Potent hemithioindigo-based antimitotics photocontrol the microtubule cytoskeleton in cellulo

Article abstract of DOI:10.3762/bjoc.16.14

Background: Hemithioindigo is a promising molecular photoswitch that has only recently been applied as a photoswitchable pharmacophore for control over bioactivity in cellulo. Uniquely, in contrast to other photoswitches that have been applied to biology,

Full text of DOI:10.3762/bjoc.16.14

Potent hemithioindigo-based antimitotics photocontrol the
microtubule cytoskeleton in cellulo
Alexander Sailer, Franziska Ermer, Yvonne Kraus, Rebekkah Bingham,
Ferdinand H. Lutter, Julia Ahlfeld and Oliver Thorn-Seshold*
Full Research Paper
Open Access
Address:
Beilstein J. Org. Chem. 2020, 16, 125–134.
Department of Pharmacy, Ludwig Maximilian University of Munich,
Butenandtstraße 5-13, Munich 81377, Germany
doi:10.3762/bjoc.16.14
Received: 31 July 2019
Email:
Accepted: 13 November 2019
Published: 27 January 2020
Oliver Thorn-Seshold* - oliver.thorn-seshold@cup.lmu.de
* Corresponding author
This article is part of the thematic issue "Molecular switches".
Guest Editor: W. Szymanski
Keywords:
antimitotics; cytoskeleton; hemithioindigo; photopharmacology;
photoswitch
© 2020 Sailer et al.; licensee Beilstein-Institut.
License and terms: see end of document.
Abstract
Background: Hemithioindigo is a promising molecular photoswitch that has only recently been applied as a photoswitchable phar-
macophore for control over bioactivity in cellulo. Uniquely, in contrast to other photoswitches that have been applied to biology,
the pseudosymmetric hemithioindigo scaffold has allowed the creation of both dark-active and lit-active photopharmaceuticals for
the same binding site by a priori design. However, the potency of previous hemithioindigo photopharmaceuticals has not been
optimal for their translation to other biological models.
Results: Inspired by the structure of tubulin-inhibiting indanones, we created hemithioindigo-based indanone-like tubulin inhibi-
tors (HITubs) and optimised their cellular potency as antimitotic photopharmaceuticals. These HITubs feature reliable and robust
visible-light photoswitching and high fatigue resistance. The use of the hemithioindigo scaffold also permitted us to employ a para-
hydroxyhemistilbene motif, a structural feature which is denied to most azobenzenes due to the negligibly short lifetimes of their
metastable Z-isomers, which proved crucial to enhancing the potency and photoswitchability. The HITubs were ten times more po-
tent than previously reported hemithioindigo photopharmaceutical antimitotics in a series of cell-free and cellular assays, and
allowed robust photocontrol over tubulin polymerisation, microtubule (MT) network structure, cell cycle, and cell survival.
Conclusions: HITubs represent a powerful addition to the growing toolbox of photopharmaceutical reagents for MT cytoskeleton
research. Additionally, as the hemithioindigo scaffold allows photoswitchable bioactivity for substituent patterns inaccessible to the
majority of current photopharmaceuticals, wider adoption of the hemithioindigo scaffold may significantly expand the scope of cel-
lular and in vivo targets addressable by photopharmacology.
125

Beilstein J. Org. Chem. 2020, 16, 125–134.
Introduction
The cytoskeletal scaffolding protein tubulin, a heterodimer
consisting of α and β subunits, each of various isotypes, revers-
ibly assembles into giant non-covalent polymeric microtubules
(MTs), which play a pivotal role as a dynamic scaffold for a
multitude of cellular processes. These include mechanostasis,
the completion of mitosis, cell motility, and cargo trafficking in
all cell types, as well as cell-type-specific roles, such as polari-
zation, cargo sorting, and trafficking in neurons; the regulation
and functioning of these processes is still not satisfactorily
understood [1-4]. The MT cytoskeleton is a finely tuned com-
plex system that is highly conserved through evolution. Direct
genetic modifications of tubulin that affect its functions risk
causing a diversity of effects, due to its many survival-critical
roles, as well as non-functionality of the modified tubulin prod-
uct. For example, knockout approaches have only been de-
scribed for single isoforms of α/β-tubulin, and these cannot
deliver the dynamic reversibility and effect-specificity that is
required for understanding MT biology; and optogenetic modi-
fications of tubulin have never succeeded. Instead, studies of
the roles of MTs in these processes overwhelmingly rely on
small molecule tubulin inhibitors [1].
Due to the non-invasiveness and high spatiotemporal precision
with which optical stimulation can be applied, photopharmaco-
logy has drawn great interest for studies of crucial biological
processes in a range of fields, from neuroscience [5,6] and
G-protein-coupled receptor (GPCR) function [7,8] to antibiotic
research [9]. Particularly in the context of MT biology, photo-
pharmacology is an attractive development beyond classical
small molecule inhibitors; since the spatiotemporal complexity
inherent to the diversity of tubulin-dependent cellular processes
may finally yield to studies that can leverage high-spatiotem-
poral-specificity optical control to deliver cell-specific, time-re-
versible modulation of native cytoskeleton function.
We and others have reported on photoswitchable azobenzene-
based inhibitors of tubulin polymerisation [10-13] that have
since been used in studies of neuronal trafficking [14] and
embryonic development [15,16], and we have recently reported
biologically robust heterostilbenes that deliver green fluores-
cent protein (GFP)-orthogonal MT photocontrol [17]. However,
in both azobenzene and heterostilbene scaffolds, the steric prop-
erties of the E- and Z-isomer are so different that the protein
binding site shape determines that the Z-isomer (the lit-form) is
the more bioactive one, without the possibility of sign inver-
sion by substituent shifts. To overcome this conceptual limita-
tion, we recently reported on the first use of hemithioindigos
Figure 1: a) The potent tubulin inhibitor colchicine as a lead scaffold
led to the development of the HOTub generation of HTI-based anti-
mitotics (e.g., HOTub-31). Changing the lead scaffold to indanocine
led to the development of up to ten times more potent HITubs (e.g.,
HITub-4). b) Straightforward, short, and high-yielding synthesis of
HITub-4.
(HTIs) as photoswitchable pharmacophores for optical control pseudosymmetry of hemithioindigos can be used to enable a
of tubulin dynamics in vitro (cell-free) and MT-dependent pro- priori design of HTI-based pharmacophores for a single binding
cesses in cellulo [18]. We showed for the first time that the site, with higher bioactivity as either the lit-form E- or the dark-
126

Beilstein J. Org. Chem. 2020, 16, 125–134.
form Z-isomer, just by changing substituent patterns, and de- colchicine/combretastatin A-4. The middle methoxy group of
veloped HTI-based antimitotics with cytotoxic potencies in the colchicine’s trimethoxy-substituted “south ring” (Figure 1a)
low micromolar range (Figure 1a) [18].
makes a beneficial polar contact in the binding pocket via the
oxygen atom, but upon demethylation, the potency is much
In this work, we wished to enhance the bioactivity of the reduced, presumably from insufficient desolvation in the
distinctive dark-active HTI-based tubulin-binding antimitotics colchicine site (which is known from work on podophyllotoxin
while retaining the benefits of the HTI scaffold, namely robust, derivatives [20]). We therefore chose to keep that methoxy
fatigue-resistant, all-visible-light photoswitching.
group intact. However, colchicine’s methoxy group on the
"north ring" establishes a non-polar spacefilling interaction,
which can be replaced equipotently by an ethyl group. Thus, we
considered that the HTI scaffold could best be reduced in
Results and Discussion
Design strategy for HTIs
The HTI-based colchicinoid HOTubs (e.g., HOTub-31) that volume by "shortening" this substituent, maintaining the above-
we previously explored had the HTI photoswitch embedded mentioned non-polar interaction.
inside a methoxylation pattern, such that one isomer obeyed the
structure–activity relationship (SAR) of colchicine or its ana- Secondly, since the torsion angle of the HTI is far lower than
logue combretastatin A-4 and was bioactive, while the other that of (Z)-stilbenes or biaryl compounds, we considered that
isomer clashed with their SAR and was less active [18]. That even with shortening, their SARs might not directly match.
approach of directly embedding a photoswitch motif inside the Mainly, we assumed that re-orientation of the substituent
pharmacophore seemed to be more promising for photopharma- pattern on one or both rings (e.g., re-orientation of the
cology than the synthetically more straightforward attachment archetypal 3,4,5-trimethoxyphenyl south ring pattern to a 4,5,6-
of photoswitches on the pharmacophore periphery. We ex- trimethoxyaryl pattern) might be needed to occupy a similar
pected that embedding (which is referred to as azologization in space.
the case of azobenzene-based photopharmaceuticals [19])
should in general lead to more significant alterations of the A wealth of CDIs have been reported, including scaffolds such
binding-relevant structure, and increase the differential potency as aurones that apparently reproduce the substituent pattern
between isomers of the resulting photopharmaceutical, than SAR of combretastatins [21] while having closely similar scaf-
peripheral attachment (referred to as azo-extension in the case fold steric properties to HTIs. However, in light of the consider-
of azobenzenes [18]). We therefore desired to maintain the ations above, we rather selected indanocine as a starting point
embedding strategy, yet to improve potency we chose to break for alternative substituent patterning (Figure 1a). Indanocine is
with substitution patterns strictly based on colchicine. It a cytotoxic indanone-based CDI (EC50 ≈ 10–40 nM) [22] with
is not the case that colchicine (or any other small-molecule similar cell culture potency to colchicine (EC50 ≈ 3–20 nM)
inhibitor) represents an ideal structure that colchicine domain [23] that likewise disrupts MTs, arrests cells in the G2/M phase,
inhibitors (CDIs) should reproduce. Thus, our design focus was and induces apoptosis. Although the size and geometry of
to introduce reversible photoresponse to a CDI rather than thioindoxyl and indanone rings differ because of the S/CH2
developing compounds with high similarity to colchicine per se, replacement, we assumed that "mapping" the substitution
aiming at compounds where one isomer would be almost pattern of indanocine onto a hemithioindigo core should result
biologically inactive such that light can be used to effect in a lead structure for tubulin-binding (Z)-HTIs, namely the
a photoisomerisation-based switch-on/switch-off of bioactivity. class of HITubs.
The end-to-end distance of the HTI scaffold is significantly
longer than that in either the biaryl colchicine or the stilbene The para-hydroxy substitution of indanocine suggested that
combretastatin, and the torsion angle between the aryl blades HTI might be a more desirable photopharmaceutical scaffold
of the HTI is nearly planar (up to 4°), while that between the than the widely used azobenzene motif. While para-
rings of (Z)-combretastatin or colchicine is approximately hydroxyazobenzenes feature negligibly short cis-to-trans ther-
50–60° [20]. Thus, we assumed that the length and the near- mal relaxation half- lives in aqueous media in the range of μs
planarity of the HTI could suit it to different substituent patterns [24,25], probably making them unsuitable for robust
to those of colchicine, and attempted to rationally determine photoswitching applications against intracellular targets, data
these.
for para-hydroxy HTIs have not been reported so far. We
considered that if the para-hydroxy-HITubs featured
Firstly, since the HTI scaffold is longer than a biaryl motif but photoswitchable bioactivity in cellulo, implying suitable (E)-
should occupy a similar volume in the pocket, we assumed that HTI stability under cellular conditions, this would more gener-
we would have to reduce the substituent bulk present on ally commend them as a scaffold of choice for cellular photo-
127

Beilstein J. Org. Chem. 2020, 16, 125–134.
pharmaceutical use with strong electron-donating substituents, this SAR study are explained below in the section on
such as amino or hydroxy groups in ortho- or para-position, bioactivity.
aiming at intracellular targets. This is an important scope of
substituents to address, since these small polar groups often General synthetic access
establish high-affinity ligand–target interactions, but otherwise Synthetic routes to HTIs are well established [26] and typically
represent an obstacle to photoswitchability with azobenzene involve aldol condensation of benzaldehyde onto thioindoxyls.
compounds.
However, the key step is the formation (and where necessary,
isolation) of the thioindoxyl species. In our studies, the elec-
tron-rich dimethoxy- and trimethoxy-substituted thioindoxyls
Preview: design of target HITubs
Our SAR-driven compound development path is described in were noted to be unstable to air, base, and silica gel during
full later in the section on bioactivity, but in brief, we began the chromatography, so it was sought to minimise their exposure to
series of indanocine-inspired HTI designs by replacing the these conditions during synthesis. In the end, we used two
south ring amino unit of indanocine (which is attached in ortho- routes to the thioindoxyls: either Friedel–Crafts acylation of
position to the key south ring methoxy group) by a hydroxy α-phenylthioacetic acids (which are easily accessible from
function, giving HITub-1 (Figure 2).
thiophenols by alkylation using 2-chloroacetic acid, Figure 1b)
or else, lithium diisopropylamide (LDA)-mediated cyclisation
When HITub-1 later proved less bioactive than we had wished, of 2-(methylthio)benzamides, which were obtained by directed
we explored steric and polarity changes to this south ring ortho-metalation of the respective benzamides followed by
hydroxy group by methylation (HITub-2), methylation quenching with dimethyl disulfide [27] (Supporting Informa-
and shifting on the ring (HITub-3), or even its removal tion File 1, Scheme S1). In general, we found the LDA-medi-
(HITub-4). We additionally controlled against our design logic ated cyclisation more convenient, as it generated fewer side
of north ring substituent shortening (HITub-5). We also con- products and enabled faster, easier workup and purification. We
trolled for the SAR observation that CDIs should not tolerate a used these routes to synthesise the HITubs typically in good
non-polar central north ring substituent (HITub-6), but can (32% for HITub-4, Figure 1b) to excellent (93% for HITub-7,
support removal of this substituent altogether, with only small Scheme S5) overall yields from commercial building blocks
potency loss (HITub-7) [20]. The progression and results of (see Supporting Information File 1).
Figure 2: Chemical structures of HITubs. Key variations with respect to HITub-4 are highlighted in dashed boxes.
128

Beilstein J. Org. Chem. 2020, 16, 125–134.
Photocharacterisation
(for more detailed analysis see Supporting Information File 1).
Although some para-hydroxy-substituted HTIs have been de- In polar aprotic solvents, the HITub-4 Z-isomer (λmax ≈ 380,
scribed [28,29], we are unaware of any report of the solvent- 460 nm) showed robust, reliable, and fully reversible
and pH-dependency of their photochromism and thermal relaxa- photoswitching (λ = 450 nm for Z → E and 530 nm for E → Z
tion. Dube and co-workers have reported that in general, in- switching), with the high fatigue resistance characteristic of
creasing the electron-donating strength of groups in the HTIs. The E-isomer's thermal half-life in EtOAc or DMSO was
hemistilbene para-position of HTIs correlates to (a) a batho- ca. 40 s (Figure 3 and Supporting Information File 1, Figure
chromic shift of the S0 → S1 absorption band (up to S3). Its absorption spectra and photoswitchability were
λmax ≈ 500 nm with julolidine substitution) and (b) decreased unaltered by the addition of acid, however, addition of base led
thermal stability of the metastable E-isomer, i.e., faster thermal to a remarkable bathochromic and hyperchromic shifts of the
relaxation [30]. However, they also reported that introducing absorption band at ca. 550 nm, and no observed photoswitchab-
electron-donating groups (methoxy, dimethylamino) in ility (Supporting Information File 1, Figure S2). We assumed
para-position to the thioindoxyl sulfur atom restored that this spectrum resulted from a quinoidal structure, formed
E-stability while maintaining red-shifted absorption maxima. after deprotonation of the hydroxy group, and that the lack of
This para-position was occupied by the key methoxy group in observable photoswitchability arose due to fast free rotation
all our HITub designs. With scant information available, around the C–C single bond connecting the thioindigo and
we could not predict the thermal stability of (E)-HITubs hemistilbene motifs. Interestingly, in neutral or acidic aqueous
in cellular conditions, so we turned to experimental measure- media where the quinoidal structure is not present (λmax ca.
ment.
370, 480 nm), photoswitching could not be observed either,
which we presumed to be due to fast thermal relaxation. How-
Since we found no substantial differences between the photo- ever, noting that typical CDIs are substantially biolocalised into
chemical properties of the para-hydroxylated compounds lipid environments within cells [31], we decided to explore
(Figure 3 and Supporting Information File 1, Figure S1), we photoswitching-based cellular assays with these compounds
here describe the photocharacterisation of HITub-4 as a repre- nonetheless (further discussion in Supporting Information
sentative example of the photoswitchable bioactive compounds File 1).
Figure 3: Photocharacterisation of HITub-4. a) Photochemical and thermal isomerisation. b) UV–vis spectra after saturating illumination at λ = 450,
505, and 530 nm, respectively. c) Photoreversible switching upon alternating illumination (λ = 450 and 530 nm). d) Thermal relaxation at room temper-
ature after reaching the photostationary state at λ = 450 nm.
129

Beilstein J. Org. Chem. 2020, 16, 125–134.
We found that the photochemical properties of the non-para-
hydroxylated control HITub-6 were similar to those of previ-
ously reported non-para-hydroxylated HOTubs [18], with
satisfactory photoswitching in both DMSO and phosphate-
buffered saline (PBS)/DMSO mixtures (Figure S1f,g in Sup-
porting Information File 1).
Bioactivity: SAR study of HITubs in cellulo
To begin evaluating the isomer-dependent bioactivity of HITub
photopharmaceuticals in cellulo, we first performed resazurin
(resorufin N-oxide) antiproliferation assays under different
lighting conditions (Figure 4). Inhibitors of tubulin polymerisa-
tion act as antimitotic cytotoxins in cell culture by preventing
formation of a functional mitotic spindle, resulting in mitotic
arrest and eventually cell death. The reduction of resazurin by
viable cells serves as a fluorogenic proxy readout for anti-
mitotic potency in cellulo, since the degree of resazurin
turnover scales to the number of cells still viable after com-
pound treatment, although the mechanism behind antiprolifera-
tive activity must later be determined using more specific
assays. We used the HeLa human cervical cancer cell line to
assess the biological activity of HITubs in all cellular experi-
ments shown in this work. Since tubulin is a highly conserved
protein target critical for survival in all cell types, we expected
that, as for other colchicine domain tubulin inhibitors, trends in
potency and in photoswitchability of potency determined in this
representative mammalian cell line can be translated to other
cell types, although their specific response (e.g., EC50 values)
would need individual determination. Nocodazole was used as a
benchmarking reference and mechanistic positive control in all
Figure 4: a) Resazurin reduction assay for HITub-4 and nocodazole in
HeLa cells (n = 3), demonstrating the difference in antiproliferative po-
tencies at λ = 450 and 530 nm/dark conditions (75 ms pulsing every
15 s). b) EC50 values of HITubs from cellular antiproliferation assays
under dark conditions and at λ = 450 nm. The ratio of lit/dark EC50
values shows the fold change in photoswitchable bioactivity.
in cellulo assays since it is a potent inhibitor of the colchicine hydroxy and amino groups have similar size and polarity, and
binding site (EC50 ≈ 40 nM) with both appropriate solubility can both act as H-bond donors or acceptors. Therefore, we ex-
and straightforward handling.
pected HITub-1 to allow reliable evaluation of the indanocine
substituent pattern. (Z)-HITub-1 was already strongly bioac-
Self-made low-intensity LED arrays with relatively narrow tive (EC50 ≈ 500 nM, Figure 4b), although one order of magni-
bandwidth were used for illumination of cells during assays, tude less so than indanocine (EC50 ≈ 10–40 nM, depending on
with a pulsing regime of 75 ms every 15 s to maintain photosta- the cell line). Pleasingly, although we had not observed its
tionary state equilibria in cellulo [10]. We cross-checked differ- photoswitching in pure aqueous media, in the heterogeneous
ent illumination wavelengths in cellular toxicity assays; in cellular environment, we found that its overall toxicity under lit
accordance with the DMSO photoswitching studies, we ob- conditions was reliably and reproducibly halved.
served that 530 nm (ca. 97% Z-configuration, but additionally
controls for non-specific phototoxicity) delivered results equiv- We then explored changes to the substituent pattern to deter-
alent to dark conditions (exclusively Z-configuration), to which mine whether we could improve both Z-isomer potency in an
450 nm (lit conditions, ca. 70% E-configuration) gave the absolute sense and the overall photoswitchability of potency
greatest difference in antiproliferative potencies.
comparing E- and Z-isomers. We began by methylating the
south ring hydroxy group (HITub-2) to see how changes in size
We began our studies with HITub-1. This is a HTI analogue of and polarity affect the bioactivity; surprisingly, the potency loss
indanocine in which the indanocine amino function (in ortho- was not dramatic (indicating that this position is not a key
position to the key south ring methoxy group) has been replaced determinant of bioactivity), but the photoswitchability in-
by a synthetically more accessible hydroxy group (delivered via creased substantially (3-fold with regard to the lit/dark ratio).
demethylation of a trimethoxy precursor through BBr3). The We took this as an encouraging indicator of the overall polarity
130

Beilstein J. Org. Chem. 2020, 16, 125–134.
required for binding, and now examined re-orienting the south (Supporting Information File 1, Figure S4), especially by the
ring substituents by shifting the trimethoxy pattern on the ring potency and photoswitchability of HITub-4.
(HITub-3), which improved the potency dramatically
(EC50 ≈ 150 nM for the Z-isomer) while retaining the 3-fold These results indicated that indanocine-inspired HTI-based
photoswitchability of bioactivity.
reagents are a potent, cellularly bioactive class of photoswitch-
ably antiproliferative agents, with the most potent light-con-
Since comparison of HITub-2 and HITub-3 showed that the trolled antimitotic bioactivity reported for photoswitches de-
potency can be retained without substituents in ortho-position signed for tubulin: 10-fold enhancement compared to the prede-
to either the carbonyl group or the sulfur atom, for maximal cessor HTI generation HOTubs [18] and styrylbenzothiazole-
simplicity, we tested whether both substituents could be deleted based SBTubs [17], and 5-fold superior to azocombretastatins
simultaneously (HITub-4). This proved to be the strongest-per- [10-12]. In view of the generally limited solubility of photo-
forming compound of our studies, with the Z-isomer possessing pharmaceuticals (associated with their extended flat aromatic
an EC50 value of ca. 110 nM and a 4-fold difference of structures), this increase in potency renders the HITub com-
bioactivity between lit and dark conditions (λ = 450 nm). This pound class a promising addition to the toolbox of photoswitch-
difference was surprisingly high, given that even in aprotic able antimitotics, which might prove valuable for future in vivo
media (e.g., lipid environment reservoirs within cells), there studies.
should be ca. 30% residual Z-isomer at photoequilibrium [18],
and we had expected that any (E)-HITub entering the cytosol We also noted that the para-hydroxy HITubs featured ca. 30%
(aqueous environment) would quickly relax to its more bioac- residual (Z)-HITub at PSS λ = 450 nm in cell-free measure-
tive Z-isomer before encountering its cytosolic protein target. ments, and that for HITub-2–4, the PSS isomer mixture's cellu-
We theorised that fast cytosolic relaxation of the para-hydroxy lar cytotoxicity at that wavelength was on average 3.3-fold
HITubs to their bioactive Z-isomer may actually be a decisive lower than the cytotoxicity of the corresponding (Z)-HITub
factor in preventing the simple equilibration of the extracellular (Figure 4). This can be interpreted as indicative that the
HITub concentration (exclusively Z-configuration due to fast E-isomers are essentially biologically inactive, similar to what
relaxation, irrespective of illumination conditions) with the has been observed for heterostilbene SBTubs [17] and azoben-
cytosolic (Z)-HITub concentration available to bind to tubulin zene photostatins (PSTs) [10]. If substantiated, HTI-like ana-
(and which all experiments show is reduced under λ = 450 nm logues for which photostationary state (PSS) with enhanced
illumination). Examining this in detail is beyond the scope of proportions of E-isomer can be photogenerated would represent
this study, however, see Supporting Information File 1 for a an exciting advance: they could, in contrast to the other
discussion on isomer-dependent subcellular biolocalisation photoswitch types, allow all-visible, photoreversible, high-po-
effects.
tency switching while reproducing similarly beneficial
photoswitchability of bioactivity.
We first controlled against our design logic of north ring sub-
stituent shortening by changing the north ring apolar-contact Mechanistic assessment of HITub action
methyl groups to methoxy groups (HITub-5), and were satis- We now determined to confirm the mechanism of action of the
fied when this abolished bioactivity. We also controlled for the HITub compounds. To evaluate the biological mechanism of
result, known from extensive SAR work at the colchicine site action behind the HITubs’ photoswitchable antimitotic activity,
[18,32], that CDIs should not tolerate a non-polar central north we first checked their inhibition of polymerisation of purified
ring substituent (HITub-6) but can support removal of this sub- tubulin in a cell-free assay. The results showed almost identical
stituent altogether, with only small potency loss (HITub-7). We inhibition potency for HITub-4 at c = 10 μM as for the
considered that if the screened compounds obeyed this prin- archetypal CDI colchicine at c = 20 μM (Figure S5, Supporting
ciple, it would reinforce our mechanistic understanding of them Information File 1), which we took to indicate that (Z)-HITub-
as CDIs. Indeed, HITub-6 proved inactive until it reached its 4 exerted its bioactivity by specifically binding to tubulin
solubility limit, but (Z)-HITub-7 was relatively potent and directly in the cell-free system. This suggests that the same spe-
featured an only 4-fold reduction of bioactivity as compared to cific direct action can be reproduced in cellulo, and that effects
its hydroxylated parent (Z)-HITub-4. Interestingly, however, on auxiliary cellular systems dependent upon the MT cytoskel-
HITub-7 displayed no difference between dark (all-Z) and lit eton can likely be downstream effects of MT depolymerisation.
(mostly E) conditions, and we were unable to rationalise this We next investigated the HITubs’ isomer-dependent effects on
with reference to either polarity or structure, in light of our prior the MT network inside cells, focusing on the active analogues
work on apolar HOTubs [18] (see also Supporting Information HITub-4 and HITub-2 in comparison with inactive HITub-5
File 1). We were, however, overall satisfied by these findings as a control (Figure 5). By reducing tubulin polymerisation dy-
131

Beilstein J. Org. Chem. 2020, 16, 125–134.
Figure 5: Confocal microscopy images of immunofluorescently labelled MT networks after treatment with HITubs for 24 h under lit (λ = 450 nm) and
dark conditions. HITub-4 and HITub-2 caused light-dependent MT disruption at low concentrations (c = 1 and 5 μM, respectively) matching their cyto-
toxicities (EC50 = 1 and 5 μM, respectively). High concentrations of the structurally related negative control HITub-5, like the DMSO co-solvent
control, produced no effects under either illumination condition. MTs (anti-α-tubulin) are visualized in green while the nuclear stain 4′,6-diamidino-2-
phenylindole (DAPI) is visualized in blue. Scale bar length = 20 μm.
namics, CDI treatment should first disorganise and then depoly- File 1, Figure S7). HeLa cells were treated for 24 h with
merise the cellular MT network. We performed immunofluores- HITub-4 under λ = 530 and 450 nm irradiation, respectively,
cence staining of the MT network within cells treated with with the synthetic tubulin-binding agent nocodazole (Noc) used
HITubs and documented the resulting disruption of the physio- as a reference. As expected, HITub-4 showed highly light-de-
logical MT network integrity, and consequently also changes in pendent bioactivity with near-complete G2/M phase arrest at a
cell morphology by confocal microscopy. Cells exposed to concentration of 6 μM and λ = 530 nm irradiation (Figure 6a
HITub-4 (1 μM) in the dark or under λ = 530 nm illumination and Figure 6b), but nearly no cell cycle interference at the same
(to maintain exclusively Z-configuration) show near-complete concentration and λ = 450 nm irradiation (Figure 6c and
disruption of MT structures after 24 h, while treatment with Figure 6d).
λ = 450 nm illumination caused no significant disruption of the
MT network compared to an untreated dark control (Figure 5). Conclusion
Less potent HITub-2 also showed similar light dependency of Taken together, these results indicate that the HITubs had
its biological effects at higher concentrations (Supporting achieved their design aims, being a rationally-designed, poten-
Information File 1, Figure S6). Pleasingly, SAR control cy-enhanced set of HTI-based tubulin-inhibiting photopharma-
HITub-5 showed no impact on MT integrity at the highest ceuticals with photoswitchable bioactivity across cell biology
tested concentration under lit or dark conditions, which we took assays, allowing reliable photocontrol over tubulin polymerisa-
as a promising indication for the absence of phototoxicity or of tion, MT network structure, cell cycle, and cell survival. They
other effects non-specific to tubulin disruption.
feature mid-nanomolar potency in cellulo, the highest yet re-
ported for photopharmaceutical tubulin inhibitors, as well as
Lastly, to substantiate the causative link between the observa- satisfactory photoswitchability of potency. We expect that due
tions on MT disruption and cellular toxicity, we examined the to the HITubs’ potency of tubulin inhibition, they will prove a
impacts of HITub-4 on the cell cycle. Tubulin-binding agents powerful reagent system for biological studies on MT, espe-
whose major cellular mechanism of toxic action is the disrup- cially where dark-isomer activity (compared to the currently
tion of MT dynamics or structure should cause cell cycle arrest known, lit-active azobenzenes or styrylbenzothiazoles) is desir-
in the G2/M phase by preventing the completion of mitosis [1]. able, in particular for cell-free mechanistic studies [33]. More
We examined cell cycle repartition by quantification of cellular broadly, this work also shows that the HTI scaffold robustly
DNA content via propidium iodide (PI) incorporation, which enables the photoswitchable use of resonance-capable substitu-
was analysed by flow cytometry (Supporting Information ents that can establish high-affinity interactions (such as para-
132

Beilstein J. Org. Chem. 2020, 16, 125–134.
Figure 6: Cell cycle analysis of HITub-4-treated cells. a) and b) (Z)-HITub-4 caused significant G2/M arrest already at a concentration of 6 μM
(λ = 530 nm). c) and d) Under irradiation at λ = 450 nm, where a majority of (E)-HITub-4 was present, no mitotic arrest could be observed at compar-
able concentrations. a) and c) Representative histograms showing cell cycle arrest, with population binning as indicated. In b) and d), nocodazole
(Noc) was used as a positive control (c = 1 μM).
References
hydroxy groups), which are otherwise problematic for current
1. Peterson, J. R.; Mitchison, T. J. Chem. Biol. 2002, 9, 1275–1285.
doi:10.1016/s1074-5521(02)00284-3
photopharmaceutical scaffolds to tolerate without loss of
photoswitchability. In the broader sense, this is of interest for
cell biology and further highlights the potential of HTIs as a
pharmacophore scaffold for expanding the scope of cellular
photopharmacology.
2. Guedes-Dias, P.; Nirschl, J. J.; Abreu, N.; Tokito, M. K.; Janke, C.;
Magiera, M. M.; Holzbaur, E. L. F. Curr. Biol. 2019, 29, 268–282.e8.
doi:10.1016/j.cub.2018.11.065
3. Armstrong, M. T.; Armstrong, P. B. Exp. Cell Res. 1979, 120, 359–364.
doi:10.1016/0014-4827(79)90395-1
4. Kelliher, M. T.; Saunders, H. A.; Wildonger, J. Curr. Opin. Neurobiol.
2019, 57, 39–45. doi:10.1016/j.conb.2019.01.003
Supporting Information
5. Banghart, M.; Borges, K.; Isacoff, E.; Trauner, D.; Kramer, R. H.
Nat. Neurosci. 2004, 7, 1381–1386. doi:10.1038/nn1356
6. Caporale, N.; Kolstad, K. D.; Lee, T.; Tochitsky, I.; Dalkara, D.;
Trauner, D.; Kramer, R.; Dan, Y.; Isacoff, E. Y.; Flannery, J. G.
Mol. Ther. 2011, 19, 1212–1219. doi:10.1038/mt.2011.103
7. Riefolo, F.; Matera, C.; Garrido-Charles, A.; Gomila, A. M. J.;
Sortino, R.; Agnetta, L.; Claro, E.; Masgrau, R.; Holzgrabe, U.;
Batlle, M.; Decker, M.; Guasch, E.; Gorostiza, P. J. Am. Chem. Soc.
2019, 141, 7628–7636. doi:10.1021/jacs.9b03505
Supporting Information File 1
Full experimental protocols for chemical syntheses,
photocharacterisation, biochemistry, and cell biology,
including NMR spectra.
[https://www.beilstein-journals.org/bjoc/content/
supplementary/1860-5397-16-14-S1.pdf]
8. Agnetta, L.; Kauk, M.; Canizal, M. C. A.; Messerer, R.; Holzgrabe, U.;
Hoffmann, C.; Decker, M. Angew. Chem., Int. Ed. 2017, 56,
7282–7287. doi:10.1002/anie.201701524
9. Velema, W. A.; van der Berg, J. P.; Hansen, M. J.; Szymanski, W.;
Driessen, A. J. M.; Feringa, B. L. Nat. Chem. 2013, 5, 924–928.
doi:10.1038/nchem.1750
10.Borowiak, M.; Nahaboo, W.; Reynders, M.; Nekolla, K.; Jalinot, P.;
Hasserodt, J.; Rehberg, M.; Delattre, M.; Zahler, S.; Vollmar, A.;
Trauner, D.; Thorn-Seshold, O. Cell 2015, 162, 403–411.
doi:10.1016/j.cell.2015.06.049
11.Engdahl, A. J.; Torres, E. A.; Lock, S. E.; Engdahl, T. B.; Mertz, P. S.;
Streu, C. N. Org. Lett. 2015, 17, 4546–4549.
Acknowledgements
This research was supported by funds from the German
Research Foundation (DFG: SFB1032 Nanoagents for Spati-
otemporal Control project B09, SFB TRR 152 project P24
number 239283807, and an Emmy Noether grant) and the
Munich Centre for Nanoscience (CeNS) to O.T.-S. A.S. particu-
larly thanks Linda Pettersson for her great assistance during
photocharacterisation.
ORCID® iDs
doi:10.1021/acs.orglett.5b02262
Alexander Sailer - https://orcid.org/0000-0003-4140-9956
Rebekkah Bingham - https://orcid.org/0000-0001-5740-0353
Julia Ahlfeld - https://orcid.org/0000-0002-4879-4159
12.Sheldon, J. E.; Dcona, M. M.; Lyons, C. E.; Hackett, J. C.;
Hartman, M. C. T. Org. Biomol. Chem. 2016, 14, 40–49.
doi:10.1039/c5ob02005k
13.Müller-Deku, A.; Loy, K.; Kraus, Y.; Heise, C.; Bingham, R.; Ahlfeld, J.;
Trauner, D.; Thorn-Seshold, O. bioRxiv 2019, No. 778993.
doi:10.1101/778993
Oliver Thorn-Seshold - https://orcid.org/0000-0003-3981-651X
Preprint
14.Eguchi, K.; Taoufiq, Z.; Thorn-Seshold, O.; Trauner, D.; Hasegawa, M.;
Takahashi, T. J. Neurosci. 2017, 37, 6043–6052.
A non-peer-reviewed version of this article has been previously published
as a preprint doi:10.26434/chemrxiv.9176747
doi:10.1523/jneurosci.0179-17.2017
133

Beilstein J. Org. Chem. 2020, 16, 125–134.
15.Zenker, J.; White, M. D.; Templin, R. M.; Parton, R. G.;
Thorn-Seshold, O.; Bissiere, S.; Plachta, N. Science 2017, 357,
925–928. doi:10.1126/science.aam9335
License and Terms
This is an Open Access article under the terms of the
Creative Commons Attribution License
(https://creativecommons.org/licenses/by/4.0). Please note
that the reuse, redistribution and reproduction in particular
requires that the authors and source are credited.
16.Singh, A.; Saha, T.; Begemann, I.; Ricker, A.; Nüsse, H.;
Thorn-Seshold, O.; Klingauf, J.; Galic, M.; Matis, M. Nat. Cell Biol.
2018, 20, 1126–1133. doi:10.1038/s41556-018-0193-1
17.Gao, L.; Kraus, Y.; Wranik, M.; Weinert, T.; Pritzl, S. D.;
Meiring, J. C. M.; Bingham, R.; Olieric, N.; Akhmanova, A.;
Lohmüller, T.; Steinmetz, M. O.; Thorn-Seshold, O. bioRxiv 2019,
No. 716233. doi:10.1101/716233
18.Sailer, A.; Ermer, F.; Kraus, Y.; Lutter, F. H.; Donau, C.; Bremerich, M.;
Ahlfeld, J.; Thorn-Seshold, O. ChemBioChem 2019, 20, 1305–1314.
doi:10.1002/cbic.201800752
19.Hüll, K.; Morstein, J.; Trauner, D. Chem. Rev. 2018, 118,
10710–10747. doi:10.1021/acs.chemrev.8b00037
20.Tron, G. C.; Pirali, T.; Sorba, G.; Pagliai, F.; Busacca, S.;
Genazzani, A. A. J. Med. Chem. 2006, 49, 3033–3044.
doi:10.1021/jm0512903
The license is subject to the Beilstein Journal of Organic
Chemistry terms and conditions:
(https://www.beilstein-journals.org/bjoc)
The definitive version of this article is the electronic one
which can be found at:
doi:10.3762/bjoc.16.14
21.Lawrence, N. J.; Rennison, D.; McGown, A. T.; Hadfield, J. A.
Bioorg. Med. Chem. Lett. 2003, 13, 3759–3763.
doi:10.1016/j.bmcl.2003.07.003
22.Leoni, L. M.; Hamel, E.; Genini, D.; Shih, H.; Carrera, C. J.;
Cottam, H. B.; Carson, D. A. J. Natl. Cancer Inst. 2000, 92, 217–224.
doi:10.1093/jnci/92.3.217
23.Dose Response Curves for NSC 757. National Cancer Institute -
Developmental Therapeutics Program: Rockville, MD,
https://dtp.cancer.gov/services/nci60data/colordoseresponse/pdf/757
(accessed Nov 6, 2019).
24.Garcia-Amorós, J.; Sánchez-Ferrer, A.; Massad, W. A.; Nonell, S.;
Velasco, D. Phys. Chem. Chem. Phys. 2010, 12, 13238–13242.
doi:10.1039/c004340k
25.Dunn, N. J.; Humphries, W. H., IV; Offenbacher, A. R.; King, T. L.;
Gray, J. A. J. Phys. Chem. A 2009, 113, 13144–13151.
doi:10.1021/jp903102u
26.Wiedbrauk, S.; Dube, H. Tetrahedron Lett. 2015, 56, 4266–4274.
doi:10.1016/j.tetlet.2015.05.022
27.Mukherjee, C.; De, A. Synlett 2002, 325–327.
doi:10.1055/s-2002-19752
28.Takeo, Y.; Takahiro, S.; Takashi, T.; Kunihiro, I. Bull. Chem. Soc. Jpn.
1992, 65, 649–656. doi:10.1246/bcsj.65.649
29.Eggers, K.; Fyles, T. M.; Montoya-Pelaez, P. J. J. Org. Chem. 2001,
66, 2966–2977. doi:10.1021/jo0056848
30.Kink, F.; Collado, M. P.; Wiedbrauk, S.; Mayer, P.; Dube, H.
Chem. – Eur. J. 2017, 23, 6237–6243. doi:10.1002/chem.201700826
31.Bisby, R. H.; Botchway, S. W.; Hadfield, J. A.; McGown, A. T.;
Parker, A. W.; Scherer, K. M. Eur. J. Cancer 2012, 48, 1896–1903.
doi:10.1016/j.ejca.2011.11.025
32.Nguyen, T. L.; McGrath, C.; Hermone, A. R.; Burnett, J. C.;
Zaharevitz, D. W.; Day, B. W.; Wipf, P.; Hamel, E.; Gussio, R.
J. Med. Chem. 2005, 48, 6107–6116. doi:10.1021/jm050502t
33.Gaspari, R.; Prota, A. E.; Bargsten, K.; Cavalli, A.; Steinmetz, M. O.
Chem 2017, 2, 102–113. doi:10.1016/j.chempr.2016.12.005
134