H2S-Donating trisulfide linkers confer unexpected biological behaviour to poly(ethylene glycol)-cholesteryl conjugates

Article abstract of DOI:10.1039/c9tb02614b

Inspired by the properties of the naturally occurring H₂S donor, diallyl trisulfide (DATS, extracted from garlic), the biological behaviour of trisulfide-bearing PEG-conjugates was explored. Specifically, three conjugates comprising an mPEG tail and a cholesteryl head were investigated: conjugates bridged by a trisulfide linker (T), a disulfide linker (D) or a carbamate linker (C), and a fourth comprising two mPEG tails bridged by a trisulfide linker (P). H₂S testing using both a fluorescent chemical probe in HEK293 cells and an amperometric sensor to monitor release in suspended cells, demonstrated the ability of the trisulfide conjugates,TandP, to release H₂S in the presence of cellular thiols. Cytotoxicity and cyto-protective capacity on HEK293 cells showed thatTwas the best tolerated of the conjugates studied, and remarkably more so thanDorC. Moreover, it was noted that application ofTconferred a protective effect to the cells, effectively abolishing the toxicity associated with co-administeredC. The interaction of conjugates and combinations thereof with the cell membrane of HEK cells, as well as ROS generation were also investigated. It was found thatCcaused significant membrane perturbation, correlating with high losses in cell viability and pronounced generation of ROS, especially in the mitochondria.T, however, did not disturb the membrane and was able to mitigate the generation of ROS, especially in the mitochondria. The interplay of the cholesteryl group and H₂S donation for conferring cytoprotective effects was clearly demonstrated asPdid not display the same beneficial characteristics asT

Full text of DOI:10.1039/c9tb02614b

Journal of
Materials Chemistry B
View Article Online
View Journal
PAPER
H2S-Donating trisulfide linkers confer unexpected
biological behaviour to poly(ethylene glycol)–
cholesteryl conjugates†
Cite this:DOI: 10.1039/c9tb02614b
a
ab
Francesca Ercole,‡a Yuhuan Li,‡a Michael R. Whittaker,
Thomas P. Davis
*
ac
and John F. Quinn
*
Inspired by the properties of the naturally occurring H2S donor, diallyl trisulfide (DATS, extracted from
garlic), the biological behaviour of trisulfide-bearing PEG-conjugates was explored. Specifically, three
conjugates comprising an mPEG tail and a cholesteryl head were investigated: conjugates bridged by a
trisulfide linker (T), a disulfide linker (D) or a carbamate linker (C), and a fourth comprising two mPEG
tails bridged by a trisulfide linker (P). H2S testing using both a fluorescent chemical probe in HEK293
cells and an amperometric sensor to monitor release in suspended cells, demonstrated the ability of the
trisulfide conjugates, T and P, to release H2S in the presence of cellular thiols. Cytotoxicity and cyto-
protective capacity on HEK293 cells showed that T was the best tolerated of the conjugates studied, and
remarkably more so than D or C. Moreover, it was noted that application of T conferred a protective effect
to the cells, effectively abolishing the toxicity associated with co-administered C. The interaction of
conjugates and combinations thereof with the cell membrane of HEK cells, as well as ROS generation were
also investigated. It was found that C caused significant membrane perturbation, correlating with high losses
in cell viability and pronounced generation of ROS, especially in the mitochondria. T, however, did not
disturb the membrane and was able to mitigate the generation of ROS, especially in the mitochondria. The
interplay of the cholesteryl group and H2S donation for conferring cytoprotective effects was clearly
demonstrated as P did not display the same beneficial characteristics as T.
Received 20th November 2019,
Accepted 19th March 2020
DOI: 10.1039/c9tb02614b
rsc.li/materials-b
demonstrated the positive impacts of garlic and other alliums
in mediating risk factors for cardiovascular disease, including
Introduction
Organic polysulfides, (R1–(S)n–R2, n 4 2) exist in nature in a reducing platelet aggregation and oxidative stress.6 Some of the
variety of plants and fungi, primarily as molecules containing benefits, especially those associated with vasoactivity, are
tri- and tetrasulfide bonds.1,2 Such compounds can reveal thought to be mediated by hydrogen sulfide, H2S, an endogenous
potent antibiotic, antifungal and even anti-cancer activity, signaling molecule which elicits numerous important actions in
generating interest for medical and agricultural applications.3 the body (e.g., relaxation of vascular smooth muscle, reduction in
The health-promoting benefits provided by a diet rich in garlic blood pressure).7,8 Related impacts include gastroprotective,9,10
and other alliums have been attributed, at least in part, to cardioprotective,11 hepatoprotective12,13 and neuroprotective14,15
polysulfides such as diallyl trisulfide (DATS).4,5 Studies have effects. The protective effect of H2S in the cardiovascular system
has been demonstrated against damage induced by agents such
as doxorubicin16 and in the gastrointestinal system, by damage
a ARC Centre of Excellence in Convergent Bio-Nano Science and Technology,
induced by non-steroidal anti-inflammatory drugs.17 Specifically,
Drug Delivery, Disposition and Dynamics Theme, Monash Institute of
exogenous H2S was shown to reduce myocardial necrosis and
Pharmaceutical Sciences, Monash University, Parkville, VIC 3052, Australia.
rescue contractile activity in isoproterenol-damaged rat hearts.18 It
E-mail: john.f.quinn@monash.edu, thomas.p.davis@monash.edu
b Australian Institute for Bioengineering and Nanotechnology,
The University of Queensland, St Lucia, Queensland 4072, Australia
c Department of Chemical Engineering, Faculty of Engineering, Monash University,
Wellington Road, Clayton, Victoria 3800, Australia
was also found to amend hyper-homocysteine induced myocardial
tissue damage by inhibiting ER stress-associated apoptosis in
rats,19 and by eliciting a cardioprotective effect against H2O2-
induced oxidative stress in an MI mouse model.20 Synthetic
small molecule donors have been demonstrated to protect
primary H9c2 cardiomyocyte cells from hypoxia-induced
injury21 and cardiomyoblast apoptosis.20
† Electronic supplementary information (ESI) available: Full synthetic details of
P, 1H NMR spectra with spectral assignments for T, D, C and P; full descriptions
of all methods. See DOI: 10.1039/c9tb02614b
‡ These authors contributed equally to this work.
J. Mater. Chem. B
This journal is © The Royal Society of Chemistry 2020

View Article Online
Paper
Journal of Materials Chemistry B
carbamate (C) were synthesized as previously described,38
1
(ESI,† Schemes S1–S3 and H NMR Fig. S1–S3). MeO-PEG-SSS-
OMe (P) was synthesised using a different method for the
generation of the trisulfide bridge39 (ESI† synthetic protocol,
Scheme S4 and 1H NMR Fig. S4). The fluorescent probe for H2S
detection in cells, SF4, was synthesized as described.40
Scheme 1 Thiol exchange reaction of DATS (diallyl trisulfide) with glu-
tathione (GSH) to generate H2S via persulfide intermediates.
The possibility of applying H2S therapeutically has motivated the
MeO-PEG-NH2 MW 2000 g molꢀ1 (Linear Monofunctional
development of numerous small molecule22,23 and high molecular PEG Amine NH2) was purchased from Creative PEGWorks.
weight H2S donors.24–26 In particular, polymeric H2S donors offer Sodium sulfide (Na2S) anhydrous was purchased from Alfa Aesar
certain advantages, such as increased aqueous solubility, modulated and hydrochloric acid 32% was purchased from Ajax FineChem.
release kinetics and the potential to tailor the pharmacokinetics of Solvents (except anhydrous solvents) were purchased from
the agent.27,28
Merck Millipore and used as received. All other chemicals, such
As shown in Scheme 1, H2S is generated from the trisulfide DATS as reagents and anhydrous solvents for synthesis, were pur-
initially as a result of nucleophilic attack of the sulfur–sulfur bond by chased from Sigma-Aldrich at the highest purity available and
glutathione (GSH), to produce GSS–allyl and allyl persulfide. The used without further purification (unless otherwise stated).
persulfide (RSSH) then reacts further with GSH to produce H2S. A Reveleris Flash Chromatography System fitted with silica gel
Alternatively, the exchange reaction results in another mixed trisul- (230–400 mesh) cartridges was used for purification of monomer
fide, GSSS–allyl, which releases H2S in a similar fashion, i.e., via an and intermediates. TLC was performed on Merck Silica 60F254
intermediate persulfide.5,29,30 Notably, in a closed system the reac- plates. PBS 7.4 (phosphate buffered saline, pH 7.4) was recon-
tion between H2S and an oxidized thiol, RSSR, can also occur, to stituted from powder (Sigma-Aldrich) which, when dissolved in
generate a persulfide31 in equilibrium with all the species, H2S, RSH, 1 L of deionized water, yielding 0.01 M phosphate buffered
RSSR, RSSH and RSSSR.32 Persulfide generation on critical protein saline (NaCl 0.138 M, KCl 0.0027 M), pH 7.4 at 25 1C.
cysteines constitutes a post-translational modification and a mecha-
nism by which H2S is thought to signal.33,34 The emerging role of
Characterization and methods
1H (400 MHz) and 13C NMR (100 MHz) spectroscopy. Spectra
were obtained with a Bruker UltraShield 400 MHz spectrometer
persulfide species as possible mediators of H2S signaling and their
biochemistry is of increasing interest and scrutiny.35–37
Our interest is in nanomaterials incorporating polysulfide
bonds, with an emphasis on their ability to generate H2S and
biological benefits thereof. In a previous communication38 we
described the synthesis and physical properties of a polymeric
conjugate incorporating a trisulfide linkage, bridging hydrophilic
and hydrophobic sections. The mPEG-SSS-cholesteryl conjugate
known as T, was shown to be capable of self-assembling into
micelles and to biodegrade in a cellular environment to concurrently
deliver H2S. In this work we delve further by focusing our attention
on probing the activity of T in a biological environment. Namely, we
carry out cellular in vitro assays investigating cytotoxicity, ROS
mitigation, membrane effects and H2S donation. As previously, we
apply the controls, D, a disulfide version which is also capable of
exchanging with thiols (but without concomitant release of H2S) and
a non-cleavable, non H2S-donating version, C. In this study we
include another control, P, an H2S-donating, water-soluble version,
incorporating a trisulfide bridge yet no lipid tail. The comprehensive
study presented here builds a more complete picture regarding
activity of T, beyond physical properties which were probed in the
initial publication. Importantly, the biological tests conducted here
give a better understanding of designed trisulfide nanomaterials
which is a necessary step towards realizing their potential to be
applied in a pharmaceutical setting.
at 25 1C running Bruker Topspin Software. Spectra were
recorded for samples dissolved in deuterated solvent and
chemical shifts are reported as parts per million from external
tetramethylsilane.
CMC determination. 990 mL of 13 different polymer solu-
tions of decreasing concentration (1.0 to 0.000001 mg mLꢀ1
)
were added individually to 13 vials. These are prepared by
diluting a 5.0 mg mLꢀ1 polymer stock solution with MilliQ
water. A pyrene stock solution (c = 5 ꢁ 10ꢀ5 M) was prepared in
THF and 10 mL aliquots were added to the 13 vials containing
the polymer solutions. The final pyrene concentration in each
vial was 5 ꢁ 10ꢀ7 M. Vials were shaken gently for two hours at
room temperature, shielded from the light with aluminium
foil. The fluorescence spectra were recorded using an excitation
wavelength of 336 nm and emission spectra were recorded
ranging from 350 to 450 nm. From the pyrene emission spectra,
the intensities (peak height) of the 393 nm peak (I1) and 384 nm
peak (I3) were used to calculate the I3/I1 ratio.
A CMC value was determined from plots of log C vs. pyrene
I3/I1 ratio by fitting the curves to a Boltzmann sigmoid type
equation41 given by:
A1 ꢀ A2
y ¼
þ A2
1 þ eðxꢀx Þ=Dx
0
where y corresponds to the pyrene I3/I1 ratio value; the inde-
pendent variable (x) is the concentration of polymer; A1 and A2
are the upper and lower limits of the sigmoid respectively; x0 is
Experimental
Materials
Cholesteryl (((2-methoxyPEGamino)carbonyloxyethyl)trisulfan-yl- the centre of the sigmoid and Dx is directly related to the
ethyl)carbamate (T), cholesteryl (((2-methoxyPEGamino)carbonyl- independent variable (x) range where there is an abrupt change
oxyethyl)disulfanylethyl)carbamate (D), cholesteryl-PEG-OMe of the dependent (y) variable. A CMC value (CMC1) was taken to be,
J. Mater. Chem. B
This journal is © The Royal Society of Chemistry 2020

View Article Online
Journal of Materials Chemistry B
Paper
x0 the centre of the sigmoid, as determined using KaleidaGraph samples the media was removed, and cells were incubated with
curve fitting software for the series of data points. 10% v/v AlamarBluet Cell Viability Reagent (ThermoFisher)
Glutathione-mediated H2S release from the conjugates T in cell culture media containing 1% (v/v) antibiotic, for 4 h
and P in cell media as measured using the amperometric at 37 1C and 5% CO2. A microplate reader (CLARIOstar, BMG
sensor. Polymeric solutions of P and T were prepared in Milli LABTECH) was used to read the fluorescence at 560 nm excitation
Q water to a final polymer concentration of 7.2 mg mLꢀ1 and and 590 nm emission. Background values (10% AlamarBlue in
5 mg mLꢀ1 respectively. The sensor tip was immersed into a cell culture medium with 1% (v/v) antibiotic, with no cells) were
vial containing cell culture media only (Dulbecco’s Modified subtracted from each well and the average fluorescence intensity
Eagle Media, DMEM, high glucose, Sigma-Aldrich, supple- of the triplicates was calculated.
mented with 10% (v/v) Fetal Bovine Serum, FBS) (500 mL) for
Cell based high-content SF4 time-course study. Solutions of
the first 10 minutes, then P (50 mL of 7.2 mg mLꢀ1 stock polymers were made to a final concentration of 1 mg mLꢀ1 for T
solution, 8.2 ꢁ 10ꢀ8 moles) or T (50 mL aliquot of 5 mg mLꢀ1 and 1.4 mg mLꢀ1 for P. The solutions were then diluted 1:10 in
stock solution, 8.2 ꢁ 10ꢀ8 moles) is added. As with the calibra- PBS 7.4 prior to addition to the cells (final concentration
tion, in order to minimize exposure to air and loss of H2S to the 0.1 mg mLꢀ1 for T and 0.14 mg mLꢀ1 for P). The solutions of
environment, Parafilm was used to cover the mixture during SF4 were prepared fresh in DMF/DMSO (5 mM) and diluted to
the measurements. During the measurements the mixture was give a 5 mM solution in Hank’s Balanced Salt Solution (HBSS).
stirred with a magnetic stirrer bar. H2S release was monitored
HEK293 cells (1.5 ꢁ 104 cells per well) were seeded in black,
for a period of 10–12 minutes. Glutathione (GSH) was then optically clear 96-well plates and grown to 90% confluency.
injected into the stirring solution using a micro-syringe (20 mL Cells were washed with HBSS, then loaded with 5 mM SF4 probe
of a 250 mM solution in water, 5.0 ꢁ 10ꢀ7 moles) and then H2S (in 90 mL HBSS) for 10 minutes at 37 1C. Fluorescence imaging
release was monitored for another period of 20–25 minutes. For was performed using a high-content PerkinElmer Operetta
each release curve over time, the peaking [H2S] and the corres- with an Olympus LUCPlanFLN 20ꢁ (NA 0.45) objective. SF4
ponding peaking [sulfides]tot was calculated and displayed, as fluorescence was visualized using the EGFP filter set (excitation
derived from the calibration plot (refer to ESI† for calculations, 460–490 nm, emission 500–550 nm). Images were taken every
as well as previous work42).
4 minutes. After baseline fluorescence was determined for
Cell culture. HEK 293 (Human embryonic kidney cells, 8 minutes, the polymeric solution (10 mL) was added into wells,
ATCC) were grown in DMEM supplemented with 10% FBS. then images were taken every 4 minutes for 1 hour at 37 1C.
Cells were cultured at 37 1C in a humidified incubator with After that, images were taken every 1 hour for 23 hours at 37 1C.
5% atmospheric CO2.
Data were automatically analyzed by determining the mean SF4
Cell-mediated H2S release from the conjugates, T and P, as fluorescence per well using Harmony High Content Imaging
measured using the amperometric sensor. HEK293 (cultured as and Analysis software (v3.5.2). SF4 fluorescence was PBS vehicle
described above and maintained in T75 flasks) were detached subtracted and expressed relative to the baseline fluorescence
with trypsin, isolated as a plug, re-dispersed in complete media for each experimental condition. Data are expressed as the
and then counted. Two individual groups of equal volume and mean ꢂ SD from three independent experiments.
cell concentration were dispensed in separate vials. The sensor
ROS detection
was immersed in each cell population, a signal was measured
(for approx. 2 minutes) and then a T or P aliquot was added: HEK293 cells (1.5 ꢁ 104 cells per well) were exposed to
50 mL, 5 mg mLꢀ1 stock solution, 8.2 ꢁ 10ꢀ8 moles for T, and materials for 1, 2 and 4 hours in 96-well plates. Cell culture
50 mL, 7.2 mg mLꢀ1 stock solution, 8.2 ꢁ 10ꢀ8 moles for P. For medium was used as a control. The CellROXt Deep Red
each release curve over time, the peaking [H2S] and the corres- Reagent was added to the cells at a final concentration of
ponding peaking [sulfides]tot was calculated and displayed, as 10 mM followed by incubation for 40 minutes at 37 1C. The
derived from the calibration plot (refer to ESI† for calculations, medium was then removed, and the cells were washed three
as well as previous work42).
times with PBS. Fluorescence imaging was performed using
Alamar blue assay. All assay plates were coated with poly-L- a high-content PerkinElmer Operetta with an Olympus LUC-
lysine solution (mol wt 150 000–300 000, 0.01%, sterile-filtered, PlanFLN 20ꢁ (NA 0.45) objective. The fluorescence was visua-
BioReagent). HEK293 cells (cultured as described above) were lized using the mCherry filter set (excitation 640 nm, emission
seeded in a 96-well plate (1.5 ꢁ 104 cells per well) in the cell 665 nm). Background values (PBS only) were subtracted from
medium overnight and subsequently incubated with serial each well and the average fluorescent intensity of the triplicates
dilutions of polymer samples (1000, 500, 250, 125, 63 and was calculated.
31 mg mLꢀ1 for T, C and D) and (1440, 720, 360, 180, 90 and
45 mg mLꢀ1 for P) made up in cell culture medium containing
MitoSOX (mitochondrial superoxide indicator) test
1% (v/v) antibiotic (penicillin–streptomycin solution), for 24 h HEK293 cells (4.0 ꢁ 104 cells per well) were seeded in 8 well
at 37 1C and with 5% CO2. Cell culture medium containing chamber and cultured overnight. Cells were washed with HBSS,
1% (v/v) antibiotic was used as a control. Each polymer was then loaded with 2.5 mM MitoSOXt Red Mitochondrial Super-
tested in triplicate at each concentration to obtain representa- oxide Indicator, in 300 mL HBSS for 30 minutes at 37 1C and the
tive cell viability values. After exposure of cells to polymer plates were kept protected from light. Cells were washed gently
J. Mater. Chem. B
This journal is © The Royal Society of Chemistry 2020

View Article Online
Paper
Journal of Materials Chemistry B
three times with warm HBSS buffer, then stained with Hoechst
33342 stain for 1 minute, followed by imaging.
PI staining experiment
HEK293 cells (3.0 ꢁ 104 cells per well) were seeded in 8 well
chamber and cultured overnight. Cells were washed with HBSS,
then loaded with C, D, T and P combinations (T 500 mg mLꢀ1
;
C 31.25 mg mLꢀ1; D 125 mg mLꢀ1; P 720 mg mLꢀ1) for 24 hours at
37 1C. After that, PI was added into the wells at a final
concentration of 1 mM for 30 minutes, and then imaging was
started.
Ratiometric imaging
HEK293 cells (1.5 ꢁ 104 cells per well) were seeded in 8 well
chamber coated with poly-D-lysine and cultured overnight. Prior to
imaging, old media was removed and replaced with serum-free
Opti-MEM (Life Technologies), with cells washed in between
using Opti-MEM. Laurdan dye was added to each well to a final
concentration of 5 mM and allowed to equilibrate with the cell
membranes for at least 30 minutes. Cells were transferred to a
Leica SP8 inverted confocal fluorescence microscope housed in a
humidified, 37 1C, 5% CO2 environment. Appropriate areas of
each well, containing at least 10 viable cells, were identified. C, D,
Fig. 1 OMe-PEG–(S)n–CHOL and OMe-PEG-(S)3-PEG-MeO conjugates
investigated in the study (n = 0, 2 or 3).
A commonality between each is a hydrophilic PEG tail and a
hydrophobic cholesteryl moiety. However, C does not contain a
degradable bridge between both sections. The water-soluble
construct P, which incorporates a trisulfide bridge yet no
cholesteryl tail, was also synthesized as part of this study to
elucidate any effects of the cholesteryl tail. The reaction of a
trisulfide bridge with intracellular thiols not only imparts
cleavability to the synthetic conjugate, but also enables the
delivery of H2S, both of which could be of interest in a
pharmaceutical setting.
In our previous work, we demonstrated, using Nile Red as
both the cargo and probe, that micelles composed of T could be
disassembled using the thiol, L-cysteine, as a result of cleavage
of the trisulfide bridge. We also demonstrated that this was
concurrent with release of H2S. Herein, we demonstrate that in
a biological setting the trisulfide carriers are also able to
undergo thiol-stimulated cleavage to deliver H2S. We applied
the fluorescent probe for H2S detection, SF4,40 which under-
goes reduction of its azides into amines by H2S, to ascertain
that both T and P are capable of donation in a cellular
environment. The fluorescence was measured over time in
HEK293 cells which had been incubated with either T or P
along with SF4 probe, without the addition of exogenous thiols.
As can be seen in Fig. 2 (showing the average fluorescence
result) both trisulfide donors caused a change in fluorescence
over time, demonstrating their ability to donate H2S to live
cells via reaction with endogenous thiols. The representative
fluorescent images are displayed in ESI† (Fig. S5).
T and P (T 500 mg mLꢀ1; C 31.25 mg mLꢀ1; D 125 mg mLꢀ1
;
P 720 mg mLꢀ1) and combinations thereof were added to respec-
tive wells. Imaging was undertaken at 0 minutes (i.e. before
adding the material) and at 20 minutes timepoints with a
40ꢁ oil objective. Laurdan dye integrated into cell membranes
was excited along the 405 nm laser line and emission read at
430–470 nm (representing the lipid membrane at the gel/liquid
ordered phase) or 480–550 nm (representing the lipid membrane
at the liquid disordered phase). To calibrate dye background
levels, a well containing dye only was excited on the 405 nm laser
line using 0.5 and 2ꢁ laser power. Images were false-coloured and
any adjustments made using LAS X software (Leica).
Determination of generalised polarization (GP)
The acquisition of GP images was performed using the ImageJ
(National Institute of Health) software and custom-written
macro by Owen et al.43 with some modifications. GP values
were then calculated for each pixel of a cell membrane accord-
ing to the following equation:
I400ꢀ460 ꢀ I470ꢀ530
GP ¼
I400ꢀ460 þ I470ꢀ530
where I represents the intensity of pixels in the areas of interest in
the image acquired in the ordered (430–470 nm) and disordered
(480–550 nm) spectral channels. The GP shift was observed by
subtraction of the GP distribution peak maximum of each sample
with 20 minutes of incubation time from the GP value of the image
taken at the beginning of the experiment (at 0 minutes).
The H2S-donating ability of T and P was further demon-
strated in a cellular environment using amperometry. In this
method, an amperometric microsensor (Unisense) was immersed
in cell media (DMEM with 10% FBS) containing cultured HEK293
cells of known concentration. The cells were previously detached
and re-suspended in fresh media prior to measurement. During the
measurement the cells were gently stirred in the media to reduce
Results and discussion
The polymeric constructs, T, D and C, shown in Fig. 1, were sedimentation. Given that cells may have become compromised
synthesized as previously described (ESI,† Schemes S1–S3).38 during the test, it was stopped at 60 minutes with further testing
J. Mater. Chem. B
This journal is © The Royal Society of Chemistry 2020

View Article Online
Journal of Materials Chemistry B
Paper
Fig. 2 Fluorescence intensity of SF4 probe over time in HEK293 cells on
addition of MeOPEG-SSS-CHOL (T) and MeOPEG-SSS-PEGOMe (P). SF4
fluorescence is PBS-vehicle subtracted.
Fig. 3 Representative H2S release curve for P (red) and T (black) in cell
culture medium (Dulbecco’s Modified Eagle Media, DMEM) with (10% Fetal
Bovine Serum, FBS), as measured using amperometric microsensor
(Unisense). Note: the first 10 minutes is measured in cell culture media only,
then P (50 mL of 7.2 mg mLꢀ1 solution in water added, 8.2 ꢁ 10ꢀ8 moles) or
T (50 mL aliquot of 5 mg mLꢀ1 solution in water is added, 8.2 ꢁ 10ꢀ8 moles) is
added. Glutathione (GSH) is then added where indicated (20 mL of a
250 mM solution in water, 5.0 ꢁ 10ꢀ7 moles). The values for peaking [H2S]
and peaking [sulphides]tot shown are derived from calibration plot.
unlikely to be meaningful. ESI,† Fig. S6 displays the individual
release profiles of T and P in three cell populations, with graphs in
the top row (a1–c1) representing release from cells exposed to T and
graphs in the bottom row (a2–c2) representing release profiles from
cells exposed to P. In this test the extracellular H2S was detected,
however we cannot determine whether this involves trisulfides
being taken up by the cells, to react with intracellular thiols, or
whether thiols existing in the extracellular space are reacting with
the trisulfides to generate H2S. Notably, however, no exogenous doses of C (ESI,† Fig. S8) which showed a sharp decrease in cell
thiols were added indicating that both T and P are capable of viability occurs at concentrations 415 mg mLꢀ1. It is also
donating H2S to cells via reaction with endogenous thiols. This is in important to understand the influence of chemical structure
line with our previous cellular tests using SF4. Furthermore, since on toxic or cell-protective effects. For this reason, we also tested
H2S was detected only after the addition of both T and P conjugates the impact on cell viability of the water-soluble construct P which
to cells and media, this means that the sensor was not simply incorporates a trisulfide bridge, yet no cholesteryl tail (Fig. 4d).
detecting endogenous H2S production by the cells. Therefore, it was Since P is still capable of releasing H2S, any associated influence
established qualitatively using the amperometric method, that both of the cholesteryl tail on viability may also become evident. Due
trisulfide conjugates, T and P, can generate H2S via their reaction to having two conjugated PEG tails, the molecular weight of
with cellular thiols.
construct P (4400 g molꢀ1) is higher than C, D and T and
We also explored the reactivity of the trisulfides with cell
culture medium only, as well as with glutathione (GSH), a
biologically relevant thiol which exists in the intracellular
environment. Amperometry was again used to measure the
release of H2S, as shown in Fig. 3, which is a representative
curve (refer to Fig. S7, ESI,† for subsequent run). In cell culture
media alone, there was no release when both P and T were
added, however when GSH was added (B6 equivalents vs. T/P),
H2S was generated immediately. At face value it appears that
the release from P is higher compared to T, however more tests
would be required in order to draw any quantitative conclusions
regarding release capacity.
To demonstrate the suitability of the trisulfide conjugates
for biomedical application, we tested the dose-dependent
effects on cell viability. All constructs were evaluated against
the HEK 293 (human embryonic kidney) cell line using Alamar
Blue assay. Cell viability was tested after 24 hours of exposure of cells
to each conjugate. Results, displayed in Fig. 4a–c, show a striking
disparity in viability results between T, D and C, with T showing
negligible cytotoxicity at concentrations up to 1000 mg mLꢀ1
(0.33 mM) and C showing the highest dose-dependent toxicity
profile. The cell viability assay was also conducted using lower
Fig. 4 Cytotoxicity evaluation of (a) MeO-PEG-CHOL, C (b) MeO-PEG-
SS-CHOL, D (c) MeO-PEG-SSS-CHOL, T and (d) MeO-PEG-SSS-PEG-
OMe, P conjugates against HEK-293 cells, as determined by Alamar Blue
assay after 24 hours of exposure of cells to conjugates.
J. Mater. Chem. B
This journal is © The Royal Society of Chemistry 2020

View Article Online
Paper
Journal of Materials Chemistry B
Fig. 5 (a) Protective effect of T on C-induced cytotoxicity in HEK293 cells. Cell were exposed to increasing concentrations of T along with two different
toxic doses of C (125 mg mLꢀ1 and 62.5 mg mLꢀ1) for 24 hours and subjected to Alamar Blue cell viability assay. Control group is without T and C and ‘‘0’’
point refers to C only (without T); (b) protective effect of T on D-induced cytotoxicity in HEK293 cells. Cells were exposed to increasing concentrations of
T along with two different toxic doses of D (250 mg mLꢀ1 and 500 mg mLꢀ1) for 24 hours and subjected to Alamar Blue cell viability assay. Control group is
without T and D and ‘‘0’’ point refers to D only (without T); (c) protective effect of P on C and D-induced cytotoxicity in HEK293 cells. Cells were exposed
to increasing concentrations of P along with two different toxic doses of C (62.5 mg mLꢀ1) and D (250 mg mLꢀ1) for 24 h and subjected to Alamar Blue cell
viability assay. Control group is without P and C and D and ‘‘0’’ point refers to C or D only (without P).
therefore the concentrations were tested up to 1440 mg mLꢀ1 group likely has an important role in the cyto-protection, and
(0.33 mM). P was found to be well tolerated by HEK cells, that H2S is not solely responsible for this effect.
although marginally less so than T. These results therefore
Further experiments, aimed at elucidating the mechanism
suggest some interplay between toxicity and the cholesteryl tail. by which T protects HEK cells from C/D-induced toxicity, led us
An investigation into the cyto-protective capacity of the to investigate their effects on membrane integrity. Propidium
H2S-donors, T and P, was also conducted. This was assessed by iodide (PI) is a useful dye for such purposes since it binds to
undertaking a cell rescue test using an Alamar Blue cell viability double-stranded DNA but cannot cross an intact plasma
assay. In this investigation the T or P-dose dependent effect membrane. Therefore, it will only be present in the DNA of
on viability was screened when HEK cells were co-incubated cells which have had their plasma membranes compromised
(for 24 hours) with an otherwise toxic dose of C (either 62.5 or and permeabilized. In this study HEK293 cells were loaded with
125 mg mLꢀ1) and D (either 250 mg mLꢀ1 or 500 mg mLꢀ1). As C, D, T and P combinations for 24 hours, after which time PI
shown in Fig. 5a, on increasing the T concentration from 31.25 to was added for 30 minutes before imaging (Fig. 6). The doses of
1000 mg mLꢀ1 for an exposure time of 24 hours, the protective C (31.25 mg mLꢀ1) and D (125 mg mLꢀ1) chosen for the
effect of T against C-induced cytotoxicity was found to increase in combinations with T and P were lower than the previous
a dose-dependent manner. The same effect of T was demon- experiment to maintain a higher population of cells for
strated on D-induced cytotoxicity (Fig. 5b). However, for P, a imaging.
protective effect was only demonstrated against D, with C-induced
cytotoxicity not able to be diminished across all doses of P (Fig. 5c).
The minimum concentration of T or P required to affect an
improvement in viability to 50%, vs. the ‘0’ control group (i.e.,
with no T or P present), was estimated for each group as
displayed in Table 1. As expected, the higher the dose of C or
D that the cells were exposed to, the higher the concentration of
T or P that was required to rescue the cells. It is also notable
that T is an effective cell protector against C, however P is
completely ineffective. This result indicates that the cholesteryl
Table 1 Protective effect of T and P on cell viability in the presence of C
and D, as assessed by Alamar Blue assay
Concentration of T/P at which
Co-incubation groups
T + C (62.5 mg mLꢀ1
50% of cells remain viable (mg mLꢀ1
)
)
250
640
200
890
—
T + C (125 mg mLꢀ1
)
)
T + D (250 mg mLꢀ1
T + D (500 mg mLꢀ1
Fig. 6 Propidium Iodide (PI) detection in HEK293 cells exposed to com-
binations of: T (500 mg mLꢀ1) and C (31.25 mg mLꢀ1); T (500 mg mLꢀ1) and D
)
)
(125 mg mLꢀ1); P (720 mg mLꢀ1) and C (31.25 mg mLꢀ1); and P (720 mg mLꢀ1
and D (125 mg mLꢀ1). Merged Confocal Microscope images.
)
P + C (62.5 mg mLꢀ1
P + D (250 mg mLꢀ1
)
590
J. Mater. Chem. B
This journal is © The Royal Society of Chemistry 2020

View Article Online
Journal of Materials Chemistry B
Paper
Fig. 6 shows a moderate infiltration of PI in the D-only
images and a small amount into the C-only sample. Both D-only
and C-only images have a high number of round cells that
have departed from their normal cell morphology. However,
comparing these results to the cell viability results (Fig. 4), it is
reasonable to conclude that necrosis by loss of membrane
integrity does not fully account for toxicity effects, especially
for C. Both T-only and P-only images show no evidence of
infiltration, and preservation of normal cell morphology, in line
with their cell viability results which showed high viability of
cells maintained even at high doses. The P–C combination
image in Fig. 6 shows moderate infiltration of PI and a high
incidence of round cells, in contrast to the P–D combination
image, in which cells are uncompromised. These distinct
results are in line with the previous cell rescue tests (Fig. 5c)
which showed that P could rescue D-exposed cells but was
unable to rescue cells exposed to C.
Fig. 7 Plots of pyrene 1 : 3 ratio versus concentration of C, T and D in Milli
Q water, experimental data were fitted by a decreasing sigmoid of the
Boltzmann type to derive CMC1, taken to be the critical micelle
concentration.
This difference likely arises from the degradability of P and
the interaction of the degradation products with D. As mentioned
previously,32,35–37 an important reaction studied in literature,
partly due to its significance to post-translational modification self-assemble into micelles in aqueous solution. The determina-
of proteins, is the reaction of H2S with oxidized thiol derivatives to tion of the CMC for each conjugate was carried out by the pyrene
generate a persulfide (R-SSH), as in:
1 : 3 ratio method.41 Briefly, a plot of the pyrene fluorescence
intensity ratio (I1/I3, measured at the wavelengths corresponding
to the bands located near 373 and 384 nm) as a function of the
total surfactant concentration, displays around the CMC, a
H2S + RSSR $ RSSH + RSH
As such, there is the potential for H2S generated from P to sigmoidal decrease. Below the CMC, the pyrene 1 : 3 ratio value
react with the disulfide bridge of D, resulting in a loss of its corresponds to a polar environment, and as the surfactant
amphiphilic structure. This could result in a diminished or less concentration increases, the pyrene 1 : 3 ratio decreases rapidly
harmful interaction with the cell membrane. With a range to indicate that the pyrene is sensing a more hydrophobic
of products generated from the GSH-mediated cleavage of environment. Above the CMC, the pyrene 1 : 3 ratio reaches a
trisulfides, P and T, such as, H2S, RSH, RSSR, RSSH and RSSSR roughly constant value because of the incorporation of the probe
(ESI,† Fig. S9), the ongoing reaction of H2S with D could into the hydrophobic region of the micelles.
produce persulfides such as CHOL-SSH, which could be pertinent
As displayed in Fig. 7 (and ESI,† Table S1), the CMC value of
mediators involved in protection. Importantly, CHOL-SSH would conjugate C (B13 mg mLꢀ1) was found to be well below that of T
only be produced in combinations where cyto-protection is and D, and also to be closest in value to the concentration range
observed, but this would not be the case for the P–C combination. at which it displays toxicity (415 mg mLꢀ1). This is in line with
Cell membrane destabilization and/or destruction are events studies conducted on HeLa cells using non-ionic surfactants in
that could take place when a surfactant comes into contact with which they were found to be more toxic above or close to their
cells and these events can be influenced by their self-assembly CMC values.45 For D and T, which carry either two or three
behaviour.44 The degradability of the conjugate (which would successive sulfur atoms in their structures, the CMC values
be the case for T and D) and loss of amphiphilicity in a cellular were found to be significantly higher (67 and 90 mg mLꢀ1). This
environment could also be associated factors to consider with would be expected, as sulfur is much larger than carbon and
respect to plasma membrane interactions and toxic effects. oxygen and multiple sulfurs are likely to affect the packing-
Indeed, studies on amphiphilic, non-ionic surfactants such as ability and stability of the self-assembled system. However,
Triton and Tween series have shown that surfactants can since D and T degrade to lose their surfactant structure, the
integrate-with and even solubilise lipid membrane compo- interplay of their degraded structures with the membrane are
nents, depending on concentration and surfactant structure. also important considerations. The degradability of D in a
Specifically, the toxicity of non-ionic surfactants to HeLa cells cellular environment may explain the higher cell viability
were found to increase with increasing chain length of the results compared to C. However, it is notable that its measured
hydrophobic tail. Additionally, toxic doses were found to exist CMC value (67 mg mLꢀ1) was also found to occur relatively close
above the critical micelle concentration (CMC) of the surfac- to its toxic concentration range (462.5 mg mLꢀ1).
tants evaluated in these studies.46,47 This would suggest that
It has been demonstrated in literature that non-ionic
a strong tendency to self-assemble, may also increase the surfactants can induce apoptosis by way of insertion into the
likelihood of a surfactant to destabilize the cell membrane. cell membrane, and that this process can occur at low surfactant
We therefore determined the CMC of the conjugates, which concentrations. Membrane insertion can trigger an apoptotic
J. Mater. Chem. B
This journal is © The Royal Society of Chemistry 2020

View Article Online
Paper
Journal of Materials Chemistry B
signal inside the cell by way of transduction. At higher concentra- GP, (see ESI† for methodology), where a negative GP shift indicates
tions the disruption of the cell membrane can occur via decreased membrane lipid order, i.e., pertubation.51
mixed micelle formation between surfactant and membrane
phospholipids, leading to cell necrosis (known as a detergent change in control cells (no conjugate added), nor when cells
effect).47
were incubated with T. The largest GP shift occurred with C
As can be seen in Fig. 8, membrane lipid order did not
Membrane damage via a detergent effect, with subsequent (GP shift = ꢀ0.12), followed by P (ꢀ0.059) and D (ꢀ0.049).
permeabilization does not account for the toxic effect of C since Notably, at 20 minutes, the morphology of the cells exposed to
this was not demonstrated by the PI test. However, a physical C showed an unhealthy, rounded appearance (Fig. 9, left
association between the cholesteryl conjugate and the cell panels), as was observed for D-exposed and P-exposed cells.
membrane, such as membrane insertion, is conceivable as has When C was combined with T, the GP shift reduced signifi-
been demonstrated for other cholesteryl-based self-assemblies.48 cantly (from ꢀ0.12 to ꢀ0.012), as was seen, although to a lesser
Such events could potentially activate a pathway that co-elicits a extent, for the D and T combination (Fig. 9). These results
cytotoxic effect.49 We therefore applied the lipophilic dye, indicate that membrane perturbation may be involved with
6-dodecanoyl-2-dimethylaminonaphthalene (Laurdan), to probe toxic effects and subsequent cell death.
the physical association of the conjugates on membrane lipid
The specific mechanism for C-induced toxicity and T-induced
order.43,50 We investigated the interaction of all conjugates and protection is still uncertain, however, one possibility is that it
combinations thereof with the cell membrane of HEK cells using involves reactive oxygen species, ROS, which are known to
ratiometric imaging. The lipophilic Laurdan dye undergoes contribute to cell dysfunction. We therefore sought to investigate
spectral red-shifting upon an increase in membrane fluidity, ROS generation in HEK cells incubated with the conjugates C, D,
thus acting as a reporter for cell membrane lipid order. Changes T and P and then determined whether ROS production or
in cell membrane lipid order, namely between the gel/liquid mitigation correlated with cell viability results. This was
ordered phase (I0, predominantly fluorescent at 430–470 nm) analysed using the fluorogenic dye, CellROXt Deep Red
and the liquid disordered phase (Id, predominantly fluorescent Reagent. The cell-permeant dye is non-fluorescent while in a
at 480–550 nm), were determined by generalised polarisation, reduced state and exhibits bright fluorescence upon oxidation
Fig. 8 Ratiometric imaging of HEK293 cells treated for 20 minutes with T (500 mg mLꢀ1); C (62 mg mLꢀ1); D (125 mg mLꢀ1) and P (720 mg mLꢀ1). The shift
in GP values for HEK cells labelled with Laurden dye was recorded for 20 minutes.
J. Mater. Chem. B
This journal is © The Royal Society of Chemistry 2020

View Article Online
Journal of Materials Chemistry B
Paper
Given that the ROS results are well correlated with toxicity, it
is likely that ROS production plays a role in C-mediated toxicity
and that the cyto-protective effect of T (on C) involves the
mitigation of ROS. That is, when T was combined with C,
toxicity was reduced, as was ROS. These related effects are also
likely to involve interactions with the membrane, since results
of membrane perturbation tests were well correlated with both
toxicity and ROS generation. Specifically, association of C with
the cell membrane caused the largest GP shift and induced the
highest levels of oxidative stress in cells, as well as cell death.
Furthermore, combination of T with C caused a significant
reduction of membrane perturbation, compared to what was
seen for C alone. Whilst P is a potent donor, able to generate
H2S in the presence of cellular thiols, its activity did not match
that of T. Viability of HEK cells remained high in the presence
of high concentrations of P, however, it did not adequately
protect cells in the presence of C, nor mitigate the production
of ROS. Furthermore, P caused some membrane disturbances,
albeit only to a minimal level compared to control group. It is
reasonable to suggest an interplay of H2S and cholesteryl group
is necessary for protection.
Fig. 9 Ratiometric imaging of HEK293 cells treated for 20 minutes with
combinations: T (500 mg mLꢀ1) and C (62.5 mg mLꢀ1) and D (125 mg mLꢀ1
)
and T (500 mg mLꢀ1) combinations. The shift in GP values for HEK cells
labelled with Laurden dye was recorded for 20 minutes.
Mitochondria are a major source of ROS, such as superoxide
radical anion (ꢃO2ꢀ) and are also particularly susceptible to
by reactive oxygen species. Initially, low doses of C, D and T oxidative damage caused by accumulation of such species.52
were tested (31.25 mg mLꢀ1), as in Fig. 10 (left panel). Con- This can affect mitochondrial components such as, electron
jugates C and D caused an increase in ROS, even at low transport complexes, mitochondrial DNA, proteins and membrane,
concentration, whilst T caused a small drop in ROS from basal potentially leading to dysfunction and cell death.53 Notably, at low
levels after 4 hours. Higher doses of T and C were tested, along concentrations ROS species are also thought to be involved in
with the other H2S donor, P, with results shown in Fig. 10 (right redox signaling.54 Further, constant exposure to accumulated
panel). P was found to be ineffectual in reducing ROS, whilst T ROS can be compensated to a degree by an antioxidant defence
caused a significant drop compared to the control. Taken system to maintain redox balance. Cellular antioxidants include
together, these results suggest that it is a combination of enzymes such as superoxide dismutase (SOD) and catalase
cholesteryl group and H2S that is involved in the eliciting a (CAT) and non-enzymatic chemical antioxidants which include
ROS-mitigating effect. Notably, there is a significant reduction vitamins C and E, and GSH.
in ROS when T was combined with C, that is a reduction of
B50% compared to C alone and B25% reduction compared to unlikely to account for the antioxidant effects ascribed to
the control group. endogenous H2S. This is due to low intracellular concentrations
The ability to scavenge reactive oxygen species directly is
Fig. 10 ROS production in HEK293 cells treated with 31.25 mg mLꢀ1 C, D and T (left) and C (62.5 mg mLꢀ1), P (720 mg mLꢀ1), T (500 mg mLꢀ1), T + C
combinations (right), as measured with CellROXt Deep Red Reagent which exhibits bright fluorescence upon oxidation by reactive oxygen species, ROS,
RFI = relative fluorescence intensity *P o 0.05, **P o 0.005, ***P o 0.001.
J. Mater. Chem. B
This journal is © The Royal Society of Chemistry 2020

View Article Online
Paper
Journal of Materials Chemistry B
Fig. 11 Fluorescence images (left) in HEK293 cells treated with T (500 mg mLꢀ1) and P (720 mg mLꢀ1) and MitoSOXt Red mitochondrial superoxide
indicator. Fluorescence generated upon oxidation of fluorogenic dye by superoxide anion radical in the mitochondrial matrix of live cells. Quantitative
mean fluorescence intensity data from images, **P o 0.005, ***P o 0.001.
and therefore rates of reaction with ROS.55 An emerging Moreover, levels were found to be substantially diminished for
mechanism to explain the plethora of biological effects, including T vs. the control after 4 hours. After 24 hours P only caused a
antioxidant behaviour, is the reaction of H2S with oxidized thiols, minor reduction of mitochondrial ROS while T was able to
namely protein disulfides and sulfenic acids, to form persulfides. maintain low levels of ROS vs. the control.
Referred to as protein persulfidation, this post-translational
Given that T and P can release H2S under the influence of
modification not only alters protein function and activity but cellular thiols, this data indicates that the cholesteryl group is an
could also play a role in the protection of particular proteins important factor involved in ROS mitigation. One possibility is that
from irreversible damage induced by ROS and/or reactive nitrogen the lipid targets the H2S release to the mitochondrial membrane or
species.33
to a specific protein that is involved in ROS mitigation in the
The predominant antioxidant properties of H2S are therefore mitochondria itself. Another possibility is that the formed persul-
likely brought about indirectly. Signalling occurs via stimulation fide intermediate, CHOL-SSH, may have high propensity to react
of cellular enzymatic or non-enzymatic antioxidants, as well with ROS generated at the mitochondria. Although the importance
as various proteins responsible for inhibiting mitochondrial of the cholesteryl group in the mitigation of ROS production is
and membrane/cytosol ROS production. For example, H2S was clear, further work is required to elucidate its specific role in the
shown to increase intracellular GSH concentrations by enhanc- mediation of these biochemical processes.
ing the activity of cystine–cysteine transporters and enhancing
the redistribution of GSH to mitochondria.56 Mitochondria are
thought to play a key role in cell death pathways and although
Conclusions
various types of pro-apoptotic signals start the signaling cascade,
these are thought to converge on the mitochondria.57
This study involved conducting various cellular assays using
The concentrations of H2S generated by exogenous administra- HEK293 cells, exposed to four different PEG-based conjugates,
tion (vs. endogenous production) may be high enough to result in either independently or in combinations with one another. By
direct ROS scavenging. So, this could be the case for our donors, as comparing the results of tests conducted on the conjugates,
well as indirect antioxidant activity by stimulation of signalling several unexpected aspects regarding trisulfide conjugate T,
proteins. However, regardless of the specific events implicated, it revealed its potential to be applied beneficially in a pharma-
was of interest to us to investigate whether the ROS mitigating ceutical setting:
property of T was associated with mitochondria which are a major
– Using both a fluorescent probe for H2S detection (SF4) and
source of ROS. To enable this, we used MitoSOXt Red mitochon- amperometry, the H2S-donating ability of trisulfides, T and P
drial superoxide indicator, a fluorogenic dye for selective detection was demonstrated in a cellular environment, without the need
of superoxide in the mitochondria of live cells. Comprised of a to add exogenous thiols.
3,8-diamino-5-ethyl-6-phenylphenanthridinium moiety linked by
– Using Alamar Blue assay, the dose-dependent effects on
a hexyl chain to a triphenylphosphonium cation, the dye is able cell viability were tested, which showed a striking disparity in
to be taken up by mitochondria and accumulate into the matrix viability results between the trisulfides (T and P) and non-
in response to the negative membrane potential.58 Referring trisulfides (D and C). The trisulfides, T and P, showed negli-
to Fig. 11, which shows fluorescence images as well as the gible cytotoxicity, C showed the highest dose-dependent toxicity
mean fluorescence intensity data, significantly elevated mito- profile, and the results of D were in between. P was found to be
chondrial ROS was found to be generated by P vs. the control. well tolerated by cells, although marginally less so than T, to
J. Mater. Chem. B
This journal is © The Royal Society of Chemistry 2020

View Article Online
Journal of Materials Chemistry B
Paper
suggest a beneficial effect on cells, by the cholesteryl group T. P. Davis is grateful for the award of an Australian Laureate
(vs. solely the trisulfide group). Fellowship from the ARC (FL140100052) and J. F. Quinn for an ARC
– Using Alamar Blue assay, a cyto-protective effect was Future Fellowship (FT170100144). The authors acknowledge sup-
demonstrated by T (against C-induced cytotoxicity and against port from the Australian Research Council (DP170102950) and the
D-induced cytotoxicity). However, for P, a protective effect was Monash University Institute of Pharmacy and Pharmaceutical
only demonstrated against D, but not against C. These results, Sciences (MIPS).
demonstrating cyto-protection, also support the notion that the
cholesteryl group has an important role to play, and that the
trisulfide group is not solely implicated. The protective effect
was found for P (against D-induced toxicity vs. C-induced
Notes and references
toxicity) is likely due to the reaction of generated H2S
with disulfides, which produces reactive persulfides, such as
CHOL-SSH. These could also be pertinent mediators involved
in cyto-protection as they are also produced in the case of T.
– Membrane effects: membrane destabilization and/or destruc-
tion are events that could take place when a self-assembled surfac-
tant encounters cells. The CMC value of C (B13 mg mLꢀ1) indicates
a tendency of C to interact with the membrane, whether T and D,
with degradable structures and with CMC values significantly higher
(67 and 90 mg mLꢀ1 respectively), are less likely to do so. Membrane
permeabilization does not account for the toxic effect of C, as
demonstrated by propidium iodide test. However, by way of ratio-
metric imaging, using the lipophilic dye (Laurdan), membrane
perturbation was instead demonstrated for C. Both D and C
were also found to cause cells to depart from their normal cell
morphology, indicating a physical interaction with the cell
membrane, with potential consequences on cell viability.
– ROS generation: a comparison of ROS levels to basal levels,
1 E. C. Estevam, L. Faulstich, S. Griffin, T. Burkholz and
C. Jacob, Curr. Org. Chem., 2016, 20, 211–217.
2 R. Steudel, Chem. Rev., 2002, 102, 3905–3945.
3 P. W. Ford, M. R. Narbut, J. Belli and B. S. Davidson, J. Org.
Chem., 1994, 59, 5955–5960.
4 U. Munchberg, A. Anwar, S. Mecklenburg and C. Jacob, Org.
Biomol. Chem., 2007, 5, 1505–1518.
5 G. A. Benavides, G. L. Squadrito, R. W. Mills, H. D. Patel, T. S.
Isbell, R. P. Patel, V. M. Darley-Usmar, J. E. Doeller and D. W.
Kraus, Proc. Natl. Acad. Sci. U. S. A., 2007, 104, 17977–17982.
6 Z. Wei and B. H. S. Lau, Nutr. Res., 1998, 18, 61–70.
7 R. Wang, Physiol. Rev., 2012, 92, 791–896.
8 L. Li, P. Rose and P. K. Moore, in Ann. Rev. Pharmacol.
Toxicol., ed. A. K. Cho, 2011, vol. 51, pp. 169–187.
9 J. V. R. Medeiros, V. H. Bezerra, A. S. Gomes, A. L. R. Barbosa,
R. C. P. Lima, P. M. G. Soares, G. A. C. Brito, R. A. Ribeiro,
F. Q. Cunha and M. Souza, J. Pharmacol. Exp. Ther., 2009, 330,
764–770.
using a fluorescent probe for detection, showed that conjugates C 10 J. L. Wallace, M. Dicay, W. McKnight and G. R. Martin,
and D cause an increase in ROS, P is ineffectual, whilst T causes a FASEB J., 2007, 21, 4070–4076.
significant drop in ROS. Mitochondrial ROS was found to be 11 X. L. Wang, Q. Wang, W. Guo and Y. Z. Zhu, Biosci. Rep.,
substantially diminished for T vs. the control, especially after 4 hours. 2011, 31, 87–98.
– Overall, membrane perturbation, leading to ROS generation, 12 S. Jha, J. W. Calvert, M. R. Duranski, A. Ramachandran and
are likely events that lead to cell death for conjugate C, which has a
D. J. Lefer, Am. J. Physiol. Heart Circ. Physiol., 2008, 295,
high tendency to self-assemble and no opportunity to degrade in a
H801–H806.
cellular environment. This contrasts with T, which has a much 13 D.-D. Wu, D.-Y. Wang, H.-M. Li, J.-C. Guo, S.-F. Duan and
higher CMC value and a strong tendency to degrade in a cellular X.-Y. Ji, Oxid. Med. Cell Longev., 2019, 3831713.
environment. A high capacity to reduce ROS in the mitiochondria is 14 K. Qu, S. W. Lee, J. S. Bian, C. M. Low and P. T. H. Wong,
a likely reason for cyto-protection offered by T. The importance of a Neurochem. Int., 2008, 52, 155–165.
cholesteryl group in combination with the ability to generate H2S 15 H. Kimura, Y. Nagai, K. Umemura and Y. Kimura, Antioxid.
was demonstrated by the superior results achieved by T vs. P. Redox Signaling, 2005, 7, 795–803.
However, we are unable to conclude whether persulfide intermedi- 16 K. Chegaev, B. Rolando, D. Cortese, E. Gazzano, I. Buondonno,
ates, such as CHOL-SSH, which are potentially generated on degra-
dation, or H2S itself are the mediators of such protection, however,
these would be interesting factors for consideration in further
studies.
L. Lazzarato, M. Fanelli, C. M. Hattinger, M. Serra, C. Riganti,
R. Fruttero, D. Ghigo and A. Gasco, J. Med. Chem., 2016, 59,
4881–4889.
17 J. L. Wallace, G. Caliendo, V. Santagada, G. Cirino and
S. Fiorucci, Gastroenterology, 2007, 132, 261–271.
18 B. Geng, L. Chang, C. Pan, Y. Qi, J. Zhao, Y. Pang, J. Du and
C. Tang, Biochem. Biophys. Res. Commun., 2004, 318, 756–763.
19 H. Wei, R. Zhang, H. Jin, D. Liu, X. Tang, C. Tang and J. Du,
Antioxid. Redox Signaling, 2010, 12, 1079–1091.
Conflicts of interest
There are no conflicts to declare.
20 H. Yao, S. Luo, J. Liu, S. Xie, Y. Liu, J. Xu, Z. Zhu and S. Xu,
Chem. Commun., 2019, 55, 6193–6196.
21 Z. L. Yang, C. T. Yang, L. C. Xiao, X. X. Liao, A. P. Lan,
X. Y. Wang, R. X. Guo, P. X. Chen, C. H. Hu and J. Q. Feng,
Int. J. Mol. Med., 2011, 28, 397–403.
Acknowledgements
This work was carried out within the ARC Centre of Excellence in
Convergent Bio-Nano Science and Technology (CE140100036).
J. Mater. Chem. B
This journal is © The Royal Society of Chemistry 2020

View Article Online
Paper
Journal of Materials Chemistry B
22 C. M. Levinn, M. M. Cerda and M. D. Pluth, Acc. Chem. Res., 40 V. S. Lin, A. R. Lippert and C. J. Chang, Proc. Natl. Acad. Sci.
2019, 52, 2723–2731. U. S. A., 2013, 110, 7131–7135.
23 Y. Zhao, A. K. Steiger and M. D. Pluth, J. Am. Chem. Soc., 41 J. Aguiar, P. Carpena, J. A. Molina-Bolivar and C. C. Ruiz,
2019, 141, 13610–13618. J. Colloid Interface Sci., 2003, 258, 116–122.
24 U. Hasegawa and A. van Der Vlies, Bioconjugate Chem., 2014, 42 F. Ercole, F. M. Mansfeld, M. Kavallaris, M. R. Whittaker,
25, 1290.
J. F. Quinn, M. L. Halls and T. P. Davis, Biomacromolecules,
2016, 17, 371–383.
43 D. M. Owen, C. Rentero, A. Magenau, A. Abu-Siniyeh and
K. Gaus, Nat. Protoc., 2012, 7, 24–35.
25 J. C. Foster, S. C. Radzinski, X. Zou, C. V. Finkielstein and
J. B. Matson, Mol. Pharmaceutics, 2017, 14, 1300.
26 Z. Xiao, T. Bonnard, A. Shakouri-Motlagh, R. A. L. Wylie,
J. Collins, J. White, D. E. Heath, C. E. Hagemeyer and 44 A. S. Inacio, K. A. Mesquita, M. Baptista, J. Ramalho-Santos,
L. A. Connal, Chem. – Eur. J., 2017, 23, 11294–11300. W. L. C. Vaz and O. V. Vieira, PLoS One, 2011, 6, 1–15.
27 M. C. Urquhart, F. Ercole, M. R. Whittaker, B. Boyd, T. P. 45 R. Ernst and J. Arditti, Toxicology, 1980, 15, 233–242.
Davis and J. F. Quinn, Polym. Chem., 2018, 9, 4431–4439.
28 L. A. Connal, J. Mater. Chem. B, 2018, 6, 7122–7128.
29 M. M. Cerda, M. D. Hammers, M. S. Earp, L. N. Zakharov
and M. D. Pluth, Org. Lett., 2017, 19, 2314–2317.
30 D. Liang, H. X. Wu, M. W. Wong and D. J. Huang, Org. Lett.,
2015, 17, 4196–4199.
46 K. Jahan, S. Balzer and P. Mosto, in Environmental Toxicol-
ogy II, ed. A. Kungolos and M. Zamorano, 2008, vol. 110,
pp. 281–290.
47 A. Perani, C. Gerardin, G. Stacey, M. R. Infante, P. Vinardell,
L. Rodehuser, C. Selve and M. Maugras, Amino Acids, 2001,
21, 185–194.
31 N. E. Francoleon, S. J. Carrington and J. M. Fukuto, Arch. 48 F. Ercole, M. R. Whittaker, J. F. Quinn and T. P. Davis,
Biochem. Biophys., 2011, 516, 146–153. Biomacromolecules, 2015, 16, 1886–1914.
32 C. L. Bianco, T. Akaike, T. Ida, P. Nagy, V. Bogdandi, J. P. 49 N. Menard, N. Tsapis, C. Poirier, T. Arnauld, L. Moine,
Toscano, Y. Kumagai, C. F. Henderson, R. N. Goddu, J. Lin
and J. M. Fukuto, Br. J. Pharmacol., 2019, 176, 671–683.
F. Lefoulon, J. M. Pean and E. Fattal, Eur. J. Pharm. Sci.,
2011, 44, 595–601.
33 J. Zivanovic and M. R. Filipovic, Biochemistry, 2016, 38, 50 S. A. Sanchez, M. A. Tricerri and E. Gratton, Proc. Natl. Acad.
12–17. Sci. U. S. A., 2012, 109, 7314–7319.
34 A. K. Mustafa, M. M. Gadalla, N. Sen, S. Kim, W. Mu, S. K. 51 T. Parasassi, G. De Stasio, A. Dubaldo and E. Gratton,
Gazi, R. K. Barrow, G. Yang, R. Wang and S. H. Snyder,
Biophys. J., 1990, 57, 1179–1186.
Sci. Signaling, 2009, 2, ra72.
52 R. S. Balaban, S. Nemoto and T. Finkel, Cell, 2005, 120, 483–495.
35 E. Cuevasanta, M. Lange, J. Bonanata, E. L. Coitino, 53 M. Schieber and N. S. Chandel, Curr. Biol., 2014, 24, R453–R462.
G. Ferrer-Sueta, M. R. Filipovic and B. Alvarez, J. Biol. Chem., 54 L. A. Sena and N. S. Chandel, Mol. Cell, 2012, 48, 158–167.
2015, 290, 26866–26880.
36 M. R. Filipovic, J. Zivanovic, B. Alvarez and R. Banerjee,
Chem. Rev., 2018, 118, 1253–1337.
55 D. Benchoam, E. Cuevasanta, M. N. Moller and B. Alvarez,
Antioxidants, 2019, 8(48), 1–23.
56 Y. Kimura, Y. I. Goto and H. Kimura, Antioxid. Redox
Signaling, 2010, 12, 1–13.
¨
37 E. Cuevasanta, M. N. Moller and B. Alvarez, Arch. Biochem.
Biophys., 2017, 617, 9–25.
57 E. Murphy and C. Steenbergen, Annu. Rev. Physiol., 2007, 69,
51–67.
58 K. M. Robinson, M. S. Janes, M. Pehar, J. S. Monette,
M. F. Ross, T. M. Hagen, M. P. Murphy and J. S. Beckman,
Proc. Natl. Acad. Sci. U. S. A., 2006, 103, 15038–15043.
38 F. Ercole, M. R. Whittaker, M. L. Halls, B. J. Boyd, T. P. Davis
and J. F. Quinn, Chem. Commun., 2017, 53, 8030–8033.
39 A. B. Sullivan and K. Boustany, Int. J. Sulfur Chem., Part A,
1971, 1, 207–210.
J. Mater. Chem. B
This journal is © The Royal Society of Chemistry 2020