Structure-activity relationship studies of phenothiazine derivatives as a new class of ferroptosis inhibitors together with the therapeutic effect in an ischemic stroke model

Article abstract of DOI:10.1016/j.ejmech.2020.112842

Ferroptosis is a new type of programmed cell death discovered recently and has been demonstrated to be involved in a number of human diseases such as ischemic stroke. Ferroptosis inhibitors are expected to have potential to treat these diseases. Herein, we report the identification of promethazine derivatives as a new type of ferroptosis inhibitors. Structure-activity relationship (SAR) analyses led to the discovery of the most potent compound 2-(1-(4-(4-methylpiperazin-1-yl)phenyl)ethyl)-10H-phenothiazine (51), which showed an EC₅₀ (half maximal effective concentration) value of 0.0005 μM in the erastin-induced HT1080 cell ferroptosis model. In the MCAO (middle cerebral artery occlusion) ischemic stroke model, 51 presented an excellent therapeutic effect. This compound also displayed favorable pharmacokinetic properties, in particular, a good ability to permeate the blood-brain barrier. Overall, 51 could be a promising lead compound for the treatment of ferroptosis related diseases and deserves further investigations.

Full text of DOI:10.1016/j.ejmech.2020.112842

Journal Pre-proof
Structure-activity relationship studies of phenothiazine derivatives as a new class of
ferroptosis inhibitors together with the therapeutic effect in an ischemic stroke model
Wei Yang, Xiaolong Liu, Chunli Song, Sen Ji, Jianhong Yang, Yang Liu, Jing You, Jie
Zhang, Shenzhen Huang, Wei Cheng, Zhenhua Shao, Linli Li, Shengyong Yang
PII:
S0223-5234(20)30814-X
DOI:
https://doi.org/10.1016/j.ejmech.2020.112842
EJMECH 112842
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 8 June 2020
Revised Date: 23 August 2020
Accepted Date: 9 September 2020
Please cite this article as: W. Yang, X. Liu, C. Song, S. Ji, J. Yang, Y. Liu, J. You, J. Zhang, S. Huang,
W. Cheng, Z. Shao, L. Li, S. Yang, Structure-Activity Relationship Studies of Phenothiazine Derivatives
as A New Class of Ferroptosis Inhibitors together with The Therapeutic Effect in An Ischemic Stroke
Model, European Journal of Medicinal Chemistry, https://doi.org/10.1016/j.ejmech.2020.112842.
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition
of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of
record. This version will undergo additional copyediting, typesetting and review before it is published
in its final form, but we are providing this version to give early visibility of the article. Please note that,
during the production process, errors may be discovered which could affect the content, and all legal
disclaimers that apply to the journal pertain.
©
2020 Elsevier Masson SAS. All rights reserved.

Graphical Abstract

Structure-Activity Relationship Studies of Phenothiazine Derivatives as A New Class
of Ferroptosis Inhibitors together with The Therapeutic Effect in An Ischemic
Stroke Model
†
,
⊥
†,
⊥
§,
⊥
†,
⊥
§
†
Wei Yang , Xiaolong Liu , Chunli Song , Sen Ji , Jianhong Yang , Yang Liu ,
†
†
†
†
†
§
Jing You , Jie Zhang , Shenzhen Huang , Wei Cheng , Zhenhua Shao , Linli Li ,
†
,
Shengyong Yang *
†
State Key Laboratory of Biotherapy and Cancer Center, West China Hospital, Sichuan
University, Chengdu, Sichuan 610041, China.
§
Key Laboratory of Drug Targeting and Drug Delivery System of the Education Ministry,
West China School of Pharmacy, Sichuan University, Chengdu, Sichuan 610041, China.
*
To whom correspondence should be addressed. Tel:+86-28-85164063; Fax:
+
86-28-85164060; E-mail:yangsy@scu.edu.cn.
⊥
These authors contributed equally to this work.

ABSTRACT
Ferroptosis is a new type of programmed cell death discovered recently and has been
demonstrated to be involved in a number of human diseases such as ischemic stroke.
Ferroptosis inhibitors are expected to have potential to treat these diseases. Herein, we
report the identification of promethazine derivatives as a new type of ferroptosis
inhibitors. Structure-activity relationship (SAR) analyses led to the discovery of the most
potent compound 2-(1-(4-(4-methylpiperazin-1-yl)phenyl)ethyl)-10H-phenothiazine (51),
which showed an EC50 (half maximal effective concentration) value of 0.0005 µM in the
erastin-induced HT1080 cell ferroptosis model. In the MCAO (middle cerebral artery
occlusion) ischemic stroke model, 51 presented an excellent therapeutic effect. This
compound also displayed favorable pharmacokinetic properties, in particular, a good
ability to permeate the blood-brain barrier. Overall, 51 could be a promising lead
compound for the treatment of ferroptosis related diseases and deserves further
investigations.
Key words: Ferroptosis inhibitor; Structure-activity relationship; Ischemic stroke;
Promethazine derivatives

1
. INTRODUCTION
Ferroptosis is a newly recognized form of regulated cell death caused by
iron-dependent lipid peroxidation[1-3]. It is genetically and biochemically distinct from
other forms of regulated cell death such as apoptosis[4-6], necroptosis[7,8] and
pyroptosis[9,10]. Ferroptosis is triggered by inactivation of cellular glutathione
(GSH)-dependent antioxidant defenses, resulting in unchecked lipid peroxidation and
eventual cell death[2,11]. Typical characteristics of ferroptosis include the reduced
volume of mitochondria with condensed mitochondrial membrane density, and reduction
or vanishing of mitochondria crista, as well as outer mitochondrial membrane
rupture[12,13]. Ferroptosis has been demonstrated to be involved in many pathological
processes, such as stroke[14,15], Parkinson’s disease[16,17], periventricular
leukomalacia (PVL)[18], Huntington’s disease[18,19], Alzheimer’s diseases[3,20],
ischemia-reperfusion injury[21,22], cardiovascular[23-25], renal failure[26-28], liver
injury[26,29], hear injury[30], cardiomyopathy[31] and diabetes[32]. Ferroptosis
inhibitors are expected to treat these kinds of diseases.
Though iron chelators can prevent ferroptotic cell death, possible toxicity limits their
clinical application because iron is an essential element of life and widely exists in the
body. Development of non-chelating ferroptosis inhibitors has thus attracted much
attention recently in both industry and academia. Since Dixon et al found the phenomena
of ferroptosis in 2012[1], a number of non-chelating ferroptosis inhibitors have been
publicly reported, including Fer-1[1,18], trolox[1,18], cycloheximide[1], SRS11-92[18],
SRS16-86[21], liproxstatin-1[26], baicalein[33], zileuton[34], 49[35], ebselen[1,36],
U0126[1,37] and butylated hydroxytoluene[1,38] (see Figure 1). However, many of these

compounds suffer from poor activity, unfavorable pharmacokinetic properties, or toxicity,
hence being not suitable for clinical use. Discovery of more potent ferroptosis inhibitors
with favorable pharmacokinetic propertiesand low toxicity is in urgent need.
Figure 1. Chemical structures of previously reported non-chelating ferroptosis inhibitors
In an attempt to discover new potent ferroptosis inhibitors, we used the
erastin-induced HT1080 cell ferroptosis model (a detailed description regarding the
screening model see Experimental Section) to screen a bioactive compound library from
Selleck (America) containing about 2,000 compounds. From this screening, we found
two phenothiazine derivatives, promethazine and phenothiazine, promethazine was used
as sedatives in clinic[39,40], to protect cells from ferroptotic cell death with an EC50 (half
maximal effective concentration) value of 0.264 and 0.049 μM, respectively (see Figure
2
). According to literature review, phenothiazine and its analogues are antioxidants that
can inhibit lipid peroxidation[41-43]. Besides, Shah et al[44] reported a number of
diarylamine radical-trapping antioxidants as inhibitors of ferroptosis, which include

phenothiazine. Based on these studies, we shall describe the structural optimization and
SAR analyses of the promethazine derivatives, and the therapeutic effect of the most
potent compound on an ischemic stroke model, as well as the mechanism of ferroptosis
inhibition and a preliminary evaluation of drugability.
Figure 2. Chemical structures of promethazine (left) and phenothiazine (right).
2
. CHEMSTRY
For the structural optimization and SAR analyses, a total of 71 new phenothiazine
derivatives (1-71) were synthesized. Additionally, we also purchased 24 phenothiazine
derivatives (P1-P24) from commercial compound providers SPECS and ChemDiv. This
section describes the chemical syntheses of the 71 new phenothiazine derivatives.
Synthetic routes of compounds 1 and 2 are depicted in Scheme 1. Reaction of
phenothiazine with 1-bromopropane and 2-ethylhexyl bromide, respectively, provided
molecules 1 and 2,.
a
Scheme 1. Synthetic Routes for Compounds 1 and 2
a
Reagents and conditions: (a) NaH, DMF, 1 h, 0°C.

Scheme 2 outlines the synthetic routes for compounds 3-4 and 21. Reduction of
-cyano phenothiazine with diisobutylaluminium hydride (DIBAL) gave compound 3,
2
followed by further reduction to produce compound 4. Reaction of
0H-phenothiazine-2-carbaldehyde with 4-methylbenzenesulfonhydrazide gave
1
intermediate a1, and a1 was coupled with benzyl boric acid to provide compound 21.
a
Scheme 2. Synthetic Routes for Compounds 3−4 and 21
a
Reagents and conditions: (b) DIBAL, CH Cl , -78℃ ; 16 h, rt; (c) NaBH CN, THF, rt; (d)
2
2
3
4
-methylbenzenesulfonhydrazide, AcOH, MeOH, 65℃ ; (e) benzyl boric acid, K
2 3
CO ,
dioxane, reflux.
Compounds 5-11 and 22 were prepared following the route outlined in Scheme 3.
Treatment of 2-acetyl phenothiazine with Et SiH resulted in target compound 5. Reaction
3
of 2-acetyl phenothiazine with corresponding substituted acehydrazide or 4-methyl
benzene sulfonyl hydrazine gave compounds 6-11. Commercially available 2-acetyl
phenothiazine was reduced to generate intermediate a2. A condensation reaction between
a2 and benzoyl chloride produced compound 22.
a
Scheme 3. Synthetic Routes for Compounds 5−11 and 22

a
Reagents and conditions: (f) Et
3
SiH, CH
2
Cl
2
, rt, 20 h, H
2
SO
4
, NH
4
F; (g) K
2
CO
3
, dioxane,
reflux; (h) NaBH , THF, rt; (i) Et
4
3
N, CH Cl
2
2
, rt.
Scheme 4 illustrates the synthetic routes of compounds 12-20 and 23-71.
Compounds 12-20 were prepared by reacting 10 with various secondary amines in the
presence of 1,4-diazabicyclo[2.2.2]octane-1,4-diium-1,4-disulfinate (DABSO) in
dimethyl sulfoxide (DMSO). Compound 10 was coupled with various substituted benzyl
boric acid to produce 23-71.
a
Scheme 4. Synthetic Routes for Compounds 12-20 and 23-71
a
2 3
Reagents and condition: (j) secondary amine, DABSO, DMSO, 100℃ , Ar; (k) K CO ,
dioxane, reflux.

3
3
3
. s AND DISCUSSIONS
.1. Structure-activity relationship (SAR) analyses of phenothiazine derivatives
1
.1.1. Impact of substitution at the 10-position of phenothiazine (R )
To explore the possible influence of the 10-position’s substitution of phenothiazine
1
(
R ), we purchased 18 phenothiazine derivatives (P1-P18) with different substituent
1
groups at R from SPECS and ChemDiv, and synthesized 2 new compounds (1 and 2);
the substituent groups cover short and long alkyl/aryl chains, as well as hydrophobic and
hydrophilic fragments. Chemical structures and bioactivities of these compounds are
displayed in Table 1. Most of these compounds demonstrated very weak or no activity
except P3 and P4, which exhibited comparable activity as promethazine but obviously
lower potency than the unsubstituted phenothiazine. These results indicate that the
introduction of any substituents at the 10-position of phenothiazine is not favored for the
activity of ferroptosis inhibition.
Table 1. Chemical Structures and Bioactivities of Compounds P1-P18, 1-2, Promethazine, and
Phenothiazine.
1
a
1
a
Compound
P1
R
EC (µM)
Compound
P11
R
EC (µM)
5
0
50
>10
>10
>10
P2
P3
P12
P13
4.419±1.284
4.041±1.037
0.145±0.002

P4
P5
P6
P7
0.439±0.083
>10
P14
P15
P16
P17
4.305±1.233
5.241±1.231
>10
4.467±0.937
>10
>10
P8
P9
1.132±0.073
2.761±1.202
P18
1
>10
>10
P10
1.154±0.229
0.264±0.001
2
>10
promethazine
phenothiazine
0.049±0.009
a
Each compound was tested in triplicate; the data are presented as the means ± SD.
3
.1.2. Influence of the 2-position substitution of phenothiazine
2
We then explored the effect of 2-position (R ) substitution of phenothiazine with the
1
0-position unsubstituted (see the right of Figure 2). To this end, we synthesized 21 new
compounds (3-23) and purchased 6 phenothiazine derivatives (P19-P24) from Specs and
ChemDiv. Chemical structures and bioactivities of these compounds are shown in Table 2.
2
Compounds P19-P24 and 3-5 with smaller substituents at R presented considerable
activity of ferroptosis inhibition, but their potencies do not exceed that of phenothiazine
2
except P22 and 5. Compounds 6-23, which contain longer or larger substituents at R ,
also displayed considerable activity. These results prove that introduction of the group at
the 2-position of phenothiazine is clearly favorable. Among them, 23 bearing a
2
1
-phenylethyl group at R , is the most potent compound, whose EC value is 0.004 μM.
5
0

Table 2. Chemical Structures and Bioactivities of Compounds P19-P24 and 3-23.
2
a
2
a
Compound
Fer-1
R
EC50 (µM)
Compound
10
R
EC50 (µM)
0.088±0.002
0.209±0.089
0.028±0.001
0.005±0.001
P19
P20
-CN
11
12
-CF
3
0.047±0.002
0.195±0.014
0.080±0.006
0.080±0.011
P21
P22
P23
13
14
15
0.038±0.018
1.748±0.263
0.057±0.020
0.078±0.025
P24
3
0.064±0.023
0.142±0.016
16
17
0.067±0.009
0.034±0.001
4
5
6
0.063±0.004
0.030±0.003
0.017±0.005
18
19
20
0.025±0.002
0.032±0.003
0.035±0.001

7
8
9
0.018±0.0007
0.065±0.003
0.071±0.005
21
22
23
0.017±0.001
0.013±0.008
0.004±0.001
a
Each compound was tested in triplicate; the data are presented as the means ± SD.
3
.1.3. Effect of various substituents at the phenyl ring of 23
The preceding SAR studies led to the identification of compound 23, which
displayed the highest potency with an EC50 value of 0.004 µM. To further improve the
potency of 23 and examine the possible effect of various substituents on the phenyl ring
of 23, a series of compounds (24-53) with different substituents at 3- or 4- positions of
the phenyl ring were synthesized. Firstly, compounds 24-37, containing a single
3
substituent at the 3-position (R ) of phenyl ring, displayed excellent bioactivity with EC50
less than 0.1 µM (Table 3). Among them, introduction of simple groups such as fluorine,
chlorine and hydroxyl (24-26) maintained inhibitory potency compared to 23, while
introduction of relatively bulk groups (33-35) led to a decrease of potency. Secondly,
4
compounds 38-53 with a single substituent at the 4-position (R ) of phenyl ring also
displayed excellent activity. Compounds 38, 40, 41 and 43, which contained fluorine,
cyano, trifluoromethyl and isopropyl, respectively, exhibited similar activity compared
with compound 23. Of special note is 51 bearing a methylpiperazine, which presented the
most potent activity (EC : 0.0005 µM).
5
0
Table 3. Chemical Structures and Bioactivities of Compounds 24-53.

3
a
4
a
Compound
R
EC50 (µM)
Compound
38
R
EC50(µM)
2
4
-F
0.006±0.001
-F
0.006±0.001
2
2
5
6
-Cl
0.005±0.001
0.005±0.001
39
40
-Cl
0.016±0.001
0.006±0.001
-OH
-CN
2
2
2
3
3
7
8
9
0
1
-CN
0.015±0.002
0.005±0.003
0.018±0.003
0.012±0.001
0.014±0.002
41
42
43
44
45
-CF
3
0.006±0.001
0.020±0.001
0.006±0.001
0.073±0.012
0.009±0.004
3
3
3
2
3
4
0.015±0.001
0.025±0.004
0.017±0.001
46
47
48
0.064±0.002
0.017±0.001
0.017±0.001
3
3
3
5
6
7
0.052±0.014
0.014±0.002
0.009±0.002
49
50
51
0.062±0.001
0.014±0.002
0.0005±0.0001

Fer-1
0.088±0.002
0.049±0.009
52
53
0.013±0.001
0.039±0.002
phenothiazine
a
Each compound was tested in triplicate; the data are presented as the means ± SD.
Finally, we examined the effect of di-substitution of phenyl ring on the bioactivity.
As shown in Table 4, 54 and 57 displayed good activity, but other compounds with dual
3
4
5
substituents are still less potent than 54 and 57. Especially, at the R , R or R position,
we observed that addition of benzyloxy resulted in a decrease in potency than those
containing other substituents. Overall, the structural optimization and SAR analyses
offered the most potent phenothiazine derivative 51, which exhibited an EC50 value of 0.5
nM. Besides, we performed the chiral resolution of 51 and tested the EC50 of the two
conformations (see showrting information Figure S1), as it turned out that there was no
signification difference between the two conformations and the cell activity of 51. To the
best of our knowledge, 51 is also the most potent compound among all the publicly
reported ferroptosis inhibitors. Further studies will be carried out on compound 51.
Table 4. Chemical Structures and Bioactivities of Compounds 54-71.
3
4
5
a
Compound
R
R
R
EC (µM)
50
5
5
5
4
5
6
-Cl
-Cl
-F
-F
-H
-H
-H
0.005±0.001
0.016±0.005
0.019±0.003
-Cl
-Cl

5
5
5
6
7
8
9
0
-H
-H
-H
-H
0.007±0.001
0.022±0.003
0.019±0.001
0.025±0.002
-F
6
6
1
2
-Cl
-H
-H
0.022±0.002
0.017±0.001
-Cl
-Cl
-CF
3
6
6
3
4
-H
-H
0.021±0.006
0.071±0.023
6
5
-F
-H
0.035±0.001
6
6
6
7
-Cl
-H
-H
-Cl
0.017±0.001
0.021±0.006
6
6
8
9
-H
-H
0.013±0.001
0.024±0.002
-Cl
7
7
0
1
-F
-F
-H
-H
0.030±0.004
0.060±0.001
a
Each compound was tested in triplicate; the data are presented as the means ± SD.
3
.2. 51 specifically inhibits the ferroptotic cell death, not other forms of programmed
cell death

We firstly tested the cell activity of 51 in erastin-induced ferroptosis and compared
with known ferroptosis inhibitors, the results indicated that 51 displayed better
anti-ferroptosis activity (see Figure 3A and supporting information Table S1). To examine
the specificity of compound 51 to rescue the ferroptotic cell death, we tested the effect of
5
1 on the other three typical forms of programmed cell death, including apoptosis,
necroptosis, and pyroptosis. As shown in Figure 3B, 51 could not rescue the apoptotic
cell death in the staurosporine (STS)-induced HT1080 cell model. In the
TNFα/Smac-mimetic/ z-VAD-FMK (TSZ)-induced HT29 cell necroptosis model[7], 51
as well as Fer-1 did not exhibit protection effect, but necrostatin-1 (Nec-1), which is a
known necroptosis inhibitor, could efficiently protect cells from necroptosis (see Figure
3
C). In the nigericin-induced THP-1 cell pyroptosis model[10], 51 again did not display
protection activity, while the known pyroptosis inhibitor MCC950 had a good protection
effect (see Figure 3D). Collectively, these results demonstrated that 51 could efficiently
inhibit the ferroptotic cell death, but had no effect on other forms of programmed cell
death including apoptosis, necroptosis, and pyroptosis.

Figure 3. 51 exhibited the ability to inhibit cell ferroptosis, but no activity on other forms
of regulated cell death. (A) The dose-response curves of 51 and Fer-1 in inhibiting erastin
(10 μM)-induced ferroptosis. (B) The effects of compounds 51 (10 μM) and Fer-1 (10 μM)
on staurosporine (STS)-induced apoptosis. (C) Necroptosis of HT29 cells was induced by
TNFα (10 ng/mL), Smac-mimetic (100 nM), and Z-VAD-FMK (20 μM). The
concentrations used for Nec-1, Fer-1 and 51 are 10 μM. (D) Pyroptosis of THP-1 cells
was induced by phorbol-12-myristate-13-acetate (PMA), lipopolysaccharide (LPS), and
nigericin. The concentrations used for MCC950, Fer-1 and 51 are 10 μM. Columns, mean
(n=3); bars, SD. ns, not significant; ***, p<0.001.
3
.3. 51 inhibited cell ferroptosis in a cell-independent manor
To examine whether the protective effect of compound 51 on ferroptosis is cell line-
dependent or not, we used the other three cell lines, namely BxPC-3 (a pancreatic cancer

cell line), Calu-1 (a lung cancer cell line), and MDA-MB-231 (a triple negative breast
cancer cell line) to induce ferroptosis and test the effect of 51. As shown in Figure 4A-C,
5
1 significantly protected BxPC-3, Calu-1, and MDA-MB-231 cells from ferroptosis
induced by erastin with EC values of 2.925 nM, 1.879 nM, and 0.424 nM, respectively.
50
These results point out that the protective effect of compound 51 on ferroptosisis is
cell-independent.
Figure 4. 51 inhibited ferroptosis induced by erastin (10 μM) in Bxpc-3 cells (A), Calu-1
cells (B), and MDA-MB-231 cells (C).
3
.4. Underlying mechanism by which 51 protects cells from ferroptosis
We sought to explore the possible mechanism by which 51 protects cells from
ferroptosis. The original hit compounds (promethazine and phenothiazine) are known as
antagonists of G-protein coupled receptors (GPCR), and primarily target DRD2L,
ADRA1A, HTR2A, HTR2C, CHRM3, ADRA1B, and H1. We, therefore, in the first step,
examined whether 51 also targets these GPCRs by fluorescent imaging plate reader
(
FLIPR) assay[36]. The results exhibited that 51 presented weak or no activity against
these GPCRs (see supporting information Figure S2). Then, western blot was used to
examine the changes in key signal proteins of cell ferroptosis, including
-
cystine/glutamate antiporter system Xc (xCT), extracellular regulated protein kinases
(ERK), lipoxygenases (LOX), and glutathione peroxidase 4 (GPX4) expression levels

upon 51 treatment. As shown in Figure 5A, erastin treatment resulted in a significant
decrease in the expressions of GPX4. 51 and the positive control Fer-1 did not affect the
expression level of GPX4. Further more, 51 displayed no inhibition on the expression of
xCT, ERK, and LOX, which are key factors in lipid oxidation (Figure 5A). The
malondialdehyde (MDA) method was adopted to detect the lipid oxidation level; MDA is
a product of the oxidation of fatty acids. As displayed in Figure 5B, 51 markedly
decreased the erastin-induced accumulation of MDA, proving again that 51 reduces the
lipid oxidation. Then we measured the GSH level by the dithiodinitrobenzoic acid
method; in this method, GSH can react with dithiodinitrobenzoic acid to form
2
-nitro-5-mercaptobenzoic acid, and 2-Nitro-5- mercaptobenzoic acid has a light
absorption at the wavelength 412 nm, which could be used to detect the GSH level. As
shown in Figure 5C, 51 slightly elevated the GSH level, which was significantly reduced
by erastin.
The fluorescent probe 2',7'-dichlorofluorescein diacetate (DCFH-DA) and BODIPY
5
81/591 C11 assay were used to measure the cytosolic ROS and lipid ROS level in
HT1080 and Calu-1 cells, respectively. The results disclosed that 51 significantly
prevented erastin-induced accumulation of cytosolic ROS and lipid ROS (Figure 6A, 6B
and 6C). Moreover, the 2,2-diphenyl-1-picrylhydrazyl (DPPH) assay was utilized to
examine the intrinsic antioxidant potential of 51. As shown in Figure 6D, 51 readily
oxidized the stable radical DPPH under cell-free conditions, indicating a good antioxidant
potential. From here, we know that 51 can inhibit lipid peroxide. Does 51 also inhibit
other forms of ROS production? To answer this question, we tested the impact of 51 on
2 2
other forms of oxidative death, including rotenone- and H O -induced cell deaths. The

results revealed that 51 and Fer-1 did not exhibit inhibition activity in rotenone- and
H O -induced cell deaths (Figure 6E and 6F). Finally, iron chelator detecton assay with a
2
2
spectrophotometric reagent[45] showed that 51 is not an iron chelator (see supporting
information Figure S3). Overall, we can conclude that 51 specifically prevented
erastin-induced cell ferroptosis mainly through its antioxidant mechanism.
Figure 5. 51 did not block the expression of key protein of ferroptosis. (A) Western blot
analyses reflect the expression levels of xCT, ERK, LOX, and GPX4. (B) and (C) Impact
of 51 on the lipid oxidation level (indicated by MDA) and GSH level in erastin (10
μM)-induced HT1080 cell ferroptosis. Columns, mean (n=3); bars, SD. ***, p<0.001;
*
*, p<0.01; *, p<0.05; ns, not significance.

Figure 6. 51 inhibited ferroptosis through an antioxidant mechanism. (A) 51 inhibited the
produce of cytosolic ROS from erastin-induced ferroptosis in HT1080 and Calu-1 cells.
The results of ROS statistics are on the right. (B) and (C), Lipid ROS production assessed
by flow cytometry in HT1080 and Calu-1 cells using probe BODIPY 581/591 C11,

erastin was used at a concentration of 10 μM. (D) Cell-free antioxidant potential
monitored by the relative DPPH absorption. (E) and (F), Effect of 51 (10 μM) on
rotenone and H O -induced cell oxidative death in HT1080 cells within 24 h. Columns,
2
2
mean (n=3); bars, SD. ***, p<0.001.
3
.5. In vivo effect of 51 on an ischemic stroke model
The classical middle cerebral artery occlusion (MCAO) model was used to evaluate
the in vivo effect of 51 on ischemic stroke. In this assay, Sprague-Dawley (SD) rats were
used. After 24 hours of ischemia-reperfusion, the 2,3,5-triphenyl tetrazolium chloride
staining (TTC staining) method was used to monitor the volume of cerebral infarction.[46]
As shown in Figure 7A, 51 treatment markedly decreased the volume of cerebral
infarction compared with the model group. We then quantitatively calculated the volume
of lesioned hemisphere by Image J[14]. The results exhibited that the infarct volume of
the rat brain of the 51 treated group (5 mg/kg, iv) was significantly reduced compared
with that of the model group (see Figure 7B and 7C). In the same experiment, Fer-1
treatment (5 mg/kg, iv) also reduced the infarct volume but with a weaker activity
compared with 51. Neuroethology evaluation by Longa (the five-score method) indicated
that 51 treatment evidently decreased neurological score, implying preventing reperfusion
injury and restoring walking ability of MCAO rats (see Figure 7D). Finally, we measured
the GSH and MDA levels of the lesioned hemisphere. The results presented that 51
treatment slightly elevated the GSH level (Figure 7E) and significantly reduced the MDA
level (Figure 7F), which are consistent with those of the previous in vitro experiments
(
Figure 5B and 5C). All of the data indicated that 51 has an excellent therapeutic effect on

the MCAO model.
Figure 7. 51 protected brain tissue from ischemic stroke through inhibiting cell
ferroptosis. (A) Representative TTC-stained brain sections of rats at 24 h after
ischemia/reperfusion, where lesioned tissue stains white, N=8 per group, scale bars=1cm.
(B) Quantification of infarct volume of lesioned hemisphere indicated by TTC staining
using Image J. (C) Quantitative percentage of lesioned hemisphere indicated by TTC
staining using Image J. (D) Neurological scoring (The higher number, the more serious
impairment) was performed at 24 h of ischemia/reperfusion. (E, F) GSH level and MDA
contents of lesioned hemisphere were measured after 24 h of ischemia/reperfusion. Data
are presented as mean ± SD. ns, not significance, *, p<0.05, **, p<0.01, ***, p<0.001.
3
.6. Evaluation of pharmacokinetic properties of 51 in rats
A preliminary evaluation of pharmacokinetic (PK) properties of compound 51 was
carried out on SD rats. Both the intravenous (2.20 mg/kg) and oral (6.67 mg/kg)

administrations were used. Key PK parameters are summarized in Table 5. After oral
administration of 51 (6.67 mg/kg), the areas under the concentration-time curve AUC(0-t)
and AUC(0-∞) are 1112.0 ng/mL*h and 1214.1 ng/mL*h, respectively. The half-life (t ) is
1/2
1
9.4 h, and the maximum plasma concentration (Cmax) is 54.2 ng/mL at around 5.3 h.
After intravenous administration of 51 (2.20 mg/kg), the areas under the
concentration-time curve AUC(0-t) and AUC are 1253.2 ng/mL*h and 1325.3 ng/mL*h,
∞
(0- )
respectively. The t1/2 is 14.6 h, and the Cmax is 518.2 ng/mL at around 0.08 h. The
clearance rate (Cl), the volume of distribution at a steady state (V ), and the oral
d
bioavailability (F) are 27.7 mL/min/kg, 34.9 L/kg, and 30.2%, respectively. We also
detected the plasma stability of fer-1 (see supporting information Table S2), the results
reveal that fer-1 has poor metabolic stability. All of these demonstrated that 51 has good
pharmacokinetic properties.
Table 5. In Vivo Pharmacokinetic Parameters of 51 Obtained in a Preliminary
Assessment Experiment for the Pharmacokinetic Properties.
Parameter
PO (6.67 mg/kg)
IV (2.20 mg/kg)
14.58
a
t
1/2 (h)
19.4
5.3
b
t
max (h)
max (ng/mL)
AUC(0-t) ng/mL*h
0.08
c
C
54.2
518.2
d
1112.0
1214.1
27.7
1253.2
1325.3
27.7
e
∞
)
AUC(0- ng/mL*h
f
Cl (mL/min/kg)
g
V
d
(L/kg)
34.9
34.9

h
F
30.2%
a
b
Mean half-life associated with the terminal slope. Mean time to reach maximum plasma
c
concentration. Mean peak plasma concentration of compound in systemic circulation
d
following dose. Mean area under the plasma concentration-time curve in 72 hours.
e
f
g
Mean area under the plasma concentration-time curve. Mean body clearance. Mean
h
volume of distribution at a steady state. Oral bioavailability.
The blood-brain barrier (BBB) permeability is important for drugs targeting central
nervous system (CNS). We therefore examined the BBB permeability of 51 in rats. The
B/P value, a ratio of drug in brain to drug in plasma or blood, is usually used to
characterize the BBB permeability of a drug compound[47]. In general, a B/P value
of >0.3 indicates good capability to permeate the BBB. The calculated B/P value of 51 is
0
.442, implying good BBB permeability (Figure 8).
Figure 8. The plasma and brain concentration-timecurve of 51 in SD rats after a single
intravenous administration dose of 5 mg/kg, N=3 per group. Data are presented as means
＋SD.
3
.7. Evaluation of PAINS-liable property

Compound 51 has a phenothiazine substructure, which is suspected of pan assay
interference compounds (PAINS)[48]. Here we designed several experiments to evaluate
the PAINS-liable properties of our compound. Firstly, we examined whether our
compound is a promiscuous compound with activity to multiple biologically unrelated
targets. For this purpose, we tested the bioactivities of 51 against a panel of 388 human
protein kinases through the Eurofins kinase profiling service. The results, displayed in
supporting information Table S3, disclosed that 51 had no activity against all the tested
kinases (IC50>10 µM). Besides, we also measured the bioactivities of 51 against other
biological targets, whose bioactivity assays are available in our lab. These targets include
IDO1 (Indoleamine 2,3-dioxygenase 1), HAT1 (Histone acetyltransferase 1), and JMJD6
(Jumonji domain-containing protein 6). The results revealed again that compound 51 had
no activity against them (IC >10 µM, see supporting information Table S4). Collectively,
5
0
all of these experiments demonstrated that 51 is not a PAINS compound.
3
.8. Preliminary toxicity evaluation of 51
We firstly examined the cytotoxicity of compound 51 in vitro. For this purpose, a
total of 23 cell lines were chosen, including 7 normal cell lines and 16 cancer cell lines.
The MTT assay was used to measure the anti-viability activity of 51 against the selected
cell lines. The results exhibited that 51 treatment for 72 h did not show obvious
cytotoxicity against all of these cell lines (IC >10 µM) (see Table 6). A preliminary in
5
0
vivo acute toxicity of compound 51 was then tested. In this assay, Kun-Ming mice and
SD rats were used. 51 did not expression obvious acute toxicity on Kun-Ming mice and
SD rats (see supporting information Figure S4).

Table 6. Anti-viability Activity of 51 against Various Cell Lines
a
Cell lines
93T
Cell type
IC50 (μM)
2
Human embryonic kidney cells
Human umbilical endothelial cells
Human fibroblast cells
>10
>10
>10
>10
>10
>10
>10
HUVEC
BJ
LO2
Human liver cells
HaCaT
Hpde6-C7
L929
Human immortalized keratinocytes
Human pancreatic ductal epithelial cells
Mouse fibroblast cells
NCI-H23
NCI-H1975
PC-9
Non-small cell lung cancer
Non-small cell lung cancer
Lung adenocarcinoma
Ovarian cancer
Ovarian cancer
Prostatic carcinoma
Melanoma
>10
>10
>10
>10
>10
>10
>10
>10
>10
>10
ES-2
SK-OV-3
DU145
A375
Malme-3M
CHL-1
U87
Melanoma
Melanoma
Glioma

U251
Glioma
>10
>10
>10
>10
>10
>10
K562
Chronic myeloid leukemia
Hepatocarcinoma
Huh7
Bel7402
Hepatocarcinoma
MDA-MB-435s
MDA-MB-231
Triple negative breast cancer
Triple negative breast cancer
a
Each compound was tested in triplicate.
4
. CONCLUSION
In this investigation, structural optimization and SAR analyses towards the
ferroptosis inhibitors promethazine and phenothiazine were carried out, which led to the
discovery of a series of derivatives as new ferroptosis inhibitors. Among these
compounds, 51 is the most potent one, which demonstrated an EC value of 0.0005 μM
5
0
on the erastin-induced HT1080 cell ferroptosis model. Further studies demonstrated that
1 could specifically inhibit cell ferroptosis in a cell-independent manor. In the MCAO
5
ischemia stroke model, 51 displayed excellent therapeutic effect. 51 also showed good
pharmacokinetic properties, particularly a good ability to permeate the BBB, which is
critical for drugs targeting the neural system. In preliminary toxicity evaluation assays, it
did not exhibit obvious toxicity either in vitro or in vivo. Taken together, 51 could be a
promising agent for the treatment of ischemia, and deserves further in-depth studies.
5
. EXPERIMENTAL SECTION

5
.1. Compounds and Cell Cultures
.1.1. Chemicals. Erastin, Fer-1, promethazine, phenothiazine, and compounds
P1-P24 were purchased from commercial source (SPECS and ChemDiv). Compounds
-71 were synthesized at the State Key Laboratory of Biotherapy, Sichuan University
details for the synthesis see chemical methods). Unless otherwise indicated, all other
reagents or compounds were from Sigma-Aldrich (St. Louis, USA).
.1.2. Cell lines and cell culture conditions. All the cell lines used in this study were
5
1
(
5
obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA).
Culture conditions for all the cell lines are in accordance with the requirements of ATCC
at 37℃ in a humidified 5% CO incubator (Thermo Scientific).
2
5
.1.3. Cell viability measurements. Cell viability was evaluated by the MTT assay
method. In brief, cells were seeded into 96-well plates and exposed to various
concentrations of compounds for indicated times. Then 20 μL of the MTT (5 mg/mL in
sodium chloride) was added to each well and incubated at 37℃ in 5% CO for 2-4 hours,
2
and later 50 μL of the acidified sodium dodecyl sulfate (20%, w/v) was added to each
2
well and incubated at 37℃ in 5% CO . Finally, the absorbance at 570 nm was measured
using the Multiskan MK3 ELISA photometer (Thermo Scientific, USA) next day. The
EC50 or IC50 of each compound was calculated by GraphPad Prism software. All
experiments were repeated in triplicate.
5
.1.4. Erastin-induced ferroptosis model. Cells were seeded into 96-well plates and
exposed to both erastin (10 μM) and various concentrations of compounds for 48 hours.
Cells were seeded into 96-well plates at an appropriate density, for instance, HT1080 and
MDA-MB-231 cells were seeded at 3000 cells per well, Calu-1 and BxPC-3 cells were