Identification of Frataxin as a regulator of ferroptosis

Abstract

Ferroptosis is a newly discovered form of non-apoptotic regulated cell death and is characterized by iron-dependent and lipid peroxidation. Due to the enhanced dependence on iron in cancer cells, induction of ferroptosis is becoming a promising therapeutic strategy. However, the precise underlying molecular mechanism and regulation process of ferroptosis remains largely unknown. In the present study, we demonstrated that the molecular protein Frataxin (FXN) is a key regulator of ferroptosis by modulating iron homeostasis and mitochondrial function. Suppression of FXN expression specifically repressed the proliferation, destroyed mitochondrial morphology, impeded Fe–S cluster assembly and activated iron starvation stress. Moreover, suppression of FXN expression significantly enhanced erastin-induced cell death through accelerating free iron accumulation, lipid peroxidation and resulted in dramatic mitochondria morphological damage including enhanced fragmentation and vanished cristae. In addition, this type of cell death was confirmed to be ferroptosis, since it could be pharmacologically restored by ferroptotic inhibitor Fer-1 or GSH, but not by inhibitors of apoptosis, necrosis. Vice versa, enforced expression of FXN blocked iron starvation response and erastin induced ferroptosis. More importantly, pharmacological or genetic blocking the signal of iron starvation could completely restore the resistance to ferroptosis in FXN knockdown cells and xenograft graft in vivo. This paper suggests that FXN is a novel ferroptosis modulator, as well as a potential provided target to improve the antitumor activity based on ferroptosis.

Full text

Contents lists available at ScienceDirect
Redox Biology
journal homepage: www.elsevier.com/locate/redox
Identification of Frataxin as a regulator of ferroptosis
Jing Du
a,1
, Yi Zhou
c,f,1
, Yanchun Li
c,1
, Jun Xia
a
, Yongjian Chen
a
, Sufeng Chen
a
, Xin Wang
d
,
Weidong Sun
e,h
, Tongtong Wang
f
, Xueying Ren
a
, Xu Wang
g
, Yihan An
h
, Kang Lu
h
, Wanye Hu
h
,
Siyuan Huang
c
, Jianghui Li
g
, Xiangmin Tong
a,b,c,d,g,h,∗∗
, Ying Wang
b,d,h,∗
a
Department of Laboratory Medicine, Zhejiang Provincial People's Hospital, People's Hospital of Hangzhou Medical College, Hangzhou, Zhejiang, 310014, China
b
Phase I Clinical Research Center, Zhejiang Provincial People's Hospital, People's Hospital of Hangzhou Medical College, Hangzhou, Zhejiang, 310014, China
c
The Second Clinical Medical School of Zhejiang Chinese Medical University, Zhejiang Chinese Medical University, Hangzhou, Zhejiang, 310053, China
d
Clinical Research Institute, Zhejiang Provincial People's Hospital, People's Hospital of Hangzhou Medical College, Hangzhou, Zhejiang, 310014, China
e
Department of Hematology, Shaoxing Central Hospital, Shaoxing, Zhejiang, 312030, China
f
Department of Wangjiangshan, Zhejiang Provincial People's Hospital, People's Hospital of Hangzhou Medical College, Hangzhou, Zhejiang, 310014, China
g
School of Laboratory Medicine and Life Science, Wenzhou Medical University, Wenzhou, Zhejiang, 325035, China
h
Bengbu Medical College, Bengbu, Anhui, 233000, China
ARTICLE INFO
Keywords:
Ferroptosis
Frataxin
Iron-sulfur cluster
Mitochondria
ABSTRACT
Ferroptosis is a newly discovered form of non-apoptotic regulated cell death and is characterized by iron-de-
pendent and lipid peroxidation. Due to the enhanced dependence on iron in cancer cells, induction of ferroptosis
is becoming a promising therapeutic strategy. However, the precise underlying molecular mechanism and
regulation process of ferroptosis remains largely unknown. In the present study, we demonstrate that the protein
Frataxin (FXN) is a key regulator of ferroptosis by modulating iron homeostasis and mitochondrial function.
Suppression of FXN expression specifically repressed the proliferation, destroyed mitochondrial morphology,
impeded Fe–S cluster assembly and activated iron starvation stress. Moreover, suppression of FXN expression
significantly enhanced erastin-induced cell death through accelerating free iron accumulation, lipid peroxidation
and resulted in dramatic mitochondria morphological damage including enhanced fragmentation and vanished
cristae. In addition, this type of cell death was confirmed to be ferroptosis, since it could be pharmacologically
restored by ferroptotic inhibitor Fer-1 or GSH, but not by inhibitors of apoptosis, necrosis. Vice versa, enforced
expression of FXN blocked iron starvation response and erastin-induced ferroptosis. More importantly, phar-
macological or genetic blocking the signal of iron starvation could completely restore the resistance to ferrop-
tosis in FXN knockdown cells and xenograft graft in vivo. This paper suggests that FXN is a novel ferroptosis
modulator, as well as a potential provided target to improve the antitumor activity based on ferroptosis.
1. Introduction
Despite success in chemotherapy drugs clinically, drug toxicity and
resistance continue to be the principal limiting factor to achieving cures
in patients with cancer. Activation of regulated cell death and ex-
ploration therapy targets with less toxicity is a potential anticancer
treatment strategy. Numerous therapy targets have been discovered for
that cancer is a heterogeneous disease defined by various genetic and
epigenetic variations. But little is known about the molecular biological
characteristics of mitochondrial protein Frataxin (FXN) in cancer.
Iron is an indispensable element for various metabolic and physio-
logical functions in living organisms. It functions as a cofactor for vital
iron-containing enzymes that are involved in DNA synthesis, ATP pro-
duction, heme synthesis and many other physiological activities [1]. In
general, a certain amount of iron is crucial for cell survival, growth and
proliferation. Therefore, iron deficiency is undoubtedly deleterious,
leading to a multisystem disorder with innumerable effects, including
anemia [2]. Conversely, too much iron is also detrimental. The aberrant
accumulation of iron causes excess free radical generation through
Fenton reaction and subsequent results in significant alterations in the
https://doi.org/10.1016/j.redox.2020.101483
Received 27 November 2019; Received in revised form 15 February 2020; Accepted 28 February 2020
∗
Corresponding author. Phase I Clinical Research Center, Zhejiang Provincial People's Hospital, People's Hospital of Hangzhou Medical College, Hangzhou,
Zhejiang, 310014, China.
∗∗
Corresponding author. Phase I Clinical Research Center, Zhejiang Provincial People's Hospital, People's Hospital of Hangzhou Medical College, Hangzhou,
Zhejiang, 310014, China.
E-mail addresses: tongxiangming@hmc.edu.cn (X. Tong), wangying@hmc.edu.cn (Y. Wang).
1
These authors contributed equally to this work.
Redox Biology 32 (2020) 101483
Available online 02 March 2020
2213-2317/ © 2020 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/BY-NC-ND/4.0/).
T

cellular redox homeostasis, thus leading to the damage of DNA, proteins
or other biomolecules. Interestingly, relative to healthy cells, cancer
cells do exhibit an enhanced dependence on iron to enable its growth,
which may act as a target for cancer therapy [3]. It is noted that iron
overload can cause a specific cell death termed as ferroptosis, which is a
recently discovered form of non-apoptotic regulated necrosis [4]. Fer-
roptosis has been implicated in the pathological process associated with
carcinogenesis, degenerative diseases, stroke, and kidney ischemia/re-
perfusion (I/R) injury [5–8]. However, the precise underlying mole-
cular mechanism and potential anticancer therapeutic significance of
ferroptosis remain to be elucidated.
The execution of ferroptosis closely involves the iron-dependent
peroxidation of polyunsaturated fatty acids (PUFAs) [9]. Since mi-
tochondria employ the majority of cellular iron, it plays a critical role in
the iron homeostasis. The excess iron leads to a disturbance in mi-
tochondrial dynamics and even results in ferroptosis. The cystine/glu-
tamate antiporter inhibitor erastin and GPX4 inhibitor RSL3 accelerate
the progression of ferroptosis, which inhibited by the iron chelator,
GSH, ferrostatin-1 and its analogs. In addition, degradation of ferritin
by ferritinophagy can release free iron and initiate the process of fer-
roptosis [10,11]. Cancer cells harboring oncogenic Ras exhibit sensi-
tivity to ferroptosis and can be selectively killed. Our previous study has
provided experimental evidence that dihydroartemisinin, the semi-
synthetic derivatives, represents a promising therapeutic effect to pre-
ferentially target AML cells by inducing ferroptotic cell death [12].
Daolin Tang and co-workers also discovered the ferroptosis inducer
erastin enhanced the sensitivity of AML cells to chemotherapeutic
agents [13]. Here we focus on exploring targets to improve the sensi-
tivity for ferroptosis in cancer cells.
The ancient and conservative cofactors, iron-sulfur cluster (ISC), are
essential for biological processes such as redox reactions, iron home-
ostasis, enzymatic catalysis, heme synthesis and regulation of gene
expression [14]. More than 60 human proteins bind with different types
of ISCs to execute their functions. ISC assembly genes encode in the
nuclear genome and translocate into mitochondria for creating cofac-
tors. In humans, the de novo of Fe–S cluster biogenesis is initiated with
the removal of sulfur from cysteine by the cysteine desulfurase complex
Nfs1-Isd11, a process activated by frataxin [15]. The removed persul-
fide intermediately reduces to sulfide with the accomplished of ferre-
doxin reductase and ferredoxin 2, and delivers to scaffold protein ISCU
to form a [2Fe–2S] cluster with ferrous iron (Fe
2+
). Neosynthesized
ISCs are then transferred to mitochondrial recipients or further as-
sembled into [4Fe–4S] with the help of the HSC20/HSPA9 chaperone
system [16]. The exact molecular function of FXN is unclear, but it has
been proposed to be an iron chaperone or an allosteric activator of
cysteine binding to NFS1 [15]. Defects in the biogenesis of Fe/S protein
are associated with human disorders, which called Fe–S diseases. An
expansion of the GAA repeat in the first intron of the FXN gene would
result in the diminished expression of the encoded protein, thus leading
to Friedreich's ataxia (FRDA), a most common recessive ataxia in the
Caucasian population. Complete depletion of FXN is embryonically le-
thal in mice demonstrating its extremely vital biological function [17].
Decreased expression of FXN in FRDA is characterized as mitochondrial
iron accumulation, mitochondrial dysfunction and increased oxidative
stress [18]. The newly published paper demonstrated that ferroptosis
inhibitors might have therapeutic potential in primary FRDA patient-
derived fibroblasts [19]. Despite the central role of FXN in Fe–S as-
sembly and FRDA, it remains unclear whether it plays a pivotal role in
cancer biologic behavior and prognosis.
Indeed, our study demonstrated suppression of FXN expression
specifically destroyed mitochondrial morphology, impeded Fe–S cluster
assembly and enhanced cysteine deprivation-induced cell death in HT-
1080 cells. This type of cell death could be pharmacologically restored
by ferroptotic inhibitor Fer-1 or GSH, but not by inhibitors of apoptosis,
necrosis. Moreover, the observed phenotypes of increased lipid ROS,
accumulation of free iron and mitochondria dysfunction could be
rescued by ectopic expression of FXN. Additionally, pharmacological or
genetic blocking the signal of iron starvation could completely restore
the resistance to ferroptosis in FXN knockdown cells. This study is the
first to discover the role of FXN in molecular biological characteristics
of cancer and to identify the link between FXN and ferroptosis.
Meanwhile, our work sheds new light on the molecular mechanisms
underlying FXN induced ferroptosis through mitochondrial dysfunction
as well as activation of iron starvation response, and discoveries that
the ISC assembly is tightly correlated with the cancer progression,
which can be developed as a potential therapeutic strategy.
2. Materials and methods
2.1. Antibodies and reagents
The antibody to Frataxin (ab219414), to Aconitase 2 (ab129069), to
Aconitase 1 (ab126595), to IREB2 (ab181153), to NDUFV2
(ab183717), to FECH (ab137042), to Transferrin Receptor (ab80194),
SDHB (ab14714), to Ferritin Heavy Chain (ab65080), to beta Actin
(ab8226) and N-acetylcysteine (NAC), glutathione (GSH) and DCFH-DA
were obtained from Abcam (Cambridge, MA). CCK-8 Assay Kit was
obtained from Meilunbio (Dalian, China). CFDA-SE, GSH Assay Kit was
purchased from Beyotime (Shanghai, China). Erastin, Sorafenib,
Ferrostatin-1, Z-VAD-FMK, Necrosuifonamide were obtained from
Selleck Chemicals (Houston, TX). C11-BODIPY (581/591), TMRE,
MitoTracker were obtained from Thermo Fisher Scientific (Waltham,
MA). The Cell Cycle Staining Kit was purchased from MultiSciences
(Hangzhou, China).
2.2. Cell culture
Human fibrosarcoma HT-1080 cells were obtained from the Cell
Bank of Chinese Academy of Sciences (Shanghai, China) and cultured in
DMEM medium (Hyclone, Logan, UT, USA) containing 10% fetal bo-
vine serum (Gibco, Grand Island, NY, USA), supplementary with 100 U/
mL penicillin and 100 μg/ML streptomycin. Cells were maintained in an
incubator with a humidified atmosphere of 5% CO
2
at 37 °C.
2.3. Cell viability assay
Cell viability assay was evaluated using the CCK8 Assay Kit. HT-
1080 cells were seeded at a density of 4 × 10
4
cells/well in 100 μl
DMEM medium in the 96-well plate for overnight, followed by treat-
ment with different doses of erastin for 12 h. 10 μl CCK-8 was added to
each well and the cells were subsequently incubated at 37 °C for 2 h.
Absorbance was measured at 450 nm using the microplate reader.
2.4. PI staining
Cells were plated in 96-well plates at a density of 4 × 10
4
cells/well
and treated with the indicated concentration of erastin. Propidium io-
dide (PI) was then added to each well at a final concentration of 10 μg/
mL for 10 min in the dark, and cells were subjected to fluorescence
microscopy.
2.5. Transwell assay
HT-1080 cells (4 × 10
4
) in 200 μl serum-free DMEM medium were
seeded onto the upper transwell chambers, while 500 μl medium con-
taining 5% FBS was added to the lower chambers. The cells were cul-
tured at 37 °C in an incubator for 24 h. The transwell chambers were
then removed, washed with PBS twice, fixed in 4% paraformaldehyde
for 30 min and stained with 0.1% crystal violet for 20 min at room
temperature. Next, PBS was used to wash chambers, and the un-mi-
grated cells in the upper transwell membrane were gently wiped off
using a cotton swab. Images were observed and photographed using an
J. Du, et al. Redox Biology 32 (2020) 101483
2

inverted light microscope.
2.6. Cell growth assay
HT-1080 cells were seeded in a 96-well plate at a density of
3×10
3
cells/well in 100 μl DMEM medium containing 10% FBS. The
cells were incubated overnight at 37 °C in an incubator. From the
second day after plating, the numbers of cells were measured at the
specified time using the CCK8 Assay Kit. Next, a microplate reader was
used to measure the absorbance at 450 nm.
2.7. Cell proliferation assay
Cell proliferation was detected using a CFDA-SE probe. Briefly, HT-
1080 cells stained with CFDA-SE according to the manufacturer's pro-
tocol and seeded in 6-well plates. Then, cells were exposed to the same
concentration of erastin. CFDA-SE fluorescence was detected by the
flow cytometry.
2.8. Cell cycle analysis
Cell cycle profiles were examined by the Cell cycle staining Kit
(MultiSciences, Hangzhou, China) according to the manufacturer's in-
structions. Briefly, cells were harvested, washed and fixed in ice-cold
70% ethanol at 4 °C overnight. Then they were washed, treated with
100 μl RNase A at 37 °C for 30 min and stained with 400 μl PI at 4 °C in
the dark for 30 min. Finally, the cells were washed and analyzed by
flow cytometry.
2.9. Colony formation assay
3×10
3
cells were seeded in a 6-well plate. After incubation for 2
weeks, the colonies were washed with PBS, fixed with 4% paraf-
ormaldehyde, and then stained with crystal violet. The colonies were
photographed by the inverted microscope.
2.10. Soft agar colony formation assay
6-well plates were covered with a bottom layer of 2 ml 0.8% agar in
10% FBS-DMEM. Then 3 × 10
3
cells were suspended in 2 ml of 10%
FBS-DMEM containing 0.4% agarose and seeded to the top of a 0.8%
agarose. After two weeks, colonies were fixed with 4% paraformalde-
hyde for 30 min, stained with dilute crystal violet for 20 min. The co-
lonies were photographed by the inverted microscope.
2.11. Determination of ROS production
ROS levels in cells were measured by using 2ʹ,7ʹ-dichlorofluore scin
diacetate (DCFH-DA). After treatment with erastin(5 μM), cells were
washed with serum-free culture medium and incubated with 4 μM
DCFH-DA in the dark at 37 °C for 30 min. Cells were then washed and
harvested, suspended in serum-free culture and fluorescence intensity
was measured by flow cytometry.
2.12. Lipid peroxidation
Lipid peroxidation was measured with C11-BODIPY (581/591).
When lipid peroxidation increased, the fluorescence is undergoing a
shift from red to green fluorescence emission. HT-1080 cells in 6 well
plates were incubated with the indicated concentrations of erastin.
2.5 μM C11-BODIPY(581/591) was added and incubated at 37 °C for
30 min. Excess C11-BODIPY was removed by washing the cells twice
with Hanks buffer. Cells were harvested and resuspended for flow cy-
tometry analysis.
2.13. Confocal microscopy assay
The cells were seeded in a chamber confocal dish and treated with
the indicated concentration of erastin. Then 581/591 C11- BODIPY
(5 μM) or MitoTracker(100 nM) or RPA(4 μM) were co-staining with
DAPI(10 mg/ml) in the dark for a 30 min incubation. Representative
images were viewed under a confocal microscope.
2.14. Measurement of mitochondrial membrane potential (MMP)
The changes of mitochondrial membrane potential (MMP) were
observed by tetramethylrhodamine methyl ester (TMRE) probe. HT-
1080 cells were treated with 5 μM erastin for 12 h and then stained with
TMRE (100 nM) for 30 min. The stained cells were washed with PBS
twice, and analyzed using the flow cytometry.
2.15. Iron assay
The mitochondrial chelatable iron pool was assessed using
Rhodamine B-[(1, 10-phenanthroline-5-yl)-aminocarbonyl]benzyl ester
(RPA), a Fe
2+
specificfluorescent sensor. After treatment, cells were
harvested and incubated with 2 μM RPA for 15 min at 37 °C in Hanks
balanced salt solution(HBSS), then washed subsequently three times
with HBSS. The samples were incubated with dye-free HBSS for another
15 min following washing once with HBSS. Mitochondrial iron was
measured using flow cytometry.
2.16. Determination of iron concentrations in isolated mitochondria and
cytoplasm by ICP-MS
Total iron concentration was measured using inductively coupled
plasma mass spectrometry (ICP-MS, Agilent, Varian) as previously de-
scribed [20]. In brief, indicated cells were washed, trypsinized and
subjected to the subcellular fractions isolation kit (Beyotime, Shanghai,
China). Fractions were lysed and quantified by BCA Protein Assay Kit.
Each aliquot of the lysed sample was prepared for ICP-MS analyses.
Quadruplicate determinations of iron concentration were performed for
each sample. The results were conversions from raw ppb values and
expressed as pmol/100 mg protein.
2.17. GSH detection
Intracellular GSH levels were examined by using a GSH Assay Kit
(Beyotime, Shanghai, China). HT-1080 cells were treated with erastin
for 24 h. Then, the subsequent procedures were performed according to
the manufacturer's instructions. The experimental data were obtained
by a microplate reader.
2.18. Transmission electron microscope (TEM)
After indicated treatment, The cells were fixed by the 2.5% glutar-
aldehyde solution at 4 °C overnight. After fixation, the samples were
dehydrated by a graded series of ethanol, then dehydrated by alcohol
and eventually transferred to absolute acetone. Following Infiltration
with absolute acetone and the final Spurr resin mixture, the samples
were embedded, ultrathin sectioned and stained. Finally, the samples
were observed in the Hitachi Model H-7650 transmission electron mi-
croscope.
2.19. Plasmids
For knockdown of FXN, target shRNA sequences were subcloned
into pLVX-shRNA Lentivector (Takara). The two shRNA knockdown
sequences for FXN were forward: 5′- GATCCGCTGGACTCTTTAGCAG
AGTTTTCAAGAGAAACTCTGCTAAAGAGTCCAGCTTTTTTG-3′, and 5′-
GATCCGCAGACGCCAAACAAGCAAATTTCAAGAGAATTTGCTTGTTTG
J. Du, et al. Redox Biology 32 (2020) 101483
3

GCGTCTGCTTTTTTG-3’. Human full-length FXN or FTH cDNA was
amplified by RT–PCR using HEK-293 mRNA and verified by sequen-
cing. Then the cDNA was subcloned into pLVX -Neo lentivirus vector
(Takara, Dalian, China) by ClonFast Seamless Cloning kit (obio,
Nanjing, China). The plasmid ofshRNA resistant form of FXN (Res-FXN)
was generated according to the described methods by introduced silent
changes in the coding region targeted by the shRNA [21].
2.20. Lentiviral transduction
The recombinant lentiviral plasmids were verified by sequencing
and co-transfected with pMD2G, pSPAX2 into HEK293 cells to produce
recombinant lentiviral. Lentivirus infections were carried out as de-
scribed previously [12]. Briefly, the cell seeded in 24-well plates
reached 70–80% confluence, the 10%-DMEM medium was removed.
Cells were then transfected with the corresponding lentivirus. After two
days, puromycin or G418 were added for screening . Then the stable
cells were maintained in puromycin or G418. The expression efficiency
was evaluated by RT-PCR and western blot analysis.
2.21. Western blotting
Following treatment, the cells were lysed in RIPA buffer after
washing with PBS and incubated on ice for 30 min. Then cellular debris
was removed by centrifugation and the protein concentration was
quantified with BCA Protein Assay Kit. Subsequently, equal amounts of
protein were separated by SDS–PAGE and transferred to PVDF mem-
branes. The membranes were blocked with 5% skim milk for 1 h and
incubated with the primary antibodies at 4 °C overnight. After washing
three times with TBST, the membranes were incubated with the sec-
ondary antibodies at room temperature for 1 h and washed again. The
blots were visualized using a chemiluminescence detection kit ECL-
PLUS.
2.22. RNA isolation and quantitative real-time PCR (RT-PCR)
Cells were lysed using Trizol reagent (Invitrogen, USA) and total
RNA was extracted with chloroform and isopropyl alcohol. cDNA was
then synthesized using a reverse transcription reagent kit (TaKaRa,
Dalian, China) according to the manufacturer's protocols. The SYBR
Green Master Mix Kit was used for relative quantification of RNA levels
according to the manufacturer's instructions. GAPDH was chosen as an
internal control. The sequences of the primers were as follows: GAPDH,
forward, 5′-GCACCGTCAAGGCTGAGAAC, reverse, 5′-ATGGTGGTGAA
GACGCCAGT; FXN, forward, 5′-TAGCAGAGGAAACGCTGGAC, reverse,
5′-ACGCTTAGGTCCACTGGATG. The expression level was normalized
to the internal control and determined by a 2
-ΔΔCT
method.
2.23. Determining mitochondrial DNA (mtDNA) copy number
Quantitation of the mitochondrial DNA copy number relative to the
nuclear DNA was carried out by using real-time PCR. Primer specific for
HGB1 genes were used for the determination of nuclear DNA (nDNA).
This primer sequences were used as follows: forward primer, 5′-GTGC
ACCTGACTCCTGAGGAGA-3′; reverse primer, 5′-CCTTGATACCAACCT
GCCCAG-3′. And another primer (ND-1) for the detection of mtDNA.
The primer sequences were as follows: forward primer, 5′-CCCTAAAA
CCCGCCACATCT-3′; reverse primer, 5′-GAGCGATGGTGAGAGCTAA
GGT-3′. Q-PCR was performed and the mtDNA copy number was cal-
culated. The thermal cycling conditions for the nDNA and mtDNA
amplification were 95 °C for 5 min, followed by 40 cycles of 95 °C for
15 s, 55 °C for 15 s, and 72 °C for 1 min.
2.24. Mouse xenograft model
4–6 weeks old male BALB/c nude mice were used to construct
xenograft models. 2.5 × 10
6
HT-1080 cells suspended in 0.1 mL PBS
were injected subcutaneously into the nude mice. After 7 days, tumor
growth was detectable and monitored every 2 days. Tumor volume in
mm
3
was determined by measuring the longest diameter (a) and
shortest width (b) and calculated by using the following formula: vo-
lume (mm
3
)=0.5×a×b
2
. On the 12th day, mice were euthanized
and tumors were isolated.
2.25. H&E analysis
Tumors collected from mice were fixed in 4% paraformaldehyde.
The paraffin-embedded samples were cut to 4 μm thickness and stained
with H&E (Sigma). Stained sections were viewed and photographed
under a microscope.
2.26. Immunohistochemistry (IHC)
Tissues were fixed with 4% paraformaldehyde and embedded in
paraffin. The paraffin-embedded block tissues were cut into 4 μm sec-
tions and followed dewaxed, hydrated and antigen retrieval. After
washing with PBS three times, the slides were treated with 3% hy-
drogen peroxide for 15 min, washed with PBS, blocked with BSA for
15 min at room temperature. Subsequently, anti-FXN antibody (1:200),
anti-FTH antibody (1:100) were added to the sections at 4 °C for
overnight. The streptavidin peroxidase method was used for signal
detection and then stained by diaminobenzidine (DAB) and counter-
stained with hematoxylin. The sections were observed and photo-
graphed under the light microscope. All slides were scored by two in-
dependent observers in a blinded fashion.
2.27. Statistical analysis
All statistical calculations were performed using GraphPad Prism
(version 7.0). All results were presented as the mean ± standard de-
viation (SD). The differences between the two groups were performed
by the Student's t-test. Comparisons among multiple groups were ana-
lyzed by the one-way ANOVA. P< 0.05 was considered to be sig-
nificant.
3. Results
3.1. FXN knockdown repressed the proliferation of HT-1080 cells
FXN stable knockdown cells were generated through lentivirus
transduction by two separate short hairpin RNA sequences. Western
blot and real-time PCR were used to detect FXN expression. As ex-
pected, the knockdown approach dramatically reduced RNA levels and
protein expression of FXN in the shRNA cells (Fig. 1A, B). Using these
cells, we found that FXN depletion did not lead to cell death, but instead
inhibited cellular proliferation (Fig. 1C). To further determine whether
the decreased cell proliferation was affected by the cell cycle distribu-
tion or not. Cell cycle distribution assay was performed followed by
flow cytometry and showed cells were arrested at G0/G1 phase
(Fig. 1D, E). Furthermore, the capacity of long-term cell viability be-
tween the stably FXN knockdown and control cells was detected by
colony formation and soft agar assay (Fig. 1F-I). The results showed
FXN suppressed significantly inhibited the capacity of colony forma-
tion. Besides, transwell assays were performed to determine the mi-
gration of HT-1080 cells and revealed that FXN knockdown dramati-
cally inhibited the migration of HT-1080 cells (Fig. 1J, K). Notably, the
high FXN expression indicated a significantly poorer overall survival
rate in sarcoma, acute myeloid leukemia, bladder urothelial carcinoma,
and adrenocortical carcinoma patients based on The Cancer Genome
Atlas (TCGA) datasets (Fig. S1). Taken together, these results discover
that FXN plays a vital role in the molecular biological characteristics of
cancer and might be a potential indicator of poor prognosis.
J. Du, et al. Redox Biology 32 (2020) 101483
4

3.2. FXN knockdown induced the dysfunction of mitochondria and
accumulation of free iron
As FXN is a conserved mitochondrial protein, we next examined the
mitochondrial function in the FXN suppressed HT-1080 cells. Firstly,
mitochondria were labeled with the mitotracker probe and observed by
a confocal laser scanning microscope. The control cells showed a net-
work of elongated mitochondria, in contrast, the FXN knockdown cells
appeared fragmentation and accumulation around the nucleus, in-
dicating that FXN depletion resulted in the dysfunction of mitochon-
drial homeostasis (Fig. 2A). For the close association of mitochondrial
morphology with functionality, we next monitored the changes of mi-
tochondrial membrane potential (MMP) through TMRE staining fol-
lowed by flow cytometry. The results confirmed that the MMP of HT-
1080 cells with two different FXN knockdown sequences appeared
significant hyperpolarization (Fig. 2B). Earlier published paper has
proved that excessive elevating of MMP does not help to accelerate ATP
production, but assist the production of free radicals exponentially
[22].
The above studies demonstrate that FXN plays a role in maintaining
mitochondrial function; it is unclear the regulation loop between FXN
and the activity of oxidative phosphorylation. We next tested the mi-
tochondrial oxygen consumption rate (OCR) by extracellular flux ana-
lyzer, which reflected the main mitochondrial function of energy re-
spiration. As showed in Fig. 2C, knockdown of FXN remarkably
decreased the oxygen consumption in mitochondria. The basal re-
spiration, maximal respiration and spare respiration, three indices that
represent mitochondrial respiration flux revealed a significant decline,
demonstrating that FXN knockdown disrupted the process of oxidative
phosphorylation (Fig. 2D, E, F). Moreover, ATP production was also
calculated and a pronounced loss of ATP exhibited in the FXN knock-
down cells, which was unable to meet the cellular ATP demands for
supporting cell growth (Fig. 2G). Collectively, the mitochondrial OCR
tested by Seahorse XF96 Analyzer uncovered an important link between
FXN and oxidative phosphorylation.
Besides energy respiration, ISC assembly is another important pro-
cess that takes place in mitochondria and considered to be essential for
viability. Thus, ISC turnover in iron-sulfur protein was monitored in
several ways. Above all, the activity of mitochondrial and cytoplasmic
aconitase, important [4Fe–4S] proteins that catalyze the reaction from
Fig. 1. FXN knockdown repressed the proliferation
of HT-1080 cells. (A) Western blot analysis of FXN
expression in indicated FXN knockdown and control
HT-1080 cells. Tom20 was served as internal
loading control. (B) Quantitative real-time PCR
analysis of the mRNA of FXN in the three indicated
HT-1080 cells. (C) The ability of proliferation was
detected by CCK8 assay between control and sh-
FXN cells. Induction of cycle arrest in HT-1080 cells
was detected by flow cytometry (D-E). Effects of
FXN suppression in HT-1080 cells on the prolifera-
tion capacity were measured by plate clone forma-
tion assay (F-G) and soft agar colony formation
assay (H-I), corresponding histograms were shown
on the right . Migration of HT-1080 cells was eval-
uated by transwell assay; corresponding histograms
were shown on the right (J-K). All histograms were
represented as mean ± SD.
★
P < 0.05,
★★
P < 0.01 versus control.
J. Du, et al. Redox Biology 32 (2020) 101483
5

citric acid to isocitrate, were analyzed by the in-gel activity assay. And
the breakdown of mitochondrial ISC assembly resulted in a more dra-
matic decrease of cytoplasmic aconitase activity than m-aconitase ac-
tivity, without obvious changes of protein level, demonstrating FXN
depletion promoted the loss of the aconitase ISCs (Figs. 2H, I, S2A). It
has been reported cellular iron-responsive protein1 (IRP1) and IRP2
play a central role in the regulation of iron metabolism. Apo form of
cytoplasmic aconitase transformed into IRP1, which could promote the
degradation of ferritin heavy chain (FTH1) mRNA and stabilize trans-
ferrin receptor 1 (TFR1) mRNA through IRP-IRE mechanism, thus
leading to the iron starvation response. Increased iron starvation re-
sponse was also confirmed by the upregulated IRP2, TFR and down-
regulated FTH (Fig. 2J). For the importance of FXN in the de novo ISC
biogenesis and impaired mitochondrial OXPHOS in FXN suppressed
cells, we also measured the expression of SDHB and NDUFV2, subunits
of complexes I and II whose electron transfer function rely on the intact
ISC cluster. Pronounced depletion of SDHB and NDUFV2 was detected
in FXN knockdown cells. In addition, the [2Fe–2S] binding FECH, a
rate-limiting enzyme in heme biosynthesis, and lipoic acid harbored in
PDH E2, KGDH E2 which represented the activity of [4Fe–4S] binding
lipoic acid synthase were also significantly decreased (Fig. 2K). These
data conclude that FXN depletion reduce steady-state levels of Fe–S
cluster dependent proteins as well as reactions and subsequent activa-
tion of the iron starvation stress. Besides, in line with previous studies
that NRF2, an important transcription factor in regulating cellular
redox homeostasis, was downregulated in FXN suppressed cells
[23,24]. The mechanism involved may be transcription-independent
(protein-protein interaction), for the increased levels of cytosolic Kelch-
like ECH-associated protein 1 and a similar level of NRF2 mRNA under
FXN suppression (Fig. S 2B, C).
Fig. 2. Suppression of FXN induced the dysfunction
of mitochondria and accumulation of free iron. (A)
To evaluate mitochondrial morphology, HT-
1080 cells were stained with MitoTracker probe
(100 nM) and DAPI (10 mg/ml), and then photo-
graphed by confocal laser microscope. Scale bars:
75 μm. (B) Flow cytometry was performed to mea-
sure mitochondrial membrane potential (MMP) by
TMRE probe (100 nM) in indicated FXN knockdown
and control HT-1080 cells. (C) OCR rate was carried
out by extracellular flux analyzer after the additions
of oligomycin, FCCP and antimycin A. (D-G) The
basal respiration, maximal respiration, spare re-
spiration and ATP production were calculated. (H)
Mitochondrial and cytoplasmic aconitase activity
was analyzed by the in-gel activity assay. (I) The
protein levels of mitochondrial and cytoplasmic
aconitase were detected by western blot assay. β-
actin and Tom20 were used as a loading control. (J-
K) Western blot analysis of iron-starvation stress
and ISC turnover in iron-sulfur protein. (L) HT-
1080 cells were stained with Rhodamine B-[(1,10-
phenanthroline-5-yl)-aminocarbonyl]benzyl ester
(RPA) probe (4 μM) and DAPI (10 mg/ml) to detect
the cellular labile iron by confocal laser microscope.
Scale bars: 75 μm. (M) The quantitative PCR ana-
lysis was performed to measure mitochondrial DNA
copy number. Primer specific for HGB1 genes were
used for the determination of nuclear DNA (nDNA)
and another primer (ND-1) for the detection of
mtDNA.
★
P < 0.05,
★★
P < 0.01 versus control.
J. Du, et al. Redox Biology 32 (2020) 101483
6

We next used the selective yield fluorescence probe RPA, whose
cationic fluorophore could rapidly quench by Fe
2+
ions, to detect the
cellular labile iron. As showed in Fig. 2L, FXN knockdown induced a
pronounced accumulation of free Fe
2+
ions with the decreased fluor-
escence of RPA. The quantitation of the RPA fluorescence was shown in
Fig. S2D. For the observed reduction of OCR, we subsequently mon-
itored the mtDNA copy number, which encodes 13 polypeptides playing
crucial roles in the OXPHOS. Consistent with this finding, the mtDNA
copy number appeared a clearly decrease in the FXN knockdown cells
(Fig. 2M). Collectively, these findings provide a pivotal role of FXN in
maintaining the homeostasis of mitochondrial and keep the balance of
iron metabolism.
3.3. Suppression of FXN enhanced erastin-induced ferroptosis
We wondered whether suppression of FXN accelerated the process
of ferroptosis cell death. Erastin, system X
c
–
cystine/glutamate anti-
porter (xCT) inhibitor, which can depress cysteine levels, was applied to
induce ferroptosis in HT-1080 cells. FXN knockdown and control cells
were exposed to erastin treatment for 12 h and suppression of FXN
expression significantly enhanced erastininduced cell death. The con-
centrations of IC50 values for control were evidently higher than that of
two FXN knockdown cells, which calculated results were 4.130 μM and
5.527 μM, respectively (Fig. 3A). We obtained a similar phenomenon in
U266 and Kasumi-1 cells (Fig. S3 A, B), indicating that FXN sensitized
cells to ferroptosis was not restricted to only a single cell lineage. Of
note, IKE, another cystine/glutamate antiporter inhibitor, also co-
operate with FXN suppression to induce cell death. On the contrary,
suppression of FXN expression did not affect cell death induced by other
ferroptosis inducers, including RSL3 (GPX4 inhibitor) and BSO (GSH
synthase inhibitor) (Fig. S3 C).
Erastin-induced ferroptosis would lead to a disruption of cell
membrane permeability, therefore enabled propidium iodide (PI)
staining to monitor the progress. The PI staining assay manifested that
suppression of FXN expression remarkably induced HT-1080 cell death
following erastin treatment. Phase-contrast microscopy also showed
FXN depleted cells became shrinking followed by condensation of cy-
toplasmic constituents, like a “ballooning”phenotype, upon erastin
treatment (Fig. 3B). These findings indicated FXN knockdown sig-
nificantly augmented erastininduced ferroptotic cell death. In addition,
we investigated the proliferation rate of diverse cells exposed to a low
concentration of erastin and carboxyfluorescein diacetate succinimidyl
ester (CFDA-SE) fluorescence intensity which exponentially decreased
with cell proliferation and division was analyzed by flow cytometry.
The result indicated that erastin obviously attenuated cell proliferation
of HT-1080 cells, which was much more significant upon FXN depletion
cells (Fig. 3C). Furthermore, Dramatic morphological changes of mi-
tochondria were observed by transmission electron microscopy in FXN
depletion cells comprising mitochondrial fragmentation, vacuolization
and cristae enlargement, which was aggravated under treatment with
erastin (Fig. 3D).
As the accumulation of lipid peroxidation and free iron level were
two critical drivers of ferroptosis. The fluorescent staining of BODIPY
C11, a sensitive fluorescent reporter for lipid peroxidation, was used to
quantify the formation of lipid peroxidation in live cells by flow cyt-
ometer and confocal laser microscope. Increasing levels of lipid per-
oxides were monitored in indicated HT-1080 cells after erastin treated
for 12 h, and FXN deletion significantly promoted lipid peroxide for-
mation upon erastin treatment with red fluorescence of the BODIPY
shifted to green fluorescence. But FXN depletion alone could not sti-
mulate the accumulation of lipid peroxidation (Fig. 3E, H). Quantita-
tion of the confocal fluorescence was provided in Fig. S3 D. We also
discovered similar results that FXN accelerated the accumulation of
lipid peroxidation and ROS under the exposure of erastin in U266 cells
(Fig. S3 E, F). We next asked whether alteration of GSH level partici-
pated in the FXN dependent ferroptosis, and found FXN depletion alone
was not enough to reduce GSH level. But under ferroptotic stress, such
as erastin treatment, FXN depletion accelerated the decrease of GSH
(Fig. S3 G). Given that iron enables to produce superoxide radicals via
the Fenton reaction and result in lipid peroxidation. We next assessed
the cellular labile iron pool by RPA probe and found FXN suppression
robustly activated the labile iron level under the erastin exposure (Fig.
S3 H). In consistent with RPA fluorescence, we also found that FXN
depletion activated the iron-starvation response, which was far more
robust under the treatment of erastin (Fig. S3 I).
Mitochondrial dysfunction and oxidative stress are the other two
characteristics of ferroptosis. We hypothesized that FXN accelerated
ferroptosis was associated with the increased of MMP and cytoplasm
ROS. Indeed, the elevated levels of MMP and cytoplasm ROS induced by
erastin were significantly aggravating by the suppression of FXN
(Fig. 3F, G). To further validate that FXN knockdown specifically en-
hanced the ferroptotic process, we used different signal pathway in-
hibitors to observe the rescue effect. Notably, erastin-induced cell death
was able to reverse by the ferroptosis inhibitor ferrostatin-1 and GSH,
but not by the apoptosis inhibitor ZVAD-FMK and necroptosis inhibitor
necrosulfonamide (Fig. 3I).
Several studies have revealed that autophagy-dependent degrada-
tion of ferritin, a process named ferritinophagy, takes place in ferrop-
tosis. To distinguish whether FXN regulates ferroptosis through mod-
ulating autophagy. The conversion of LC3 was detected in FXN
depletion and control cells in the absence or presence of erastin.
Alterations of FXN expression did not affect the erastin-induced for-
mation of lipidated LC3B (Fig. S3 J). A similar phenomenon was
showed that pharmacological inhibition of autophagy by BafA1 could
not reverse the FXN induced ferroptosis (Fig. 3I), indicating autophagy
was not involved in the FXN dependent ferroptosis. Together, these
studies demonstrate that FXN depletion is capable of accelerating era-
stin induced ferroptosis, most likely due to the breakdown of the ISC
biosynthetic machinery and mitochondrial dysfunction.
3.4. Overexpress of FXN contributed to ferroptosis resistance
To shed more light on the specific order of events in modulating
mitochondrial function and ferroptosis by FXN, ectopic expression of
FNX cell was generated and the proliferation rate was analyzed by the
CFDA-SE probe. The results showed that FXN overexpression obviously
increased cell proliferation of HT-1080 cells (Fig. 4A). In contrast with
mitochondrial dysfunction in FXN suppression cells, ectopic expression
of FXN markedly increased mtDNA copy number and decreased the
liable iron level (Figs. 4B, D, S4A). Next, we asked whether over-
expression of FXN in HT-1080 cells could diminish erastin-induced
ferroptosis. As shown in Fig. 4C, cell viability assays of the indicated
cells experienced that ectopic expression of FXN sensibly abolished
erastin-induced growth inhibition compared to the control and FXN
depletion cells. To further verify the specificity of the observed phe-
nomenon, we generated an shRNA-resistant FXN cDNA (Res-FXN) and
co-transduced with shFXN. Indeed, overexpression of an shRNA re-
sistant form of FXN efficiently restored the growth inhibition of erastin
(Figs. S4B and C). Similarly, expression of Res-FXN conferred resistance
to xCT inhibitors, IKE, but not to RSL3 and BSO (Fig. S4D). In consistent
with the results of RPA, mitochondrial and cytoplasmic iron measured
by ICP-MS demonstrated that knockdown of FXN induced the accu-
mulation of iron in both fractions, while ectopic expression of Res-FXN
decreased the level of iron(Figs. S4E and F). Collectively, these findings
indicate that expression of FXN is directly responsible for the observed
phenotypes and excluding the possibility of off-target effects. More
details were found with the mitochondrial morphological changes
through MitoTracker fluorescence staining and transmission electron
microscopy. The results showed that FXN expression could keep mi-
tochondrial morphology integrity, without severe mitochondrial frag-
mentation or vacuolization changes under erastin treatment (Fig. 4E,
F).
J. Du, et al. Redox Biology 32 (2020) 101483
7

As FXN expression robustly decreased cellular free iron contents,
another important question was whether FXN expression could rescue
the erastininduced accumulation of lipid peroxidation. In line with
previous findings, FXN expression restored the erastinchallenged lipid
peroxidation (Figs. 4G, H, S4G). In addition, ferroptosis inhibitors
(ferrostatin-1 and GSH) and DFO restored much less in the FXN over-
expression cells than knockdown cells (Fig. 4I). Collectively, these re-
sults imply that FXN expression significantly suppresses erastin-induced
ferroptosis.
3.5. Free iron depletion enhanced the resistance to ferroptosis in the FXN
knockdown cells
The above results suggested that FXN suppression dramatically in-
duced mitochondria dysfunction and activated the iron-starvation re-
sponse, thus accelerating erastin-challenged ferroptosis. Since ferritin
heavy chain (FTH) has ferroxidase activity and oxidizes Fe
2+
to
catalytically inactive Fe
3+
and plays a vital role in maintaining iron
homeostasis by storing iron in a soluble, non-toxic form. We, therefore,
hypothesized that overexpression of FTH might restore dysregulated
iron homeostasis in the FXN suppression HT-1080 cells. To test this
hypothesis, we reconstituted the expression of iron storage protein FTH
in FXN suppression HT-1080 cells and firstly explored the effects on the
cell proliferation and mitochondrial function. As expected, the FTH
overexpressed cells eliminated the proliferation arrest induced by FXN
suppression and accelerated the formation of cell colony, manifesting
that iron homeostasis regulation could modulate the injury derive from
FXN knockdown (Fig. 5A, B). The quantification of free iron level re-
vealed that FTH expression sensibly blocked the iron accumulation
stimulated by FXN suppression (Figs. 5C, D, S5A). Similarity, mito-
tracker staining also showed a significantly increased network of tu-
bule-shaped morphology, which characteristic for healthy and func-
tional mitochondria under the recombinant expression of FTH in FXN
suppression cells (Fig. 5E). Additionally, we evaluated the
Fig. 3. Suppression of FXN accelerated erastin in-
duced ferroptosis. (A) HT-1080 cells were exposed
to various concentrations of erastin for 12 h re-
spectively and followed by CCK8 assay. (B)
Propidium iodide positive cells were stained and
observed by fluorescent microscopy after erastin
exposure for 12 h. Scale bars: 30 μm. (C) CFDA-SE
probe (5 μM) stained HT-1080 cells were exposed to
2.5 μM erastin and cultured for 3 days, then sub-
jected to flow cytometry. (D) The morphological
changes of mitochondria were detected by trans-
mission electron microscopy (TEM) in the absence
or presence of erastin. Flow cytometry was per-
formed to measure lipid peroxides (E), cytoplasm
ROS (F), mitochondrial membrane potential(G)
after erastin treatment for 12 h in HT-1080 cells. (H)
Representative images of BODIPY staining in erastin
treated HT-1080 cells which observed by confocal
laser microscope. Scale bars: 75 μm. (I) HT-
1080 cells were treated with erastin (10 μM) with or
without small molecule inhibitor for 12 h and re-
spective cell viability was detected by CCK8. (fer-
roptosis inhibitor ferrostatin-1,0.5 μM; GSH, 1 mM;
apoptosis inhibitor Z-VAD-FMK, 4 μM; necroptosis
inhibitor Necrosulfonamide, 0.5 μM; autophagy in-
hibitor BafA1, 20 nM ). All histograms were re-
presented as mean ± SD.
★★
P < 0.01 versus control.
J. Du, et al. Redox Biology 32 (2020) 101483
8

mitochondrial DNA copy and found that FTH recombinant expression
could reverse the decrease of mitochondrial DNA copy (Fig. 5F). In
summary, these data emphasize the fact that FTH expression restores
mitochondrial impairment induced by FXN suppression.
Based on the observed results, we next studied whether FTH ex-
pression could rescue the ferroptosis induced by erastin in FXN sup-
pression cells. And the results showed suppression of FXN increased
erastin-induced cell death, while reconstituted expression of FTH sig-
nificantly restored the ferroptosis cell death (Fig. 6A). CCK8 assay was
used to evaluate the potency of FTH to restored erastin-induced cell
death and results showed FTH fully restored the erastin-induced phe-
notypes (Fig. 6B). Cells transduced with shFXN displayed noticeable
altered mitochondrial shape and size, including mitochondrial frag-
mentation, vacuolization and loss of cristae enlargement, whereas these
phenotypes were reversed by FTH overexpression through reducing
cellular iron even with the treatment with erastin (Fig. 6C).
In addition, overexpression of FTH was able to rescue the accu-
mulation of lipid peroxidation and cytoplasm ROS in the FXN knock-
down cells upon erastin exposure (Fig. 6D, E). Consistent with these
results, pharmacologically chelate free iron by DFO also prevented cell
death of FXN suppression cells to a similar extent as the control under
the treatment of erastin (Fig. 4I). In summary, our results emphasize the
decisive role of FXN in the regulation of cysteine deprivation-induced
ferroptosis through modulating iron homeostasis, which alleviating by
the iron depleted.
3.6. Knockdown of FXN inhibited the growth of xenograft in vivo
To evaluate the effect of FXN knockdown on proliferation in vivo,
FXN suppression HT-1080 cells with or without FTH recombinant ex-
pression were subcutaneously injected into nude mice. The growth
condition of xenograft tumors was traced, which manifested that FXN
silencing indeed blocked the proliferation of tumor in vivo, while
overexpression of FXN or FTH accelerated the tumor growth (Fig. 7A).
Further, H&E and IHC staining were performed and results showed that
there were large necrosis areas in FXN silenced tumor tissues, accom-
panied by nuclear condensation, fragmentation and a significantly de-
creased Ki-67 expression (Fig. 7B). Inversely FXN overexpression not
only maintained the tumor morphology but also promoted the expres-
sion of Ki-67. FTH expression was also detected by IHC staining and
Fig. 4. Enforced expression of FXN contributed to
ferroptosis resistance. (A) The proliferation rate was
analyzed by the CFDA-SE probe between the in-
dicated cells. (B) Quantitative PCR analysis was
performed to measure mtDNA copy number.
Primers specific for the HGB1 gene were used for
the determination of nuclear DNA (nDNA) and pri-
mers specific for ND-1 were used to detect mtDNA.
(C) Indicated HT-1080 cells were incubated with
different concentrations of erastin for 12 h and cell
viability was assayed by CCK8 assay. (D)
Intracellular Fe
2+
was measured by the staining of
RPA and photographed by the confocal microscope.
Scale bars: 75 μm. (E) MitoTracker Red labeled HT-
1080 cells were subjected to the confocal micro-
scope for observing the changes of mitochondrial
morphology. Scale bars: 75 μm. (F) The morpholo-
gical changes of mitochondria were detected by
transmission electron microscopy (TEM) in the ab-
sence or presence of erastin. Lower scale bars:
0.2 μm. (G-H) To assess lipid ROS production, HT-
1080 cells were treated with 5 μM erastin and
loaded with BODIPY C11 probe for 30 min followed
by flow cytometry measurement (G) and confocal
laser microscope (H). Scale bars: 75 μm. (I)
Indicated HT-1080 cells were treated with erastin
(10 μM) with or without small molecule inhibitor
and respective cell viability was detected by CCK8.
(ferrostatin-1,0.5 μM; GSH, 1 mM; iron chelator
DFO, 100 μM; Z-VAD-FMK, 4 μM;
Necrosulfonamide, 0.5 μM). All histograms were
represented as mean ± SD. (For interpretation of
the references to colour in this figure legend, the
reader is referred to the Web version of this article.)
★
P < 0.05,
★★
P < 0.01 versus control.
J. Du, et al. Redox Biology 32 (2020) 101483
9

consistent with the data in vitro that FXN knockdown robustly activated
iron starving stress and decreased the FTH expression, which sig-
nificantly restored by FXN overexpression. In addition, FTH expression
could also reverse the phenomenons exhibited in FXN knockdown xe-
nograft, including inhibited tumors growth and decreased Ki-67 ex-
pression. These observations lead to the hypothesis that FXN or the
downstream iron homeostasis associated protein can be developed as a
therapeutic target, leaving cancer cells at increased risk of ferroptotic
cell death.
4. Discussion
Iron is an essential metallic element for all eukaryotes with im-
portant functions in the synthesis of iron-sulfur cluster (Fe–S), heme
and other cofactors [25–27]. In mammalian cells, Iron homeostasis is
tightly regulated by iron regulatory proteins 1 (IRP1) and 2(IRP2)
through post-transcriptionally controlling iron metabolism genes via
the IRE-IRP mechanism. As shown in several studies, iron excess is
closely related with tumor genesis in some types of human cancers [28].
There is a phenomenon termed iron addiction, which is to say, cancer
cells do exhibit an enhanced dependence on iron relatively to healthy
cells to enable its growth [3,29]. Supersaturation ferric ion or iron
exposure may increase cancer risk. Like two sides of a coin, excess iron
loading may also lead to toxicity due to the ability of Fe participating in
the production of ROS via Fenton reactions, which may be more pro-
nounced in the highly redox-active mitochondrion. Therefore, the high
level of iron may act as a target for cancer therapy [8].
Ferroptosis is a recently identified form of regulated cell death that
is distinct from the other types of cell death at morphological, bio-
chemical, and genetic levels and efficiently recovered by ferrostatin-1,
GSH and DFO, et, al [4]. Recent study demonstrated that cancer cells
depended on high levels of the ISC biosynthetic enzyme NFS1, which
was indispensable for Fe–S clusters present in multiple cell essential
proteins upon exposure to oxidative damage. Suppression of NFS ro-
bustly activates the accumulation of iron and triggers ferroptosis co-
operates with cystine/glutamate antiporter inhibitor [30]. Our present
Fig. 5. Overexpress of FTH eliminated the pro-
liferation arrest and mitochondria dysfunction in-
duced by FXN suppression. (A-B) The ability of cell
proliferation was measured by CCK8 assay (A) and
plate clone formation assay(B) between the three
indicated cells. (C-D) To assess intracellular Fe
2+
,
cells were stained with RPA probe for 30 min and
photographed by a confocal laser microscope(C) or
subjected to flow cytometry (D). (E) Cells were
stained by MitoTracker Red and mitochondrial
morphology changes were observed under the con-
focal microscope. Scale bars: 75 μm. (F) The quan-
titative PCR analysis was performed to measure
mtDNA copy number. (For interpretation of the re-
ferences to colour in this figure legend, the reader is
referred to the Web version of this article.)
★
P < 0.05,
★★
P < 0.01 versus control.
J. Du, et al. Redox Biology 32 (2020) 101483
10

study demonstrated that FXN was a key regulator of ferroptosis by
modulating the iron homeostasis and mitochondrial function. Genetic
inhibition of FXN significantly repressed the proliferation, destroyed
mitochondrial morphology, impeded Fe–S cluster assembly and ex-
acerbated iron accumulation. Erastin further induced dramatic mi-
tochondria morphological damage in FXN suppressed cells such as en-
hanced fragmentation and vanished cristae, a phenotype as the
hallmarks of ferroptosis, and was consistent with early research. Vice
versa, enforced expression of FXN blocked the iron starvation response,
MMP hyperpolarization, mitochondrial fragmentation and erastin-in-
duced ferroptosis, confirming that the function of FXN shared beneficial
mechanisms for cell survival which acted at upstream of ferroptosis. In
particular, both pharmacological and genetic inhibition of iron starva-
tion stress also wholly restored the resistance to ferroptosis in FXN
knockdown cells. These results supporte a mechanistic hypothesis that
FXN deficiency accelerate erastin activated ferroptotic cell death.
FXN is a highly conserved protein localizes in the mitochondrial
matrix and participates in the biosynthesis of Fe–S cluster. A recent
study shows that FXN can activation NFS1 and accelerating a rate-
limiting sulfur transfer step of Fe–S cluster assembly by inducing an
unusual rearrangement of protein subunits in the de novo assembly
complex [15]. In line with the previous finding, suppression of NFS1
make cancer cell sensitive to ferroptosis in vitro and slow tumor growth
[30]. The main mechanism of FXN medicated ferroptosis may be the
breakdown of the ISC biosynthetic machinery. Mutation of human FXN
causes FRDA, a devastating, multi-systemic inherited degenerative
syndrome that affects 1 in 50,000 people. Although there is clear evi-
dence demonstrating that FRDA is associated with mitochondrial
dysfunction, mitochondrial iron accumulation, and increased oxidative
stress [31], the exact mechanism remains uncertain. Kevin Kemp [32]
and Irazusta V et al. [33] demonstrated that cells derived from FRDA
patients had deficiencies in defenses against oxidative stress and de-
creased in superoxide dismutase (SOD) activity. Therefore, Hongting
Zhao et al. [34] discovered that mitochondrion-targeted peptide SS-31,
a novel mitochondrion-targeted antioxidant, improved the function of
mitochondria in FRDA patient-derived cells and might potentially be a
new drug for the early treatment of FRDA. Furthermore, Amy Anzovino
et al. [23] revealed significant alterations in the cellular redox home-
ostasis were mediated by Nrf2 deficiency through increased cytosolic
Keap1 levels in FRDA mouse model. Similar to earlier studies, Tslil Ast
et al. [35] also verified that FXN depletion resulted in the activation of
ATF4-dependent integrated stress response and loss of anti-oxidant
NRF2 signaling, which was compromised under hypoxia. NRF2 is an
important transcription factor in regulating cellular redox homeostasis
through binding antioxidant response elements (AREs). It has been
reported NRF2 plays a critical role in mitigating lipid peroxidation
[36,37] and its inhibition reverses the resistance to ferroptosis [6,38].
We also identified the downregulation of NRF2 in FXN suppression
cells, which might be another factor giving rise to the cells sensitive to
ferroptosis. Despite the central role of FXN in oxidative metabolism, it
remains unclear whether it plays a central role in cancer progression. In
our work, we examined the expression of FXN based on The Cancer
Genome Atlas (TCGA) datasets and confirmed that the expression of
FXN was much higher in various cancer tissue. And high FXN expres-
sion level indicated poor survival of Sarcoma, ACC, BLCA, LAML pa-
tients. Overall, the present study exposes FXN as the molecular link
Fig. 6. FTH expression restored the resistance to
ferroptosis in FXN knockdown cells. (A) Indicated
cells were treated with erastin (0–10 μM) for 12 h
and cell viability was assayed by CCK8. (B) HT-
1080 cells were treated with erastin (10 μM) for
12 h with or without small molecule inhibitor and
respective cell viability was detected by CCK8. (C)
The morphological changes of mitochondria were
detected by transmission electron microscopy
(TEM) in the absence or presence of erastin. Lower
scale bars: 0.2 μm. (D) Cells were treated with 5 μM
erastin and loaded with DCFH-DA (4 μM) probe for
30 min followed by flow cytometry to assess cyto-
plasm ROS formation. (E) Indicated HT-1080 cells
were treated with erastin (5 μM) for 12 h. Levels of
lipid ROS production were detected with BODIPY
C11 (2.5 μM) probe by flow cytometry measure-
ment.
★★
P < 0.01 versus control.
J. Du, et al. Redox Biology 32 (2020) 101483
11

between the oxidative stress signaling and ferroptosis.
It is currently under debate whether mitochondria involved in fer-
roptosis. Dixon and his co-workers found that mtDNA depleted cancer
cell lines did not show a significant difference in ferroptosis sensitivity
[39]. Gaschler confirmed that ferrostatin analogs executed the anti-
death potency without located in mitochondria [40]. These observa-
tions provided evidence arguing against the potential involvement of
mitochondria in ferroptosis. By contrast, we evaluated mitochondria
from both its morphology and function. Abnormal mitochondrial
morphology with decreased or vanished mitochondria cristae was de-
tected in the FXN knockdown cells. The subsequent copy number of
mtDNA, encodes 13 polypeptides which play crucial roles in the OX-
PHOS, was significantly decreased following the knockdown of FXN. As
published before, ISCs undergo spontaneous degradation upon ex-
posure to oxygen and other oxidative stress conditions [41]. Upon FXN
suppression, breakdown of the ISC biosynthetic machinery cannot meet
the increased demand, ultimately leading to the reduced steady-state
levels of Fe–S cluster dependent proteins such as the Fe–S cluster
binding subunits of mitochondrial complex and aconitase. All the above
data were consistent with the finding FXN deficiency led to dramati-
cally decline of mitochondrial oxygen consumption rate, thus resulting
in the insufficient generation of ATP. The detailed investigation of
mitochondrial dysfunction provide novel insights into the role of mi-
tochondrial damage in ferroptosis.
It's interesting that FXN suppression stimulates the hyperpolariza-
tion of MMP, followed by the eventual collapse leading to cell death,
Fig. 7. Knockdown of FXN inhibited the growth of xenograft in vivo. (A) The tumor volumes were measured every 2 days and significant inhibition of tumor growth
was observed after FXN knockdown. Representative photographs showed the tumor size in mice after the mice were sacrificed. (B) H&E and IHC analysis for Ki67,
FXN and FTH in indicated tumor specimens. Scale bars: 50 μm. (C-E) The Specimens were scored by relative integrated optical density (IOD) value.
★
P < 0.05,
★★
P < 0.01 versus control.
J. Du, et al. Redox Biology 32 (2020) 101483
12

which is in accordance with previous study [35]. It has been reported
that excess hyperpolarization of MMP did not accelerate ATP produc-
tion, but help to the production of free radicals [22]. In addition, upon
pharmacological inhibition of GPX4 or GSH synthase, the downstream
components of the ferroptosis signal pathway, ferroptosis can still be
triggered independently of FXN depletion. It is likely that iron-starva-
tion stress and mitochondrial dysfunction, associated with FXN defi-
ciency, act at the upstream of ferroptosis, making FXN dispensable for
GPX4 inhibition induced ferroptosis. This is in general agreement with
earlier reports published by Yang et al., which revealed that mi-
tochondria was involved in the cysteine deprivation-induced ferroptosis
but not in glutathione peroxidase-4 (GPX4) inhibition-induced ferrop-
tosis [42]. Collectively, the oxidative stress caused by mitochondrial
dysregulation in FXN suppressed cells together with the defective uti-
lization of mitochondrial iron enhancing the erastin-induced ferrop-
tosis, which might potentially be a new drug target.
Declaration of competing interest
The authors declare that there is no conflict of interest.
Acknowledgments
This research was supported by National Science and Technology
Major Project for New Drug (No. 2017ZX301033), National Natural
Science Foundation of China (81971172), The Key Research and
Development Program of Zhejiang Province (WKJ-ZJ-1914), Zhejiang
Public Welfare Technology Application Research Project (Grant No.
LGF19H080006, LGF20H080005, LY18C090004, 2017C33091),
Medical and Health Science and Technology Project of Zhejiang
Province (No. 2019RC014, 2019RC115, 2018KY003, 2017KY006,
2017KY209), Zhejiang Students' Science and Technology Innovation
Activity Plan (No. 2019R410053).
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://
doi.org/10.1016/j.redox.2020.101483.
List of abbreviations
FXN frataxin
shRNA short hairpin small interfering RNA
PUFAs polyunsaturated fatty acids
FRDA Friedreich's ataxia
ISC Iron-sulfur cluster
ROS reactive oxygen species
DFO deferoxamine
GSH glutathione
TCGA The Cancer Genome Atlas
MMP mitochondrial membrane potential
OCR oxygen consumption rate
IRP1 iron responsive protein 1
FTH Ferritin heavy chain
TFR1 transferrin receptor 1
CFDA-SE carboxyfluorescein diacetate succinimidyl ester
Fer-1 ferrostatin-1
AREs antioxidant response elements
GPX4 glutathione peroxidase-4
PI propidium iodide
SOD superoxide dismutase
References
[1] S. Puig, L. Ramos-Alonso, A.M. Romero, M.T. Martinez-Pastor, The elemental role
of iron in DNA synthesis and repair, Metallomics 9 (2017) 1483–1500.
[2] K. Furuyama, K. Kaneko, Iron metabolism in erythroid cells and patients with
congenital sideroblastic anemia, Int. J. Hematol. 107 (2018) 44–54.
[3] D.H. Manz, N.L. Blanchette, B.T. Paul, F.M. Torti, S.V. Torti, Iron and cancer: recent
insights, Ann. N. Y. Acad. Sci. 1368 (2016) 149–161.
[4] Scott J. Dixon, Kathryn M. Lemberg, Michael R. Lamprecht, R. Skouta, Eleina
M. Zaitsev, Caroline E. Gleason, Darpan N. Patel, Andras J. Bauer, Alexandra
M. Cantley, Wan S. Yang, et al., Ferroptosis: an iron-dependent form of non-
apoptotic cell death, Cell 149 (2012) 1060–1072.
[5] P. A, I. Fl, M. Jm, C. Pj, Regulated necrosis in kidney ischemia-reperfusion injury,
Kidney Int. 96 (2019) 291–301.
[6] D. Shin, E.H. Kim, J. Lee, J.L. Roh, Nrf2 inhibition reverses resistance to GPX4
inhibitor-induced ferroptosis in head and neck cancer, Free Radical Biol. Med. 129
(2018) 454–462.
[7] Y. Mou, J. Wang, J. Wu, D. He, C. Zhang, C. Duan, B. Li, Ferroptosis, a new form of
cell death: opportunities and challenges in cancer, J. Hematol. Oncol. (2019),
https://doi.org/10.1186/s13045-019-0720-y.
[8] B. Hassannia, P. Vandenabeele, T. Vanden Berghe, Targeting ferroptosis to iron out
cancer, Canc. Cell 35 (2019) 830–849.
[9] T. Hirschhorn, B.R. Stockwell, The development of the concept of ferroptosis, Free
Radical Biol. Med. (2018), https://doi.org/10.1016/j.freeradbiomed.2018.09.043.
[10] M. Gao, P. Monian, Q. Pan, W. Zhang, J. Xiang, X. Jiang, Ferroptosis is an autop-
hagic cell death process, Cell Res. 26 (2016) 1021–1032.
[11] W. Hou, Y. Xie, X. Song, X. Sun, M.T. Lotze, H.J. Zeh 3rd, R. Kang, D. Tang,
Autophagy promotes ferroptosis by degradation of ferritin, Autophagy 12 (2016)
1425–1428.
[12] J. Du, T. Wang, Y. Li, Y. Zhou, X. Wang, X. Yu, X. Ren, Y. An, Y. Wu, W. Sun, DHA
inhibits proliferation and induces ferroptosis of leukemia cells through autophagy
dependent degradation of ferritin, Free Radical Biol. Med. (2018), https://doi.org/
10.1016/j.freeradbiomed.2018.12.011.
[13] Y. Yu, Y. Xie, L. Cao, L. Yang, M. Yang, M.T. Lotze, H.J. Zeh, R. Kang, D. Tang, The
ferroptosis inducer erastin enhances sensitivity of acute myeloid leukemia cells to
chemotherapeutic agents, Mol. Cell. Oncol. 2 (2015) e1054549.
[14] C. Jc, L.B. Ne, Redox-sensing iron-sulfur cluster regulators, Antioxidants Redox
Signal. 29 (2018) 1809–1829.
[15] S. Patra, D.P. Barondeau, Mechanism of activation of the human cysteine de-
sulfurase complex by frataxin, Proc. Natl. Acad. Sci. U.S.A. 116 (2019)
19421–19430.
[16] S. Pp, V. V, B. K, D.S. P, Iron-sulfur protein assembly in human cells, Rev. Physiol.
Biochem. Pharmacol. 174 (2018) 25–65.
[17] C. M, P. H, G. A, K. H, D. A, L. M, F. K, D. P, K. M, Inactivation of the Friedreich
ataxia mouse gene leads to early embryonic lethality without iron accumulation,
Hum. Mol. Genet. 9 (2000) 1219–1226.
[18] F. Lupoli, T. Vannocci, G. Longo, N. Niccolai, A. Pastore, The role of oxidative stress
in Friedreich's ataxia, FEBS Lett. 592 (2018) 718–727.
[19] M.G. Cotticelli, S. Xia, D. Lin, T. Lee, L. Terrab, P. Wipf, D. Huryn, R. Wilson,
Ferroptosis as a novel therapeutic target for Friedreich's ataxia, J. Pharmacol. Exp.
Therapeut. (2019), https://doi.org/10.1124/jpet.118.252759.
[20] U. Mühlenhoff, J.A. Stadler, N. Richhardt, A. Seubert, T. Eickhorst, R.J. Schweyen,
R. Lill, G. Wiesenberger, A specific role of the yeast mitochondrial carriers Mrs3/4p
in mitochondrial iron acquisition under iron-limiting conditions, J. Biol. Chem. 278
(2003) 40612–40620.
[21] Z. L, B. U, R. Jl, An efficient one-step site-directed and site-saturation mutagenesis
protocol, Nucleic Acids Res. 32 (2004), https://doi.org/10.1093/nar/gnh110.
[22] M. Hüttemann, I. Lee, A. Pecinova, P. Pecina, K. Przyklenk, J.W. Doan, Regulation
of oxidative phosphorylation, the mitochondrial membrane potential, and their role
in human disease, J. Bioenerg. Biomembr. 40 (2008) 445–456.
[23] A. Anzovino, S. Chiang, B.E. Brown, C.L. Hawkins, D.R. Richardson, M.L. Huang,
Molecular alterations in a mouse cardiac model of Friedreich ataxia: an impaired
Nrf2 response mediated via upregulation of Keap1 and activation of the Gsk3beta
Axis, Am. J. Pathol. 187 (2017) 2858–2875.
[24] Y. Shan, R.A. Schoenfeld, G. Hayashi, E. Napoli, T. Akiyama, M. Iodi Carstens,
E.E. Carstens, M.A. Pook, G.A. Cortopassi, Frataxin deficiency leads to defects in
expression of antioxidants and Nrf2 expression in Dorsal Root Ganglia of the
Friedreich's Ataxia YG8R mouse model, Antioxidants Redox Signal. 19 (2013)
1481–1493.
[25] S. Toyokuni, F. Ito, K. Yamashita, Y. Okazaki, S. Akatsuka, Iron and thiol redox
signaling in cancer: an exquisite balance to escape ferroptosis, Free Radical Biol.
Med. 108 (2017) 610–626.
[26] H. Harigae, K. Hino, S. Toyokuni, Iron as soul of life on earth revisited: from che-
mical reaction, ferroptosis to therapeutics, Free Radical Biol. Med. 133 (2019) 1–2.
[27] D.A. Stoyanovsky, Y.Y. Tyurina, I. Shrivastava, I. Bahar, V.A. Tyurin,
O. Protchenko, S. Jadhav, S.B. Bolevich, A.V. Kozlov, Y.A. Vladimirov, et al., Iron
catalysis of lipid peroxidation in ferroptosis: regulated enzymatic or random free
radical reaction? Free Radical Biol. Med. 133 (2019) 153–161.
[28] S.V. Torti, F.M. Torti, Iron and cancer: more ore to be mined, Nat. Rev. Canc. 13
(2013) 342, https://doi.org/10.1038/nrc3495.
[29] C. Raggi, E. Gammella, M. Correnti, P. Buratti, E. Forti, J.B. Andersen, G. Alpini,
S. Glaser, D. Alvaro, P. Invernizzi, Dysregulation of iron metabolism in cholangio-
carcinoma stem-like cells, Sci. Rep. 7 (2017) 17667.
[30] S.W. Alvarez, V.O. Sviderskiy, E.M. Terzi, T. Papagiannakopoulos, A.L. Moreira,
S. Adams, D.M. Sabatini, K. Birsoy, R. Possemato, NFS1 undergoes positive selection
in lung tumours and protects cells from ferroptosis, Nature (2017), https://doi.org/
10.1038/nature24637.
[31] R. Abeti, M.H. Parkinson, I.P. Hargreaves, P.R. Angelova, C. Sandi, M.A. Pook,
P. Giunti, A.Y. Abramov, Mitochondrial energy imbalance and lipid peroxidation
cause cell death in Friedreich's ataxia, Cell Death Dis. 7 (2016) e2237, https://doi.
J. Du, et al. Redox Biology 32 (2020) 101483
13

org/10.1038/cddis.2016.111.
[32] K. Kemp, R. Dey, A. Cook, N. Scolding, A. Wilkins, Mesenchymal stem cell-derived
factors restore function to human frataxin-deficient cells, Cerebellum 16 (2017)
840–851.
[33] I. V, O. E, M.-C. A, C. E, R. J, T. J, Yeast frataxin mutants display decreased su-
peroxide dismutase activity crucial to promote protein oxidative damage, Free
Radical Biol. Med. 48 (2010) 411–420.
[34] H. Zhao, H. Li, S. Hao, J. Chen, J. Wu, C. Song, M. Zhang, T. Qiao, K. Li, Peptide SS-
31 upregulates frataxin expression and improves the quality of mitochondria: im-
plications in the treatment of Friedreich ataxia, Sci. Rep. 7 (2017) 9840, https://
doi.org/10.1038/s41598-017-10320-2.
[35] T. Ast, J.D. Meisel, S. Patra, H. Wang, R.M.H. Grange, S.H. Kim, S.E. Calvo,
L.L. Orefice, F. Nagashima, F. Ichinose, et al., Hypoxia rescues frataxin loss by re-
storing iron sulfur cluster biogenesis, Cell 177 (2019) 1507–1521.
[36] M. Dodson, R. Castro-Portuguez, D.D. Zhang, NRF2 plays a critical role in miti-
gating lipid peroxidation and ferroptosis, Redox Biol. (2019), https://doi.org/10.
1016/j.redox.2019.101107.
[37] X. Sun, Z. Ou, R. Chen, X. Niu, D. Chen, R. Kang, D. Tang, Activation of the p62-
Keap1-NRF2 pathway protects against ferroptosis in hepatocellular carcinoma cells,
Hepatology 63 (2016) 173–184.
[38] J.L. Roh, E.H. Kim, H. Jang, D. Shin, Nrf2 inhibition reverses the resistance of
cisplatin-resistant head and neck cancer cells to artesunate-induced ferroptosis,
Redox Biol. 11 (2017) 254–262.
[39] D. Sj, L. Km, L. Mr, S. R, Z. Em, G. Ce, P. Dn, B. Aj, C. Am, Y. Ws, et al., Ferroptosis:
an iron-dependent form of nonapoptotic cell death, Cell 149 (2012) 1060–1072.
[40] G. Mm, H. F, F. H, L. A, M. W, S. Br, Determination of the subcellular localization
and mechanism of action of ferrostatins in suppressing ferroptosis, ACS Chem. Biol.
13 (2018) 1013–1020.
[41] D.R. Crooks, N. Maio, A.N. Lane, M. Jarnik, R.M. Higashi, R.G. Haller, Y. Yang,
T.W. Fan, W.M. Linehan, T.A. Rouault, Acute loss of iron-sulfur clusters results in
metabolic reprogramming and generation of lipid droplets in mammalian cells, J.
Biol. Chem. 293 (2018) 8297–8311.
[42] M. Gao, J. Yi, J. Zhu, A.M. Minikes, P. Monian, C.B. Thompson, X. Jiang, Role of
mitochondria in ferroptosis, Mol. Cell. 73 (2019) 354–363.
J. Du, et al. Redox Biology 32 (2020) 101483
14