroscopy. The photostationary states of the PIAS
under physiological conditions could not be determined directly for the same
reason.
This article is protected by copyright. All rights reserved.

Choice of Yarrowia lipolytica ATP synthase as test system. The crystal
structure of trans-resveratrol and trans-piceatannol in complex with the bovine
F-type ATPase shows a distinct binding pocket for polyphenols formed
between an -subunit, the TP-subunit and the C-terminal part of the subunit
(Fig. 1A, [29]). The residues of the binding pocket of polyphenols in the bovine
enzyme are well conserved both within eukaryotic and bacterial types of ATP
synthases (Fig. S2, alignment), suggesting that they harbor an identical
binding pocket in the Yarrowia lipolytica ATP synthase except for three amino
acids in the -subunit of which two (263V and E264D) are conservative and
in the case of K260N the interaction occurs via hydrophobic interactions,
while the positively charged nitrogen points away from the resveratrol
molecule. Given that and the fact that the purification procedure of Y. lipolytica
ATP synthase was already available in our lab [8], we decided to use this
ATPase as a model system to study the effect of the PIAS on the ATPase
hydrolytic activity. First, the fully assembled, monomeric form of Y. lipolytica
ATP synthase was purified (Fig. 1D, E and F). Then the ATP hydrolysis
activity of the enzyme was determined to be 7 U/mg with a variance of 3 U/mg
due to three biological replicates measured. The measured activities are well
in the range of reported literature values [8]. To test the F1-Fo coupled activity
of the enzyme, we used the inhibitor oligomycin, which showed that 95±5% of
this ATPase's activity could be inhibited [33].
Inhibition of Y. lipolytica ATP hydrolysis activity by resveratrol and
trans-PIAS 1-4. Using the fully active, coupled ATP synthase from
Y. lipolytica, we next tested the capability of resveratrol and resveratrol
derivates, PIAS 1-4, to inhibit the ATPase activity in a concentration-
dependent manner and with all compounds in their trans form (Fig. 2). First,
the control experiment was performed, using resveratrol at concentrations
from 0.1 to 1 mM. Resveratrol was able to reduce the ATP hydrolytic specific
activity to 1.4 U/mg, corresponding to 16% of the initially uninhibited activity.
The inhibition experiments were then performed using different concentrations
from 0.1 to 200 µM (PIAS-1, PIAS-4) and 0.1 to 500 µM (PIAS-2-3) and the
inhibitory concentration that inhibits 50% of the initial activity (IC50) was
This article is protected by copyright. All rights reserved.

calculated and determined, showing an IC50 of 184.7±17.7 µM for resveratrol
in the Y. lipolytica ATP synthase.
Further, all four PIAS showed a concentration-dependent inhibitory effect on
ATP hydrolysis activity, however to various degrees. Among these
derivatives, PIAS-4, the azo-polyphenol with an additional hydroxy group
(R4 = OH), was the most potent inhibitor, showing 100% ATPase inhibition at
a concentration of 500 µM (IC50 = 14±2.7 µM). PIAS-1, the azo-resveratrol
derivative, was less potent and inhibited the ATPase with an
IC50 = 139.6±56.6 µM. The methylated derivatives PIAS-2 and PIAS-3, finally,
showed the lowest inhibitory activity, with an IC50 above 200 µM in both
cases.
Optical control of ATP synthase activity in vitro. The PIAS derivatives
undergo an isomerization from the stable trans isomer to the cis isomer upon
irradiation with UV-A light (Fig. 1B/C). Next, to study the impact of the cis
isomers on ATPase, we used a 190 mW (Min) UV-A laser at a wavelength of
365 nm to irradiate the samples for 3 minutes and determined their ATP
hydrolysis activity shortly thereafter (Fig. S1). The PIAS concentrations were
chosen (PIAS-1: 500 µM, PIAS-2-3: 200 µM and PIAS-4: 50 µM) according to
their previously determined IC50 values. The chosen concentrations should
ensure inhibition up to 60% while not blocking ATP hydrolysis completely, to
be able to observe photoswitching effects (Fig. 3A). While the non-irradiated
PIAS-1-4 inhibited the Y. lipolytica ATP hydrolysis activities to different
degrees, the samples irradiated with 365 nm light showed a reduced
capability to inhibit the ATPase hydrolytic activity, comparable to the yeast
wild-type sample (compare purple columns of PIAS 1-4 with YF1Fo in Fig. 3A).
Specifically, the non-irradiated PIAS-4 inhibited ATP hydrolysis activity best
and up to 90% at 50 µM after 3 minutes (black column of PIAS-4 in Fig. 3A) ,
while irradiated for 3 minutes, PIAS-4 lost its potency for inhibition and
showed only 55% of ATP hydrolysis inhibition (purple column of PIAS-4 and
YF1Fo in Fig. 3A). Remarkably, while all other three PIAS compounds (PIAS-
1, PIAS-2 and PIAS-3) were less potent in ATPase inhibition, they were able
to regain the full level of wild-type activity, showing no more inhibitory effect
(purple columns of PIAS-1-3 compared with YF1Fo in Fig. 3A). Hence the 3
This article is protected by copyright. All rights reserved.

minute irradiation procedure completely abrogated their capability to inhibit
ATPase.
ATPase activity can be reversibly controlled by light. Next, we determined
to what extent the described effect of PIAS 1-4 on ATP hydrolysis activity can
be reversed to restore ATPase activity (Fig. 3B). To test this, we used an
irradiation protocol that was applied on three differently treated 1 g samples
of Y. lipolytica ATPase: the first sample was kept in the dark for 4 minutes, a
second sample was irradiated for just 1 minute followed by 3 minutes in the
dark and the third sample was irradiated continuously for 4 minutes. After
these treatments, the ATPase activity was determined (Fig. 3B). Generally, all
compounds, PIAS 1-4, showed the lowest effects on ATP hydrolysis activity
while being in the dark (Fig. 3B; 0-4 min: dark) but the highest ATPase
activities upon continuous UV irradiation (Fig. 3B; 0-4 min: UV). Remarkably,
ATP hydrolysis activity for the measured samples with 1 minute irradiation
(Fig. 3B; 0-1 min UV, 1-4 min: dark) is lower than the activity for the measured
samples with 4 minutes irradiation time, which exemplifies the reversibility of
ATPase inhibition using PIAS 1-4.
The observed results can be rationalized the following way: The PIAS 1-4
compounds in the cis conformation do not inhibit the ATP hydrolysis activity
during the first minute of irradiation, however, they all isomerize back into their
trans-isoforms under dark conditions. It is this trans-isoform in which PIAS 1-4
are again capable of inhibiting the enzyme. The differences in ATP hydrolysis
activities for the three measured samples using PIAS 1-4 are due to their
different potency to inhibit the ATPase: PIAS-4 inhibits up to 90% under dark
conditions, approximately 80% with 1 min irradiation and 65% with continuous
irradiation, while the non-irradiated PIAS 1-3 inhibit between 40-50%,
between 20-30% with 1 min irradiation and do not show any inhibition effect
under continuous irradiation.
This article is protected by copyright. All rights reserved.

Discussion
This study reports about the synthesis of new, reversible photoswitches
derived from natural stilbenoid polyphenols such as resveratrol, for the optical
control of the yeast Y. lipolytica ATPase activity. Resveratrol itself can be
photoisomerized, but this requires short, cytotoxic wavelengths (<300 nm)
and eventually results in the formation of a photochemical byproduct
(resveratrone) [34]. We therefore synthesized four azobenzenes, PIAS 1-4,
and evaluated them for their ability to reversibly inhibit ATPase activity, with
resveratrol as reference inhibitor. The trans isomers indeed inhibit the ATP
hydrolytic activity, while isomerization to the cis isomers decreased enzymatic
activity inhibition. The study also demonstrates that PIAS-1-4 can be used as
reversible ON/OFF switches for Y. lipolytica ATPase.
The high sequence homology between bovine and yeast Y. lipolytica ATPase
and the isosteric nature of resveratrol and trans-PIAS-1 suggests that the
molecular mechanism of binding and ATPase inhibition of the two compounds
is identical. To support this notion we further explored this possibility with
molecular modeling using Maestro ([35], Fig. 4C). The template yeast F1-
PIAS-4 bound structure used for the molecular modeling, was created in
Pymol [36] by merging the bovine F1-resveratrol structure (pdb ID 2jiz, [29])
with the Y. lipolytica ATPase (pdb ID 5fl7, [8]) and PIAS-4. Our model of
trans-PIAS-4 bound to the Y. lipolytica enzyme shows hydrophobic
interactions of the azobenzene with residues from three different subunits of
the F1 complex. The residues V283, TPV310 and TPA309 are equivalent to
residues I263, TPV279 and TPA278 in bovine F1. Additionally, non-polar
interactions are formed by side chains that involve the two residues A319
and A276. The trans-PIAS-4 binding to the Y. lipolytica F1 complex appears
to be further stabilized by a hydrogen bond network involving the hydroxy
groups of trans-PIAS-4 and F1 complex intrinsic water molecules. It has been
shown in purified E. coli F1 and F1Fo ATPase in membrane vesicles that the
relative positions of hydroxy groups of polyphenols appear to be critical for the
degree of inhibition of ATPase hydrolysis [37]. The four compounds PIAS 1-4,
This article is protected by copyright. All rights reserved.

which have either methoxy or hydroxy groups at positions 1-4 inhibited the Y.
lipolytica ATP synthase to different degrees. Given that and the findings made
for the E. coli ATPase, we therefore suggest that the bulkier methoxy groups
of PIAS-2 and PIAS-3 at positions 2 and 3 provide a rationale to understand
the decreased inhibitory effects on Y. lipolytica ATP synthase by these two
compounds. Their larger side chains may cause steric hindrance in the
inhibitor binding side (Fig. 4D). Without further experimental structural
information available, one cannot exclude the alternative possibility that
decreased inhibition results from the absence of hydrogen bond donors at the
ligand sites R1, R2 and R4.
While our studies were ongoing, Hoersch published the optical control of E.
coli F1-ATPase using a photoswitchable cross linker [38]. Crosslinking with an
azobenzene bis-maleimide between engineered cysteines in the - and -
subunits reduced the ATP hydrolysis activity in a light-dependent fashion. This
approach requires genetic engineering of the ATPase, and covalent
attachment with maleimides. In contrast, our study uses the complete and
genetically unmodified, native F1Fo ATP synthase holoenzyme (Fig. 1D, E and
F), hence it has the potential to be used with broader applicability, e.g. in
genetically non-modified host cells.
Finally, our work extends the reach of photopharmacology to an important
new target class and provides a blueprint for the development of
photoswitches that enable to spatio-temporally control ATP dependent
reactions, for example in in vitro biotechnological applications. The exact
mechanism by which resveratrol promotes a wide range of beneficial effects
in humans is still unclear. As PIAS 1-4 inhibit ATPase in an analogous
fashion, future photochemical experiments with the PIAS on other resveratrol
targets and pathways such as AMPK and SIRT1, which are the key metabolic
effectors of resveratrol, could provide more insights on the fundamental
biochemical actions of resveratrol. This work could be explored to other
resveratrol targets such as cyclooxygenases [39], phosphodiesterases [40]
and estrogen receptors [41]. PIAS-1 and PIAS-4 provide a proof of principle
This article is protected by copyright. All rights reserved.

and a basis for further chemical modifications that potentially can fulfill the
requirements of in vivo studies, e.g. high binding affinities and mitochondrial
targeting motifs. As the studied PIAS activities can be switched off, future
work could shed more light on the molecular mechanisms governing the
biochemical properties of resveratrol.
Materials and Methods
Chemical synthesis. All reactions of the chemical synthesis of PIAS 1-4
were performed with standard Schlenk techniques under an atmosphere of
nitrogen in oven-dried glassware (100 °C oven temperature) that was further
dried using a heat gun (set to 650 °C) for all water-sensitive reactions.
Dichloromethane (CH2Cl2) was distilled from calcium hydride. Reagents were
purchased from Sigma-Aldrich, TCI or Acros Organics and used without
further purification. Reaction progress was monitored by analytical thin-layer
chromatography (TLC), which was carried out using pre-coated glass plates
(silica gel 60 F254) from Merck. Visualization was achieved by exposure to
ultraviolet light (UV, 254nm) where applicable followed by staining with
potassium permanganate solution. Flash column chromatography was
performed using Merck silica gel (40–63 μm particle size). Proton nuclear
magnetic resonance (1H NMR) spectra were recorded on a Varian 300,
Varian 400, Inova 400 or Varian 600 spectrometer. Chemical shifts (δ scale)
are expressed in parts per million (ppm) and are calibrated using residual
protic solvent as an internal reference (CHCl3: δ = 7.26 ppm, CD3OD:
1
δ = 3.31 ppm). Data for H NMR spectra are reported as follows: chemical
shift (ppm) (multiplicity, coupling constants (Hz), integration). Couplings are
expressed as: s = singlet, d = doublet, t = triplet, m = multiplet or combinations
thereof. Carbon nuclear magnetic resonance (13C NMR) spectra were
recorded on the same spectrometers at 75, 100 and 150 MHz (±1 MHz
variance). Carbon chemical shifts (δ scale) are also expressed in parts per
million (ppm) and are referenced to the central carbon resonances of the
solvents (CDCl3: δ = 77.16 ppm, CD3OD: δ = 49.00 ppm). Infrared (IR)
This article is protected by copyright. All rights reserved.

spectra were recorded on a Perkin Elmer Spectrum BX II (FTIR System)
equipped with an attenuated total reflection (ATR) measuring unit. IR data is
reported in frequency of absorption (cm-1). Mass spectroscopy (MS)
experiments were performed on a Thermo Finnigan MAT 95 (electron
ionization, EI) or on a Thermo Finnigan LTQ FT (electrospray ionization, ESI)
instrument.
4-((3,5-dimethoxyphenyl)diazenyl)phenol (PIAS-2). To an ice-cold solution
of 3,5-dimethoxyaniline (306 mg, 2.00 mmol, 1.00 eq.) in THF (5 mL) and HCl
(1 M, 6 mL) an aqueous solution (5 mL) of sodium nitrite (166 mg, 2.40 mmol,
1.20 eq.) was added dropwise, resulting in the formation of a dark-red
suspension. After 30 min, a solution of phenol (226 mg, 2.40 mmol, 1.20 eq)
in aqueous NaOH (1 M, 6 mL) was added dropwise. The mixture was stirred
at 0 °C for 1 h before the bulk of the solvent was removed in vacuo. The
residue was redissolved in ethyl acetate (15 mL) and water (15 mL). After
phase separation, the aqueous phase was further extracted with ethyl acetate
(3 × 20 mL). The combined organic layers were washed with water (20 mL)
and aqueous saturated sodium chloride (20 mL), then dried over MgSO4 and
concentrated under reduced pressure. Purification by flash column
chromatography (5:1 hexane: ethyl acetate) afforded PIAS-2 (387 mg,
1.50 mmol, 75%) as a yellow solid.
1H NMR (400 MHz, CD3OD) δ = 7.81–7.71 (m, 2H), 6.95–6.86 (m, 2H), 6.82
(d, J = 2.2, 2H), 6.38 (t, J = 2.2, 1H).
HRMS (ESI) m/z calculated for C14H15N2O3 259.1077; found 259.1080.
(M+H+).
Analytical data were in good agreement with literature values.[30]
4-((3,5-dimethoxyphenyl)diazenyl)-2-methoxyphenol (PIAS-3). To an ice-
cold solution of 3,5-dimethoxyaniline (306 mg, 2.00 mmol, 1.00 eq.) in THF
(5 mL) and HCl (1 M, 6 mL) an aqueous solution (5 mL) of sodium nitrite
(166 mg, 2.40 mmol, 1.20 eq.) was added dropwise, resulting in the formation
of a dark-red suspension. After 30 min, a solution of guaiacol (298 mg,
2.40 mmol, 1.20 eq) in aqueous NaOH (1 M, 6 mL) was added dropwise. The
mixture was stirred at 0 °C for 1 h before the bulk of the solvent was removed
This article is protected by copyright. All rights reserved.

in vacuo. The residue was redissolved in ethyl acetate (15 mL) and water
(15 mL). After phase separation, the aqueous phase was further extracted
with ethyl acetate (3 × 20 mL). The combined organic layers were washed
with water (20 mL) and aqueous saturated sodium chloride (20 mL), then
dried over MgSO4 and concentrated under reduced pressure. Purification by
flash column chromatography (5:1 hexane:ethyl acetate) afforded PIAS-3
(243 mg, 1.40 mmol, 70%) as a yellow solid.
1H NMR (400 MHz, CDCl3 δ = 7.61 (dt, J = 8.4, 1.5, 1H), 7.50 (t, J = 1.5, 1H),
7.10–7.08 (m, 2H), 7.06 (dd, J = 8.4, 1.1, 1H), 6.57 (q, J = 2.0, 1H), 5.95 (s,
1H), 4.00 (d, J = 1.1, 3H), 3.88 (d, J = 1.1, 6H).
HRMS (ESI) m/z calculated for C15H17O4N2 289.1183; found 289.1187
(M+H+).
Analytical data were in good agreement with literature values.[30]
5-((4-hydroxyphenyl)diazenyl)benzene-1,3-diol (PIAS-1). To a solution of
dimethyl PIAS-2 (77 mg, 0.30 mmol, 1.00 eq) in CH2Cl2 (8 mL) was added
boron tribromide solution in CH2Cl2 (1.0 M, 2.1 mL, 2.1 mmol, 7.0 eq)
dropwise at 0 °C. After warming to room temperature over 14 h, LC-MS
analysis indicated full conversion. Saturated aq. NaHCO3 (15 mL) was added
at room temperature and the mixture was poured on water (10 mL). After
extraction of the aq. phase with EtOAc (4 × 20 mL), the combined organic
phases were washed with aqueous saturated sodium chloride (20 mL), dried
and evaporated to give a black oil that was purified by column
chromatography (9:1 CH2Cl2:MeOH) to give PIAS-1 as a dark-red solid
(44 mg, 0.19 mmol, 64%).
1H NMR (400 MHz, CD3OD) δ = 7.81–7.71 (m, 2H), 6.95–6.86 (m, 2H), 6.82
(d, J = 2.2, 2H), 6.38 (t, J = 2.2, 1H).
HRMS (ESI) m/z calculated for C12H10O3N2 231.0764; found 231.0763
(M+H+).
Analytical data were in good agreement with literature values [30].
This article is protected by copyright. All rights reserved.

4-((3,5-dihydroxyphenyl)diazenyl)benzene-1,2-diol (PIAS-4). To a solution
of PIAS-3 (115 mg, 0.40 mmol, 1.00 eq) in CH2Cl2 (8 mL) was added boron
tribromide solution in CH2Cl2 (1.0 M, 3.6 mL, 3.6 mmol, 9.0 eq) dropwise at
0 °C. After warming to room temperature over 14 h, LC-MS analysis indicated
full conversion. Saturated aq. NaHCO3 (15 mL) was added at room
temperature and the mixture was poured on water (10 mL). After extraction of
the aq. phase with EtOAc (4 × 20 mL), the combined organic phases were
washed with aqueous saturated sodium chloride (20 mL), dried and
evaporated to give a black oil that was purified by column chromatography
(9:1 CH2Cl2:MeOH) to give PIAS-4 as a dark-red amorphous solid (32 mg,
0.13 mmol, 32%).
Rf 0.52 (water:MeOH 3:2)
1H NMR (400 MHz, CD3OD) δ = 7.37–7.33 (m, 2H), 6.92–6.88 (m, 1H), 6.79
(d, J = 2.2, 2H), 6.37 (t, J = 2.2, 1H).
13C NMR (101 MHz, CD3OD) δ = 160.0, 156.1, 150.4, 147.7, 147.0, 120.2,
115.9, 107.6, 105.4, 102.1.
IR (ATR) 3258, 1675, 1601, 1379, 1284, 1156, 1001 cm-1.
HRMS (ESI) m/z calculated for C12H11O4N2 247.0641; found 247.0713.
See Figure S3 and Figure S4 for NMR spectra of PIAS-4.
Purification of Yarrowia lipolytica F1Fo-ATP synthase. Monomeric form of
Yarrowia lipolytica F1Fo-ATP synthase was purified as described in [8]. Briefly,
the ATP synthase was isolated from mitochondria prepared from large-scale
Y. lipolytica cultures [42]. The isolation, solubilisation and collection of
solubilised material from mitochondrial membranes was carried out as
previously described [43]. After the removal of Complex I by metal affinity
purification [43], glycerol was added to a final concentration of 20% (v/v) to
the solubilised membranes, which were rapidly frozen in liquid nitrogen for
storage at -80°C. The solubilisate was thawed on ice and supplied with
50 mM MgCl2. To the slow stirring suspension on ice, 3% (w/w) polyethylene
glycol (PEG) 6’000 was added from a 50% (w/w) stock solution to induce
This article is protected by copyright. All rights reserved.

protein precipitation. After 15 min of continued stirring on ice, the precipitated
proteins were removed by centrifugation (15 min and 20’000 g at 4°C). The
ATP synthase containing fraction was then precipitated by the increase of the
PEG 6’000 concentration to 6% (w/w), again with continuous stirring on ice for
15 min and then collected by centrifugation for 15 min, 20’000 g at 4°C. The
pellet was then dissolved in 2 ml of buffer A [30 mM 3-(N-morpholino)-
propanesulfonic acid (MOPS/NaOH pH 7.4), 4 mM MgCl2, 2 mM
ethylenediaminetetraacetic acid (EDTA) and 0.1% (w/v) DDM], the sample
was applied on a density-based discontinuous glycerol gradient (1 ml steps
with 15, 20, 25, 28, 30, 35, 40, 45, 50% glycerol in buffer A) and run for 16 h
at 4°C and 34’600 rpm in a SW40 rotor (Beckman Coulter, USA). After the
run, 1 ml fractions were collected from top using a pipette and the ATP
synthase-containing fractions, as judged by high resolution Clear Native
PAGE (hrCN-PAGE [44]), were pooled and directly loaded onto an anion
exchange chromatography using a POROS GoPure HQ 50 anion exchange
column (Life Technologies, USA), which was previously equilibrated with 1
column volume (CV) of buffer A, using an ÄKTAexplorer chromatography
system (GE Healthcare, USA). The column was then washed with 1 CV of
buffer A and the Y. lipolytica ATP synthase was eluted by a continuous
gradient using buffer B (buffer A with 1 M NaCl). The ATP synthase as judged
by high resolution Clear Native PAGE (hrCN-PAGE [44]), was concentrated to
1 mg/ml by ultrafiltration using Vivaspin PES membranes with a molecular
weight cut-off of 100 kDa at 1’500 g and 4°C to 2 ml final volume. The protein
concentration was determined by using the bicinchoninic acid (BCA) method
(Pierce, ThermoFisher). Bovine serum albumin was used as a standard
between 2 and 2000 g/ml.
Determination of ATP hydrolysis activity using the malachite green
assay. The ATP hydrolysis activity was determined using the malachite green
assay as described in [45-47]. Briefly, 1 mg/ml Y. lipolytica ATP synthase
protein solution was prepared in a reaction tube and supplemented with
10 µg/ml cardiolipin and 40 µg/ml yeast extract lipids and diluted in the
reaction buffer (50 mM Tricine-NaOH pH 8.0, 5 mM MgCl2; 150 µl per
This article is protected by copyright. All rights reserved.

experiment). The malachite green stock reagent (320 µl per experiment) was
prepared [0.08% (w/v) malachite green: 2.3% (v/v) polyvinyl alcohol: 5.7%
(w/v) ammonium molybdate: 6 N HCl in distilled water (ratio 2:1:1:2) and
incubated for 30 minutes at room temperature before use. A time course for
ATPase activity was measured, taking sample aliquots from the reaction tube
and stopping the reaction after 4 time points, in 1 minute intervals: Each
reaction was started by the addition of 5 mM Na2-ATP (pH 7.4) using a 0.2 M
stock solution. At each taken time point, an aliquot of 20 µl of reaction mix
was transferred into 80 µl malachite green reagent and immediately mixed.
After 75 seconds each reaction was quenched by the addition of 34% (w/v)
sodium citrate. Each time course was measured in triplicates, a calibration
curve (0, 3, 6 and 12 nmol Pi) was determined in duplicates for each
measurement. Note: The malachite green assay is a colorimetric method for
measuring Pi in aqueous solutions and was performed on aliquots taken from
the reaction tube. As such, it does not interfere with photoswitching.
Conversely, the azobenzene has no absorption beyond 600 nm, which could
potentially falsify the assay.
Inhibition of ATP hydrolysis activity by resveratrol and PIAS 1-4. PIAS 1-
4 stock solutions (concentrations: 0.01, 0.1, 0.5, 5, 10, 20 and 50 mM,
dissolved in methanol) and resveratrol (concentration 0.01, 0,1, 0.5, 5, 10, 50
and 100 mM (Sigma-Aldrich, D) in pure ethanol, were diluted to 1% (v/v) and
added to the Y. lipolytica ATP synthase sample (1 mg/ml diluted 1/100 in
reaction buffer) and incubated at room temperature for 1 h. As a control, 1%
(v/v) of methanol/ethanol was added only. The ATP hydrolysis activities were
then determined using the malachite green assay. To next study the effect of
UV irradiation on the samples containing the azo-compounds (absorption
maxima at 365 nm) and their effect on the change in ATP hydrolysis activity
due to UV irradiation, the UV irradiation experiments were performed the
following way: a 365 nm- laser [Thorlabs M365L2 - UV (365 nm) mounted
LED, 700 mA, 190 mW (Min)] was used to constantly irradiate the reaction
mix in each sample in the cuvette at a distance of 91,4 mm in a homemade
setup as shown in Fig. S1. The irradiation time for the whole reaction time of
4 min. After that, the malachite green assay was performed the same as
This article is protected by copyright. All rights reserved.

described for non-irradiated samples. Table S1 shows the results of each
measurement before and after irradiation.
Statistical analysis. The results (Fig. 3) were represented as the mean ±
SEM of the three replicates from three independent experiments (n = 9). A
calibration curve for the absorption/Pi was generated to calculate the Pi
concentration at each time point (Pi/time) from the absorption of the malachite
green complex at 620 nm (absorption/time). The Pi concentration per time
corresponds to the ATP hydrolysis activity in Units/min. The percentage of
inhibition was determined by normalizing the ATP hydrolysis activity of
solutions with inhibitors with the native, monomeric Y. lipolytica ATP hydrolytic
enzyme's activity. The values of the inhibitor concentration at which 50% of
the ATP hydrolysis activity was inhibited (IC50 values) were calculated using
GraphPad Prism® 5, version 5.01 (GraphPad Software, Inc., USA) by plotting
the log concentration of the azo-compounds versus percentage inhibition of
ATP hydrolysis activities.
Acknowledgements
We thank Jürgen Reichert (Max-Planck-Institute of Biophysics, Frankfurt am
Main, Germany), who designed and constructed the device for the UV
measurements (Fig. S1). This work was financially supported by the Deutsche
Forschungsgemeinschaft (SFB 807 and SFB 1032) and the Wellcome Trust
(WT110068/Z/15/Z).
Author contributions
T.M. and D.T. conceived and directed the study. B.E. designed and carried
out biological experiments and analyzed data, F.H. designed and carried out
chemical syntheses. All authors contributed to writing the manuscript.
Conflict of interest statement. The authors declare no competing financial
interests.
This article is protected by copyright. All rights reserved.

References
[1]
Junge, W. and Nelson, N. (2015). ATP synthase. Annu Rev Biochem 84, 631-
57.
[2]
Meier, T., Faraldo-Gómez, J.D. and Börsch, M. (2011) ATP Synthase – A
Paradigmatic Molecular Machine. In Molecular Machines in Biology:
Workshop of the Cell (Frank, J., ed.), pp. 208-238. Cambridge University
Press, Cambridge.
[3]
[4]
[5]
[6]
Schulz, S., Wilkes, M., Mills, D.J., Kuhlbrandt, W. and Meier, T. (2017).
Molecular architecture of the N-type ATPase rotor ring from Burkholderia
pseudomallei. EMBO Rep 18, 526-535.
Watt, I.N., Montgomery, M.G., Runswick, M.J., Leslie, A.G. and Walker, J.E.
(2010). Bioenergetic cost of making an adenosine triphosphate molecule in
animal mitochondria. Proc Natl Acad Sci U S A 107, 16823-7.
Allegretti, M., Klusch, N., Mills, D.J., Vonck, J., Kühlbrandt, W. and Davies,
K.M. (2015). Horizontal membrane-intrinsic alpha-helices in the stator a-
subunit of an F-type ATP synthase. Nature 521, 237-40.
Morales-Rios, E., Montgomery, M.G., Leslie, A.G.W. and Walker, J.E.
(2015). Structure of ATP synthase from Paracoccus denitrificans determined
by X-ray crystallography at 4.0 Å resolution. Proc Natl Acad Sci USA 112,
13231-6.
[7]
[8]
Zhou, A., Rohou, A., Schep, D.G., Bason, J.V., Montgomery, M.G., Walker,
J.E., Grigorieff, N. and Rubinstein, J.L. (2015). Structure and conformational
states of the bovine mitochondrial ATP synthase by cryo-EM. Elife 4, e10180.
Hahn, A., Parey, K., Bublitz, M., Mills, Deryck J., Zickermann, V., Vonck, J.,
Kühlbrandt, W. and Meier, T. Structure of a complete ATP synthase dimer
reveals the molecular basis of inner mitochondrial membrane morphology.
Mol Cell 63, 445-456.
[9]
Abrahams, J.P., Leslie, A.G.W., Lutter, R. and Walker, J.E. (1994). Structure
at 2.8 Å resolution of F1-ATPase from bovine heart mitochondria. Nature 370,
621-8.
[10] Boyer, P.D. (1997). The ATP synthase--a splendid molecular machine. Annu
Rev Biochem 66, 717-49.
[11] Hong, S. and Pedersen, P.L. (2008). ATP synthase and the actions of
inhibitors utilized to study its roles in human health, disease, and other
scientific areas. Microbiol Mol Biol Rev 72, 590-641.
[12] Gledhill, J.R. and Walker, J.E. (2006). Inhibitors of the catalytic domain of
mitochondrial ATP synthase. Biochem Soc Trans 34, 989-92.
[13] Bhullar, K.S. and Hubbard, B.P. (2015). Lifespan and healthspan extension by
resveratrol. Biochim Biophys Acta 1852, 1209-18.
[14] Szymański, W., Beierle, J.M., Kistemaker, H.A.V., Velema, W.A. and
Feringa, B.L. (2013). Reversible photocontrol of biological systems by the
incorporation of molecular photoswitches. Chem. Rev 113, 6114-6178.
[15] Schoenberger, M., Damijonaitis, A., Zhang, Z., Nagel, D. and Trauner, D.
(2014). Development of a new photochromic ion channel blocker via
azologization of fomocaine. ACS Chem. Neurosci 5, 514-518.
[16] Bandara, H.M.D. and Burdette, S.C. (2012). Photoisomerization in different
classes of azobenzene. Chem Soc Rev. 41, 1809-1825.
This article is protected by copyright. All rights reserved.

[17] Dong, M., Babalhavaeji, A., Samanta, S., Beharry, A.A. and Woolley, G.A.
(2015). Red-shifting azobenzene photoswitches for in vivo use. Acc. Chem.
Res. 48, 2662-2670.
[18] Konrad, D.B., Frank, J.A. and Trauner, D. (2016). Synthesis of redshifted
azobenzene photoswitches by late-stage functionalization. Chem. Eur. J 22,
4364-4368.
[19] Broichhagen, J., Frank, J.A. and Trauner, D. (2015). A roadmap to success in
photopharmacology. Acc. Chem. Res. 48, 1947-1960.
[20] Lerch, M.M., Hansen, M.J., van Dam, G.M., Szymanski, W. and Feringa, B.L.
(2016). Emerging targets in photopharmacology. Angew. Chem. Int. Ed 55,
10978-10999.
[21] Velema, W.A., Szymanski, W. and Feringa, B.L. (2014). Photopharmacology:
Beyond proof of principle. J. Am. Chem. Soc 136, 2178-2191.
[22] Borowiak, M. et al. (2015). Photoswitchable inhibitors of microtubule
dynamics optically control mitosis and cell death. Cell 162, 403-411.
[23] Frank, J.A. et al. (2016). Photoswitchable diacylglycerols enable optical
control of protein kinase C. Nat Chem Biol 12, 755-762.
[24] Hartrampf, F.W.W., Barber, D.M., Gottschling, K., Leippe, P., Hollmann, M.
and Trauner, D. (2017). Development of a photoswitchable antagonist of
NMDA receptors. Tetrahedron 73, 4905.
[25] Laprell, L. et al. (2015). Optical control of NMDA receptors with a diffusible
photoswitch. Nature Communications 6, 8076.
[26] Pittolo, S. et al. (2014). An allosteric modulator to control endogenous G
protein-coupled receptors with light. Nat Chem Biol 10, 813-815.
[27] Reis, S.A. et al. (2016). Light-controlled modulation of gene expression by
chemical optoepigenetic probes. Nat Chem Biol 12, 317-323.
[28] Szymanski, W., Ourailidou, M.E., Velema, W.A., Dekker, F.J. and Feringa,
B.L. (2015). Light-controlled histone deacetylase (HDAC) inhibitors:
Towards photopharmacological chemotherapy. Chem. Eur. J 21, 16517-
16524.
[29] Gledhill, J.R., Montgomery, M.G., Leslie, A.G.W. and Walker, J.E. (2007).
Mechanism of inhibition of bovine F1-ATPase by resveratrol and related
polyphenols. Proc Natl Acad Sci USA 104, 13632-7.
[30] Song, Y.M. et al. (2012). Synthesis of novel azo-resveratrol, azo-
oxyresveratrol and their derivatives as potent tyrosinase inhibitors. Bioorg.
Med. Chem. Lett 22, 7451-7455.
[31] Garcia-Amoros, J., Sanchez-Ferrer, A., Massad, W.A., Nonell, S. and
Velasco, D. (2010). Kinetic study of the fast thermal cis-to-trans isomerisation
of para-, ortho- and polyhydroxyazobenzenes. Phys. Chem. Chem. Phys 12,
13238-13242.
[32] Kojima, M., Nebashi, S., Ogawa, K. and Kurita, N. (2005). Effect of solvent
on cis-to-trans isomerization of 4-hydroxyazobenzene aggregated through
intermolecular hydrogen bonds. J. Phys. Org. Chem 18, 994-1000.
[33] Symersky, J., Osowski, D., Walters, D.E. and Mueller, D.M. (2012).
Oligomycin frames a common drug-binding site in the ATP synthase. Proc
Natl Acad Sci U S A 109, 13961-5.
[34] Yang, I., Kim, E., Kang, J., Han, H., Sul, S., Park, S.B. and Kim, S.K. (2012).
Photochemical generation of a new, highly fluorescent compound from non-
fluorescent resveratrol. Chem. Commun 48, 3839-3841.
This article is protected by copyright. All rights reserved.

[35] maestro Schrödinger Release 2016-3: Maestro, Schrödinger, LLC, New York,
NY, 2016ed.)
[36] The PyMOL Molecular Graphics System, Version 1.2r1 Schrödinger,
LLC.ed.^eds)
[37] Ahmad, Z., Ahmad, M., Okafor, F., Jones, J., Abunameh, A., Cheniya, R.P.
and Kady, I.O. (2012). Effect of structural modulation of polyphenolic
compounds on the inhibition of Escherichia coli ATP synthase. Int J Biol
Macromol 50, 476-86.
[38] Hoersch, D. (2016). Engineering a light-controlled F1 ATPase using structure-
based protein design. PeerJ 4, e2286.
[39] MacCarrone, M., Lorenzon, T., Guerrieri, P. and Agro, A.F. (1999).
Resveratrol prevents apoptosis in K562 cells by inhibiting lipoxygenase and
cyclooxygenase activity. Eur J Biochem 265, 27-34.
[40] Park, S.J. et al. (2012). Resveratrol ameliorates aging-related metabolic
phenotypes by inhibiting cAMP phosphodiesterases. Cell 148, 421-33.
[41] Gehm, B.D., McAndrews, J.M., Chien, P.Y. and Jameson, J.L. (1997).
Resveratrol, a polyphenolic compound found in grapes and wine, is an agonist
for the estrogen receptor. Proc Natl Acad Sci U S A 94, 14138-43.
[42] Kashani-Poor, N., Kerscher, S., Zickermann, V. and Brandt, U. (2001).
Efficient large scale purification of his-tagged proton translocating
NADH:ubiquinone oxidoreductase (complex I) from the strictly aerobic yeast
Yarrowia lipolytica. Biochim Biophys Acta 1504, 363-70.
[43] Hunte, C., Zickermann, V. and Brandt, U. (2010). Functional modules and
structural basis of conformational coupling in mitochondrial complex I.
Science 329, 448-51.
[44] Ladig, R. et al. (2011). A high-definition native polyacrylamide gel
electrophoresis system for the analysis of membrane complexes. Plant J 67,
181-94.
[45] Chan, K.M., Delfert, D. and Junger, K.D. (1986). A direct colorimetric assay
for Ca2+ -stimulated ATPase activity. Anal Biochem 157, 375-80.
[46] Maehama, T., Taylor, G.S., Slama, J.T. and Dixon, J.E. (2000). A sensitive
assay for phosphoinositide phosphatases. Anal Biochem 279, 248-50.
[47] Rowlands, M.G., Newbatt, Y.M., Prodromou, C., Pearl, L.H., Workman, P.
and Aherne, W. (2004). High-throughput screening assay for inhibitors of
heat-shock protein 90 ATPase activity. Anal Biochem 327, 176-83.
This article is protected by copyright. All rights reserved.

Figure Legends
Figure 1. Design and synthesis of the PIAS. (A) Structure of resveratrol bound to
bovine F1 ATP synthase (from PDB 2JIZ) shown in cartoon representation. Green: 
and  subunits of F1. Blue:  subunit. The DP-site containing subunit is removed to
provide an unobstructed view of resveratrol wedged between the  rotor subunit and
the ,  stator subunits. Resveratrol (CPK colors, sphere model) is bound in two
overlapping orientations. ATP is shown as a stick model in the TP site. (B)
Azologization of resveratrol affords PIAS-1. (C) Chemical synthesis of PIAS-1-4. (D)
Carton representation of the Y. lipolytica F1Fo ATPase ([8]) and (D) its biochemical
characterization by (E) clear native PAGE, Coomassie stained gel and (F) SDS
polyacrylamide gel electrophoresis (PAGE), silver stained gel.
Figure 2. Inhibition of ATPase activity of Y. lipolytica F1Fo-ATP synthase by
PIAS 1-4. The inhibition of the ATP hydrolysis activity by PIAS 1-4 was determined
using the malachite green assay. The results of the ATP hydrolysis activity
measurements were plotted and fitted against the inhibitor concentration. The IC50
values are listed below in the table. Each measurement was done in triplicates and
three biological replicates.
Figure 3. Optical control of ATPase activity in vitro using purified Y. lipolytica
ATP synthase (A) and reversibility (B). For each measurement, a sample of 0.01
mg/ml ATP synthase in the reaction buffer was used to measure the initial ATP
hydrolytic activity. The activities were normalized against the Y. lipolytica F1Fo
ATPase activity under dark conditions. The ATPase activity was inhibited by adding
1: 500 µM, 2: 200 µM, 3: 200 µM, 4: 50 µM of compound PIAS-1, 2, 3 and 4,
respectively. All concentrations used were higher than the previously determined IC50
for each compound. The reactions were activated by switching UV laser light (365
nm) for 3 min at room temperature. The ATP hydrolysis activity was determined by
the malachite green assay and normalized against the ATP hydrolysis activity of the
native enzyme without UV irradiation. All compounds showed an inhibitory effect on
ATP hydrolysis activity under dark conditions, which could be reduced by UV-A
This article is protected by copyright. All rights reserved.

irradiation. The assays were repeated 4 times and performed in triplicates. (B) For
each inhibitor measured series were performed under dark conditions (indicated by 4
black boxes), 1 min UV irradiation followed by 3 minutes dark conditions (1 white
box, 3 black boxes) and continuous UV-A irradiation (4 white boxes). The
measurement was performed similarly to the UV measurement shown in Fig. 3, apart
from the UV irradiation times: The ATP hydrolysis activity was determined by the
malachite green assay and normalized against the ATP hydrolysis activity of the
native enzyme without UV irradiation. All compounds showed inhibition of ATP
hydrolysis activity under dark conditions, higher activities with 1 min UV irradiation
and the highest activity in case of 4 minutes UV irradiation. The assays were
repeated 3 times and performed in triplicates.
Figure 4. Model PIAS-1 bound to the Y. lipolytica F1 ATPase domain. The bovine
F1-resveratrol structure (pdb ID 2jiz, [29]) was used to create a model with the
Y. lipolytica ATPase (pdb ID 5fl7, [8]) and the azo-polyphenols used in this study.
Colors:  dark green,  light green, blue) (A) Tilted view into the F1-ATPase
upwards from the inner mitochondrial membrane. The azo-resveratrol (stick
representation, highlighted by a black box) binds in a pocket made by the subunits ,
 and . (B) Zoom (boxed in (A)) to interaction site of azo-polyphenol bound to the F1-
ATPase. Hydrogen bonds are indicated by black dashed lines. (C) 2-dimensional
interaction plot of the azo-polyphenol interaction with the F1-ATPase (created using
Maestro [35]) Interaction distances are color coded. The interaction network of the
polyphenol hydrogen bonds as well as hydrophobic and polar interactions contribute
to the binding affinity of the polyphenol. (D) Modeling of cis- and trans-polyphenols in
F1 ATPase. Left panel: Side-view of resveratrol-F1 binding pocket. Various
orientations of cis-polyphenols (in red) are modeled and shown in the four right
panels along with the trans-configurations (black). All cis-molecules generate steric
clashes with the Van-der-Waals radii of the F1 subunits, indicated by the dashed
circles in the left panel.
Figure S1. UV irradiation experiment setup. (A) Side view and top view (B) on
experimental setup. (C) Horizontal cross section through (B) indicated by C-C. (D)
Cross section showing the beam path. A 365 nm- laser [Thorlabs M365L2 - UV (365
nm) mounted LED, 700 mA, 190 mW (Min)] was used to constantly irradiate the
reaction mix in each sample in the cuvette at a distance of 91.4 mm.
This article is protected by copyright. All rights reserved.

Figure S2. Resveratrol binding site in the ATP synthase F1 complex. Alignment
of amino acid sequence of  and  subunit for B. taurus, Y. lipolytica and E. coli.
The residues of the binding pocket of polyphenols in the bovine enzyme are well
conserved (red letters) both within eukaryotic and bacterial types of ATP synthases.
Figure S3. 1H NMR spectrum of PIAS-4. The spectrum was recorded in CD3OD.
Figure S4. 13C NMR spectrum of PIAS-4. The spectrum was recorded in CD3OD.
Table S1. Reversible optical control of ATPase in vitro using purified
Y. lipolytica ATP synthase. The ATP hydrolysis activities with PIAS-1-4 were
determined by the malachite green assay and are shown for irradiation times of 0, 1
and 4 minutes. The ATP hydrolysis activity measurements were performed in
triplicates using four independent biological samples.
This article is protected by copyright. All rights reserved.

FIGURES
Fig. 1
This article is protected by copyright. All rights reserved.

Fig. 2
This article is protected by copyright. All rights reserved.

Fig. 3
This article is protected by copyright. All rights reserved.

Fig. 4
This article is protected by copyright. All rights reserved.