Novel dual-action prodrug triggers apoptosis in glioblastoma cells by releasing a glutathione quencher and lysine-specific histone demethylase 1A inhibitor

Article abstract of DOI:10.1111/jnc.14655

Targeting epigenetic mechanisms has shown promise against several cancers but has so far been unsuccessful against glioblastoma (GBM). Altered histone 3 lysine 4 methylation and increased lysine-specific histone demethylase 1A (LSD1) expression in GBM tumours nonetheless suggest that epigenetic mechanisms are involved in GBM. We engineered a dual-action prodrug, which is activated by the high hydrogen peroxide levels associated with GBM cells. This quinone methide phenylaminecyclopropane prodrug releases the LSD1 inhibitor 2-phenylcyclopropylamine with the glutathione scavenger para-quinone methide to trigger apoptosis in GBM cells. Quinone methide phenylaminocyclopropane impaired GBM cell behaviours in two-dimensional and three-dimensional assays, and triggered cell apoptosis in several primary and immortal GBM cell cultures. These results support our double-hit hypothesis of potentially targeting LSD1 and quenching glutathione, in order to impair and kill GBM cells but not healthy astrocytes. Our data suggest this strategy is effective at selectively targeting GBM and potentially other types of cancers. Open science badges: This article has received a badge for *Open Materials* because it provided all relevant information to reproduce the study in the manuscript. The complete Open Science Disclosure form for this article can be found at the end of the article. More information about the Open Practices badges can be found at https://cos.io/our-services/open-science-badges/. (Figure presented.).

Full text of DOI:10.1111/jnc.14655

DR. MARTIN ENGEL (Orcid ID : 0000-0001-7602-1340)
PROF. GILLES J. GUILLEMIN (Orcid ID : 0000-0001-8105-4470)
Article type : Original Article
Novel Dual-Action Prodrug Triggers Apoptosis in Glioblastoma Cells by
Releasing a Glutathione Quencher and Lysine-specific Histone Demethylase
1A Inhibitor
Running title: Novel prodrug reduces survival of glioblastoma cells
Authors: Martin Engel1,2†, Yi Sing Gee3†, Dale Cross1,2, Alan Maccarone4, Benjamin Heng5, Amy
Hulme1,2, Grady Smith1,2, Gilles J. Guillemin5, Brett W. Stringer6, Christopher J. T. Hyland3*, Lezanne
Ooi1,2,
*
1Illawarra Health and Medical Research Institute, Wollongong, NSW 2522, Australia
2School of Biological Sciences, University of Wollongong, Wollongong, NSW 2522, Australia
3School of Chemistry, University of Wollongong, Wollongong, NSW 2522, Australia.
4Mass Spectrometry User Resource and Research Facility, School of Chemistry, University of
Wollongong, Wollongong, NSW 2522, Australia
5Faculty of Medicine and Health Sciences, Macquarie University, NSW 2109, Australia
6QIMR Berghofer Medical Research Institute, 300 Herston Road, Herston QLD 4006, Australia
† Joint first authors: M.E. led the biology experiments and Y.S.G. led the chemistry experiments
*Corresponding Authors:
Christopher Hyland, chrhyl@uow.edu.au, Tel: + 61 24221 4953, Fax: +61 24221 4287
Lezanne Ooi, lezanne@uow.edu.au, Tel: +61 24221 5865, Fax: +61 24221 8130
This article has been accepted for publication and undergone full peer review but has not
been through the copyediting, typesetting, pagination and proofreading process, which may
lead to differences between this version and the Version of Record. Please cite this article as
doi: 10.1111/jnc.14655
This article is protected by copyright. All rights reserved.

Keywords: glioblastoma, LSD1A, methylation, apoptosis, oxidative stress
Abbreviations: 2-PCPA (trans-2-phenylcyclopropylamine), 4-OHT (4-hydroxytamoxifen), CPA
(cyclopropylamine), GBM (glioblastoma), GSH (glutathione), H3K4 (histone 3 lysine 4), LSD1 (lysine-
specific histone demethylase 1A), NMR (nuclear magnetic resonance), Q-PAC (quinone-methide-
phenylaminocyclopropane), QAC (quinone-methide-aminocyclopropane), QM (quinone methide),
ROS (reactive oxygen species).
Abstract
Targeting epigenetic mechanisms has shown promise against several cancers but has so far been
unsuccessful against glioblastoma (GBM). Altered histone 3 lysine 4 methylation and increased
lysine-specific histone demethylase 1A (LSD1) expression in GBM tumours nonetheless suggests that
epigenetic mechanisms are involved in GBM. We engineered a dual-action prodrug, which is
activated by the high hydrogen peroxide levels associated with GBM cells. This quinone-methide-
phenylaminecyclopropane (Q-PAC) prodrug releases the LSD1 inhibitor 2-phenylcyclopropylamine
with the glutathione (GSH) scavenger para-quinone methide to trigger apoptosis in GBM cells. Q-
PAC impaired GBM cell behaviours in 2-dimensional and 3-dimensional assays, and triggered cell
apoptosis in several primary and immortal GBM cell cultures. These results support our double-hit
hypothesis of potentially targeting LSD1 and quenching GSH, in order to impair and kill GBM cells but
not healthy astrocytes. Our data suggest this strategy is effective at selectively targeting GBM and
potentially other types of cancers.
Introduction
Glioblastoma (GBM) is the most common primary and malignant brain tumour in adults. GBM is an
aggressive tumour that proliferates and migrates rapidly (Demuth and Berens 2004), leaving patients
with a median survival of 12-15 months (McLendon and Halperin 2003; Verhaak et al. 2010). To
improve treatment outcomes, the search for new targets against GBM has turned towards the
contribution of epigenetic mechanisms to GBM formation and progression (Mack et al. 2015).
Transcriptional activation and repression of tumorigenesis-relevant genes is influenced by local
histone methylation. For example, methylation at histone 3 lysine 4 (H3K4) residues, which is
associated with increased gene expression (Liang et al. 2004; Santos-Rosa et al. 2002; Schneider et
al. 2004), is reduced in severe GBM cases (Liu et al. 2010). The catalysing enzymes of these non-
permanent epigenetic markers are thus of particular interest for therapeutic approaches.
Lysine-specific histone demethylase 1A (LSD1) belongs to the flavin adenine dinucleotide-dependent
amine oxidase family and catalyses the demethylation of monomethylated (me1) and dimethylated
(me2) H3K4 residues (Forneris et al. 2005; Shi et al. 2004). In combination with transcription factors,
such as repressor element 1-silencing transcription factor, LSD1 activity and the removal of H3K4
methylation is associated with gene repression (Ooi and Wood 2007). In concert with reduction in
H3K4 methylation, GBM cells have increased LSD1 protein levels (Sareddy et al. 2013; Singh et al.
2011; Zheng et al. 2015), with evidence for the functional involvement of LSD1 in GBM proliferation
This article is protected by copyright. All rights reserved.

(Suvà et al. 2014). Positive results from pharmacological inhibition of LSD1 in other tumours with
increased LSD1 expression, such as leukaemia, lung and breast cancers (Fiskus et al. 2014;
Kumarasinghe and Woster 2014; Mohammad et al. 2015; Murray-Stewart et al. 2014), indicate a
promising treatment potential for GBM by targeting LSD1 inhibition.
To increase the prospective success of a LSD1 inhibitor-based GBM drug, employment of a cancer-
selective mechanism is needed to reduce undesired effects in healthy tissue. Cancer cells produce
higher levels of reactive oxygen species (ROS), including H2O2, than non-cancerous cells (Reuter et al.
2010), while being able to withstand 10-100 fold higher H2O2 concentrations (Hagen et al. 2012).
Activation of a prodrug by high H2O2 levels can therefore reduce off-target effects in healthy tissue
(Peng and Gandhi 2012; Singer et al. 2015). Hagen et al. have elegantly shown the activation of an
aminoferrocene-based anti-cancer prodrug by H2O2 (Hagen et al. 2012), which takes advantage of
high extracellular H2O2 levels in close proximity to cancer cells to increase selectivity (Lim et al. 2005;
Zieba et al. 2000). Their prodrug undergoes oxidative break down to release glutathione (GSH)-
scavenging quinone methide (QM) and an Fe-catalyst for ROS-generation, which work in concert to
amplify accumulation of ROS in cancer cells (Hagen et al. 2012). Similarly, Noh et al. developed a
dual stimuli-responsive anticancer prodrug activated by H2O2 and acidic conditions to generate QM
and ROS-generating cinnamaldehyde (Noh et al. 2015). Importantly, cancer cells survive high ROS by
upregulation of the ROS scavenger GSH (Ogunrinu and Sontheimer 2010). Both these prodrugs and
other anticancer drugs therefore directly or indirectly target the GSH mechanism to further elevate
intracellular ROS and kill their target cells (Noh et al. 2015; Alexandre et al. 2006; Badr et al. 2013;
Kohsaka et al. 2013). However, cancer-cell selective prodrugs based on the common LSD1 inhibitor
trans-2-phenylcyclopropylamine (2-PCPA/tranylcypromine) are almost unexplored to date. The
group of Suzuki made a seminal contribution in this field, with their development of a prodrug that
conjugated 2-PCPA with the anti-estrogen agent 4-hydroxytamoxifen (4-OHT) (Ota et al. 2016). Their
prodrug selectively released 4-OHT in the presence of LSD1, which has increased expression in
breast cancer tissue (Lim et al. 2010). As such, the prodrug was able to inhibit the growth of breast
cancer cells via inhibition of LSD1 and estrogen receptor alpha, while not exhibiting cytotoxicity
towards normal cells.
Advancing on existing (pro)drugs that focus primarily on amplification of oxidative stress, we have
developed a dual-action prodrug that releases the LSD1 inhibitor 2-PCPA and the GSH scavenger QM
following H2O2 activation via an aryl boronate trigger. This dual-action prodrug was more effective
than either 2-PCPA or QM precursors alone or when applied together as independent compounds.
The quinone-methide-phenylaminocyclopropane prodrug (Q-PAC) impaired key GBM cell behaviours
and triggered cell apoptosis through its hybrid action in several primary and immortal GBM cell
cultures. Our data supports the double-hit hypothesis of targeting LSD1 and scavenging GSH, in
order to selectively impair and ultimately kill GBM cells over healthy astrocytes.
This article is protected by copyright. All rights reserved.

Materials and Methods
Chemical Synthesis and Analysis
All reactions were conducted in oven-dried glassware under nitrogen atmosphere. Reaction solvents
were dried by passing through a column of activated alumina and then stored over 4Å molecular
sieves. Progress of reactions was tracked by thin layer chromatography (TLC) and was performed on
aluminium backed silica gel sheets (Grace Davison, UV254). TLC plates were visualised under UV
lamp at 254 nm and/or by treatment with one of the following TLC stains: Phosphomolybdic acid
(PMA) stain: PMA (10 g), absolute EtOH (100 mL); Potassium permanganate stain: KMnO4 (1.5 g),
10% NaOH (1.25 mL), water (200 mL); Vanillin stain: Vanillin (15 g), concentrated H2SO4 (2.5 mL),
EtOH (250 mL). Column chromatography was performed using silica gel (40 – 75 μm) as the solid
phase. For NMR spectroscopy, analytes were dissolved in deuterated chloroform unless stated
otherwise. NMR spectra for each compound were collected from one of the following instruments:
Mercury 2000 spectrometer operating at 500 and 125 MHz for 1H and 13C NMR respectively, or a
Varian spectrometer operating at 300 and 75 MHz for 1H and 13C NMR respectively. NMR data are
expressed in parts per million (ppm) and referenced to the residual chloroform in the deuterated
solvent (7.26 ppm for 1H NMR and 77.16 ppm for 13C NMR). The following abbreviations are used to
assign the multiplicity of the 1H NMR signal: s = singlet; bs = broad singlet; d = doublet; t = triplet; q =
quartet; quin = quintet; dd = doublet of doublets; m = multiplet. For mass spectrometry analytes
were dissolved in HPLC grade methanol. High-resolution mass spectra were collected from a Waters
Xevo G1 QTOF mass spectrometer (ESI or ASAP) or Thermo Scientific LTQ Orbitrap XL (ESI). Infrared
spectra were obtained from a Shimadzu IRAffinity-1 Fourier transform infrared spectrophotometer
with an ATR attachment. The log BB value for Q-PAC was calculated using ChemDraw Professional
15.0.
Synthesis details of Q-PAC (4-(4,4,5,5-Tetramethyl-1,3,2-dioxaborolan-2-yl)benzyl (2-
phenylcyclopropyl)carbamate)
Triethylamine (0.4 mL, 290.4 mg, 2.87 mmol, 1.2 equiv) and diphenylphosphoryl azide (0.58 mL,
742.4 mg, 2.70 mmol, 1.1 equiv) were added to a solution of 2-phenylcyclopropane-1-carboxylic acid
(399.6 mg, 2.46 mmol, 1 equiv) and 4-(hydroxymethyl)phenylboronic acid pinacol ester (637.6 mg,
2.72 mmol, 1.1 equiv) in dry dioxane (5 mL). The reaction solution was heated at 105 °C for 4 h then
cooled to room temperature. Solvent was evaporated under reduced pressure and the compound
was purified by column chromatography (25% ethyl acetate in hexane). The title compound was
obtained as a colourless oil (513.6 mg, 1.31 mmol) in 53% yield.
1H NMR (500 MHz, CDCl3): δ 7.79 (d, J = 7.5 Hz, 2H), 7.33 (d, J = 7.5 Hz, 2H), 7.25 – 7.23 (m, 2H), 7.17
– 7.09 (m, 3H), 5.23 (bs, 1H), 5.12 (s, 2H), 2.75 (bs, 1H), 2.06 (bs, 1H), 1.33 (s, 12H), 1.18 (bs, 2H)
ppm. 13C NMR (75 MHz, CDCl3)a: δ 156.8, 140.5, 139.5, 135.1, 128.4, 127.2, 126.6, 126.2, 83.9, 66.7,
32.7, 24.9b, 16.2 ppm. IR (Neat): 3318, 2977, 1706 cm-1. HRMS (ESI) Found: M+, 393.2102.
C23H28BNO4 requires M+, 393.2111 a The carbon directly attached to boron was not detected, likely
due to quadropolar relaxation (Wray 1979). b Overlap of 2 signals.
Synthesis of quinone-methide-aminocyclopropane (QAC) is included in the supplemental methods.
4-(hydroxymethyl) PhenylBoronic acid pinacol Ester (PBE) was purchased from AKScientific
(AMTB135; Ca, USA), 2-PCPA from Sigma Aldrich (P8511, Castle Hill, AU).
This article is protected by copyright. All rights reserved.

Prodrug Activation Assay
Following the procedure of Hagen et al. (Hagen et al. 2012) a solvent system of 9:1
acetonitrile:water (v:v) was used to prepare a solution of Q-PAC and triethylamine (both 0.9 mM) for
activation with hydrogen peroxide (9 mM). At 5 minute intervals out to 30 minutes aliquots were
diluted 90-fold in solvent for analysis via electrospray ionization mass spectrometry on either a
Thermo LTQ ion-trap or LTQ Orbitrap XL. Both utilized an Ion Max ESI source operated in positive
mode with nitrogen as the desolvation gas. The following conditions were employed on the single-
trap instrument: 5 µL/min infusion rate, 3.5 kV source voltage; sheath, auxiliary and sweep gases set
to 12, 0 and 0 (arbitrary flow), respectively; capillary temperature 200 °C and voltage 46 V; tube lens
130 volts. Settings for Orbitrap analysis: 10 µL/min infusion rate, 4.2 kV source voltage; sheath,
auxiliary and sweep gases set to 10, 0 and 0 (arbitrary flow), respectively; capillary held at 275 °C and
50 volts; tube lens 150 volts. The infusion syringe, tubing and ESI probe were rinsed with solvent
until the ionized Q-PAC signal was reduced to background levels prior to analysis of a particular
sample. Spectra reported here constitute the average between 50 and 100 scans and were analysed
to monitor reaction species relative to Q-PAC as a function of reaction time.
LSD1 inhibition assay
The effect of Q-PAC on the demethylase activity of LSD1 was assessed in vitro using the fluorometric
LSD1 Assay Kit (# 700120, Cayman Chemical) according to the manufacturer's instructions. Each
concentration was assessed in duplicate alongside no inhibitor controls and enzyme-only reactions,
with the average fluorescence intensity from three consecutive measurements used (FLUOstar
Optima, BMG labtech, excitation/emission 540 nm/580 nm).
Cell Culture Assays
Culture Details
U87MG cells (ECACC Cat# 89081402, RRID:CVCL_0022, Acc Nr.: 89081402, obtained in 2014, Female
astrocytoma, identity confirmed via short tandem repeat profiling by Garvan Institute (Sydney, AU)
in 2015) were maintained in Dulbecco’s Modified Eagle Medium with F12 supplement (Life
Technologies, #10565-018), 10% foetal bovine serum (Bovogen, #SFBS-F), and seeded at 20,000
cells/cm2. Cells were used between passage 8 and 15, absence of mycoplasma confirmed every
three months (MycoAlert, Lonza, AU). The U87MG line is listed by ICLAC for contamination of the
ATCC version. We employused the ECACC version in this study, which shows do bewas not identical
to the ATCC version, nor was it contaminated with other cell lines based on our STR analysis.
Primary glioblastoma cultures provided by the Brain Cancer Research Unit of the QIMR Berghofer
Medical Research Institute (2015) were established from untreated biopsy samples of different
glioblastoma subtypes (Verhaak et al. 2010; Day et al. 2013) (SJH1: 72 years male, neural; RN1: 56
years male, classical; JK2: 75 years male, proneural). Approval for this study was obtained from the
Human Research Ethics Committee of The University of Wollongong (HE16/324). Cells were
maintained in Knockout-DMEM/F12 (Life Technologies, #12660-012) with StemPro supplement (Life
Technologies, #A10508-01), human EGF (20 ng/ml) (Life Technologies, #PHG0314) and human FGF2
(10 ng/ml) (Life Technologies, #PHG0024), and seeded at 35,000 cells/cm2 on matrigel (Corning,
This article is protected by copyright. All rights reserved.

#354277, 1/100 dilution). Cells were used between passage 5 and 13 in 2015 and 2016, absence of
mycoplasma confirmed every three months (MycoAlert, Lonza, AU).
Human astrocyte cultures were generated from human foetal brain tissue, which was obtained from
17 to 20-week-old foetuses collected after therapeutic termination following informed consent.
Approval for this study was obtained from the Human Research Ethics Committee of Macquarie
University (#5201200411). Written informed consent was obtained from the participants. Astrocytes
were prepared using a protocol adapted from previously described methods (Guillemin et al. 1997)
with slight modification. 1 g of brain was washed thrice with PBS containing 1%
antibiotic/antimycotic to remove contaminating blood. Visible blood vessels were removed with
sterile scissors. Next, the tissue was placed in RPMI medium (Sigma-Aldrich, Australia) supplemented
with 10% FBS and 2% antibiotic/antimycotic and dissociated mechanically by pipetting with a
serological pipette. After one week in culture, the medium was removed and the culture was
washed with PBS to remove unattached tissue fragments followed by addition of fresh medium.
Once confluent, the culture was subjected to successive passage with trypsin-EDTA (0.25%) (Life
Technologies) to remove contaminating cells and seeded at 20,000 cells/cm2 for experiments. Cells
were used between passage 2 and 5 in 2016, absence of mycoplasma confirmed after collection
(MycoAlert, Lonza, AU).
Confluence Assay
Culture confluence was monitored in 96 well plates (Greiner Bio-One CELLSTAR®) imaged every 2 h
using an IncuCyte Zoom (Essen Bioscience) at 10x magnification (1.22 µm/pixel resolution) with
three images per well. Pre- and post-treatment confluence was quantified through the inbuilt basic
analyser algorithm (Essen Bioscience) adjusted to the individual morphology of each culture type.
Migration Assay
For migration assays, cells were seeded into ImageLock 96-well Plates (Essen Biosciences) and
maintained until 70% confluent. The 700-800 µm scratch wounds were made in each well using the
96 well WoundMaker (Essen Biosciences) directly prior to drug treatment. Plates were imaged every
2 h and migration into the wound area was quantified using the inbuilt Scratch Wound algorithm
(Essen Biosciences), adjusted to the individual morphology of each culture type.
Invasion Assay
Cell invasion was examined in real time using the xCELLigence RTCA DP System (Roche Applied
Science). The xCELLigence system allows continuous quantitative monitoring of cellular behaviour
including invasion by measuring electrical impedance at a porous membrane (pore size 8 μm). U87
cells were seeded at 22,500 cells/well into specialised two-layer Cell Invasion and Migration (CIM)
plates coated with 20 µl matrigel (Corning, 1:30 dilution) and cultured without FBS for 24 in the
presence or absence of Q-PAC. Lower layer wells were filled with DMEM/F12 with 10% FBS as
chemoattractant. Invasion was continuously monitored in real time over a period of 24 h. Data
analysis was carried out using RTCA Software 1.2.1 supplied with the instrument.
This article is protected by copyright. All rights reserved.

Apoptosis Assay
For apoptosis assays, the culture media was supplemented with caspase 3/7 NucView 488 enzyme
substrate (2.5 µM final concentration, Biotium, #10402) 2 h prior to drug treatment. Phase contrast
and fluorescent images were captured using an IncuCyte Zoom (green emission/excitation at 460
nm/524 nm) at 2 h intervals and 10x magnification. Caspase substrates were quantified using the in-
built basic analyser algorithm (Table 1) from a minimum of three images per well and time point.
Table 1: Mask parameters for Incucyte Basic analyser image analysis
Targets
Channel Exposure
(ms)
Background
Correction
Edge
Sensitivity
Minimum Maximum
Particle Particle Size
Size (µm2) (µm2)
Caspase
3/7
Green
400
Top-Hat (10
µm, 2 GCU)
0
10
∞
Substrate
s
MCM2
Green
Red
400
800
Top-Hat (20
µm, 0.4 GCU)
0
7
∞, maximum
eccentricity:
0.96
Reddot2
Top-Hat (20
µm, 0.3)
-11
15
∞
Cell Viability Assay
Culture viability was assessed with the resazurin-based Presto Blue cell viability reagent (Life
Technologies, #A13261) according to the manufacturer’s instructions (2 h incubation) and quantified
on a FLUOstar Optima (BMG labtech, excitation/emission 540 nm/580 nm).
Immunocytochemistry
Cultures were fixed (4% paraformaldehyde, 15 min) and blocked (5% goat serum, 1 h), before
incubation with MCM2 polyclonal rabbit antibody (Cell Signalling Technologies, #4007,
RRID:AB_2142134, 1/500 in 5% BSA) overnight at 4 °C. This was followed by incubation with goat
anti-rabbit IgG conjugated to Alexa 488 (Life Technologies, #A11008, RRID:AB_143165, 1/1000,
1%BSA) for 1 hr at room temperature and reddot2 (Biotium, #40061-1, 1/200) as a nuclear
counterstain. Images were captured on an Incucyte Zoom in phase-contrast, green and red
(emission/excitation 585 nm/635 nm) at 20x magnification with three images per well. The fraction
of MCM2-positive cells was determined through automatic counting of reddot2 and MCM2 positive
cells per image (see Table 1 for mask parameters).
Western Blot
For histone modification quantification, cultures were lysed in triton extraction buffer (PBS
containing 0.5% Triton X 100 (v/v), 1% protease inhibitor cocktail (P8340-1ML, Sigma-Aldrich) and
0.02% (w/v) NaN3) and histones extracted in 0.2 M HCl at 4 °C over 16 h. Reduced samples were
separated on 15% polyacrylamide gels and transferred onto PVDF membranes (Millipore).
Membranes were immunoblotted at 4 °C over 16 h for monomethylated H3K4 (5% milk block,
Abcam #ab8895, RRID:AB_306847, 1:10,000 in 1% BSA), dimethylated H3K4 (5% milk block, Abcam
#ab7766, RRID:AB_2560996, 1:10,000 in 1%BSA) and acetylated H4 (3% milk block, Millipore #06-
This article is protected by copyright. All rights reserved.

866, RRID:AB_310270, 1:4000 in 3% milk) followed by goat anti-rabbit IgG-HRP (Millipore, #AP307P,
RRID:AB_11212848, 1:2500 in 1% milk) and detected by chemiluminescence. LSD1 expression was
quantified in whole-cell lysates, separated on Criterion TGX Stain-Free Precast Gels (4-20%, Biorad
#567-8095) and transferred onto PVDF membranes. Total protein loading was quantified by UV-
imaging of trihalo transferred from the gels. Membranes were immunoblotted at 4 °C over 16 h for
LSD1 (5% milk block, Cell Signalling #C69G12, RRID:AB_2070132, 1:1,500 in 5% milk), followed by
goat anti-rabbit IgG-HRP (Sigma, #A0545, RRID:AB_257896, 1:3,000 in 2.5% milk) and detected by
chemiluminescence.
GSH Assay
Reduced glutathione (GSH) of cell lysates was measured with a fluorometric kit, according to the
manufacturer’s instructions (Abcam, #ab138881) and fluorescence intensity monitored on a
FLUOstar Optima (excitation/emission 490 nm/520 nm). Sample GSH concentration was determined
through a serial-diluted GSH calibration curve (150 nM to 20 µM).
Reactive oxygen species (ROS) quantification
Oxidative stress in live cultures was assessed with CellROX Green (Life Technologies, #C10444),
which remains non-fluorescent until oxidized by intracellular reactive oxygen species (ROS). The
fluorescent signal intensity is proportional to the levels of intracellular free radicals. Cultures were
plated in black optical bottom plates (Thermo Fisher, #NUN165305) and incubated in CellROX Green
for 30 min after treatment, followed by 2x PBS washes prior to imaging. Images were captured on an
Incucyte Zoom in phase-contrast and green at 20x magnification (0.61 µm/pixel resolution) with
three images per well. Mean Green Intensity was normalised to culture confluence for each image.
Blinding and statistical Analysis
Experimenter conducting the sample analysis for the GSH assay and western blot assays were
blinded to the sample treatment details. Confluence, migration and caspase substrate experiments
were automatically quantified through standardised algorithms and therefore not blinded. No
randomization was performed to allocate samples in this study. There were no differences in sample
size between the beginning of the experiments and their conclusion. This study was not pre-
registered.
Analyses were performed using Graphpad Prism (Graphpad Software Inc.). Treatment effects were
assessed using one-way or two-way analysis of variance (ANOVA) as relevant, followed by
Bonferroni’s multiple comparisons test where appropriate. All cell culture experiments were
conducted with at least three independent biological replicates and at least two technical replicates
each. Significance was accepted at p<0.05 and data presented as mean ± standard error of mean
(SEM) for biological replicates. Number of replicates (n) in figure legends indicates independent
experiments, with averages of technical replicates within each experiment.
This article is protected by copyright. All rights reserved.

Results
Design and synthesis of Q-PAC
The prodrug Q-PAC was designed so that it would simultaneously release the LSD1 inhibitor 2-PCPA
and QM in the presence of high concentrations of H2O2. Q-PAC was synthesised using a Curtius
rearrangement as the key step and its chemical structure assigned by NMR and mass spectrometry
(Fig. 1a and Figure S1). The presence of the carbamate linkage in Q-PAC was confirmed by a 13C
resonance at 156.8 ppm (Fig. S1a) and the four methyl groups of the pinacol boronic ester were
observed as a 12H singlet at 1.33 ppm in the 1H NMR (Fig. S1b). This spectral data, in combination
with a shift of the methylene protons from 4.72 ppm in the starting 4-(hydroxymethyl)phenylboronic
acid pinacol ester to 5.12 ppm in Q-PAC provided confirmation that the coupling of the two active
fragments of the prodrug had taken place. Using the same route we also prepared control
compound QAC, which would generate quinone methide like Q-PAC, but releases LSD1 inactive
cyclopropylamine in place of 2-PCPA. Similar to Q-PAC, the formation of QAC was indicated by the
OCH2 signal at 5.11 ppm in 1H NMR (Fig. S2a) and carbamate signal at 157.0 ppm in 13C NMR (Fig.
S2b). In QAC the aminocyclopropane-generating fragment lacks a phenyl group, which is essential
for LSD1 inhibition, and QAC is therefore used as a control to demonstrate that the QM and LSD1
inhibitor components of Q-PAC act synergistically.
Q-PAC is activated by hydrogen peroxide
It was envisaged that prodrug Q-PAC activation with H2O2 would yield QM and 2-PCPA (Fig. 1a). In
the absence of GSH, QM can react with 2-PCPA to yield adduct QMA1, which can then further react
with another QM to form adduct QMA2 (Fig. 1a). We used positive-mode electrospray ionisation
mass spectrometry (+ESI-MS) to examine this activation process. Prior to H2O2 addition to a Q-PAC
solution a signal detected at m/z 416 is assigned to [Q-PAC + Na]+ (Fig. 1b). Under identical
instrument conditions the appearance of m/z 107 and 134 following 5 minutes treatment with H2O2
indicates the presence of initial products QM and 2-PCPA, respectively (Fig. 1b). Longer reaction
times (Fig. 1b, 10 min and 25 min) show further relative increase in QM and 2-PCPA and production
of QMA1 and QMA2 adducts evidenced by m/z 240 and 346 signals, respectively. These results are
consistent with the activation mechanism and adduct formation illustrated in Fig. 1a. Experiments
were repeated on a LTQ Orbitrap XL for high-resolution mass analysis to further support assignment.
Fig. 1a includes the exact mass for each compound and the corresponding errors calculated using
the accurate masses acquired from averaged Orbitrap analyses (data not shown); all errors fall
within 2 ppm.
The prodrug Q-PAC reduces migration and viability of U87 glioblastoma cells
To investigate the two-pronged approach of Q-PAC against GBM, we first explored its anti-cancer
properties in the immortal U87 cell line, commonly used to assess novel GBM treatments. Q-PAC
treatment dose-dependently reduced U87 confluence within 48 h (F(7, 42)=73.94, P<0.0001; Fig. 2a).
Q-PAC treatment also reduced the migration (F(6, 13)=16.93, P<0.0001; Fig. 2b) and invasion
through matrigel (F(4, 14)=5.324, P = 0.0081, Fig. S3e) of U87 cells, and reduced viability after 48 h
treatment (F(6,28)=9.47, P<0.0001, Fig. S3f). Cells responded within 4 h to Q-PAC, showing a
prominent change in morphology (Fig. 2c) and migration at concentrations above 1 µM.
To identify the contribution of the distinct components of Q-PAC to the observed responses, we
assessed the need for QM (using 2-PCPA) and the requirement for LSD1 inhibition (using a phenyl-
This article is protected by copyright. All rights reserved.

free form of Q-PAC (QAC) that would release QM and non-LSD1 inhibitor cyclopropylamine (CPA)).
Neither 2-PCPA, CPA nor QAC affected the culture viability (Fig. S4b), confluence (Fig. S4c-e), or
migration ability (Fig. S4f-h) at concentrations up to 300 µM for up to 48 h. This data demonstrates
that combining QM with 2-PCPA provides the resulting compound with anti-cancer properties, which
neither of the individual components possesses.
Q-PAC shows higher selectivity for glioblastoma cells over healthy astrocytes
To probe for the selectivity of Q-PAC against GBM cells, we evaluated the treatment response in
primary GBM cultures and primary cerebral astrocyte cultures (human and mouse). The GBM
cultures were grown from untreated biopsy samples of three different GBM subtypes (RN1: classical,
JK2: proneural, SJH1: neural), which have been characterised previously (Day et al. 2013). Q-PAC
dose-dependently reduced the culture confluence in the primary GBMs (P<0.001 for each line, Fig.
2d and Fig. S3a and c), and impaired migration in the scratch-wound assay (P<0.001 for each line,
Fig. 2e and Fig. S3b and d). While the three primary GBMs show different proliferation and migration
rates, Q-PAC treatment impaired both characteristics at concentrations above 10 µM. GBM culture
viability dropped within 48 h of Q-PAC treatment at concentrations of 30 µM and above (Fig. S3f).
Importantly, the potent LSD1 inhibitor Triazole 6 (Kutz et al. 2014) had no effect on viability,
confluence, or caspase activity in the primary GBM RN1 cells, and only reduced migration at 100 µM
(Fig. S5). In contrast to the GBM cultures, healthy astrocytes treated with Q-PAC at concentrations
up to 300 µM for 48 h showed no reduction in cell viability (F(6, 53)=0.56, P=0.76; Fig. S3f), nor a
change in their migratory behaviour (F(6, 12)=0.47, P = 0.82; Fig. 2h). Confluence of primary
astrocyte cultures differed following Q-PAC treatment (F(6, 38)=15.3, P<0.0001; Fig. 2g), at 30 µM Q-
PAC (80.1 % of vehicle after 48 h, P<0.001; Fig. 2g), but to a lesser extent than for any of the GBM
cultures at the same concentration (RN1: 32.7%, JK2: 64.6% and SJH1: 62.7% of vehicle). These
results demonstrate the higher vulnerability of GBM cells to Q-PAC treatment compared to healthy
astrocytes in vitro.
Q-PAC causes apoptosis through GSH reduction following H3K4 methylation
Q-PAC combines the LSD1 inhibitor 2-PCPA with the glutathione scavenger QM. While GBM cultures
reacted strongly to Q-PAC treatment, neither 2-PCPA nor QAC treatment resulted in such a
response. We therefore investigated the mechanisms underlying the Q-PAC treatment effect,
whether it affects the epigenetic profile of histones or performs its functions through other
mechanisms.
Q-PAC increases H3K4 mono & dimethylation without affecting H4 acetylation
Q-PAC was designed to be selective for GBM over healthy cells via LSD1 inhibition since increased
LSD1 protein levels have been observed in GBM (Sareddy et al. 2013; Singh et al. 2011; Zheng et al.
2015). We assessed the LSD1 inhibition properties of Q-PAC, in comparison to 2-PCPA and the
potent LSD1 inhibitor Triazole 6, showing that Q-PAC has a similar LSD1 inhibition profile as 2-PCPA
(Fig. S6). We consequently quantified LSD1 levels in our cultures. In line with published results, the
protein expression levels differed between culture types (F(5, 12)=17.24, P < 0.0001; Fig. S7a-b), with
RN1 and SJH1 cultures expressing higher levels of LSD1 compared to healthy astrocytes, while U87
and JK2 expression levels did not differ from healthy astrocytes. We quantified H3K4 mono (me1)
and dimethylation (me2), as well as H4 pan acetylation, after treating U87 (low LSD1 levels) and
primary RN1 (high LSD1 levels) cultures for 4 h, since changes in migration and confluence were
observed within 4 h in response to Q-PAC treatment (Fig. 2). In U87 cells Q-PAC had no effect on
This article is protected by copyright. All rights reserved.

H3K4me1 (F(5, 18)=1.024, P=0.43; Fig. 3a), H3K4me2 (F(5, 12)=0.25, P=0.93; Fig. 3a) nor H4
acetylation (F(5, 17)=1.13, P=0.38; Fig. 3a). In primary glioblastoma RN1 cells however, 4 h of Q-PAC
treatment showed a concentration-dependent effect on H3K4me1 (F(5,12)=3.18, P=0.05; Fig. 3b-c),
H3K4me2 (F(5,12)=2.74, P=0.07; Fig. 3b-c), but not pan H4 acetylation (F(5, 12)=1.51, P=0.26; Fig. 3b-
c). Both me1 and me2 of H3K4 peaked at 10 µM Q-PAC, while not differing to control levels at lower
or higher concentrations (Fig. 3b). When assessing the selectivity of Q-PAC for GBM, we found that
10 µM Q-PAC after 4 h did not affect H3K4me2 (t=0.26, df=6, P=0.814; data not shown) in healthy
astrocytes (H3K4me1 was not detected in the healthy astrocytes).
Q-PAC triggers concentration-dependent caspase 3/7 activity increase in GBM cells
LSD1 inhibitors have been shown to cause cell cycle arrest in breast cancer cells (Pollock et al. 2012).
We therefore investigated whether cell cycle arrest is contributing to the reduction in cell viability
and confluence observed in the GBM cultures after Q-PAC treatment. After quantifying the
expression of minichromosome maintenance 2 (MCM2), part of the DNA replication machinery that
is only expressed in proliferating cells (Williams and Stoeber 2007), we found that 48 h of Q-PAC
treatment did not alter the proportion of MCM2-positive U87 or primary GBM cells (Fig. S7c). Thus,
the reduction in viable cells appears not to be caused by driving cancer cells out of the cell cycle.
Apoptosis is a possible alternative explanation for the reduction in cell viability after Q-PAC
treatment, especially since increased apoptosis following the inhibition of LSD1 has been shown in
colon and blood cancer cells (Ding et al. 2013; Wen et al. 2012). To assess whether Q-PAC treatment
results in apoptosis, we monitored caspase 3/7 activity in treated GBM cells and healthy astrocytes.
Q-PAC dose-dependently increased caspase 3/7 activity in U87 (F(6, 11)=24.30, P < 0.0001; Fig. 4a),
RN1 (F(5, 22)=6.2, P=0.001; Fig. 4b), SJH1 (F(6, 7)=4.83, P=0.028; Fig. S7d) and JK2 (F(5, 6)=5.7, P =
0.027; Fig. S7e) cultures, consistent with an increase in apoptosis in these cell types, without
affecting caspase 3/7 activity in healthy astrocytes (F(5, 12)=2.67, P=0.08; Fig. 4c).
Q-PAC reduces GSH levels in GBM cells
Work by Noh et al. (Noh et al. 2015) shows that an increase in caspase 3/7 activity in cancer cells can
be the result of GSH depletion through scavengers such as QM, a component of Q-PAC (Fig. 1). We
consequently quantified GSH and ROS levels following Q-PAC treatment for 4 h, as this time point
reflects the onset of treatment effects on confluence and migration (Fig. 2). GSH levels dose-
dependently decreased in U87 and RN1 cells, but not in healthy astrocytes (Fig. 4d), after Q-PAC
treatment. At the same time, Q-PAC treatment increased ROS levels in U87 (F(7, 33)=5.03, P<0.001;
Fig. 4e) and RN1 cultures (F(6, 28)=5.76, P<0.001; Fig. 4e) after 4 h in a concentration-dependent
manner.
Discussion
Here we describe the synthesis of the novel anti-GBM drug Q-PAC and its characterization in vitro.
We provide mechanistic evidence that the prodrug Q-PAC undergoes oxidation of its boronate
functionality to release the active components QM and 2-PCPA. Our in vitro study shows that Q-PAC:
(1) reduces proliferation, migration, invasion, GSH levels and viability of GBM cells; (2) increases
caspase 3/7-mediated apoptosis, intracellular oxidative stress and H3K4 mono- and dimethylation;
and (3) shows a higher potency against GBM cells than healthy astrocytes. We provide the first in
This article is protected by copyright. All rights reserved.

vitro evidence for a dual-function strategy of possible LSD1 inhibition and GSH scavenging after
activation by H2O2 as a selective treatment avenue against GBM.
The 2-PCPA-based compound Q-PAC impaired several aspects of U87 cells at concentrations as low
as 10 µM, while even 300 µM 2-PCPA was ineffective against the immortal GBM cells, despite a
similar potency against LSD1. 2-PCPA treatment effects against breast cancer and neuroblastoma
cells have been shown by several studies (Lim et al. 2010; Schulte et al. 2009), but required
concentrations up to 20-fold higher than the IC50 for 2-PCPA (20.7 µM). Similarly, high
concentrations of 2-PCPA have been shown to be ineffective against other immortal GBM cultures,
despite high LSD1 protein levels (Singh et al. 2011). Q-PAC not only performed better than 2-PCPA
against GBM viability, but also reduced the migration and invasion phenotype of U87 cultures.
Aggressive migration and invasion are key treatment challenges of GBM tumours (Demuth and
Berens 2004). The suppression of cell migration and invasion following Q-PAC application is thus a
promising treatment characteristic of this novel compound. Critically, treatment with the two base
components (2-PCPA or QAC) was not effective against U87 cells, highlighting the need for the local
activation of the dual-action prodrug Q-PAC.
Considering the marked difference in treatment response of GBM cells to Q-PAC compared with 2-
PCPA, we questioned whether Q-PAC had retained its LSD1 inhibitor function. While the enzymatic
assay confirmed the inhibition of LSD1 by Q-PAC, surprisingly, H3K4me1 and H3K4me2 (the primary
targets of LSD1 (Lee et al. 2006)) appeared unaffected by Q-PAC treatment in U87 cells. An inverted
u-shape concentration-dependent effect on H3K4me1 and H3K4me2 was however observed in
primary GBM cultures, with the peak difference at 10 µM. One reason for inhibiting LSD1 was to
reduce the epigenetic suppression of tumour suppressor genes, as previously shown for LSD1
inhibitors in colorectal cancer (Huang et al. 2009). A reduction in cell proliferation would have been
expected but did not occur in U87 or primary GBM cells following Q-PAC treatment as suggested by
a lack of an effect on MCM2 (Supp. Fig. 5C). Together with the limited effect on global H3K4
methylation levels, this finding suggests that Q-PAC causes only minor global methylation changes at
these sites, but that H3K9me and gene-specific methylation changes are likely (Kidder et al. 2014;
Schenk et al. 2012; Zheng et al. 2015). In addition, LSD1 suppresses tumour suppressor protein 53
(p53) activity (Huang et al. 2007), with loss of LSD1 resulting in reduced colon cancer proliferation
(Jin et al. 2013). Q-PAC inhibition of LSD1 could thus result in reduced p53 promoter and protein
methylation, enabling p53-mediated apoptosis (Schuler et al. 2000; Scoumanne and Chen 2008),
which could be assessed via chromatin immunoprecipitation sequencing in future experiments.
We designed Q-PAC to be activated by H2O2, levels of which are higher in close proximity to GBM
cells, and to separate into 2-PCPA and QM. QM reportedly alkylates GSH, preventing its antioxidant
function (Hagen et al. 2012; Marzenell et al. 2013). GBM cells upregulate GSH to compensate for
their increased ROS production (Ogunrinu and Sontheimer 2010). Q-PAC and other anti-tumour
drugs (including anti-GBM compounds) therefore target this mechanism as a possible treatment
avenue (Noh et al. 2015; Alexandre et al. 2006; Badr et al. 2013; Kohsaka et al. 2013). In our mass
spectrometry assay, while not being a kinetic assay, we confirmed that prodrug Q-PAC was able to
undergo H2O2 induced oxidative cleavage of its boronate functionality to QM, 2-PCPA and QM-
derived adducts. This was critical given the desired dual action of the drug that required QM to
reduce the antioxidant GSH while 2-PCPA simultaneously inhibits LSD1. Our result is consistent with
the report of Hagen et al., who demonstrated a similar breakdown of aminoferrocene-based
prodrugs with H2O2 (Hagen et al. 2012). Unlike Hagen et al. however, we were able to directly detect
This article is protected by copyright. All rights reserved.

QM, as well as its conjugate addition products. Given the concentration of GSH in cells is higher than
the effective concentration of Q-PAC (Khan et al. 2012), the desired reaction between GSH and QM
would negate the formation of adducts QMA and therefore reduce the antioxidant ability of the cell.
In support of this mechanism, we show that Q-PAC concentration-dependently reduces GSH levels in
U87 and primary GBM cultures within 4 h of treatment, but not in primary astrocytes. Oxidative
stress levels increased within the same treatment timeframe in U87 and primary GBM cells. Similar
results have been reported from glioblastoma, prostate and colon cancer cell studies, where GSH
scavenger treatment caused a reduction in GSH followed by an increase in ROS levels and apoptosis
(Noh et al. 2015; Badr et al. 2013; Khan et al. 2012). Q-PAC treatment likewise resulted in apoptosis,
causing a concentration-dependent increase in caspase 3/7 activity in U87 and primary GBM cultures
without affecting caspase activity in primary astrocytes. Caspase 3/7 activity increased at Q-PAC
concentrations of 10 µM and above, while a ten-fold higher concentration was required for the ROS
level increase, indicating that GSH depletion can trigger apoptosis without measurable changes in
ROS production (Khan et al. 2012). The lack of any treatment effects following the phenyl ring-free
Q-PAC structure (QAC) treatment provides some support for the required combination of LSD1
inhibition and GSH quenching of this hybrid anti-cancer drug. Furthermore, the lack of any 2-PCPA
effects on GBM cells, a potent monoamine amine oxidase inhibitor (Yang et al. 2007), suggests that
monoamine oxidase inhibition is not the primary mode of action of Q-PAC.
A marked feature of Q-PAC across all culture assays has been the selectivity for GBM cells over
healthy astrocytes. The limited effect against healthy astrocytes is likely due to three related
differences in comparison to GBM cells: 1) Q-PAC has been designed to be activated by high local
H2O2 levels, which are found in the proximity of GBM cells, but not healthy cells (Reuter et al. 2010;
Peng and Gandhi 2012); 2) the high metabolic activity of GBM cells results in increased ROS
production and heavy reliance on the antioxidant properties of GSH (Noh et al. 2015; Ogunrinu and
Sontheimer 2010), making GBM cells highly susceptible to GSH scavenging; and 3) LSD1 expression is
higher in GBM cells than healthy cells, as shown in the present and previous studies, increasing the
vulnerability to LSD1 inhibitor treatment (Sareddy et al. 2013; Singh et al. 2011; Schulte et al. 2009).
While GBM cells would thus be affected by lower LSD1 inhibitor concentrations than healthy cells,
irreversible LSD1 inhibitors would nonetheless run the risk of affecting LSD1 function in healthy cells,
with cognitive deficits being reported after prolonged administration in mice (Neelamegam et al.
2012). Connecting the LSD1 inhibitor with the GSH scavenging mechanism through an aryl boronate
H2O2 activation trigger appears to add a beneficial selectivity layer to Q-PAC. LSD1 inhibitors for lung
cancer have progressed to human trials, while preclinical studies with cell and animal GBM models
have been of mixed success (Singh et al. 2011; Sareddy et al. 2016), despite high LSD1 expression in
the affected tissue. After observing the successful activation of Q-PAC, we evaluated its ability to
pass through the blood brain barrier by calculating the log BB parameter (Clark 1999; Goodwin and
Clark 2005). The calculated log BB value for Q-PAC is 0.10, above the -0.3 cut-off for blood brain
barrier permeability (Rodríguez-Rodríguez et al. 2009), suggesting that Q-PAC will pass through the
blood brain barrier. Our promising work thus needs to be followed up by assessing Q-PAC in
xenograft in vivo GBM mouse models for pharmacodynamics and pharmacokinetics as well as
further assessing the potential role of 2-PCPA.
In conclusion, we have designed and synthesised a novel prodrug that differs in concept from
previous QM-generating prodrugs in that it does not only rely on redox disruption in cancer cells.
Rather, Q-PAC targets the redox mechanism of the cancer cells, with a possible role for LSD1
This article is protected by copyright. All rights reserved.

inhibition that is subject to further investigation. We show that Q-PAC reduces viability, migration
and invasion, and triggers apoptosis selectively in glioblastoma cells but not healthy astrocytes.
Thus, our double-hit strategy exploiting LSD1 inhibition and GSH reduction in GBM forms the first
fundamental step for a promising treatment approach.
--Human subjects --
Involves human subjects:
If yes: Informed consent & ethics approval achieved:
=> if yes, please ensure that the info "Informed consent was achieved for all subjects, and the
experiments were approved by the local ethics committee." is included in the Methods.
ARRIVE guidelines have been followed:
Yes
=> if No or if it is a Review or Editorial, skip complete sentence => if Yes, insert "All experiments were
conducted in compliance with the ARRIVE guidelines." unless it is a Review or Editorial
Acknowledgements
We thank Dr Haibo Yu for assistance with calculating the log BB value of Q-PAC.
Open Science Badges
This article has received a badge for *Open Materials* because it provided all relevant information
to reproduce the study in the manuscript. The complete Open Science Disclosure form for this article
can be found at the end of the article. More information about the Open Practices badges can be
found at https://cos.io/our-services/open-science-badges/.
Author contributions
Conceptualization: C.H., L.O., M.E., Y.S.G.,; Methodology: M.E., L.O., Y.S.G., C.H., A.M.; Validation:
M.E., C.H., Y.S.G., A.M.; Formal Analysis: M.E., Y.S.G; Investigation: M.E., L.O., C.H., Y.S.G., A.M., D.C.,
A.H., G.S.; Resources: L.O., C.H., B.H., G.G., B.S.; Writing – Original Draft: M.E., L.O., C.H., Y.S.G.;
Writing – Review & Editing: A.M., B.H., B.S.; Visualization: M.E., C.H., Y.S.G.; Supervision: L.O., C.H.,
M.E.; Project Administration: L.O., C.H.; Funding Acquisition: L.O., C.H.
Conflict of interest
LO is a handling editor for the Journal of Neurochemistry. The authors declare no further potential
conflicts of interest.
Financial Support
This work was partially supported by a project grant (APP1071250 awarded to LO) from the National
Health and Medical Research Council (NHMRC) and a UOW-ANSTO Joint Seed Project Grant
This article is protected by copyright. All rights reserved.

(awarded to LO and CH). LO is supported by an NHMRC Boosting Dementia Research Leadership
Fellowship (APP1135720).
References
Alexandre J., Batteux F., Nicco C., Chéreau C., Laurent A., Guillevin L., Weill B., Goldwasser F. (2006)
Accumulation of hydrogen peroxide is an early and crucial step for paclitaxel-induced cancer
cell death both in vitro and in vivo. Int. J. Cancer 119, 41–48.
Badr C. E., Van Hoppe S., Dumbuya H., Tjon-Kon-Fat L.-A., Tannous B. A. (2013) Targeting cancer cells
with the natural compound obtusaquinone. J. Natl. Cancer Inst. 105, 643–653.
Clark D. E. (1999) Rapid calculation of polar molecular surface area and its application to the
prediction of transport phenomena. 2. Prediction of blood-brain barrier penetration. J.
Pharm. Sci. 88, 815–821.
Day B. W., Stringer B. W., Wilson J., Jeffree R. L., Jamieson P. R., Ensbey K. S., Bruce Z. C., et al. (2013)
Glioma surgical aspirate: a viable source of tumor tissue for experimental research. Cancers
5, 357–371.
Demuth T., Berens M. E. (2004) Molecular mechanisms of glioma cell migration and invasion. J.
Neurooncol. 70, 217–228.
Ding J., Zhang Z.-M., Xia Y., Liao G.-Q., Pan Y., Liu S., Zhang Y., Yan Z.-S. (2013) LSD1-mediated
epigenetic modification contributes to proliferation and metastasis of colon cancer. Br. J.
Cancer 109, 994–1003.
Fiskus W., Sharma S., Shah B., Portier B. P., Devaraj S. G. T., Liu K., Iyer S. P., Bearss D., Bhalla K. N.
(2014) Highly effective combination of LSD1 (KDM1A) antagonist and pan-histone
deacetylase inhibitor against human AML cells. Leukemia 28, 2155–2164.
Forneris F., Binda C., Vanoni M. A., Mattevi A., Battaglioli E. (2005) Histone demethylation catalysed
by LSD1 is a flavin-dependent oxidative process. FEBS Lett. 579, 2203–2207.
Goodwin J. T., Clark D. E. (2005) In silico predictions of blood-brain barrier penetration:
considerations to “keep in mind.” J. Pharmacol. Exp. Ther. 315, 477–483.
Guillemin G., Boussin F. D., Croitoru J., Franck-Duchenne M., Le Grand R., Lazarini F., Dormont D.
(1997) Obtention and characterization of primary astrocyte and microglial cultures from
adult monkey brains. J. Neurosci. Res. 49, 576–591.
Hagen H., Marzenell P., Jentzsch E., Wenz F., Veldwijk M. R., Mokhir A. (2012) Aminoferrocene-based
prodrugs activated by reactive oxygen species. J. Med. Chem. 55, 924–934.
Huang J., Sengupta R., Espejo A. B., Lee M. G., Dorsey J. A., Richter M., Opravil S., et al. (2007) p53 is
regulated by the lysine demethylase LSD1. Nature 449, 105–108.
Huang Y., Stewart T. M., Wu Y., Baylin S. B., Marton L. J., Perkins B., Jones R. J., Woster P. M., Casero
R. A. (2009) Novel oligoamine analogues inhibit lysine-specific demethylase 1 and induce
reexpression of epigenetically silenced genes. Clin. Cancer Res. Off. J. Am. Assoc. Cancer Res.
15, 7217–7228.
Jin L., Hanigan C. L., Wu Y., Wang W., Park B. H., Woster P. M., Casero R. A. (2013) Loss of LSD1
(lysine-specific demethylase 1) suppresses growth and alters gene expression of human
colon cancer cells in a p53- and DNMT1(DNA methyltransferase 1)-independent manner.
Biochem. J. 449, 459–468.
Khan M., Yi F., Rasul A., Li T., Wang N., Gao H., Gao R., Ma T. (2012) Alantolactone induces apoptosis
in glioblastoma cells via GSH depletion, ROS generation, and mitochondrial dysfunction.
IUBMB Life 64, 783–794.
Kidder B. L., Hu G., Zhao K. (2014) KDM5B focuses H3K4 methylation near promoters and enhancers
during embryonic stem cell self-renewal and differentiation. Genome Biol. 15, R32.
This article is protected by copyright. All rights reserved.

Kohsaka S., Takahashi K., Wang L., Tanino M., Kimura T., Nishihara H., Tanaka S. (2013) Inhibition of
GSH synthesis potentiates temozolomide-induced bystander effect in glioblastoma. Cancer
Lett. 331, 68–75.
Kumarasinghe I. R., Woster P. M. (2014) Synthesis and evaluation of novel cyclic Peptide inhibitors of
lysine-specific demethylase 1. ACS Med. Chem. Lett. 5, 29–33.
Kutz C. J., Holshouser S. L., Marrow E. A., Woster P. M. (2014) 3,5-Diamino-1,2,4-triazoles as a novel
scaffold for potent, reversible LSD1 (KDM1A) inhibitors. MedChemComm 5, 1863–1870.
Lee M. G., Wynder C., Bochar D. A., Hakimi M.-A., Cooch N., Shiekhattar R. (2006) Functional
interplay between histone demethylase and deacetylase enzymes. Mol. Cell. Biol. 26, 6395–
6402.
Liang G., Lin J. C. Y., Wei V., Yoo C., Cheng J. C., Nguyen C. T., Weisenberger D. J., et al. (2004) Distinct
localization of histone H3 acetylation and H3-K4 methylation to the transcription start sites
in the human genome. Proc. Natl. Acad. Sci. U. S. A. 101, 7357–7362.
Lim S. D., Sun C., Lambeth J. D., Marshall F., Amin M., Chung L., Petros J. A., Arnold R. S. (2005)
Increased Nox1 and hydrogen peroxide in prostate cancer. The Prostate 62, 200–207.
Lim S., Janzer A., Becker A., Zimmer A., Schüle R., Buettner R., Kirfel J. (2010) Lysine-specific
demethylase 1 (LSD1) is highly expressed in ER-negative breast cancers and a biomarker
predicting aggressive biology. Carcinogenesis 31, 512–520.
Liu B., Cheng J., Zhang X., Wang R., Zhang W., Lin H., Xiao X., Cai S., Chen X., Cheng H. (2010) Global
histone modification patterns as prognostic markers to classify glioma patients. Cancer
Epidemiol. Biomark. Prev. Publ. Am. Assoc. Cancer Res. Cosponsored Am. Soc. Prev. Oncol.
19, 2888–2896.
Mack S. C., Hubert C. G., Miller T. E., Taylor M. D., Rich J. N. (2015) An epigenetic gateway to brain
tumor cell identity. Nat. Neurosci. 19, 10–19.
Marzenell P., Hagen H., Sellner L., Zenz T., Grinyte R., Pavlov V., Daum S., Mokhir A. (2013)
Aminoferrocene-Based Prodrugs and Their Effects on Human Normal and Cancer Cells as
Well as Bacterial Cells. J. Med. Chem. 56, 6935–6944.
McLendon R. E., Halperin E. C. (2003) Is the long-term survival of patients with intracranial
glioblastoma multiforme overstated? Cancer 98, 1745–1748.
Mohammad H. P., Smitheman K. N., Kamat C. D., Soong D., Federowicz K. E., Van Aller G. S., Schneck
J. L., et al. (2015) A DNA Hypomethylation Signature Predicts Antitumor Activity of LSD1
Inhibitors in SCLC. Cancer Cell 28, 57–69.
Murray-Stewart T., Woster P. M., Casero R. A. (2014) The re-expression of the epigenetically silenced
e-cadherin gene by a polyamine analogue lysine-specific demethylase-1 (LSD1) inhibitor in
human acute myeloid leukemia cell lines. Amino Acids 46, 585–594.
Neelamegam R., Ricq E. L., Malvaez M., Patnaik D., Norton S., Carlin S. M., Hill I. T., Wood M. A.,
Haggarty S. J., Hooker J. M. (2012) Brain-penetrant LSD1 inhibitors can block memory
consolidation. ACS Chem. Neurosci. 3, 120–128.
Noh J., Kwon B., Han E., Park M., Yang W., Cho W., Yoo W., Khang G., Lee D. (2015) Amplification of
oxidative stress by a dual stimuli-responsive hybrid drug enhances cancer cell death. Nat.
Commun. 6, 6907.
Ogunrinu T. A., Sontheimer H. (2010) Hypoxia Increases the Dependence of Glioma Cells on
Glutathione. J. Biol. Chem. 285, 37716–37724.
Ooi L., Wood I. C. (2007) Chromatin crosstalk in development and disease: lessons from REST. Nat.
Rev. Genet. 8, 544–554.
Ota Y., Itoh Y., Kaise A., Ohta K., Endo Y., Masuda M., Sowa Y., Sakai T., Suzuki T. (2016) Targeting
Cancer with PCPA-Drug Conjugates: LSD1 Inhibition-Triggered Release of 4-
Hydroxytamoxifen. Angew. Chem. Int. Ed. 55, 16115–16118.
Peng X., Gandhi V. (2012) ROS-activated anticancer prodrugs: a new strategy for tumor-specific
damage. Ther. Deliv. 3, 823–833.
This article is protected by copyright. All rights reserved.

Pollock J. A., Larrea M. D., Jasper J. S., McDonnell D. P., McCafferty D. G. (2012) Lysine-Specific
Histone Demethylase 1 Inhibitors Control Breast Cancer Proliferation in ERα-Dependent and
-Independent Manners. ACS Chem. Biol. 7, 1221–1231.
Reuter S., Gupta S. C., Chaturvedi M. M., Aggarwal B. B. (2010) Oxidative stress, inflammation, and
cancer: how are they linked? Free Radic. Biol. Med. 49, 1603–1616.
Rodríguez-Rodríguez C., Sánchez de Groot N., Rimola A., Alvarez-Larena A., Lloveras V., Vidal-
Gancedo J., Ventura S., Vendrell J., Sodupe M., González-Duarte P. (2009) Design, selection,
and characterization of thioflavin-based intercalation compounds with metal chelating
properties for application in Alzheimer’s disease. J. Am. Chem. Soc. 131, 1436–1451.
Santos-Rosa H., Schneider R., Bannister A. J., Sherriff J., Bernstein B. E., Emre N. C. T., Schreiber S. L.,
Mellor J., Kouzarides T. (2002) Active genes are tri-methylated at K4 of histone H3. Nature
419, 407–411.
Sareddy G. R., Nair B. C., Krishnan S. K., Gonugunta V. K., Zhang Q., Suzuki T., Miyata N., Brenner A.
J., Brann D. W., Vadlamudi R. K. (2013) KDM1 is a novel therapeutic target for the treatment
of gliomas. Oncotarget 4, 18–28.
Sareddy G. R., Viswanadhapalli S., Surapaneni P., Suzuki T., Brenner A., Vadlamudi R. K. (2016) Novel
KDM1A inhibitors induce differentiation and apoptosis of glioma stem cells via unfolded
protein response pathway. Oncogene.
Schenk T., Chen W. C., Göllner S., Howell L., Jin L., Hebestreit K., Klein H.-U., et al. (2012) Inhibition of
the LSD1 (KDM1A) demethylase reactivates the all-trans-retinoic acid differentiation
pathway in acute myeloid leukemia. Nat. Med. 18, 605–611.
Schneider R., Bannister A. J., Myers F. A., Thorne A. W., Crane-Robinson C., Kouzarides T. (2004)
Histone H3 lysine 4 methylation patterns in higher eukaryotic genes. Nat. Cell Biol. 6, 73–77.
Schuler M., Bossy-Wetzel E., Goldstein J. C., Fitzgerald P., Green D. R. (2000) p53 Induces Apoptosis
by Caspase Activation through Mitochondrial Cytochrome c Release. J. Biol. Chem. 275,
7337–7342.
Schulte J. H., Lim S., Schramm A., Friedrichs N., Koster J., Versteeg R., Ora I., et al. (2009) Lysine-
specific demethylase 1 is strongly expressed in poorly differentiated neuroblastoma:
implications for therapy. Cancer Res. 69, 2065–2071.
Scoumanne A., Chen X. (2008) Protein methylation: a new mechanism of p53 tumor suppressor
regulation. Histol. Histopathol. 23, 1143–1149.
Shi Y., Lan F., Matson C., Mulligan P., Whetstine J. R., Cole P. A., Casero R. A., Shi Y. (2004) Histone
demethylation mediated by the nuclear amine oxidase homolog LSD1. Cell 119, 941–953.
Singer E., Judkins J., Salomonis N., Matlaf L., Soteropoulos P., McAllister S., Soroceanu L. (2015)
Reactive oxygen species-mediated therapeutic response and resistance in glioblastoma. Cell
Death Dis. 6, e1601.
Singh M. M., Manton C. A., Bhat K. P., Tsai W.-W., Aldape K., Barton M. C., Chandra J. (2011)
Inhibition of LSD1 sensitizes glioblastoma cells to histone deacetylase inhibitors. Neuro-
Oncol. 13, 894–903.
Suvà M. L., Rheinbay E., Gillespie S. M., Patel A. P., Wakimoto H., Rabkin S. D., Riggi N., et al. (2014)
Reconstructing and reprogramming the tumor-propagating potential of glioblastoma stem-
like cells. Cell 157, 580–594.
Verhaak R. G. W., Hoadley K. A., Purdom E., Wang V., Qi Y., Wilkerson M. D., Miller C. R., et al. (2010)
Integrated genomic analysis identifies clinically relevant subtypes of glioblastoma
characterized by abnormalities in PDGFRA, IDH1, EGFR, and NF1. Cancer Cell 17, 98–110.
Wen L., Chen Y., Zeng L., Zhao F., Li R., Liu Y., Zhang C. (2012) Triptolide induces cell-cycle arrest and
apoptosis of human multiple myeloma cells in vitro via altering expression of histone
demethylase LSD1 and JMJD2B. Acta Pharmacol. Sin. 33, 109–119.
Williams G. H., Stoeber K. (2007) Cell cycle markers in clinical oncology. Curr. Opin. Cell Biol. 19, 672–
679.
This article is protected by copyright. All rights reserved.

Wray V. (1979) Carbon-carbon coupling constants: A compilation of data and a practical guide. Prog.
Nucl. Magn. Reson. Spectrosc. 13, 177–256.
Yang M., Culhane J. C., Szewczuk L. M., Jalili P., Ball H. L., Machius M., Cole P. A., Yu H. (2007)
Structural basis for the inhibition of the LSD1 histone demethylase by the antidepressant
trans-2-phenylcyclopropylamine. Biochemistry 46, 8058–8065.
Zheng Y.-C., Ma J., Wang Z., Li J., Jiang B., Zhou W., Shi X., Wang X., Zhao W., Liu H.-M. (2015) A
Systematic Review of Histone Lysine-Specific Demethylase 1 and Its Inhibitors. Med. Res.
Rev. 35, 1032–1071.
Zieba M., Suwalski M., Kwiatkowska S., Piasecka G., Grzelewska-Rzymowska I., Stolarek R., Nowak D.
(2000) Comparison of hydrogen peroxide generation and the content of lipid peroxidation
products in lung cancer tissue and pulmonary parenchyma. Respir. Med. 94, 800–805.
Figure legends
Figure 1: Q-PAC is activated by hydrogen peroxide (a) Activation of Q-PAC by hydrogen peroxide
liberates QM and 2-PCPA. In the absence of GSH, subsequent formations of adducts QMA1 and
QMA2 were detected by MS. (b) (+)ESI-MS data was collected at selected times after Q-PAC
treatment with hydrogen peroxide. Sodiated Q-PAC resulting from analysis of the untreated prodrug
is labelled in the time-zero spectrum at m/z 416. Peaks corresponding to activation products and
adducts shown in panel (a) are also labelled. All other peaks have been accounted for as background
with the exception of m/z 306. Accurate mass data (not shown) support assignment to the phenol
derivative of Q-PAC (sodium adduct), which is the structure obtained by boronate oxidation prior to
breakdown to QM and 2-PCPA.
Figure 2: Q-PAC impairs mobility of primary GBM cells but not healthy astrocytes. Algorithm-based
confluence (n=6) and 2D migration (n=4) analysis of phase-contrast microscope images of GBM
cultures (a-f) and primary human astrocytes (g-i) treated with Q-PAC. Data represent mean ± SEM, *
P<0.05, ** P<0.01, ***P<0.001, ****P<0.0001 compared to vehicle control. (c, f, i) Representative
images of U87 cultures treated with Q-PAC. Cultures treated with vehicle (EtOH), or 100 µM Q-PAC
were captured in phase-contrast images 24 h after treatment at 10x magnification (scale bar = 50
µm).
Figure 3: Q-PAC increases H3K4 dimethylation in primary GBM cells. H3K4me1, H3K4me2 and
H4ace levels quantified via immunoblotting in U87 (a, n=4) and primary human glioblastoma cells (b,
n=3) cells after 4 h treatment with Q-PAC (0 to 100 µM), adjusted to total loaded protein. (c)
Representative immunoblot for RN1 cell samples, blotted for H3K4me1 (detected at 15 kDa),
H3K4me2 (detected at 16 kDa), and total protein (segment depicting 8 to 20 kDa) for each Q-PAC
concentration (0 to 100 µM). Data represent mean ± SEM, * P<0.05 compared to vehicle control.
This article is protected by copyright. All rights reserved.

Figure 4: Q-PAC triggers apoptosis and oxidative stress in primary GBM cells but not healthy
astrocytes. Apoptosis of U87 (a), primary human GBM (b) and primary human astrocyte cultures (c)
treated with Q-PAC. Apoptosis was quantified through counting of green-fluorescent caspase 3/7
substrates per mm2 in microscope images at 20x magnification over time (n=3 per concentration &
culture). (d) Intracellular GSH concentration was quantified via fluorometric assay 4 h after
treatment with Q-PAC (0 to 300 µM) in U87, primary GBM cells (RN1) and primary astrocytes (n=3
per concentration and cell type). (e) Oxidative stress levels were quantified via a cell-permeant
fluorogenic probe 4 h after treatment with Q-PAC (0 to 300 µM) in U87 and primary GBM cells
(RN1), normalized to vehicle treated cultures (n=3 per concentration and cell type). Data represent
mean ± SEM, * P<0.05, ** P<0.01, ***P<0.001, ****P<0.0001 compared to vehicle control; $$P<0.01
compared vehicle control (RN1); ##P<0.01 compared to vehicle control (U87).
This article is protected by copyright. All rights reserved.

This article is protected by copyright. All rights reserved.

This article is protected by copyright. All rights reserved.

This article is protected by copyright. All rights reserved.

This article is protected by copyright. All rights reserved.