Optimization of the clofazimine structure leads to a highly water-soluble C3-aminopyridinyl riminophenazine endowed with improved anti-Wnt and anti-cancer activity in vitro and in vivo

European Journal of Medicinal Chemistry 222 (2021) 113562

Contents lists available at ScienceDirect

European Journal of Medicinal Chemistry

journal homepage: http://www.elsevier.com/locate/ejmech

Optimization of the clofazimine structure leads to a highly water-

soluble C3-aminopyridinyl riminophenazine endowed with improved

anti-Wnt and anti-cancer activity in vitro and in vivo

Alexey Koval ^a^,¹, Ivan Bassanini ^b, Jiabin Xu

,

k

,

1

a

,

c

,

1

, Michele Tonelli , Vito Boido ,

d

e

,

f

,

g

e

h, i

Fabio Sparatore , Frederic Amant

, Daniela Annibali , Eleonora Leucci

,

b,

2,

**

, Vladimir L. Katanaev ^a

, j, 2, *

Anna Sparatore

a

Department of Cell Physiology and Metabolism, Translational Research Centre in Oncohaematology, Faculty of Medicine, University of Geneva, 1206,

Geneva, Switzerland

Dipartimento di Scienze Farmaceutiche, Universit aꢀ degli Studi di Milano, 20133, Milano, Italy

Department of Biomedical Sciences, Faculty of Biology and Medicine, 1011, University of Lausanne, Lausanne, Switzerland

Dipartimento di Farmacia, Universit aꢀ di Genova, 16132, Genova, Italy

Gynecological Oncology Laboratory, Department of Oncology, KU Leuven and Leuven Cancer Institute (LKI), 3000, Leuven, Belgium

Department of Obstetrics and Gynecology, University Hospitals Leuven and Department of Oncology, 3000, Leuven, Belgium

Centre for Gynecologic Oncology Amsterdam (CGOA), Antoni Van Leeuwenhoek-Netherlands Cancer Institute (AvL-NKI), University Medical Center (UMC),

b

c

d

e

f

g

1

066, Amsterdam, the Netherlands

h

i

Laboratory for RNA Cancer Biology, Department of Oncology, KU Leuven, 3000, Leuven, Belgium

Trace, LKI Leuven Cancer Institute, KU Leuven, 3000, Leuven, Belgium

School of Biomedicine, Far Eastern Federal University, 690922, Vladivostok, Russia

Istituto di Scienze e Tecnologie Chimiche "Giulio Natta", Consiglio Nazonale delle Ricerche, 20131, Milano, Italy

j

k

a r t i c l e i n f o

a b s t r a c t

Article history:

Received 13 November 2020

Received in revised form

Triple-negative breast cancer (TNBC) is a cancer subtype critically dependent upon excessive activation of

Wnt pathway. The anti-mycobacterial drug clofazimine is an efﬁcient inhibitor of canonical Wnt

signaling in TNBC, reducing tumor cell proliferation in vitro and in animal models. These properties make

clofazimine a candidate to become ﬁrst targeted therapy against TNBC. In this work, we optimized the

clofazimine structure to enhance its water solubility and potency as a Wnt inhibitor. After extensive

structure-activity relationships investigations, the riminophenazine 5-(4-(chlorophenyl)-3-((2-(piper-

azin-1-yl)ethyl)imino)-N-(pyridin-3-yl)-3,5-dihydrophenazin-2-amine (MU17) was identiﬁed as the new

lead compound for the riminophenazine-based targeted therapy against TNBC and Wnt-dependent

cancers. Compared to clofazimine, the water-soluble MU17 displayed a 7-fold improved potency

against Wnt signaling in TNBC cells resulting in on-target suppression of tumor growth in a patient-

derived mouse model of TNBC. Moreover, allowing the administration of reduced yet effective dos-

ages, MU17 displayed no adverse effects, most notably no clofazimine-related skin coloration.

10 March 2021

Accepted 17 May 2021

Available online 30 May 2021

Keywords:

Wnt signaling

Triple-negative breast cancer

Clofazimine

Riminophenazine

Medicinal chemistry

Patient-derived xenograft

©

2021 The Author(s). Published by Elsevier Masson SAS. This is an open access article under the CC BY-

NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction

Researchers and companies constantly seek novel strategies to

reduce the costs and time investments of drug discovery and

development. In this context, the use of an already approved drug

for a new clinical indication represents an attractive approach.

Unfortunately, it does not happen too often that an approved drug

turns out to be perfect for the new clinical indication [1], which

establishes the need for medicinal chemistry-based structural

optimization to exalt the novel activity. Although the chemical

*

Corresponding author. Department of Cell Physiology and Metabolism, Trans-

lational Research Centre in Oncohaematology, Faculty of Medicine, University of

Geneva, 1206, Geneva, Switzerland.

** Corresponding author.

E-mail addresses: anna.sparatore@unimi.it (A. Sparatore), vladimir.katanaev@

unige.ch (V.L. Katanaev).

1

equally contributing as ﬁrst authors.

equally contributing as last authors.

2

https://doi.org/10.1016/j.ejmech.2021.113562

nc-nd/4.0/).

A. Koval, I. Bassanini, J. Xu et al.

European Journal of Medicinal Chemistry 222 (2021) 113562

modiﬁcations introduced on the parent drug will require safety and

efﬁcacy reassessment in preclinical and clinical settings, the reuse

of the data previously collected for the parental drug approval

indeed provides a signiﬁcant mitigation of risks and costs for the

modiﬁed compound [2,3].

The Wnt pathway, a developmental signal transduction cascade

controlling tissue growth, differentiation and patterning [4,5], is

frequently switched on through mutational or epigenetic changes

from its predominantly dormant state in healthy adult cells to drive

cancer progression in organs such as breast, colon, liver, or ovaries

new derivative EB05(25), all the tested riminophenazines had

previously been reported and analyzed in our studies on the

compounds' anti-mycobacterial [19] or antiplasmodial and leish-

manicidal activities [17,18]. In the current work, we additionally

tested some synthetic intermediates (i.e. SV05 (1), SV01 (12) and

INT262 (27)). Results of the compounds' anti-Wnt activities are

summarized in Table 1, which also provides the calculated selec-

tivity index (S.I.) of each compound, deﬁned as the ratio between

the IC50 on the inhibition of the constitutively expressed Renilla

luciferase and that of the Wnt-speciﬁc (TopFlash) IC50, the former

being indicative of the compounds’ aspeciﬁc toxicity.

It can be stated that in our analysis of the effects of introduction

of different basic heads and aromatic substituents with the respect

of the original structure of clofazimine, a broad investigation of the

chemical space has been achieved.

We experimentally validated aqueous solubility of basic clofa-

zimine derivatives characterized by an aminophenyl group in po-

sition C-3 (GM05, SV06, SV12, SV13 and SV21; GM05; EB04 and

EB06, Supporting Table S1). This property was found to be

improved in comparison to clofazimine, agreeing with our expec-

tations given the structural features of the derivatives. The most

water-soluble compound of this series was the dimethylamine

derivative EB04.

[

6e9]. Our previous studies identiﬁed the approved anti-

mycobacterial drug clofazimine as a potent and speciﬁc inhibitor of

the -catenin-dependent canonical Wnt pathway and of the

b

ensuing cell proliferation in triple-negative breast cancer (TNBC)

in vitro and in vivo, providing the basis for the use of this drug

against TNBC and other Wnt-dependent cancers [10e13]. Generally

described as a highly lipophilic riminophenazine, clofazimine upon

therapeutic administration accumulates in lipid-rich tissues, pro-

ducing the most documented and psychologically harming side

effect: the red skin-coloring [14e16]. Therefore, with the aim of

developing novel clofazimine analogues endowed with the ca-

nonical Wnt signaling inhibitory activity in TNBC but possessing an

improved safety proﬁle, the current work provides a detailed me-

dicinal chemistry investigation on the structure activity relation-

ships of a panel of more than 40 riminophenazines structurally

related to clofazimine (Tables 1 and 2). The compounds were tested

in vitro for the inhibition of canonical Wnt signaling and for aspe-

ciﬁc cytotoxicity, focusing on the effects of different tertiary amines

as basic heads (introduced as a water-solubilizer at physiological

pH in place of the original isopropyl group on the imino nitrogen of

clofazimine), as well as of the concomitant modiﬁcation of clofa-

zimine's aromatic core by the replacement of the original C-3 p-

chlorophenylamine moiety with a more hydrophilic pyridin-3-

amine. These analogues were either previously described by us in

the quest for novel anti-tuberculosis and antiprotozoal drugs

Compounds from 2 to 10, which conserve the exact aromatic

substituents of clofazimine and are characterized by the insertion

of bicyclic (e.g. quinolizidine or pyrrolizidine moieties) or acyclic

aliphatic tertiary amines, were found to be mostly inactive against

the Wnt pathway, with the exception of 2-(hexahydro-1H-pyrro-

1

lizin-7a-yl)ethanaminyl and N ,N -dimethylpropane-1,3-diaminyl

derivatives SV13 (9) and EB06 (10), which showed micromolar

activities but comparable toxicities.

The ﬂuorine derivatives GM07 (11) and GM06 (12), instead, were

totally inactive.

The Wnt-speciﬁc activity of compounds from 13 to 25, whose

lipophilia were further reduced by the removal of the two chlorine

substituents of clofazimine, was strongly modulated by the nature

of the basic head inserted in the commune aromatic core. Some

derivatives (15e17, 19, 20 and 24) showed no inhibitory activity

against the Wnt pathway and, in general, a neglectable cytotoxicity;

others instead were able to strongly inhibit the pathway with

different toxicities. The most interesting compound of this series,

SV12 (21), inhibited the Wnt pathway in the low-micromolar range

with the S.I. of ~2.1.

[

17,18] or ad hoc prepared for this study.

Some of the novel and most hydrophilic compounds, charac-

terized by a C-3 aminopyridinyl riminophenazine core, showed up

to a 7-fold improved potency against the Wnt pathway in TNBC

cells as compared with the parent clofazimine, matched with an

outstanding water-solubility, exceeding 0.1 M for some of the most

potent (Supporting Table S1) - as compared to essential water

insolubility of clofazimine. This dramatic improvement in the

aqueous solubility strongly reduces deposition of the compound in

subcutaneous fat and the consequent skin pigmentation. Moreover,

the in vitro and in vivo pharmacodynamic proﬁles of the best per-

forming compound MU17 were found to outperform those of clo-

fazimine. Thus, our study discloses strong improvement of the

clofazimine scaffold for the speciﬁc anti-Wnt and anti-cancer

treatment with reduced side effects in animals.

Interestingly, among compounds 28e32 all bearing the C-3

aminopyridinyl moiety, GG08 (32), one of the most hydrophilic

compounds of Table 1, showed a micromolar Wnt-speciﬁc activity

(IC50 ¼ 7.3

mM) with the S.I. of ca. 2.2. To be noted that compounds 1,

13, and 28, all lacking a C-2 basic side chain, were devoid of any

Wnt-inhibitory activity.

The two most active and selective compounds GG08 and SV12

showed similar or slightly improved potency as compared to clo-

fazimine. However, their in vitro speciﬁcity window, as showed by

Fig. 1A, was reduced due to an overall increase of inhibition of the

Renilla luciferase. Nonetheless, when analyzed for their ability to

inhibit growth of TNBC cell lines, SV12 and GG08 outperformed

clofazimine both in the MTT survival assay (Fig. 1B) and in the

colony formation assay (Fig. 1C and D). In analogy with clofazimine

and as expected from a compound affecting the canonical rather

than non-canonical Wnt pathway [10,11], SV12 and GG08 demon-

strated low or no inhibition of the ability of different cancer cells to

migrate, with the notable exceptions of the MDA-MB-231 and

MDA-MB-468 TNBC lines (Fig. 1E and F).

2

. Results

2.1. SV12 and GG08: from an increase in aqueous solubility to

improved anti-TNBC properties in vitro

Our investigation started with the in vitro evaluation of the

ability to suppress the canonical Wnt activity by a family of rimi-

nophenazines, formally derived from clofazimine, bearing tertiary

amines as polar, hydrophilic decorations connected with the aro-

matic core via alkyl linker of different lengths (compounds of

Table 1). To this purpose, we used the Wnt3a-activated TopFlash

reporter in the TNBC BT-20 cell line. Simultaneously, the unspeciﬁc/

cytotoxic activity of the compounds was readout with the consti-

tutively expressed CMV-Renilla luciferase. With exception of the

2

A. Koval, I. Bassanini, J. Xu et al.

European Journal of Medicinal Chemistry 222 (2021) 113562

Table 1

Structures of the ﬁrst family of clofazimine analogues tested and their respective Wnt-inhibitory activity, aspeciﬁc cytotoxicity and selectivity indexes.

Label

Compound

X

Y

Z

n

R

Wnt-Inhibitory Activity IC50

(

mM)

Aspeciﬁc Cytotoxicity IC50

(

m

M)

S.I.^a

Clofazimine

SV05

CL-4

CL-5

CL-6

CL-11

CL-12

CL-13

SV06

SV13

EB06

GM07

GM06

SV01

CL-1

CH

N

Cl

F

Cl

F

0

1

2

3

1

2

3

1

2

3

4

6

0

1

2

3

1

2

3

1

2

4

6

3

1

3

0

4

3

CH(CH

3

)

2

7.2 ± 0.5

19 ± 1

2.6

1

2

3

4

5

6

7

8

H

a

b

e

f

c

H

a

b

e

c

d

c

f

N.A.^b

21.5 ± 8.9

24.3 ± 14.6

N.D.^c

b

c

N.A.

N.D.

N.A.^b

~95

~8

~73

d

N.D.^c

b

d

c

N.A.

N.D.

N.A.^b

d

N.D.^c

>1.3

N.D.^c

1.6

153.5 ± 108.8

>200

N.A.^b

~82

d

N.A.^b

~100

9

5.1 ± 1.9

2.8 ± 0.7

N.A.^b

8.0 ± 1.1

2.2 ± 0.1

14.5 ± 1.5

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

0.8

N.D.^c

b

d

c

F

N.A.

~11

~82

N.D.

N.A.^b

d

N.D.^c

1.3

H

CH

H

CH

Cl

H

Cl

21.9 ± 3.1

27.8 ± 2.6

15.8 ± 8.2

b

c

CL-2

CL-3

CL-8

CL-9

N.A.

N.D.

N.A.^b

~33

~36

d

N.D.^c

1.3

N.A.^b

d

28.8 ± 5.4

29.9 ± 8.0

46.5 ± 8.4

b

c

CL-10

SV02

SV12

GM05

GM10

GM02

EB05

SV21

EB04

INT262

GG09

GG12

GG13

GG08

N.A.

N.D.

N.A.^b

~50

d

N.D.^c

2.1

6.3 ± 0.9

13.9 ± 1.9

98.6 ± 10.3

N.A.^b

13 ± 1

17.5 ± 3.4

>100

1.3

>1.0

N.D.^c

1.4

d

~7

1.2 ± 0.1

1.6 ± 0.2

N.A.^b

~40

d

N.D.^c

1.0

3

a

f

5.7 ± 0.8

N.A.^b

5.4 ± 0.3

d

N.D.^c

H

c

f

~42

~35

~11

b

d

c

N

N.A.

N.D.

N.A.^b

d

N.D.^c

1.5

OCH

H

3

26.8 ± 1.0

7.3 ± 1.8

39.2 ± 13.7

16 ± 1

2.2

a

S.I.: selectivity index is calculated as a ratio of IC50s for aspeciﬁc cytotoxicity and Wnt Inhibition.

b

c

N.A.: not active, speciﬁc anti-Wnt activity was not appreciable up to a 200

N.D.: not deﬁned.

mM concentration.

d

IC50 was estimated through linear interpolation.

2

.2. Wnt selectivity correlates with the alkyl spacer length of the C-

side chain

example, when compared to clofazimine, MU17 solubility in water

was increased at least by a factor of ~10 .

5

The results reported in Table 2 and Fig. 2A show that not only the

nature of the basic head but also the length of the alkyl spacer plays

a role in the Wnt pathway inhibition. In general, most compounds

bearing butylene (34) or propylene-diamine (36, 38, 40, 42) side

chains inhibited the Wnt pathway more potently than the corre-

sponding ethylendiamine derivatives (33, 35, 37, 39, 41). Although

the aspeciﬁc toxicity increased as well, the selectivity indexes, with

the exception of the morfolinyl and N-methylpiperazinyl de-

rivatives, improved in the compounds bearing longer alkyl spacers.

On the contrary, the ethylenediamine-spaced N-methylpiperazinyl

derivative MU17 exhibited a ca. 2-fold better potency and a

signiﬁcantly higher S.I. than the two homologous AR18 (44) and

AR13 (45) due to the reduction of the aspeciﬁc toxicity (Fig. 2B).

It can be argued that an “optimal linker length” is needed to

produce Wnt-selective C3-pyridinyl basic riminophenazines and

Encouraged by these ﬁndings and intrigued by the possibility of

developing highly hydrophilic riminophenazine-based clofazimine

analogues endowed with potentialities as canonical Wnt-speciﬁc

chemotherapeutic agents, a family of C-3 aminopyridinyl de-

rivatives of GG08, bearing different basic heads connected to the

aromatic nucleus with alkyl chains of different length, was

designed, prepared and tested for its in vitro activity towards the

Wnt pathway (Table 2). These compounds, with the exception of

AR18 (44) and AR13 (45), had been previously tested by us for their

antiprotozoal activity [18].

We assessed the solubility of all the pyridine-containing basic

riminophenazines of our study (GG08 and GG13, MU05, MU17 and

MU23; AR13 and AR18, Supporting Table S1). They were all found

to be highly water soluble with solubility values ꢀ 0.1 M. For

3

A. Koval, I. Bassanini, J. Xu et al.

European Journal of Medicinal Chemistry 222 (2021) 113562

Table 2

Structures and in vitro screening of the second-generation C-3 aminopyridinyl clofazimine analogues prepared and tested.

Label

Compound

n

R

Wnt-Inhibitory Activity IC50

(

m

M)

Aspeciﬁc Cytotoxicity IC50

(

mM)

S.I.^a

MU12

MU21

EB01

33

34

35

36

37

38

39

40

41

42

43

44

45

1

3

1

2

1

2

1

2

1

2

1

2

3

a

b

c

d

e

f

3.50 ± 0.30

0.86 ± .06

6.28 ± 0.62

1.60 ± 0.34

1.75 ± 0.26

0.87 ± 0.19

2.15 ± 2.05

1.38 ± 0.07

10.62 ± 1.83

5.13 ± 0.57

1.1 ± 0.1

2.83 ± 0.21

1.24 ± 0.08

5.92 ± 0.68

3.96 ± 0.30

2.3 ± 0.26

1.96 ± 0.08

2.37 ± 0.18

1.93 ± 0.08

7.88 ± 3.37

2.83 ± 0.21

2.7 ± 0.2

0.8

1.4

0.9

2.5

1.3

2.3

1.1

1.4

0.7

0.6

2.5

1.7

1.8

MU05

MU09

MU23

MU10

MU24

MU25

MU15

MU17

AR18

AR13

f

3.0 ± 0.5

2.5 ± 0.2

5.1 ± 0.5

4.5 ± 0.4

a

S.I.: selectivity index is calculated as a ratio of IC50s for aspeciﬁc cytotoxicity and Wnt Inhibition.

also that this optimum is related to the nature of the amine head.

For the ﬂexible and conformationally-free diethylamine and

dimethylamine derivatives (compounds GG08 and 33e36), in fact, a

propylendiamine-spaced linker ensures S.I. > 2. The same trend

was instead not seen in relation with the more sterically

demanding piperidine and morpholine derivatives (compounds

surface (blue in Fig. 3, right) generated from the positively

charged nitrogen of the solubilizing basic heads of the two com-

pounds which could be involved in the interaction with a still-

unidentiﬁed protein target within the Wnt pathway.

2.3. Administration of MU17 signiﬁcantly inhibits TNBC tumor

3

9e42) whose selectivity appeared to be less or not at all sensitive

growth in vivo

to the length of the alkyl linker. At variance, the pyrrolidinyl de-

rivatives 37e38 were perfectly aligned with compounds 33e36.

The case of the N-methyl piperazinyl derivatives is by far the

most interesting, since the most selective derivative of the series,

MU17 (43), contains the ethylenediamine linker.

To further investigate these results, we conducted a ligand

overlay-based conformational analysis of all the C-3 pyridinyl de-

rivatives of Tables 1 and 2 using the Hermes module of the CCDC-

suite (ccdc.cam.ac.uk/) to investigate the conformational space

allowed to over-impose their most energetically favorable

conformer [20,21]. The CCDC suite was also used to generate a

putative and preliminary pharmacophore model by superposing a

set of active molecules assuming binding to the same target with

the same binding mode, as we previously reported for a panel of

potent but target-less antiprotozoal agents [22]. Fig. 3 summarizes

the obtained results.

We have next analyzed the ability of the most selective and

potent hydrophilic riminophenazine MU17 to inhibit tumor growth

in the patient-derived xenograft (PDX) model of triple-negative

breast cancer (TNBC) BRC0016. Since the solubilizing moiety, i.e.

the basic N-methylpiperazine head, would likely restrict oral

bioavailability [23] we decided to deliver the compound by intra-

peritoneal injections at 20 mg/kg daily. As shown in Fig. 4 (A-D),

this treatment for a group of 3 mice alongside with the equal

control group over 21 days resulted in a signiﬁcant decrease in the

tumor volume without incurring any changes on the body weight

of the animals or affecting any other thoroughly controlled pa-

rameters of the animal well-being, such as urine and faeces con-

ditions or overt behavior and pain signs, indicating the safety of the

new compound. Additionally, we analyzed levels of MU17 at

different doses and treatment durations in the liver and inside the

tumors (Fig. 4E). The presence of MU17 in blood was also investi-

gated at different post-injection (3h and 24h) times revealing that,

despite its impressive aqueous solubility, the compound was

virtually absent in the blood samples showing only non-

quantitative trace signals (data not shown). However, MU17 was

clearly progressively accumulated in the tumors eventually reach-

ing values well above the IC50 observed for the Wnt pathway in-

hibition in vitro. The compound levels in liver might provide an

explanation why MU17 was not found in blood: MU17 seems to be

strongly retained by this organ and thus likely to be susceptible to

hepatic clearance. These ﬁndings are also compatible with the

absence of detectable compound levels in the faeces and urine (not

shown). Interestingly, one consequence of the signiﬁcantly

As a representative example, the conformational overlay (Fig. 3,

left) of the selective and potent compounds MU17 (black) and GG08

(

gray) with the least selective of the group, MU15 (42) (blue), shows

how the positively-charged nitrogens of the former two are posi-

tioned in the same area and are mostly overlaid. The same was not

found for the oxygen-containing morpholinyl-derivative MU15

(

42) whose charged group orientation and alkyl chain folding

appear quite different, according to its different biological proﬁle.

On the basis of the ligand overlay experiment conducted (for details

see the Experimental Section), a pharmacophore model of the most

selective C-3 aminopyridinyl compounds MU17 and GG08 was also

created using their calculated surfaces of interaction. This model

further highlighted the presence of a highly directional donor-

4

A. Koval, I. Bassanini, J. Xu et al.

European Journal of Medicinal Chemistry 222 (2021) 113562

Fig. 1. Clofazimine derivatives GG08 and SV12 show inhibition of the canonical Wnt signaling and anti-cancer properties similar to the parental drug. (A) Activity of these de-

rivatives was tested in the Wnt3a-induced TopFlash assay with constitutively expressed CMV-Renilla luciferase as a non-speciﬁc inhibition and cell toxicity readout. (B) Both

compounds are also capable of inhibition of growth in the panel of TNBC cell lines, showing overall improved potency as compared to clofazimine. (C, D) Similarly, the compounds

(

all at 20

m

M) successfully inhibit growth of the cancer cells in colonies, showing higher efﬁcacy as compared to clofazimine e likely due to the improved solubility, since

5

A. Koval, I. Bassanini, J. Xu et al.

European Journal of Medicinal Chemistry 222 (2021) 113562

increased aqueous solubility of MU17, when compared to clofazi-

mine, was the absence of the skin coloration during the trials e the

feature readily observed when mice are treated with the parental

compound (Fig. 4F).

of the efﬁcient and double-edged inhibition in TNBC.

3. Discussion and conclusions

Even among the approved modern drugs, only few compounds

2.4. MU17 maintains the on-target anti-Wnt effect in vivo

can boast with their absolute selectivity towards the desired target.

Depending on the exact methods and cutoffs used for estimates,

many of the approved drugs are known to have between 1 and 10

off-target interactions in addition to its target protein [25e27].

Often left unknown and being generally undesirable due to the

ensuing potential adverse effects, these interactions might also turn

out as a surprising asset when leading to a new medical indication.

In this case, decades of prior research on the pharmacokinetics and

safety proﬁle of the drug might lead to multi-million R&D cost and

risk mitigation along with a fast-track development for the new

indication [28].

Using the extracted tumors, we were also able to conﬁrm

through the immunoﬂuorescence analysis the on-target effect of

MU17 in vivo. As shown in Fig. 5A, the animal treatment with the

compound resulted in signiﬁcant changes in the tumor b-catenin

pattern, reducing the number of cells with cytoplasmic staining of

this major Wnt signaling biomarker. This hallmark of Wnt signaling

reduction in the breast cancers [24] correlated with decrease in the

proliferation marker Ki67 staining and the corresponding increase

in the number of apoptotic cells highlighted by expression of

caspase-3 (Fig. 5B and quantiﬁcation on 5C).

This work represents a next step in our quest to transform the

clofazimine scaffold into a Wnt-targeting anti-cancer chemother-

apeutic agent [10e12]. While this compound by itself shows an

excellent anti-TNBC activity both in vitro and in vivo, no attempts

were undertaken so far to improve its anti-Wnt properties through

medicinal chemistry optimization: after the development in 1950s

[29], the full potentialities of the riminophenazine skeleton of

clofazimine have not yet been fully explored by means of modern

chemistry and molecular biology. In a wake of multidrug-resistant

strains of Mycobacterium tuberculosis, attempts to close this gap are

already ongoing by the preparation of more hydrophilic clofazi-

mine analogues as anti-tuberculosis agents [19,30].

Thus, we decided to apply a similar approach by in vitro

screening of a collection of hydrophilic clofazimine analogues

decorated with tertiary amines as basic heads [17e19] as a starting

point to develop water-soluble derivatives endowed with the

speciﬁc anti-Wnt activity. As expected, given the broad range of

biological activities showed by riminophenazines and the fact that

compounds of Table 1 were designed to be anti-tuberculosis and

antiprotozoal agents, only a handful of the derivatives was able to

efﬁciently and speciﬁcally inhibit the Wnt pathway. However, two

of them e SV12 and GG08 e were able to suppress Wnt signaling

with a potency similar to that of clofazimine. This result indicates

that the addition of tertiary amines as basic heads, which can act as

water-solubilizing groups, did not harm the clofazimine activity.

These experiments also provided valuable information about the

possible modiﬁcations of the clofazimine scaffold in view of

obtaining the Wnt-speciﬁc inhibitors. Indeed, introduction of a C3-

aminopyridinyl substituent, of a highly ﬂexible alkyl chain and of a

hydrophilic dimethyl amine boosted the compound selectivity,

potency and water solubility.

On the basis of these encouraging results, a second generation of

C3-aminopyridinyl basic clofazimine analogues structurally related

to GG08, also previously tested by us for their antiprotozoal activity

[18] was screened in vitro for the anti-Wnt activities (Table 2). The

potent N-methyl piperazinyl-based MU17, characterized by an

ethylenediamine linker resulted to be optimal for boosting the

selectivity of the Wnt pathway inhibition of the series of piper-

azinyl derivatives (compounds 43e45 in Table 2), was identiﬁed as

a prime candidate for further in vivo investigation in mice bearing

patient-derived TNBC xenografts. Accordingly, MU17 showed a 7-

fold improvement in the anti-Wnt IC50 when compared to clofa-

zimine while its selectivity index remained at the 2.5-fold level,

similar to that of parental clofazimine. It is thus reasonable to

2

.5. MU17 targets nuclear components of the canonical Wnt

signaling pathway

As we reported earlier [10,11], clofazimine inhibits canonical

Wnt signaling below the level of b-catenin, thus likely targeting one

of the multiple components of the pathway within cell nucleus. We

thus investigated if MU17 maintains the same mode of action and

conﬁrmed that MU17, just like clofazimine [10], is highly speciﬁc

towards Wnt signaling. First, we veriﬁed that MU17 and the other

potent riminophenazines GG08, SV12, AR18 and AR13 do not reveal

inhibition in the FopFlash assay, which employs a reporter plasmid

similar to TopFlash but with the mutated TCF binding site (Sup-

porting Fig. S1A). Further, we proﬁled MU17 and the listed com-

pounds against a panel of luciferase reporters for different signaling

pathways (Supporting Fig. S1B), detecting no inhibition against any

of them.

In order to further assess the mechanism of action of the clo-

fazimine compounds, we tested if and at which time point MU17

could affect cellular b-catenin. We probed for the levels of active b-

catenin in BT20 and MDA-MB-468 cells (Fig. 6A, B) at 5h and 24h of

treatment with the compound. Similarly to clofazimine [11], MU17

was unable to acutely suppress levels of the active

agreeing with the ﬁndings above that suggested that the drug acted

below -catenin. To further corroborate these ﬁndings, immuno-

staining of total -catenin and phospho- -catenin (pS33/37/T41-

catenin, indicative of the protein targeted for proteasomal degra-

dation) in the same cells stimulated by Wnt3a and treated with

MU17 for 5h shows that the drug is unable to acutely inﬂuence the

Wnt3a-induced decrease in phospho-

catenin nucleocytoplasmic accumulation (Fig. 6C, D, E). Curiously, a

strong suppression of -catenin levels is evident upon the pro-

longed 24h treatment with MU17 (Fig. 6A, B), agreeing with our

previous ﬁndings in vitro and in vivo with clofazimine [10], and our

current in vivo ﬁndings with MU17 (Fig. 5A). Thus, we conclude that

the direct molecular target of action of clofazimine and its de-

b-catenin in 5h e

b

b-

b-catenin nor the total b-

b

rivatives in the Wnt pathway is below nuclear translocation of b-

catenin. However, the prolonged inhibition of this target likely

engages a positive feedback loop, since a number of Wnt pathway

target genes are also pathway components (e.g. FZD7, Wnt3a) [6],

which further induces a decrease in the cytonuclear

levels. Future research will identify the nuclear Wnt pathway target

of clofazimine with the network properties so promising in terms

b-catenin

clofazimine is known to eventually precipitate from the culture medium. (E, F) Both clofazimine and the GG08 and SV12 derivatives (all at 20 mM) have little or no activity in

inhibiting cancer cell migration in scratch-wound assay, with MDA-MB-231 and MDA-MB-468 cell lines as exceptions. Statistical signiﬁcance was assessed by two-way ANOVA with

multiple comparisons for main effect per compound (panel A) or individually for each line (panels C and E), N ¼ 3, signiﬁcance is shown as *p < 0.05, **p < 0.01 and ***p < 0.001.

6

A. Koval, I. Bassanini, J. Xu et al.

European Journal of Medicinal Chemistry 222 (2021) 113562

Fig. 2. (A) Correlation between the Wnt-speciﬁc activity (TopFlash IC50) and selectivity of the hydrophilic derivatives of GG08 (Table 2). In respect to GG08, most of the compounds

were found to clusterize in the bottom-left section of the chart as their potency was improved by the structural changes but their selectivities were concomitantly reduced. MU23,

MU05 and MU17 instead grouped in the upper-left section of the chart in which compounds with improved potency and selectivity with the respect to GG08 are clusterized. Only

one of the novel derivatives (MU25) was less potent and less selective than GG08. (B) Variation of the linker length among AR13, AR18, and MU17, shows that 2 methylenes of MU17

are optimal due to reduced cytotoxic activity. Statistical signiﬁcance was assessed by two-way ANOVA with multiple comparisons. N ¼ 4, signiﬁcance is shown as on Fig. 1.

Fig. 3. Left) In silico section of the conformational overlay of MU17 (black), GG08 (gray) and MU15(42) (blue); Right) pharmacophore mapping of MU17 (black) and GG08 (gray): in

red, blue and yellow are highlighted the areas in which acceptor, donor and non-polar interactions can be established, respectively.

suggest that the selection of a proper linker length associated with

the nature of the basic heads inserted onto the common aromatic

riminophenazine core could affect the afﬁnity of a tested analogue

toward speciﬁc targets both within and outside the Wnt pathway

explaining the lower S.I. characterizing some of our C3-pirydinyl

riminophenazines as compared to MU17.

Comparably to clofazimine [10], MU17 demonstrated a prom-

ising proﬁle as potential chemotherapeutic agent in the model of

highly aggressive grade III invasive ductal triple-negative adeno-

carcinoma, resulting in signiﬁcant (3 to 4-fold) reduction in the

tumor volume at a comparable dose of 20 mg/kg (clofazimine itself

has shown profound effect at 25 mg/kg, as described by us [10]),

delivered intraperitoneally. Also, MU17 upon application to animals

was found not to produce any obvious adverse effects. Importantly,

the signiﬁcantly increased solubility of this N-methylpiperazinyl

analogue (>0.3 M) resulted in the absence of deposits in the sub-

cutaneous adipose tissue, eliminating one of the formally harmless,

reversible but most conspicuous adverse effects of the clofazimine

administration, skin coloration e and this is despite the fact that

the derivative maintains the strong red color characteristic of the

scaffold. Moreover, we expect that this property might lead to

drastic reduction of the other skin-related conditions known to be

associated with clofazimine such as rash, dryness and ichtiosis

[31,32].

Despite that MU17 apparently loses the ability to accumulate in

adipose tissues, the compound was still well-retained in the tumor

resulting in a profound reduction of the in-tumor Wnt signaling

levels, evaluated as loss of cytoplasmic b-catenin e a hallmark of

7

A. Koval, I. Bassanini, J. Xu et al.

European Journal of Medicinal Chemistry 222 (2021) 113562

Fig. 4. Daily IP administration of 20 mg/kg MU17 in mice bearing intramammary BRC0016 patient-derived TNBC xenografts results in signiﬁcant reduction of the tumor volume as

measured by electronic calipers (A), without any effects on animal body weight (B). (C, D) Endpoint analysis of the extracted tumors conﬁrms results of the manual measurement.

(

E) Levels of MU17 in the livers and tumors of the treated animals at different dosage and treatment duration indicate that at 20 mg/kg the compound reaches tumor levels well

above IC50. (F) Increased solubility of the compound results in reduction of subcutaneous deposition of the drug and thus absence of pigmentation in the treated animals. Each drug-

or vehicle-receiving group consisted of N ¼ 3 animals.

8

A. Koval, I. Bassanini, J. Xu et al.

European Journal of Medicinal Chemistry 222 (2021) 113562

Fig. 5. MU17 confers on-target anti-Wnt effect of clofazimine in vivo. (A) Staining of the tumor sections shows a profound reduction in the cells positive for cytoplasmic

b-catenin.

(

B) Additionally, when stained for Ki67 and caspase-3, the tumors show an increased number of apoptotic and a decreased number of proliferating cells. (C) Quantiﬁcation of the

effects of clofazimine on

b-catenin. Each cell was assigned to one of the four types depending on the observed pattern of

b-catenin localization. Cell counts reveal that MU17

9

A. Koval, I. Bassanini, J. Xu et al.

European Journal of Medicinal Chemistry 222 (2021) 113562

the Wnt signaling activation in breast cancer [24], whose dramatic

loss we have also observed during treatment with parental clofa-

zimine [10]. The tumor proliferation marker Ki67 was also

decreased, on the background of increased apoptosis. Despite this,

the high levels of MU17 found in the liver and its virtual absence in

the blood indicate that this compound is overall strongly suscep-

tible for hepatic clearance, although the metabolites have not yet

been searched for. Further improvements on this aspect, along with

more detailed in vivo pharmacokinetics studies, might in the future

generate derivatives that could require less frequent dosage regi-

mens with more steady levels reached in the tissues.

Overall, our study paves the way for the development of novel

C3-aminopyridinyl riminophenazines structurally related to clofa-

zimine with improved potency, much higher solubility and thus

reduced skin deposition, with potential for future use as perfusion

and subcutaneous or intramuscular injections and represent an

important step towards development of novel targeted anti-cancer

therapies based on the clofazimine scaffold.

(EB05 (25)) or Scheme 2 (AR18 (44) and AR13 (45)), depending on

the structure of their riminophenazine core. In both cases, target

products were obtained following the below reported general

procedure for immination reaction (step ii in Scheme 1 and step iv

in Scheme 2):

Under nitrogen atmosphere, the desired phenazine core (1

1

equiv.) and the proper N ,N -substituted alkyl-diamine (2 equiv.)

were dissolved in 1,4-dioxane ([phenazine] ¼ 65 mM) and reﬂuxed

for 6 h. After cooling, the solvent was removed in vacuo and the

obtained residue was diluted with dichloromethane (DCM) and

washed three times with water. The organic layer was dried with

anhydrous sodium sulphate and evaporated to dryness. The crude

product was puriﬁed by ﬂash column chromatography [gradient of

methanol (MeOH) in DCM] and converted, using an excess of 1 M

ethanolic solution of HCl, into the corresponding solid hydrochloric

salt which was rinsed with diethyl ether. For NMR characterization

the salts were converted again into the corresponding free bases.

EB05 (25) (isolated yield: 60%, HPLC purity > 99%) was obtained

1

by reacting N ,N -dimethylpropane-1,3-diamine with the phena-

zine core 3-imino-N,5-diphenyl-3,5,10,10a-tetrahydrophenazin-2-

4

. Experimental Section

1

amine prepared by the oxidation of the corresponding N -phenyl-

benzene-1,2-diamine with ferric chloride [29]. Compound eluates

from the ﬂash column chromatography with the 5% of MeOH in

DCM.

.1. Synthetic chemistry

.1.1. Material and methods

¹H NMR (300 MHz, CDCl

, r.t.), d: 7.76e7.66 (m, 4H), 7.39e7.33

3

All commercially available solvents and reagents were used

without further puriﬁcation, unless otherwise stated.

(

1

m, 5H), 7.23e7.09 (m, 3H), 6.99 (s, 1H), 6.53 (d, J ¼ 8.0 Hz, 1H), 5.34

s, 1H), 3.17 (t, J ¼ 6.8 Hz, 2H), 2.34 (t, J ¼ 7.5 Hz, 2H), 2.21 (s, 6H),

NMR spectra were recorded on a Varian Mercury 300 VX

.81 (quintet, J ¼ 7.2 Hz, 3H).

spectrometer in CDCl

3

, CD

3 6

OD or DMSO‑d and chemical shifts

13

C NMR (75 MHz, CD

36.1, 135.8, 132.2, 131.45, 131.37, 130.3, 129.5, 127.8, 127.4, 124.9,

22.4, 117.2, 105.9, 90.3, 55.2, 42.3, 41.1, 22.5.

3

OD, r.t.), d: 152.5, 145.5,141.4, 139.6,139.2,

were reported in ppm (

d). Peaks were assigned with 2D COSY ex-

1

periments and are in agreement with the proposed structures.

TLC was carried out on Merck precoated 60 F254 plates using UV

light and dipping with ethanol/phosphomolybdic acid 10% or a 10%

w/v ethanolic solution of ninidrine.

þ

HRMS (ESI) m/z calcd for C29

30

H N

5

[MþH] : 448.2501, found:

4

48.2506.

AR18 (44) (isolated yield: 30%, HPLC purity > 99%) was obtained

Flash column chromatography was performed using silica gel 60

by reacting 3-(4-methylpiperazin-1-yl)propan-1-amine, prepared

as reported [33], with the phenazine core 5-(4-chlorophenyl)-3-

imino-N-(pyridin-3-yl)-3,5,10,10a-tetrahydrophenazin-2-amine

(

0.040e0.063 mm, Merck).

Organic phases were dried over anhydrous sodium sulphate.

Concentrations were performed under diminished pressure

1

ꢁ

obtained from commercially available N -(4-chlorophenyl)ben-

(

1e2 kPa) at a bath temperature of 40 C.

zene-1,2-diamine, 1,5-diﬂuoro-2,4-dinitrobenzene and pyridine-3-

amine via a series of nucleophilic aromatic substitutions, reduction

and oxidative intramolecular cyclization [17,18,29]. Compound el-

uates from the ﬂash column chromatography with the 10% of MeOH

in DCM.

High-resolution mass spectra (HRMS) on FT-Orbitrap mass

spectrometer in positive electrospray ionization (ESI).

Purities of ﬁnal compounds described in the manuscript was

>

99% as determined by HPLC using CH

3

CN/H

2

O þ CF

3

COOH

gradient and a Purospher RP18 5 m column on a Hitachi Elite

m

¹H NMR (300 MHz, CD

OD, r.t.),

d

: 8.42 (d, J ¼ 7.3 Hz,1H), 8.31 (d,

3

Lachrom Instrument equipped with a DAD detector. Rt. ¼ retention

J ¼ 6.1 Hz, 1H), 7.96e7.81 (m, 5H), 7.69e7.62 (m, 4H), 7.22 (d,

time.

J ¼ 7.2 Hz, 1H), 6.01 (s, 1H), 3.62e3.46 (m, 10H), 2.94e2.91 (m, 5H),

ATR-FT-IR spectra were recorded using an Alpha spectrometer

equipped with an ALPHA's Platinum single reﬂection diamond ATR

2

.18 (bs, 2H).

¹³C NMR (75 MHz, CD

OD, r.t.), d: 161.2, 153.2, 145.0, 139.4, 137.8,

3

unit (Bruker Optics, Milan, Italy). Scans number n ¼ 40. Instru-

ꢂ1

136.4,135.8, 134.4,134.1, 133.4, 132.8,131.9, 131.7,131.0, 130.8,129.4,

mental resolution 5 cm

.

1

13

129.1, 128.2, 127.4, 117.1, 115.3, 90.7, 54.0, 50.7, 42.3, 41.1, 22.6.

H NMR, C NMR, HRMS and IR spectra of the novel compounds

EB05, AR18 and AR13 are provided in the Supporting Information.

þ

HRMS (ESI) m/z calcd for C31

38.2485.

H N

33 7

Cl [MþH] : 538.2486, found:

5

AR13 (45) (isolated yield: 33%; HPLC purity > 99%) was instead

4

.1.2. Previously reported compounds

Compounds from 1 (SV05) to 24 (GM02) and from 26 (SV21) to

2 (GG08) (Table 1) were prepared as we described [17,19]. All the

obtained by reacting 3-(4-methylpiperazin-1-yl)buthyl-1-amine,

prepared as reported [33] with the same phenazine core used for

AR18 (44). Compound eluates from the ﬂash column chromatog-

3

compounds of Table 2, with the exception of AR18 (44) and AR13

45), were instead synthetized as we previously reported [18].

raphy with the 10% of MeOH in DCM.

(

1

H NMR (300 MHz, CDCl

3

: CD

3

OD ¼ 8/2, r.t.),

d: 8.28 (s, 1H), 7.89

(

6

s, 1H), 7.77e7.76 (m, 1H), 7.68e7.66 (m, 1H), 7.27e7.18 (m, 6H),

4

.1.3. Novel compounds: EB05 (25), AR18 (44) and AR13 (45)

The novel riminophenazines were obtained following Scheme 1

.93e6.85 (m, 3H), 6.54e6.53 (bs, 1H), 5.30 (s, 1H), 3.01e2.99 (m,

treatment signiﬁcantly enriches the proportion of cells with high membrane and low cytoplasmic b-catenin, while correspondingly decreasing the share of the cells positive for the

cytoplasmic staining only. Statistical signiﬁcance was assessed by two-way ANOVA (for membranic/cytoplasmic

reported as on Fig. 1.

b-catenin quantiﬁcation) or by Student's t-test. N ¼ 3, p values are

10

A. Koval, I. Bassanini, J. Xu et al.

European Journal of Medicinal Chemistry 222 (2021) 113562

Fig. 6. MU17 maintains the mode of action of clofazimine and likely targets a nuclear component of the canonical Wnt signaling. (A,B) Analysis of active

b-catenin in TNBC cell lines

reveals no decrease of its levels in response to MU17 in 5h; however, it does reduce these levels in 24h. (C, D) Immunostaining of total and phospho- -catenin with quantiﬁcations of

b

11

A. Koval, I. Bassanini, J. Xu et al.

European Journal of Medicinal Chemistry 222 (2021) 113562

6

H), 2.72e2.64 (m, 6H), 2.31 (s, 3H), 1.38e1.36 (bs, 2H), 1.24 (bs,

Inﬁnite 200 Pro reader.

2

H).

¹³C NMR (75 MHz, CDCl

: CD

45.9, 143.1, 142.1, 141.6, 140.1, 139.0, 138.4, 137.7, 136.2, 135.0, 134.8,

32.9, 132.3, 131.6, 120.9, 118.4, 94.4, 67.1, 60.0, 47.5, 46.5, 33.4, 28.0,

OD ¼ 8/2, r.t.),

d: 157.2, 148.9,

3

4.3.3. Scratch-wound and colony formation assays with TNBC cells

1

For migration analysis, the scratch-wound assay was used. Cells

0

were seeded at 60 000 cells/well in 96-well ﬂat-bottom plates.

Next day, after attachment, the monolayer formed in each well was

wounded in two places by a single strike of a 10 ml pipette tip. The

2

5.3.

HRMS (ESI) m/z calcd for C32

52.2640.

þ

H N

35 7

Cl [MþH] : 552.2642, found:

5

detached cells were removed by 2x wash with 1xPBS. For each

experimental well, a random area of both scratches was labeled and

a phase-contrast picture was taken. The cells were left for 24h in

presence or absence of indicated concentrations of tested com-

pound. Afterwards, the pictures of the same area were taken and

the migration of the cell front was analyzed in ImageJ.

For the colony formation assay, TNBC cells were seeded at

250 cells/well for HCC1395, MDA-MB-231 and HCC38; 500/well for

MBA-MB-468; 750/well for BT-20 and 1000/well for HCC1806 in 6-

well plates. Next day, the medium was replaced with the one

containing indicated concentrations of tested compound. Colonies

of 70e100 cells were formed after 8e14 days, were ﬁxed by incu-

bation in 4% PFA in 1xPBS, pH 7.4 and visualized by staining with

the crystal violet solution, and the number of colonies was counted.

4.2. In silico calculations

Two clusters of compounds were identiﬁed from Table 2 on the

basis of compounds selectivity for Wnt-inhibition: compounds

with S.I. > 2.0 deﬁned cluster A while compounds with S.I. < 2.0

deﬁned cluster B. GG08 was added to cluster A.

At ﬁrst, compounds’ structures were reﬁned by assigning atomic

types, adding a formal positive charge on the basic group of the

solubilizing basic heads with the DiscoveryStudio and cleaned-up

with a Dreeding-like force ﬁeld. After that, the Mercury module

of the CCDC-suite was used to generate 200 conformers for each

compound which were used as the input of a ligand overlay-based

pharmacophore modeling run with the Hermes module of the

same suite. The obtained overlays were screened on the base of

their internal energy and statistical relevance. Combining the data

obtained from ligand overlay of the two clusters, a pharmacophore

model was ﬁnally visualized as a surface-ﬁeld representation.

4.3.4. Animal xenografted tumor growth

The experiments were approved by the Swiss Federal Veterinary

Ofﬁce and were carried out in accordance with the local animal

welfare act. For experiments with BRC0016 PDX xenograft, ca.

500cm3 freshly extracted tumors from 2 donor animals were

4

.3. Biological evaluation

chopped into small pieces using scissors and digested using Accu-

max for at 1 h on a rotary shaker. After sedimentation of the large

undissolved chunks, the supernatant was centrifuged at 300g for

5 min, and the resulting cell pellet was resuspended in Matrigel at

.3.1. Luciferase activity measurements

The Wnt-induced and basal luciferase activity was analyzed as

described [11]. Brieﬂy, 30

transfected with the TopFlash luciferase reporter at 6000 cells/well

were distributed in a white opaque 384-well plate. The cells were

m

l of BT-20 cells, wild type or stably

the ratio of 50 ml per recipient NOD/SCID-g (NSG) mouse (6 in total).

3

As soon as the tumor volume reached ~100 mm as measured by

electronic calipers, the mice were separated into two equal groups

each including N ¼ 3 animals. The mice were injected IP daily with

20 mg/kg MU17 solution in 10% Kolliphor (Sigma) and 10% EtOH in

water carrier solution. Tumor volume (mm3) was determined using

ꢁ

maintained in DMEM containing 10% FBS and incubated at 37 C, 5%

CO2 overnight for attachment. Afterwards, the stably transfected

BT-20 cells were additionally transfected with the plasmid consti-

tutively (under the CMV promoter) expressing Renilla luciferase

the following formula: tumor volume ¼ length ꢃ (width)2 ꢃ

p/6.

(

Addgene, Cambridge, MA, USA). Transfection was carried out as

described in manufacturer's protocol using 12 g/ml of DNA and

l/ml XtremeGENE 9 reagent (Roche), 1 l of the ﬁnal mixture

was subsequently added per well.

Next day, the medium of each well was replaced by 30

The drug treatment was started as soon as the tumor reached the

volume of ~100mm3.

m

4

0

m

4.3.5. Slide preparation and immunoﬂuorescent staining

ml (384-

After removal from the animals, tumors were photographed,

well plates) fresh medium containing Wnt3a (500 ng/ml) (puriﬁed

as described [34]) and compound concentrations indicated in the

ﬁgures (following 1h of pre-incubation with the compound). After

overnight incubation, the supernatant in each well was removed

and the luciferase activity was measured in the Tecan Inﬁnite 200

Pro reader.

sliced into fragments of ~300e400mm3 and rinsed in the excess of

ice-cold 1xPBS. Subsequently, they were ﬁxed in fresh 10% PFA

solution in 1xPBS for 3e5 days at 4 C. The tumors were embedded

ꢁ

in parafﬁn blocks and cut into slices 5 mm thick in the Histology

facility of University of Geneva. The slices were mounted on glass

ꢁ

slides and stored at 4 C. For the staining, the slides were ﬁrst

deparafﬁnized in three changes each of xylene, sequence of water

mixtures of EtOH with decreasing concentrations (95%, 70% and

50%) and ﬁnally water. Subsequently, antigen retrieval was per-

formed in 20 mM Tris-EDTA, pH 9.0 with 0.1% Tween-20 by heating

up the slides to 95 C for 20min and gradual cooling to the room

temperature. Further, the slides were blocked in 1xPBS/0.1% Triton

X-100 with 3% of normal horse serum and stained with primary

4

.3.2. MTT assay

50 ml of indicated TNBC cell lines were added into each well of a

transparent 384-well plate at density 1500 cells/well. The cells

were maintained in DMEM containing 10% FBS and incubated at

ꢁ

37 C, 5% CO2 overnight. The next day, the medium of each well was

replaced by 50

tions of compounds. After incubation for 3e4 days, the medium in

ml fresh medium containing indicated concentra-

antibodies against

b-catenin (BD Biosciences), Ki67 (BD Bio-

each well was replaced by 50

Roth) solution in 1xPBS. The plates were incubated for 3h at 37 C.

Then the solution was removed, and 30

each well. Absorbance at 570 nm was measured in the Tecan

m

l of 0.5 mg/ml Thiazolyl blue (Carl

sciences) at 1:200 dilution. Cleaved caspase-3 staining was per-

formed using corresponding antibodies (CST) at 1:1000 dilution

and developed using SuperBoost tyramide enhancer kit (Invi-

trogen). The slides were mounted using the VectaSchield mounting

ꢁ

ml DMSO was added into

nuclear

b-catenin and ratio between cytoplasmic and phospho- b-catenin (E) demonstrates that upon 5h treatment, MU17 does not affect beta-catenin phosphorylation or nuclear

translocation. Statistical signiﬁcance was assessed by one-way ANOVA. N ¼ 3, p values are reported as on Fig. 1.

12

A. Koval, I. Bassanini, J. Xu et al.

European Journal of Medicinal Chemistry 222 (2021) 113562

Scheme 1. Symmetrically substituted riminophenazines. Reagents and conditions: (i) FeCl

3

, HCl(aq) and CH

3

COOH, r.t.; (ii) Reﬂux in dioxane.

ꢁ

Scheme 2. Asymmetrically substituted riminophenazines. Reagents and conditions: (i) TEA, EtOH, r.t.; (ii) TEA, THF, 60 C; (iii) AcOH, Zn(0), 0 C / r.t.; (iv) MeOH, r.t; (v) Reﬂux in

dioxane.

medium and visualized using AiryScan LSM-880 confocal micro-

scope (Zeiss). For scoring, each of the stained slides was assigned

with perceived level of membrane or cytoplasmic staining ranging

from 0 (not observed) to 5 (maximum) and average score for all the

stained tumors was calculated for presentation.

SDS, 1x Protease inhibitor cocktail (Roche)) and incubated on ice for

10min. The cells were resuspended and then centrifuged at 18000g

ꢁ

at 4 C to remove debris. 5

ml of the supernatants each were further

analyzed by Western blot with antibodies against active

(Millipore) and -Tubulin (Sigma) at 1:1000 dilutions.

b

-catenin

a

4

.3.6. Immunoﬂuorescent staining of the cultured cells

The indicated cell lines were seeded at 200 000 cells/well in the

-well plates with poly-L-lysine-coated coverslips on the bottom.

4

.3.8. Evaluation of in vivo compound levels

On the last day of the trial, NOD/SCID-

g

(NSG) mice bearing

6

intramammary tumors were injected IP as described above and

sacriﬁced 3 h afterwards. Immediately upon sacriﬁce the tumor and

liver samples were collected and frozen on dry ice. The samples

were weighted and extracted by 2 min homogenisation on a ball

beater with at least 3 vol parts of ice cold CH CN (more in case of

3

low amounts of material to provide enough homogenisation vol-

After 5h of treatment with or without Wnt3a in presence or

absence of MU17, the cells were washed 2x by PBS, ﬁxed by 4%PFA

at room temperature and stained and visualized as described above

for tumor samples. The antibodies used for staining were against

catenin (BD Biosciences) and pS33/37/T41- -catenin (CST).

b-

b

ume) containing an internal standard (compound MU25 (40)) The

ꢁ

4.3.7. Western blotting

samples were centrifuged at 16 000g, 10min, 4 C and the super-

TNBC cell lines were seeded at 100 000 cells/well in 24 well

2 3

natant was diluted with H O to a ﬁnal concentration of 25% CH CN.

plates. The next day, the medium was replaced with fresh medium

containing drug pre-warmed at 37 C. After further 24h incubation,

After repetition of the centrifugation, the samples were quantiﬁed

by LC-MS on the Q Exactive instrument (ThermoFisher) in positive

ꢁ

the medium was removed, followed by washing with 500

PBS twice per well. The cells were lysed in the well by addition of

l of ice-cold RIPA buffer (1x TBS, 4 mM EDTA, 1% Triton, 0.1%

m

l of 1x

mode with electrospray ionization. Brieﬂy, 10 ml of sample were

injected into the reverse-phase HPLC column (Zorbax Eclipse Plus

C18, Agilent Technologies) and separated using a gradient from 25

30 m

13

A. Koval, I. Bassanini, J. Xu et al.

European Journal of Medicinal Chemistry 222 (2021) 113562

to 95% CH

the compounds ([MþH]þ) were identiﬁed and quantiﬁed using the

Xcalibur software and with the help of standard curve

0.3 Me100 M, 6 concentrations) built from known compound

3

CN with 0.1% formic acid. The peaks corresponding to

Appendix A. Supplementary data

a

Supplementary data to this article can be found online at

https://doi.org/10.1016/j.ejmech.2021.113562.

(

m

dilutions in control blood plasma and extracted in the same batch

with the samples.

References

[

1] S. Pushpakom, F. Iorio, P.A. Eyers, K.J. Escott, S. Hopper, A. Wells, A. Doig,

T. Guilliams, J. Latimer, C. McNamee, A. Norris, P. Sanseau, D. Cavalla,

M. Pirmohamed, Drug repurposing: progress, challenges and recommenda-

tions, Nat. Rev. Drug Discov. 18 (2019) 41e58, https://doi.org/10.1038/

nrd.2018.168 .

2] A. Astolﬁ, T. Felicetti, N. Iraci, G. Manfroni, S. Massari, D. Pietrella, O. Tabarrini,

G.W. Kaatz, M.L. Barreca, S. Sabatini, V. Cecchetti, Pharmacophore-based

repositioning of approved drugs as novel Staphylococcus aureus NorA ef ﬂ ux

pump inhibitors, J. Med. Chem. 60 (2017) 1598e1604, https://doi.org/

4

.4. Evaluation of compounds solubility

As all the compounds have strong absorbance in the visible part

of spectrum, their solubility was assessed by spectrophotometry.

Compounds (2e3 mg) were dissolved in PBS buffer pH 7.4 at a

concentration of 200 mg/ml (GG08, GG13, MU05, MU17, MU23) or

at a concentration of 50 mg/ml (AR13 and AR18). After 3 h of

shaking they were centrifuged at 20000g for 30 min. Supernatant

(1 ml) was taken and diluted 4000x folds to create sample. For all

other less soluble compounds (see Supporting Table S1), the com-

pound was added from DMSO stock to create a target concentration

10.1021/acs.jmedchem.6b01439 .

[3] N. Novac, Challenges and opportunities of drug repositioning, Trends Phar-

macol. Sci. 34 (2013) 267 e 272, https://doi.org/10.1016/j.tips.2013.03.004 .

[4] V.L. Katanaev, The Wnt/Frizzled GPCR signaling pathway, Biochem. Bio-

khimiia. 75 (2010) 1428e1434.

[

5] Z. Steinhart, S. Angers, Wnt signaling in development and tissue homeostasis,

Dev. Camb. Engl. 145 (2018), https://doi.org/10.1242/dev.146589.

6] A. Koval, V.L. Katanaev, Dramatic dysbalancing of the Wnt pathway in breast

cancers, Sci. Rep. 8 (2018) 7329, https://doi.org/10.1038/s41598-018-25672-6.

7] L.R. Howe, A.M.C. Brown, Wnt signaling and breast cancer, Canc. Biol. Ther. 3

(2004) 36e41, https://doi.org/10.4161/cbt.3.1.561.

of 250

After 3h of incubation on shaker, the solution was centrifuged at

0000g for 30 min. The supernatant was used to quantify the

mg/ml (ﬁnal concentration of DMSO was 0.5% in all cases).

2

compound solubility.

For quantiﬁcation of the concentration of the compound in the

supernatants, the calibration curves of the compounds ranging

[8] A. Blagodatski, D. Poteryaev, V.L. Katanaev, Targeting the Wnt pathways for

therapies, Mol. Cell. Ther. 2 (2014) 28, https://doi.org/10.1186/2052-8426-2-

28.

from 1.5 to 50

mg/ml were made on Tecan Inﬁnite M200 plate

[

9] H.V. Shaw, A. Koval, V.L. Katanaev, Targeting the Wnt signalling pathway in

cancer: prospects and perils, Swiss Med. Wkly. 149 (2019) w20129, https://

doi.org/10.4414/smw.2019.20129.

reader. For maximal sensitivity, the peak absorbance of the com-

pound was measured and the corresponding maximum wave-

length was used (490 nm for GG08, MU05, MU17, MU23, AR13 and

AR18; 470 nm for GG13; 450 nm for clofazimine and 460 nm for all

others). For the most soluble compounds (see above) the calibra-

tion curves were made directly in PBS, for all the least soluble

compounds the calibration curves were made in 50% DMSO in PBS

and for clofazimine and GM05 e in 75% DMSO in PBS. The super-

natant samples were supplied with corresponding amount of

DMSO prior to mesurement. The data was ﬁt and interpolated in

Prism 7.0 (GraphPad).

[

10] K. Ahmed, A. Koval, J. Xu, A. Bodmer, V.L. Katanaev, Towards the ﬁrst targeted

therapy for triple-negative breast cancer: repositioning of clofazimine as a

chemotherapy-compatible selective Wnt pathway inhibitor, Canc. Lett. 449

(2019) 45e55, https://doi.org/10.1016/j.canlet.2019.02.018.

11] A.V. Koval, P. Vlasov, P. Shichkova, S. Khunderyakova, Y. Markov, J. Panchenko,

A. Volodina, F.A. Kondrashov, V.L. Katanaev, Anti-leprosy drug clofazimine

inhibits growth of triple-negative breast cancer cells via inhibition of ca-

nonical Wnt signaling, Biochem. Pharmacol. 87 (2014) 571e578, https://

doi.org/10.1016/j.bcp.2013.12.007.

12] M. Philip, J.F. Samson, P.S. Simi, Clofazimine-induced hair pigmentation, Int. J.

Trichol. 4 (2012) 174e175, https://doi.org/10.4103/0974-7753.100088.

13] Jiabin Xu, Alexey Koval, Vladimir Katanaev L, et al., Beyond TNBC: reposi-

[

tioning of clofazimine against a broad range of Wnt-dependent cancers, Front.

͡

Oncol. 10 (2020) 2825, https://doi.org/10.3389/fonc.2020.602817. In this

Author information

issue.

[

14] L. Cui, S. Mharakurwa, D. Ndiaye, P.K. Rathod, P.J. Rosenthal, Antimalarial drug

resistance: literature review and activities and ﬁ ndings of the ICEMR network,

Am. J. Trop. Med. Hyg. 93 (2015) 57e68, https://doi.org/10.4269/ajtmh.15-

0007.

15] S.G. Franzblau, K.E. White, J.F. O'Sullivan, Structure-activity relationships of

tetramethylpiperidine-substituted phenazines against Mycobacterium leprae

in vitro, Antimicrob. Agents Chemother. 33 (1989) 2004e2005, https://

doi.org/10.1128/aac.33.11.2004 .

16] A. Barteselli, M. Casagrande, N. Basilico, S. Parapini, C.M. Rusconi, M. Tonelli,

V. Boido, D. Taramelli, F. Sparatore, A. Sparatore, Clofazimine analogs with

antileishmanial and antiplasmodial activity, Bioorg. Med. Chem. 23 (2015)

55e65, https://doi.org/10.1016/j.bmc.2014.11.028.

Corresponding authors: E-mail: vladimir.katanaev@unige.ch,

anna.sparatore@unimi.it, Phones: (þ41) 22 379 5353, (þ39) 02 503

1

9 365.

Funding

This work was supported by the Novartis Foundation for

Medical-Biological Research, grant #17C153 to V.L.K.; J.X. was

recipient of the Theodor et Gabriela Kummer PhD fellowship

University of Lausanne)

[17] I. Bassanini, S. Parapini, N. Basilico, A. Sparatore, Novel hydrophilic rimino-

phenazines as potent antiprotozoal agents, ChemMedChem 14 (2019)

(

1940e1949, https://doi.org/10.1002/cmdc.201900522.

[

18] M. Tonelli, F. Novelli, B. Tasso, A. Sparatore, V. Boido, F. Sparatore, S. Cannas,

P. Molicotti, S. Zanetti, S. Parapini, R. Loddo, Antitubercular activity of qui-

nolizidinyl/pyrrolizidinylalkyliminophenazines, Bioorg. Med. Chem. 22 (2014)

Declaration of competing interest

6837e6845, https://doi.org/10.1016/j.bmc.2014.10.035.

[

19] T.J. Cheeseright, M.D. Mackey, J.L. Melville, J.G. Vinter, FieldScreen: virtual

screening using molecular ﬁelds. Application to the DUD data set, J. Chem. Inf.

Model. 48 (2008) 2108e2117, https://doi.org/10.1021/ci800110p.

The authors declare that they have no known competing

ﬁnancial interests or personal relationships that could have

appeared to inﬂuence the work reported in this paper.

[20] A. Jahn, G. Hinselmann, N. Fechner, A. Zell, Optimal assignment methods for

ligand-based virtual screening, J. Cheminf. (2009) 14, https://doi.org/

0.1186/1758-2946-1-14.

[21] I. Bassanini, S. Parapini, C. Galli, N. Vaiana, A. Pancotti, N. Basilico, D. Taramelli,

S. Romeo, Discovery and pharmacophore mapping of a low-nanomolar in-

hibitor of P. Falciparum growth, ChemMedChem 14 (2019) 1982e1994,

https://doi.org/10.1002/cmdc.201900526.

1

Abbreviations

TNBC

MTT

triple-negative breast cancer

3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium

bromide (thiazolyl blue)

[

22] J. Karlsson, A. Ungell, J. Gråsjo, P. Artursson, Paracellular drug transport across

intestinal epithelia: inﬂuence of charge and induced water ﬂ ux, Eur. J. Pharm.

Sci. Off. J. Eur. Fed. Pharm. Sci. 9 (1999) 47e56, https://doi.org/10.1016/s0928-

€

HRMS

DCM

PDX

high-resolution mass spectrometry

dichloromethane

patient-derived xenograft

0987(99)00041-x.

[

23] E. L oꢁ pez-Knowles, S.J. Zardawi, C.M. McNeil, E.K.A. Millar, P. Crea,

E.A. Musgrove, R.L. Sutherland, S.A. O'Toole, Cytoplasmic localization of beta-

14

A. Koval, I. Bassanini, J. Xu et al.

European Journal of Medicinal Chemistry 222 (2021) 113562

catenin is a marker of poor outcome in breast cancer patients, Cancer Epi-

demiol. Biomark. Prev. Publ. Am. Assoc. Cancer Res. Cosponsored Am. Soc.

Prev. Oncol. 19 (2010) 301e309, https://doi.org/10.1158/1055-9965.EPI-09-

and in vivo activities of the riminophenazine TBI-166 against Mycobacterium

tuberculosis, antimicrob, Agents Chemother 63 (2019), https://doi.org/

10.1128/AAC.02155-18 .

0

741.

[30] M.C. Cholo, H.C. Steel, P.B. Fourie, W.A. Germishuizen, R. Anderson, Clofazi-

mine: current status and future prospects, J. Antimicrob. Chemother. (2011)

dkr444, https://doi.org/10.1093/jac/dkr444 .

[31] P. Gur ﬁ nkel, J.C. Pina, M. Ramos-e-Silva, Use of clofazimine in dermatology,

J. Drugs Dermatol. JDD 8 (2009) 846e851 .

[32] H. Tazarki, W. Zeinyeh, Y.J. Esvan, S. Knapp, D. Chatterjee, M. Schr o€ der,

[

24] M.S. Rao, R. Gupta, M.J. Liguori, M. Hu, X. Huang, S.R. Mantena,

S.W. Mittelstadt, E.A.G. Blomme, T.R. Van Vleet, Novel computational

approach to predict off-target interactions for small molecules, Front. Big

Data. 2 (2019), https://doi.org/10.3389/fdata.2019.00025 .

25] H.-H. Lin, L.-L. Zhang, R. Yan, J.-J. Lu, Y. Hu, Network analysis of drug-target

interactions:

2

1

a

study on FDA-approved new molecular entities between

000 to 2015, Sci. Rep. 7 (2017) 12230, https://doi.org/10.1038/s41598-017-

2061-8 .

A.C. Joerger, J. Khiari, B. Josselin, B. Baratte, S. Bach, S. Ruchaud, F. Anizon,

F. Giraud, P. Moreau, New pyrido[3,4-g]quinazoline derivatives as CLK1 and

DYRK1A inhibitors: synthesis, biological evaluation and binding mode anal-

ysis, Eur. J. Med. Chem. 166 (2019) 304e317, https://doi.org/10.1016/

j.ejmech.2019.01.052 .

[

26] T. Cheng, M. Hao, T. Takeda, S.H. Bryant, Y. Wang, Large-scale prediction of

drug-target interaction: a data-centric review, AAPS J. 19 (2017) 1264e1275,

https://doi.org/10.1208/s12248-017-0092-6 .

27] K. Ahmed, H.V. Shaw, A. Koval, V.L. Katanaev, A second WNT for old drugs:

drug repositioning against WNT-dependent cancers, Cancers 8 (2016), https://

doi.org/10.3390/cancers8070066 .

28] V.C. Barry, J.G. Belton, M.L. Conalty, J.M. Denneny, D.W. Edward, J.F. O’sullivan,

D. Twomey, F. Winder, A new series of phenazines (Rimino-Compounds) with

high antituberculosis activity, Nature 179 (1957) 1013e1015, https://doi.org/

[33] D. Zhang, Y. Lu, K. Liu, B. Liu, J. Wang, G. Zhang, H. Zhang, Y. Liu, B. Wang,

M. Zheng, L. Fu, Y. Hou, N. Gong, Y. Lv, C. Li, C.B. Cooper, A.M. Upton, D. Yin,

Z. Ma, H. Huang, Identiﬁcation of less lipophilic riminophenazine derivatives

for the treatment of drug-resistant tuberculosis, J. Med. Chem. 55 (2012)

8409e8417, https://doi.org/10.1021/jm300828h .

[34] A. Koval, V.L. Katanaev, Wnt3a stimulation elicits G-protein-coupled receptor

properties of mammalian Frizzled proteins, Biochem. J. 433 (2011) 435e440,

https://doi.org/10.1042/BJ20101878 .

10.1038/1791013a0 .

[29] J. Xu, B. Wang, L. Fu, H. Zhu, S. Guo, H. Huang, D. Yin, Y. Zhang, Y. Lu, In vitro

15

Article Doi

Optimization of the clofazimine structure leads to a highly water-soluble C3-aminopyridinyl riminophenazine endowed with improved anti-Wnt and anti-cancer activity in vitro and in vivo

DOI: 10.1016/j.ejmech.2021.113562

Source and publish data:

Authors:

Article abstract of DOI:10.1016/j.ejmech.2021.113562

Full text of DOI:10.1016/j.ejmech.2021.113562

Products guided by the article

R&D Labs maybe for 1332634-91-9

Relevant to this article

Hot Product