In silico/in vitro screening and hit evaluation identified new phenothiazine anti-prion derivatives

Article abstract of DOI:10.1016/j.ejmech.2020.112295

Prion diseases or transmissible spongiform encephalopathies (TSEs) are a group of rare neurodegenerative disorders. TSEs are characterized by the accumulation of prions (PrP^Sc) that represent pathological isoforms of the physiological cellular prion protein PrP^C. Although the conversion of PrP^C to PrP^Sc is still not completely understood, blocking this process may lead to develop new therapies. Here, we have generated a pharmacophore model, based on anti-prion molecules reported in literature to be effective in phenotypic assay. The model was used to conduct a virtual screen of commercial compound databases that selected a small library of ten compounds. These molecules were then screened in mouse neuroblastoma cell line chronically infected with prions (ScN2a) after excluding neurotoxicity. 1 has been identified as the therapeutic hit on the basis of the following evidence: chronic treatments of ScN2a cells using 1 eliminate PrP^Sc loaded in both Western blotting analysis and Real-Time Quaking-Induced Conversion (RT-QuIC) assay. We also proposed the mechanism of action of 1 by which it has the ability to bind PrP^C and consequentially blocks prion conversion. Herein we describe the results of these efforts.

Full text of DOI:10.1016/j.ejmech.2020.112295

Journal Pre-proof
In silico/in vitro screening and hit evaluation identified new phenothiazine anti-prion
derivatives
L. Zaccagnini, G. Rossetti, H.T. Tran, G. Salzano, A. Gandini, A. Colini Baldeschi,
M.L. Bolognesi, P. Carloni, G. Legname
PII:
S0223-5234(20)30264-6
DOI:
https://doi.org/10.1016/j.ejmech.2020.112295
EJMECH 112295
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 11 December 2019
Revised Date: 1 April 2020
Accepted Date: 1 April 2020
Please cite this article as: L. Zaccagnini, G. Rossetti, H.T. Tran, G. Salzano, A. Gandini, A. Colini
Baldeschi, M.L. Bolognesi, P. Carloni, G. Legname, In silico/in vitro screening and hit evaluation
identified new phenothiazine anti-prion derivatives, European Journal of Medicinal Chemistry (2020),
doi: https://doi.org/10.1016/j.ejmech.2020.112295.
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 Published by Elsevier Masson SAS.

Cpn 1 binds PrPC
5000 compounds
ZINC online database
PrP : 1 = 2:1
PrP : 1 = 1:1
PrP : 1 = 1:2
2029 compounds
QSAR screening
Membrane
3 kDa
10 compounds
predicted IC50 < 1 µM
Chronic treatment
1 hit molecule
after in vitro screening

In silico/in vitro screening and hit evaluation identified new phenothiazine anti-
prion derivatives
Zaccagnini L1#, Rossetti G2, Tran HT1,4, Salzano G1, Gandini A1,3, Colini Baldeschi A1, Bolognesi
ML3, Carloni P2, Legname G1*
1 Laboratory of Prion Biology, Department of Neuroscience , Scuola Internazionale Superiore di Studi Avanzati
(SISSA) , Via Bonomea 265 , I-34136 Trieste , Italy
2 Institute for Advanced Simulations (IAS)-5/Institute for Neuroscience and Medicine (INM)-9, Forschungszentrum
Jülich, 52428 Jülich, Germany
3 Department of Pharmacy and Biotechnology, Alma Mater Studiorum-University of Bologna , Via Belmeloro 6, I-
40126 Bologna , Italy
4 VNUK Institute for Research and Executive Education – The University of Danang, 158A Le Loi St., Da Nang,
Vietnam
# Current affiliation: Dementia Research Institute, University College London, Cruciform Building Gower street,
London, United Kingdom
* Corresponding author. Laboratory of Prion Biology, Department of Neuroscience, Scuola Internazionale Superiore di
Studi Avanzati (SISSA), via Bonomea 265, 34136 Trieste, Italy. Email: legname@sissa.it
Keywords: QSAR-model, prion, anti-prion agent, competition assay, RT-QuIC, chronic treatment
Abstract
Prion diseases or transmissible spongiform encephalopathies (TSEs) are a group of rare
neurodegenerative disorders. TSEs are characterized by the accumulation of prions (PrPSc) that
represent pathological isoforms of the physiological cellular prion protein PrPC. Although the
conversion of PrPC to PrPSc is still not completely understood, blocking this process may lead to
develop new therapies. Here, we have generated a pharmacophore model, based on anti-prion
molecules reported in literature to be effective in phenotypic assay. The model was used to conduct
a virtual screen of commercial compound databases that selected a small library of ten compounds.
These molecules were then screened in mouse neuroblastoma cell line chronically infected with
prions (ScN2a) after excluding neurotoxicity. 1 has been identified as the therapeutic hit on the
basis of the following evidence: chronic treatments of ScN2a cells using 1 eliminate PrPSc loaded in
both Western blotting analysis and Real-Time Quaking-Induced Conversion (RT-QuIC) assay. We
also proposed the mechanism of action of 1 by which it has the ability to bind PrPC and
consequentially blocks prion conversion. Herein we describe the results of these efforts.
1

Introduction
In recent years, our understanding of the pathogenic mechanisms of neurodegenerative diseases has
grown steadily. However, no effective therapies have yet been developed. These diseases are
recognized as neurological protein misfolding disorders (PMDs) since they are associated with
conformational changes of native proteins into disease-associated conformers [1]. PMDs include
Alzheimer’s disease, involving accumulation of misfolded amyloid β (Aβ) and tau, Parkinson’s
disease in which α-synuclein aggregates, and prion diseases where the physiological form of the
prion protein is converted into its pathological isoforms [2].
In the latter case, prion diseases or TSEs present as sporadic, inherited and infectious disorders [3].
Prions were long thought to be unique diseases. However, accumulating evidence suggests that in
other PMDs other proteins might follow a similar mechanism of seeding, self-propagation and cell-
to-cell spreading [4–8].
Neuropathological changes in TSEs are mainly gliosis, vacuolation and neuronal loss paralleled by
cognitive and motor impairments [9,10]. TSEs include kuru, Creutzfeldt-Jacob disease, Gerstmann
Sträussler-Scheinker syndrome and fatal familial insomnia in humans, bovine spongiform
encephalopathy in cattle, scrapie in sheep and goats, and chronic wasting disease in elk, deer and
other cervids [9]. The etiological agent is the scrapie prion protein (PrPSc), the abnormal, misfolded
isoform of the PrPC [11]. PrPC is anchored to the cell surface through a C-terminal moiety of
glycophosphatidyl-inositol or GPI [12]. Even though the two isoforms share the same primary
sequence, they have several different biochemical and biophysical properties: PrPC is rich in α-
helices, is soluble in nonionic detergents and sensitive to protease K (PK) digestion while PrPSc is
mostly rich in β-sheets, is insoluble in nonionic detergents and partially resistant to PK [13].
Although the conversion of PrPC to PrPSc is still not completely understood, blocking this process
may lead to the development of effective new therapies. Several studies have focused on the ability
of small molecules to interfere with the conversion process, by either binding and stabilizing PrPC
or blocking PrPSc aggregation and accumulation [14–16]. A widely employed strategy has been the
repositioning of compounds registered as antivirals [17], antimalarials [18,19], antifungals [20] and
antidepressants [21]. Drug repositioning is the application of available drugs for treating conditions
different from the original treatment purposes. By using this approach quinacrine (antimalarial) [18,
19], pentosan polysulfate (heparin mimetic) [22,23], doxycycline (antibiotic) [24,25] and flupirtine
(analgesic) [26] were tested in human clinical trials, but with no encouraging results. Other
approaches to develop anti-prion therapies rely on rational medicinal chemistry [26–28], multi-
target approaches [29] and in silico methods [16,30]. However, most of the approaches attempted so
2

far have not resulted in molecules to progress into clinical investigations and all the identified drugs
have inevitably failed [31]. Therefore, development of effective anti-prion small molecules that
have drug-likeness and therapeutic potential remains a major challenge.
Toward this aim, our experimental program began with the generation of a pharmacophore model,
based on anti-prion ligands reported to be effective in phenotypic assays, which was used to
conduct a virtual screen of commercial compounds databases. This approach led to a small target-
biased library, which was then screened in cellular model of the disease. Immortalized
neuroblastoma (N2a) and hypothalamic (GT1) mouse cell lines chronically infected with different
prion strains (RML and 22L) were used to measure anti-prion efficacy, after excluding
neurotoxicity. 1 emerged as a therapeutic hit; it is able to eliminate PrPSc after chronic treatments of
N2a-RML cells as shown in both Western blotting analysis and Real-Time Quaking-Induced
Conversion (RT-QuIC) assay. A mechanism by which prion conversion is blocked upon binding of
1 to PrPC is proposed. However, 1 suffered from poor solubility. We therefore carried out
preliminary SAR studies on this compound, particularly focused on improving this physicochemical
property.
Materials and Methods
2.1 Computational details
Two hundred compounds with known anti-prion activities (IC50) were identified by a literature
search (Table S1) and included into 5 datasets, which are distinguished by inhibiting prion
replication in cell lines and showing different incubation times in animal models of TSEs (Table
S2). Our QSAR model uses Multiple Linear Regression (MLR) to obtain a linear relationship
between the pIC50 values against the molecular descriptors of these compounds. For this, we used
ordinary least squares (OLS) algorithm in the QSARINS (QSAR-Insubria) software [32].
We calculated the molecular descriptors of the compounds using the PaDEL descriptor code [33]
and the Molinspiration (www.molinspiration.com) server.
We first eliminated the descriptors that possessed more than 95% of constant values and had more
than 90% correlation using the QSARINS software [32]. The procedure led to 33 descriptors. It
would be ideal to develop a QSAR model, which uses all combinations of available descriptors for
the calculation of the models; however, the number of combinations is here so large that it is too
computational demanding to include all models in the calculation. By applying genetic
algorithm/multiple linear regressions (GA/MLR) technique, the most relevant descriptors for our
3

dataset were selected to build the final QSAR model. GA/MLR technique explores a broad range of
possible combination of descriptors, searching for the best ones. This is done mimicking the natural
selection; where every descriptor represents a gene, and a set of descriptors represent a
chromosome. The fitness of a chromosome is related to the matching model performances. These
increase with R2 (the correlation coefficient), with the quantity Q2LOO - R2 as close to zero (where
Q2
is the R2 obtained by Leaving One value Out), Q2LMO - R2 as close to zero (where Q2
is
LMO
LOO
the R2 obtained by Leaving Many values Out, around 30% of the data) and RMSE as close to zero
(Root Mean Square Error is the standard deviation of the residuals, i.e. how far from the regression
line data points are; RMSE tells you how concentrated the data is around the line of best fit).
Starting with a pool of chromosomes, small subsets of chromosomes are picked randomly, and the
best become parents. Couples of parent chromosomes are then crossed at a random position
(crossing-over), thus obtaining the offspring, whose chromosomes are combinations of the parent
ones. If among the new chromosomes one or more of them outperform the fitness of the parent
population, these chromosomes will replace the less performing. Repeating the aforesaid procedure
many times, and introducing also random mutations (i.e. descriptor substitution) in the
chromosomes, the result at the end of the procedure is a population of models with better
performances than the models introduced at the beginning.
Our best final model features 5 descriptors. These are LogP, i.e. the partition coefficient of a
compound between aqueous and lipophilic phases (usually octanol and water), compounds’
Molecular Weight (MW), their number of Sulfur atoms (Nsulfur), their number of tricyclic
scaffolds (Ntricyclic) and their molecular size, expressed in terms of an approximate diameter (D).
2.2 Chemistry
All the commercially available reagents and solvents were purchased from Sigma-Aldrich, TCI and
Acros Organics, and used without further purification. Reactions were followed by analytical thin
layer chromatography (TLC), on pre-coated TLC plates (layer 0.20 mm silica gel60 with a
fluorescent indicator UV254, from Sigma-Aldrich). Developed plates were air-dried and analyzed
under a UV lamp (UV 254/365 nm). CEM Discover SP focused microwave reactor was used for
microwave mediated reactions. Nuclear magnetic resonance (NMR) experiments were run on
1
13
1
13
Varian VXR 400 (400 MHz for H, 100 MHz for C). H and C NMR spectra were acquired at
300 K using chloroform (CDCl3), methanol (CD3OD), and dimethyl sulfoxide ((CD3)2SO) as
solvents. Chemical shifts (δ) are reported in parts per million (ppm) relative to tetramethylsilane
(TMS) as internal reference and coupling constants (J) are reported in hertz (Hz). The spin
multiplicities are reported as s (singlet), d (doublet), t (triplet), q (quartet), and m (multiplet).
4

Compounds were named following IUPAC rules as applied by ChemBioDraw Ultra (version 14.0).
Compounds purity was assessed by LC-MS using a SB-C18 RP column (100 x 2.1 mm I.D., 3.5
µm) as the stationary phase and a mixture of 0.1% formic acid in water and 0.1% formic acid in
acetonitrile (60/40, V/V), in isocratic mode at a flow rate of 100 µL/min. Acquisition was
performed by total ion current (TIC) and ESI+. 1 showed a purity ≥98.1%. All the final 11-15
showed a purity >95%.
2.2.1 Synthesis of 7-chloro-4-hydrazynilquinoline (16). A suspension of 4,7-dichloroquinoline
(500 mg) and hydrazine monohydrate 65% (2 eq) was irradiated in a sealed tube at 150 W for 5 min
(T = 150°C, P = 200 psi, Power max = on). After cooling, the resulting solid was diluted with
water, washed, and isolated by filtration. The final compound was purified through crystallization
from ethanol, affording a pale yellow solid. Yield: 69%. 1H-NMR (400 MHz, CD3OD): δ 8.41 (d, J
= 5.6 Hz, 1H); δ 8.00 (d, J = 9.0 Hz, 1H); δ 7.80 (d, J = 1.8 Hz, 1H); δ 7.40 (dd, J = 9.0, 2.0 Hz,
13
1H); δ 6.99 (d, J = 5.6 Hz, 1H). C-NMR (100 MHz, CD3OD): δ 158.55; δ 143.28; δ 140.69; δ
140.05; δ 128.23; δ 125.54; δ 120.16; δ 114.92; δ 99.67.
2.2.2 Synthesis of (Z)-2-(1-(2-(7-chloroquinolin-4-yl)hydrazono)ethyl)-10H-phenothiazine
(compound 1). To a solution of 16 (2.60 mmol) in ethanol (20 mL), 2-acetylphenothiazine (7.80
mmol) and acetic acid (2.5 mL) were added in sequence at room temperature. The solution was
refluxed for 24h, and then evaporated under vacuum. The crude product was purified by column
chromatography on silica gel (DCM/MeOH/NH3 9.2:0.8:0.08), affording 1 as a dark yellow solid.
Yield: 81%. 1H-NMR (400 MHz, DMSO-d6): δ 8.68 (s, 1H); δ 8.40 (d, J = 8.7 Hz, 1H); δ 7.78-7.47
(m, 2H); δ 7.36 (t, J = 11.5 Hz, 2H); δ 7.24 (dd, J = 8.1, 1.8 Hz, 1H); δ 7.08 - 6.81 (m, 4H); δ 6.80 -
6.69 (m, 2H); δ 2.39 (s, 3H).
2.2.3 General procedure for the synthesis of compounds 11-15.
To a solution of 16 (0.3 mmol) in ethanol (3 mL), the corresponding phenothiazine derivatives 17-
21 (0.3 mmol) and acetic acid (50 µL) were added in sequence at room temperature. The solution
was refluxed for 24h, and then evaporated under vacuum. All the final compounds were purified
through column chromatography on silica gel (ethyl acetate/ethanol/toluene/NH3 6.8:0.2:3:0.02).
(E)-2-(2-(1-(2-(7-chloroquinolin-4-yl)hydrazono)ethyl)-10H-phenothiazin-10-yl)-N,N-
dimethylethan-1-amine (11). The title 11 was obtained as a yellow solid, according to the general
procedure using 17. Yield 50%. 1H-NMR (400 MHz, CDCl3): δ 8.62 (s, 1H); δ 8.01 (s, 1H); δ 7.87
(d, J = 8.9, 1H); δ 7.55 (s, 1H); δ 7.48 (d, J = 5.4, 1H); δ 7.44 (dd, J = 8.9, 2.0, 1H); δ 7.35 (d, J =
5

7.9, 1H); δ 7.21-7.15 (m, 3H); δ 6.98-6.95 (m, 2H); δ 4.28 (s, 2H); δ 3.00 (s, 2H); δ 2.53 (s, 6H); δ
2.44 (s, 3H). 13C-NMR (100 MHz, CDCl3): δ 206.87; δ 145.20; δ 144.72; δ 137.49; δ 135.13; δ
127.48; δ 127.15; δ 127.10; δ 125.89; δ 124.75; δ 122.74; δ 120.48; δ 115.79; δ 112.75; δ 102.53; δ
56.91; δ 53.38; δ 45.38; δ 45.02; δ 30.88; δ 29.66; δ 24.42; δ 22.65; δ 14.08; δ 12.61.
(E)-3-(2-(1-(2-(7-chloroquinolin-4-yl)hydrazono)ethyl)-10H-phenothiazin-10-yl)-N,N-
dimethylpropan-1-amine (12). The title 12 was obtained as a yellow solid, according to the general
procedure using 18. Yield 35%. 1H-NMR (400 MHz, CDCl3): δ 8.63 (s, 1H); δ 7.99 (s, 1H); δ 7.77
(d, J = 8.8, 1H); δ 7.43 (d, J = 7.1, 3H); δ 7.29 (d, J = 7.9, 1H); δ 7.24 (s, 1H); δ 7.15 (dd, J = 15.8,
8.0, 3H); δ 6.92 (d, J = 7.0, 2H); δ 4.03 (t, J = 6.3, 2H); δ 2.66-2.48 (m, 2H); δ 2.39 (s, 3H); δ 2.28
(s, 6H); δ 2.15-1.99 (m, 2H). 13C-NMR (100 MHz, CDCl3): δ 206.87; δ 145.20; δ 144.72; δ 137.49;
δ 135.13; δ 127.48; δ 127.15; δ 125.89; δ 124.75; δ 122.74; δ 120.48; δ 115.79; δ 112.75; δ 102.53;
δ 56.91; δ 53.38; δ 45.38; δ 45.02; δ 30.88; δ 29.66; δ 29.32; δ 24.42; δ 22.65; δ 14.08; δ 12.61.
(E)-2-(1-(2-(7-chloroquinolin-4-yl)hydrazono)ethyl)-10-(2-(piperidin-1-yl)ethyl)-10H-
phenothiazine (13) The title 13 was obtained as a yellow solid, according to the general procedure
using 19. Yield 22%. 1H-NMR (400 MHz, CDCl3): δ 8.55 (d, J = 5.2 Hz, 1H); δ 7.94 (d, J = 1.9 Hz,
1H); δ 7.75 (d, J = 8.9 Hz, 1H); δ 7.46 (d, J = 1.6 Hz, 1H); δ 7.41-7.32 (m, 2H); δ 7.32-7.25 (m,
1H); δ 7.18-7.05 (m, 3H); δ 6.92-6.88 (m, 2H); δ 4.12-4.06 (m, 2H); δ 3.70-3.67 (m, 4H); δ 2.82-
13
2.79 (m, 2H); δ 2.55-2.53 (m, 3H); δ 2.36 (s, 3H); δ 2.01 (s, 1H); δ 1.23 (t, J = 7.1 Hz, 2H). C-
NMR (100 MHz, CDCl3): δ 150.54; δ 148.22; δ 147.68; δ 146.90; δ 144.93; δ 144.55; δ 137.38; δ
135.32; δ 127.90; δ 127.50; δ 127.39; δ 127.05; δ 126.46; δ 125.87; δ 124.08; δ 122.72; δ 120.92; δ
120.45; δ 115.92; δ 115.46; δ 112.56; δ 102.35; δ 66.89; δ 55.93; δ 54.02; δ 45.98; δ 12.69.
(E)-4-(2-(2-(1-(2-(7-chloroquinolin-4-yl)hydrazono)ethyl)-10H-phenothiazin-10-
yl)ethyl)morpholine (14) The title 14 was obtained as a dark yellow solid, according to the general
procedure using 20. Yield 91%. 1H-NMR (400 MHz, CDCl3): δ 8.53 (d, J = 5.0, 1H); δ 8.01 (d, J =
2.0, 1H); δ 7.94 (d, J = 8.9, 1H); δ 7.51 (d, J = 1.6, 1H); δ 7.48-7.40 (m, 2H); δ 7.37 (dd, J = 8.0,
1.7, 1H); δ 7.22-7.14 (m, 3H); δ 6.98-6.94 (m, 2H); δ 4.15 (t, J = 6.7 Hz, 2H); δ 3.75-3.73 (m, 4H);
δ 2.87 (t, 2H); δ 2.67-2.56 (m, 4H); δ 2.50 (s, 3H). 13C-NMR (100 MHz, CDCl3): δ 150.54; δ
148.22; δ 147.68; δ 146.90; δ 144.93; δ 144.55; δ 137.38; δ 135.32; δ 127.90; δ 127.50; δ 127.39; δ
127.05; δ 126.46; δ 125.87; δ 124.08; δ 122.72; δ 120.92; δ 120.45; δ 115.92; δ 115.46; δ 112.56; δ
102.35; δ 66.89; δ 55.93; δ 54.02; δ 45.98; δ 12.69.
(E)-2-(1-(2-(7-chloroquinolin-4-yl)hydrazono)ethyl)-10-(2-(4-methylpiperazin-1-yl)ethyl)-10H-
phenothiazine (15) The title 15 was obtained as a dark yellow solid, according to the general
procedure using 21. Yield 37%. 1H-NMR (400 MHz, CDCl3): δ 8.59 (s, 1H); δ 7.98 (s, 1H); δ 7.83
(d, J = 8.5, 1H); δ 7.49 (s, 1H); δ 7.42-7.32 (m, 2H); δ 7.30 (d, J = 8, 1H); δ 7.18-7.11 (m, 3H); δ
6

6.94-6.92 (m, 2H); δ 4.12 (t, J = 6.8, 2H); δ 2.87 (t, J = 6.8, 2H); δ 2.71-2.60 (m, 6H); δ 2.38 (s,
3H); δ 2.35 (s, 3H) δ 1.24-1.18 (m, 2H). 13C-NMR (100 MHz, CDCl3): δ 142.36; δ 142.03; δ
134.86; δ 132.79; δ 124.94; δ 124.85; δ 124.53; δ 123.37; δ 121.67; δ 120.19; δ 117.96; δ 112.96; δ
110.09; δ 99.81; δ 52.54; δ 52.12; δ 50.14; δ 43.46; δ 42.83; 10.24.
2.2.4 Kinetic Solubility Studies.
A variation of a previously reported method was employed to determine the compounds’ solubility
in buffer at pH 7.4 [33]. Briefly, 10 mM DMSO stock solutions of 1 and derivatives 11-15 were
diluted to 200 or 600 μM in PBS (pH 7.4). The mixtures were stirred for 30 min, filtered, and then
injected onto a Jasco Corporation PU-1585 HPLC system (Solvent system: 73% H2O/27% MeCN +
0.1% trifluoroacetic acid; Column: Kinetex® 5μm EVO C18 100 Å, 150 x 4.6 mm; UV: 254 nm).
A 10 mM stock solution was prepared by completely dissolving test compounds in DMSO. This
solution was diluted to five known concentration solutions in MeCN. Each solution was analyzed
by HPLC, and a calibration curve was plotted using the peak areas from the standard
concentrations. The equilibrium solubility of test compounds was determined by quantifying the
concentration of test solutions against the calibration curve (Figure S3). Experiments were run in
triplicate, and results were tabulated and reported as mean ± standard deviation (Table S5).
2.3 Biological evaluation
2.3.1 Cell culture
Mouse neuroblastoma cell line, either non-infected (N2a) or chronically infected with Rocky
Mountain Laboratory (RML) or with 22L prion strains (ScN2a) were grown in Minimal Essential
Medium (MEM)-1% L-glutamax complemented with 10% fetal bovine serum (FBS), 1% non-
essential amino acids (NEAA), and 1% penicillin-streptomycin.
Immortalized mouse hypothalamic neurons (GT1) and chronically infected ScGt1 cells, with both
RML and 22L prion strains, were grown in Dulbecco’s modified Eagle’s medium (DMEM)-1%
GlutaMAX supplemented with 10% FBS and 1% penicillin-streptomycin.
All cell lines were cultivated in 25 cm2 flasks at 37 °C under 5% CO2 or in 10 cm2 Petri dishes.
2.3.2 Compounds
All the compounds were dissolved in 100% dimethyl sulfoxide (DMSO), to a 10 mM stock
solution. From these stock solutions, intermediate dilutions were prepared as needed. For cell
treatment, stock solutions were further diluted in 100% ethanol (EtOH) to a final concentration of 1
mM. Each molecule was then diluted in cell culture medium. In the cell medium, the final
7

concentration of DMSO was never above 0.1%. Detailed treatment conditions are provided in
following methods. Mock controls were treated with vehicle only under the same conditions.
2.3.3 Assessment of cell viability
N2a, N2a chronically infected with RML or 22L prion strain (ScN2a), GT1 and GT1 chronically
infected with RML or 22L prion (ScGT1) cells, were maintained in culture and grown to 80%
confluence. The medium was changed and the cells were detached.
The cell density was determined by cell counting using ScepterTM 2.0 Cell Counter (Millipore).
The cell density was adjusted to 2.5 × 104 cell/mL with MEM (N2a, ScN2a) or 3 × 104 with DMEM
(GT1, ScGT1).
The cell suspension was added to each well of a 96-well, tissue culture-treated, clear bottom, plate
(Costar) and the cells were allowed to settle for 1 day at 37 °C under 5% CO2 prior to the addition
of the compounds.
Each compound (dissolved in EtOH) was diluted in the cell medium to a final concentration of 0.1,
1 and 10 μM. After 24 h, cell culture medium was removed and replaced by compound-containing
medium.
The plate was incubated at 37 °C under 5% CO2 for 5 days.
2.3.3.1 MTT assay
The 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT, SIGMA) was diluted in
phosphate buffer solution (PBS) to a working dilution of 5 mg/mL.
The medium was then removed, and the cells were incubated with the MTT solution for 3 h at 37
°C under 5% CO2.
After incubation, a solution of DMSO/2-Propanol (1:1) was added to each well and the plate was
kept at room temperature (RT) for 30 min before reading. Absorbance was measured at OD = 590
nm by using a Spectramax M5 (Molecular Devices).
2.3.3.2 Cell counting KIT-8
Ten µM of CCK-8 solution (ST-8 [2-(2-methoxy-4-nitrophenyl)-3-(4-nitrophenyl)-5-(2,4-
disulfophenyl)-2H-tetrazolium, monosodium salt) (Dojindo Labroatories) were added to each well
of the plate.
The plate was then incubated for 4 hours in the humidified incubator (e.g., at 37°C, 5% CO2). After
that, the absorbance was measured at 450 nm using the microplate reader (Spectramax M5,
Molecular Devices).
8

2.3.3.3 Calcein assay
Ten μL of 2.5 μM calcein-AM were added, and the plates were incubated at 37°C for 30 min.
Fluorescence emission intensity was quantified using a SpectraMax M5 fluorescence plate reader,
excitation/emission ratio equal to 492/525 nm.
2.3.4 PrPSc and PrPC detection in cell lysates by western blotting
After the treatment with compounds, the amount of PK-resistant PrPSc was measured in ScN2a and
ScGT1 treated with 1 µM of concentration of each compound or then with a titration of 1. After 4
days of treatment, the accumulation of PrP protein was detected by immunoblotting of lysed cells
before and after PK digestion.
2.3.4.1 Cell lysates
After removing the medium and washing the cells with PBS, lysis buffer (10 mM Tris-HCl pH 8.0,
150 mM NaCl, 0.5% nonidet P-40, 0.5% deoxycholic acid sodium salt) was added to the
flasks/dishes, then the cells were detached using a cell scraper (Falcon) and the lysates were
collected and pelleted by centrifugation in a bench microfuge (Eppendorf).
2.3.4.3 PK digestion
Two hundred and fifty μg or 500 μg of protein were digested with 10 μg of PK (Roche) for 1 h at
37 °C.
The reaction was arrested with 2 mM of phenylmethylsulphonyl fluoride (PMSF, SIGMA) and the
PK-digested cell lysates were centrifuged at 55000 rpm for 75 min at 6°C in an ultracentrifuge
(Beckman Coulter).
After the centrifugation, the supernatants were discarded, and the pellets were re-suspended in 2X
or 5X sample loading buffer (125 mM Tris HCl, pH 6.8, 10% 2-mercaptoethanol, 4% SDS, 0.2%
bromophenol blue, 20% glycerol and 200mM DTT) and boiled for 10 min at 100 °C. An amount
equal to 30 μg of total proteins was used for the non-PK-digested samples, boiled for 10 min.
2.3.4.4 SDS-PAGE and immunoblotting
Samples were loaded onto a 12% Tris-Glycine SDS-PAGE gel for protein separation, then they
were transferred to a nitrocellulose membrane (GE Healthcare).
The membrane was blocked with 5% non-fat milk in TBST (Tris 200 mM, NaCl 1.5 mM, 1%
Tween-20) for 1 hour at RT and incubated with 1 μg/mL anti-PrP Fab W226, diluted in blocking
9

solution, overnight at 4 °C. After washes with TBST, the membrane was incubated for 1 h a RT
with secondary antibody (goat anti-mouse IgG F(Ab)2 conjugated with horseradish peroxidase
(HRP), DAKO) diluted 1:1000 in blocking solution. After washes, the signal was detected using
enhanced chemiluminescent system (ECL, Amersham Biosciences) and Uvitec Alliance
(Cambridge).
Densitometric analysis was performed using Uviband Analysis Software.
Data are expressed as mean ± SD, and the values of the controls are adjusted to 100%.
Each experiment was performed in triplicate.
2.3.4.5 PrPC detection in cell lysates by western blotting
The same treatment conditions reported above were used for non-infected N2a and GT1 cells. After
4 days of drug treatment, the amount of PrPC was measured by immunoblotting of lysed cells.
Thirty μg of total proteins were used for the analysis. Same protocol as above was used, without PK
digestion step. Each experiment was performed in triplicate.
2.3.4.6 PrPSc quantification by ELISA assay
PK digestion of cell lysates was performed as described above. After ultracentrifugation, pellets
were dissolved and denatured in 50 μL of 8 M guanidine hydrochloride (GdnHCl) in coating buffer
(0.1 M sodium bicarbonate, pH 8.2) for 1 h and diluted into 500 μL of coating buffer.
Twenty μL of the suspension were transferred to 96-well MaxiSorp ELISA plates (Nunc), with each
well containing 180 μL of coating buffer and the plates were sealed and incubated overnight at 4
°C. To increase the immuno-reactivity of PrPSc, coated proteins were denatured in situ. Fifty μL of
8 M GdnHCl were added to each well and incubated for 10 min at room temperature. The ELISA
plates were washed three times with TBST and blocked with 200 μL of 3% BSA in TBS (20 mM
Tris-HCl, 137 mM NaCl, pH 7.5) for 1 h at 37 °C. After three washes with TBST, the plates were
incubated with anti-PrP antibody W226 (1.5 μg/mL) in 1% BSA/TBS, at 37 °C for 2 h. After seven
washes with TBST, goat anti-mouse IgG Fab conjugated to HRP diluted 1:1000 in 1% BSA/TBS
was added to the plates and incubated at 37 °C for 1 h. Again, plates were washed seven times with
TBST, and then developed with 1-step TMB (3,3′,5,5′- tetramethylbenzidine) Turbo ELISA HRP
substrate (Pierce).
The reaction was stopped by the addition of 2 M sulfuric acid to the plates. Absorbance at 450 nm
was measured using a microplate reader (VersaMax, Molecular Devices). Dose-response curves and
IC50 values were computed using IGOR (Demo Version 6.32).
10

2.3.5 Recombinant full-length mouse PrP production and purification
The mouse construct encoding for full-length PrP was expressed in competent BL21 Rosetta2
(DE3) cells Escherichia coli (Stratagene).
Freshly transformed overnight culture was inoculated into Luria Bertani (LB) medium and 100
μg/mL ampicillin and 30 μg/mL chloramphenicol. At 0.8 OD600 expression was induced with
isopropyl b-D galactopyranoside (IPTG) to a final concentration of 1 mM. Cells were grown in a
BioStat-B plus fermentor (Sartorius). The cells were lysed by a homogenizer (PandaPLUS 2000)
and the inclusion bodies were suspended in buffer containing 25 mM Tris-HCl, 5 mM EDTA, 0.8%
TritonX100, pH 8, and then in bi-distilled water several times.
Inclusion bodies containing MoPrP (23-231) were dissolved in 5 volumes of 8 M guanidine
hydrochloride (GdnHCl), loaded onto pre-equilibrated HiLoad 26/60 Superdex 200-pg column, and
eluted in 25 mM Tris–HCl (pH 8.0), 5 mM ethylenediaminetetraacetic acid, and 5 M GdnHCl at a
flow/rate of 1.5 mL/min. Protein refolding was performed by dialysis against refolding buffer [20
mM sodium acetate and 0.005% NaN3 (pH 5.5)] using a Spectrapor membrane (molecular weight,
10000). Purified protein was analyzed by SDS-polyacrylamide gel electrophoresis under reducing
conditions and Western blot.
2.3.6 Competition assay
A mix of recombinant mouse full length PrP ( MoPrP(23-231)) and 1 was incubated at 37 °C in 350
rpm of shaking for 30 minutes.
The mixes were different:
•
•
•
•
PrP : 1 = 1:1 = 1µM : 1µM
PrP : 1 = 2:1 = 2µM : 1µM
PrP : 1 = 1:2 = 0.5 µM : 1µM
PrP : 1 = 1:5 = 0.2 µM : 1µM
After 1 hour, the mix and the other treatments (1 alone) were centrifuged at 2000 rpm in Nanosep
centrifugal device (Merk) with Omega membrane 3 kDa to pass the whole solution through. These
tubes are permeable to 1 (416.93 Da) but not to the PrP (25 kDa) and to the PrP-1 complex (Figure
S1).
The mix and the other treatments were added to the cells as a normal treatment for 4 days. Then the
cells were lysed, and results observed by SDS-page and RTQuIC.
2.3.7 RT-QuIC procedure
11

After purification, aliquots of the recombinant PrP (MoPrP(23-231)) were stored at −80 °C in 10
mM phosphate buffer (pH 5.8). Before each test, the protein solution was allowed to thaw at room
temperature and filtered using Millex-GV filter 0.22 µm (Millipore).
The final reaction volume was 100 μL loaded into the plate (ViewPlate-96 F TC/50x1B, Perkin
Elmer) and the reagents (Sigma) were concentrated as follow: 150 mM NaCl, 0.002% SDS, 1X
PBS, 1 mM EDTA, 10 μM ThT and 0.2 mg MoPrP(23-231) ml−1.
The seed consists of sonicated ScN2a-RML cells. Before the sonication the cells were collected in
100 µl of PBS 1X, after the sonication the sample was quantified using the BCA assay, in order to
use it as a seed (1 µg of the protein).
After the addition of 10 μL of seed, the plate was sealed with a sealing film (Perkin Elmer) and
inserted into a FLUOstar OPTIMA microplate reader (BMG Labtech). The plate was shaken for 1
minute at 600 rpm (double orbital) and incubated for 1 minute at 45 °C.
Fluorescence readings (480 nm) were taken every 30 minutes (30 flashes per well at 450 nm).
Given the rapid response, a specific threshold was set to decrease the likelihood of false positives.
A sample was considered positive if the mean of the highest two fluorescence values (AU) of the
replicates was higher than 10000 AU and at least two out of three replicates crossed the threshold
that was set at 30 hours. This reaction cutoff was established because in all the experiments there
were wells only with MoPrP(23-231) and in these cases no positive RT-QuIC reactions were
observed until after 30 hours.
2.3.8 Immunofluorescence of fixed cells, the surface staining
N2a cells were seeded to semi-confluence in each well of a 24-well plate containing a poly-lysine
coated coverslip and an appropriate culture medium coverslip for 24 hours.
After one day of incubation, the cells were put on ice for 15 minutes.
Then the cells were stained with the primary antibody (W226 1:200) in culture medium for 20
minutes, always in ice. After this time, the medium was removed, and the cells were washed 2 times
with PBS.
The cells were then removed from the ice and were fixed using 4 % of paraformaldehyde (PFA) for
20 minutes. The PFA was discharged and the cells were washed 3 times for 5 minutes with PBS.
Blocking buffer, consisting of 7% Normal Goat Serum (NGS) in PBS, was added to the cells for 1
hour at room temperature. After the blocking, the cells were incubated with a secondary antibody
(Goat anti-mouse [GαMo]-AlexaFluor488, Life Technologies) diluted 1:1000 in the incubation
buffer (1% NGS, 0.02% triton-100 in PBS) for 1 hour at room temperature in the dark.
12

After 2 washes of 5 minutes in PBS, the cells were incubated with 0.1-1 µg/ml of DAPI (Life
Technologies) in PBS-T for 5 minutes.
The cells were washed 3 times in PBS for 5 minutes and then the coverslips were mounted with a
drop of Fluoromount-G (Invitrogen). The coverslips were sealed with nail polish to prevent drying
and movement under the microscope.
Images were acquired with C1 confocal microscope (Nikon). FITC filter was used for detection of
PrP specific staining, while Dapi specific staining was acquired with 500 nm filter.
Results
3.1 In silico screening
3.1.1 QSAR model
We first constructed a list of ca. 200 compounds with known anti-prion activities (IC50) by
collecting data from 14 publications (Table S1). The compounds were divided into 5 datasets
depending on the type of cell line used in the assays and their incubation time (see Table S2). For
each dataset, we generated a QSAR model. Here we present the best model (according to criteria
specified in the Materials and Methods section and below) across the five datasets: this is the QSAR
model based on the ScN2a cell line dataset, see below.
For each compound in the dataset, constitutional, topological, geometric, electrostatic, hydrophobic
and steric descriptors were generated. Plus the drug-likeness (Lipinski’s rule of 5) [34] was
calculated by using PaDEL descriptor code [35] and the Molinspiration webserver
(www.molinspiration.com).
Several of these descriptors were correlated with each other or did not display sufficient variability
within the dataset and were therefore discarded (see Material and Methods section). Next, the most
relevant descriptors for our dataset were identified by applying genetic algorithm/multiple linear
regressions (GA/MLR) technique. This is commonly used with experimentally measured pIC50
[36][37][38].
The resulting best QSAR model was found to depend on the following descriptors: the compounds’
LogP (logarithm of partition coefficient of the molecule in a octanol/water system), their Molecular
Weight (MW), their number of Sulfur atoms (Nsulfur), their number of tricyclic scaffolds
(Ntricyclic) and their molecular size, expressed in terms of an approximate diameter (D).
The best QSAR model, reporting predicted pIC50 values as a function of the five descriptors, is
shown in Figure 1.
13

The expression of pIC50 reads:
pIC50= 0.2344(0.066)LogP
0.0047(0.008)MW
+
0.3597(0.1135)NSulfur
+
0.6237(0.2450)NTricyclic+ 0.2278(0.0242)D + 2.6812
n=38, R2=0.80, R2adj=0.77, s=0.34, F=26.06, Q2loo=0.72, Q2lmo,30%=0.71, R2- R2adj=0.031
where n is the number of compounds of the training set, s is standard error of estimate, F is variance
ratio, R2 is the coefficient of determination, R2adj is an “adjusted” R2 based on the number of
independent variables in the model: it is indicative of the convenience to add a new descriptor to
the model. Our value is comparable with that of R2. Therefore, the model exhibits no overfitting. In
other words, it presents a good fit with minimum number of descriptors. Q2
(Leave One Out)
LOO
and Q2
(leave many out) are values of R2 calculated by removing randomly one or 30% of the
LMO
values. The criterion based on Q2
is particularly important as it introduces a significant
LMO
perturbation on the data in contrast to Q2LOO . Both values should be rather close to R2 for the model
to be robust. This is the case here.
Figure 1. QSAR model obtained by QSARIN software on ScN2a cell line
3.1.2 Virtual screening
The values of IC50 were calculated for 2029 compounds using our QSAR model. These compounds
were identified by screening the Zinc database (http://zinc.docking.org) according to the following
criteria (i) they follow the Lipinski rule of 5; (ii) they are within the Applicability Domain of our
proposed model; (iii) they contain a moiety similar to chloroquine. The compounds with the largest
values of predicted IC50 were selected for the binding assays (Figure 2).
14

Figure 2. Distribution of the predicted pIC50 among the 2029 compounds extracted from the ZINC database.
Example of compounds with highest predicted pIC50 is highlighted in the red box.
A small set of 10 molecules (1-10 in Figure 3) was selected, spanning different bicyclic and
tricyclic scaffolds (e.g. quinoline, acridine and phenothiazine derivatives).
3.2 In vitro screening
The in vitro screening of 1-10 was performed using the N2a cell line chronically infected with RML
prion strain. The ability of all compounds to reduce PrPSc level was determined by Western blotting
followed by densitometry of the PK resistant PrPSc. The first assessment of the compounds was
performed at 1 µM concentration.
15

Figure 3. Chemical structure of the 10 selected compounds
3.2.1 Inhibition of PrPSc replication at a concentration of 1 µM
Before starting the evaluation of the anti-prion potential of 1-10, we determined their effects on cell
viability using three different assays: MTT assay, CCK-8 kit, calcein assays and setting a toxicity
threshold of 80% [39]. Positively, no toxicity was observed in the assays (Table S3). Then, their
ability to reduce the level of the resistant PrPSc in prion-infected cells was determined by Western
blotting densitometric analysis. Relative amounts of PK-resistant PrPSc were measured comparing
to untreated ScN2a-RML cell lysates.
Molecules 3, 4, 5, 7, 8 and 10 in four independent experiments did not decrease the levels of PrPSc.
2 and 9 showed mild anti-prion efficacy, while the treatment of ScN2a-RML cells with 1 resulted in
harboring 35% of PK-resistant PrPSc (Figure 4A-B).
A
16

B
Compound
−PK
+PK
Ctrl
100
100
Mock
92.4 ± 7.2
55.7 ± 6.3
63.8 ± 6.0
92.5 ± 8.9
92.6 ± 11.1
95.6 ± 12.1
109.5 ± 20.4
99.6 ± 20.7
69.1 ± 3.5
83.5 ± 14.0
85.8 ± 11.6
94.0 ± 8.1
35.6 ± 8.8
70.9 ± 17.3
94.4 ± 12.8
108.6 ± 19.3
82.3 ± 18.0
101.7 ± 3.1
82.4 ± 11.8
85.4 ± 12.1
77.1 ± 17.3
88.0 ± 19.9
1
2
3
4
5
6
7
8
9
10
Figure 4. Anti-prion effect evaluation of all compounds at 1 µM of concentration. (A) Western blot of ScN2a-RML
cell lysates depicting the presence or absence of prions following the treatment with the compounds before (down) or
after (up) PK digestion. The proteinase-K digested cell lysates were detected with W226 anti-PrP monoclonal antibody.
Bands represent di-, mono-, and un-glycosylated isoforms at approximately 32, 23 and 18 kDa, respectively. (B) The
values under –PK and +PK represent the percentage of the level of total amount of PrP (–PK) and of PrPSc (+PK)
normalized to the control (100%). Values are the means ± SD calculated from the four independent experiments
performed.
3.2.2 Prion inhibition of 1 on ScN2a cells
17

The promising activity of 1 at 1 µM was confirmed employing 8-point dose-response curves
ranging from 0 to 4 µM. In parallel, we assessed cell viability at different concentrations of 1 (up to
10 µM) and its non-toxic profile was confirmed (Table S4).
The IC50 was calculated using an ELISA assay of PK digested samples from ScN2a-RML cells and
confirmed by Western blotting analysis. We measured an IC50 (calculated with IGOR software) of
0.42 ± 0.1 µM (Figure 5).
Figure 5. IC50 determination of 1 on ScN2a-RML. (A) Dose-response curve form ELISA assay of 8 different
concentrations of 1. (B) Western blot of ScN2a-RML cell lysates treated with 1 at the same concentrations used in (A),
after digestion with proteinase K.
As 1 showed a very promising profile in terms of safety and dose-dependent activity, the next step
was to resynthesize 1, with the aim of validating the biological activity of the vendor-supplied
sample, and in order of obtaining a sufficient amount to perform further studies.
Particularly, as no synthetic protocol was reported, we aimed to develop a versatile synthetic
procedure that would also allow further chemical manipulation, starting from easily accessible
building blocks.
Thus, the synthesis of 1 was achieved through a simple two-step reaction, depicted in Scheme 1.
Hydrazine 16, was synthesized through a solvent-free, nucleophilic substitution of 4,7-
dichloroquinoline by hydrazine using microwave irradiation [40]. Then, 16 (1 eq) was condensed
18

with 2-acetylphenothiazine in acetic acid and ethanol, affording 1 in a good yield (81%). 1 identity
was confirmed by analytical (HPLC), full scan MS (ESI+) and spectroscopic data (1H-NMR),
which were identical to those of the commercial sample.
Scheme 1. Synthetic procedure for the synthesis of 1. Reagents and conditions: a) NH2NH2, 5 min, 150°C, MWI; b)
EtOH, AcOH, 24h, reflux.
3.2.3 Prion inhibition of 1 on different cell lines and prion strains
It is known from the literature that several small molecules are active only on precise cell lines and
on some prion strains and not on others, these phenomena are called cell and strain specificity [41–
43]. The strain specificity occurs with molecules that exert their interaction by either directly
binding to PrPSc or both PrPC and PrPSc, since the strain specificity seems to derive from the
alternate folding of PrPSc [44]. Therefore, we decided to test 1 on three other cell lines: ScN2a
chronically infected with 22L mouse prion strain, ScGT1 infected with two prion strains, RML and
22L.
In Figure 6A, the Western blots of cell lysates are shown. In accordance with the previous results, 1
resulted to be not toxic at 4 µM on these cells (data not shown)).
On ScN2a-RML 1 reduces the level of PrPSc to about 10%, showing a strong clearance of PrPSc.
This activity is maintained in the other cell lines (5% for N2a-22L and 10% for GT1-RML and GT1
22L) (Figure 6B). The total amount of PrP (samples non-treated with proteinase K) also seems to be
reduced in all the cell lines. This reduction is probably due to the clearance of PrPSc visible in the
PK-treated samples.
Controls are normalized to 100% of untreated PK-digested samples (upper square in Figure 6A) the
amount of PrPSc may not result the same and this reflects the difference in PrPSc protein level in
different cell lines.
19

B
Cell line
Concentration (µM)
−PK
100.0
+PK
100.00
0
4
0
4
0
4
0
4
ScN2a-RML
62.1 ± 3.4
100.0
9.4 ± 2.8
100.0
ScN2a-22L
ScGT1-RML
ScGT1-22L
48.7 ± 12.4
100.0
5.1 ± 4.5
100.0
42.0 ± 3.8
100.0
12.3 ± 5.3
100.0
39.2 ± 9.6
11.0 ± 8.0
Figure 6. Activity of 1 on the total PrP or PrPSc on ScN2a-22L, ScGT1-RML, ScGT1-22L. (A) One of the three
WBs carried out using the cell lysates of treated and untreated cells with 1, before and after PK digestion. (B)
Quantification of the three independent experiments performed. The values under –PK and +PK point the percentage of
the level of the total amount of PrP (–PK) and of PrPSc (+PK) normalized to the control (100 %). Values are the means
± SD calculated from the four independent experiments performed.
3.3 Understanding the mechanism of action of 1
Small molecules can block prion replication through three main strategies: (i) by directly blocking
prion replication (via PrPC stabilization or PrPSc destabilization); (ii) by modulating other proteins
involved in the prion replication; (iii) by increasing PrPSc clearance/degradation.
We therefore attempted to elucidate the mechanism responsible of the strong reduction of prions by
1.
20

3.3.1 Compound 1 does not affect the PrPC localization from the cell membrane
Antipsychotic drug chlorpromazine (CPZ) has been shown to inhibit prion replication by directly
binding to PrPC [45]. However, biophysical studies have revealed that its affinity to PrPC is in a
high micromolar range (KD=421 µM), while the active concentration on infected cells is 100 times
lower. Knowing that the antipsychotic effect of CPZ involves the clathrin-mediated endocytosis
(CME) and that CME allows the recycling of PrPC from the plasma membrane, a recent study
shown that CPZ promotes relocalization of PrPC from the cell surface to the intracellular
compartments [14].
CPZ and 1 are both phenothiazine derivatives. So, based on the similar chemical structure, the
following experiments were conducted to test if the mechanism of action of 1 involves a
redistribution of PrPC.
An immunofluorescence surface staining was performed using N2a cells treated with compound 1
at two different concentrations: 0.4 and 4 µM. We expected that if 1 can promote a re-localization
of PrPC as in the case of CPZ treatment (Figure S3), the samples treated with the molecule at 0.4
µM, but especially at 4 µM, should exhibit a decrease in the level of PrPC stained in respect to the
untreated and the mock controls. On the contrary, we noticed that treated samples had the same
staining pattern of controls (Figure 7).
We concluded that anti-prion effect of 1 does not involve a PrPC redistribution from the cell surface
to the intracellular compartments, as is the case for CPZ and other phenothiazine derivatives.
Figure 7. Surface immunofluorescent staining on N2a cells treated with 1. Two different concentration of 1 were
used: 0.4 µM and 4 µM. PrPC present on the surface is labeled in green, and in blue the nucleus.
3.3.2 Activity of 1 in the in vitro prion conversion
1 does not affect the levels of the cellular PrP (data not shown) and the treatment of uninfected N2a
cells does not disturb redistribution of the PrPC from the plasma membrane to intracellular
compartments.
21

In order to elucidate mechanism of action of 1, we checked whether it is actually able to bind the
PrPC.
We used the SPR to check a hypothetical binding between 1 and the PrPC and eventually calculate
the dissociation constant (KD). Unfortunately, due to interference with EtOH/DMSO solution in
which 1 was dissolved, this technique was inconclusive. For this reason, we tried to avoid the use of
SPR and we designed a new protocol to evaluate the possible binding between 1 and PrPC, that we
named competition assay (see Materials and Methods section and Figure S1).
In this assay, PrP and 1 were mixed in different ratios and incubated with constant shaking for 30
minutes. Then the mix and 1 alone were centrifuged in Nanosep centrifugal device with a 3 kDa
membrane to pass the whole solution through. These tubes are permeable for 1 (416.93 Da) but not
for the PrP (25 kDa) and the hypothetical PrP-1 complex.
All the solutions were then analyzed using two different approaches: with WB where they were
added to the N2a-RML cells and in the RT-QuIC where they were loaded into a 96-well plate. If the
binding between PrP and 1 would occur, a decrease of PK-resistant PrPSc in the WB and inhibition
of the prion aggregation in RT-QuIC are expected.
3.3.2.1 Competition assay on ScN2a cells
Then, the solution from the bottom of the tube after the centrifugation was added to the ScN2a cells
as a normal treatment for 4 days.
As a negative control, we used cells treated just with the vehicle. The cells incubated with 1 µM of
1 without the centrifugation in the Nanosep centrifugal device and the ScN2a treated with 1 µM of
1 after the use of Nanosep tubes were utilized as positive controls.
From the WB analysis (Figure 8A), we noticed that the treatment with 1 without the use of Nanosep
tube and the treatment with the solution obtained after putting 1 alone in the tube caused the same
effect on the ScN2a. In both cases, it is possible to see a strong and a very similar reduction of the
PrPSc. This result confirmed the functionality of the Nanosep device, because it means that 1 alone
is able to cross the membrane, indeed 3kDa > 416.93Da (1 MW). Once we concluded this, we
checked what happened when the PrP and 1 are put in the Nanosep tubes after a pre-incubation at
37 °C with shaking. As it is possible to see in Figure 8A, when the ratio is 2 moles of PrP and 1
mole of 1 or 1 mole of PrP and 1 of 1, the level of PK-resistant PrPSc is not affected by a strong
reduction as it happened after the addition of 1 alone.
This means that when the two molecules are incubated in the ratios 2:1 and 1:1, there is a binding
between 1 and PrP that does not allow the 1 to cross the membrane, otherwise the results would
have the same pattern of those in line 1 and 1 (N), since the 1 crosses the 3kDa membrane of the
22

Nanosep tubes. After the quantification, this result is even more evident. Treatment directly with 1
or after the use of Nanosep device produced a PrPSc inhibition around 30%, while in the 2:1(N) and
1:1(N) did not cause any or a strong inhibition, the values are around 100% and 90% (Figure 8B).
When the mixes were prepared in a ratio 1:2, 1 mole of PrP and 2 moles of 1, the level of resistant-
PK PrPSc decrease to 65%, demonstrating that a part of 1 was linked to the PrP and the other
crossed the Nanosep membrane and affected the level of PrPSc. We also used a ratio in which the
concentration of 1 is 5 times higher than the PrP (1:5). The strong inhibition of PrPSc from 100% to
23.6%, using the ratio 1:5, could be explained again through the binding between PrP and 1. In this
case the level of PrPSc is the same as in the samples treated just with 1, probably because 1 is in a
large excess causing the majority of it to cross the 3 kDa membrane.
All the results obtained using the different ratios and treating the cells with the solution present in
the Nanosep device’s bottom, led us to believe that the 1 is actually able to bind the PrP.
The level of total PrP followed the same trend of PrPSc, even if the decrease is less evident. This
shows that 1 is able to affect just the PrPSc level, although it binds the PrPC.
B
Condition
Ctrl
−PK
+PK
100
100
1
48.8 ± 4.5
51.7 ± 6.7
114.4 ± 7.1
117.6 ± 7.3
98.4 ± 5.2
41.0 ± 4.8
25.2 ± 7.8
22.6 ± 9.5
108.2 ± 3.8
90.1 ± 2.9
65.7 ± 5.7
13.6 ± 9.0
1 (N)
2:1 (N)
1:1 (N)
1:2 (N)
1:5 (N)
23

Figure 8. Competition assay in ScN2a cell line. (A) Representative WB of the cell lysates of ScN2a treated with the
different conditions. The first line corresponds to the untreated sample, then in order: sample incubated with 1 µM of 1,
samples treated with the solution obtained from the bottom part of Nanosep device after centrifugations of 1 alone, cells
incubated with the solution obtained from the bottom part of Nanosep device after centrifugations of a pre-incubated
PrP and 1 in ratios 2:1, 1:1, 1:2 and 1:5. In the upper part the cell lysates are represented after the PK-digestion (PrPSc),
in the bottom the samples without the PK treatment. (B) The table shows the quantification of the three-independent
experiment performed. The values under −PK represent the normalization of the total PrP on the actin, the values under
+PK represent the PrPSc on the total PrP. Values are the means ± SD calculated from the three independent experiments
performed.
3.3.2.2 Competition assay using RT-QuIC
To confirm our results, we decided to perform RTQuIC analysis. WB and RTQuIC are different
techniques, with advantages and disadvantages. In the WBs it is possible to see the effect of the
treatment on living cells, but it is possible to detect only the presence of PK-resistant PrPSc and only
if there is enough amount of protein. On the other side, with the RTQuIC technique it is not
possible to consider all the other potential factors that are present in cells and that can trigger the
spreading of PrPSc, however, it is able to detect very low amounts of prions.
In the RT-QuIC assay, the MoPrP (23-231) alone was used as a negative control, because it is
known that this protein is not able to aggregate in the absence of a seed [46]. Instead, the positive
control consists of PrP with a PrPSc seed. We tested two different amounts, 1 µg and 10 µg, and
tested which one gave the best result in terms of seeding activity and reproducibility. Figure 9 A
and B show the curves using the two different seeds.
24

Figure 9. Aggregation curves of MoPrP (23-231) using 1 µg or 10 µg of the seed and after the treatments with
the solution obtained with the competition assay. Three independent RT-QuIC experiments, each one represented by
a specific color. (A) The aggregation is promoted by 1 µg of the seed. (B) The aggregation of PrP is allowed by 10 µg
of the seed. (C) The dark blue line is the PrP alone, the orange is the PrP with the seed, in gray the aggregation curve of
the PrP after the addition of the solution obtained after the pre-incubation and centrifugation in the Nanosep device of
the mix consisting PrP and 1 in a ratio 2:1, in yellow 1:1 and in light blue 1:5, in green the 1 alone at 1 µM. Each curve
represents the average of three independent experiments.
Using only the seed at 1 µg, we obtained an aggregation curve, for this reason we decided to use 1
µg of seed for the later experiments.
We have set up an RT-QuIC protocol, by which the negative control consists of only PrP, that did
not show a curve aggregation and the positive control which it has a reproducible curve aggregation
after about 1050 min (17,5 hours) (Figure 10C).
As before, the same competition assay procedure was performed, but instead of cell treatment, the
solutions were added to the 96 well plate.
The experiments confirmed what we saw in the WBs, the solution obtained from the pre-incubation
of PrP:1=2:1 after several centrifugations using the Nanosep device contains prions. Indeed, it is
possible to observe that the aggregation curves have the same ThT fluorescence, around 6000 AU,
and the lag phase (1050 min) as the positive control. Also, in the WBs analysis, the level of PrPSc in
this condition was the same of the untreated control. So, these two different assays support our
hypothesis that 1 binds PrP and so it is not able to cross the 3kDa membrane present on the
Nanosep device and to inhibit PrPSc and its aggregation. When the ratio used is PrP: 1=1:1, the
aggregation curves of the three independent experiments are not comparable. Also, in this case the
RT-QuIC assay confirmed the WBs results. The increase of the lag phase and the decrease of ThT
fluorescence can mean a lower presence of prions and this could correspond to the results of the
WB, where after the treatment with the solution obtained from PrP: 1=1:1 mix, the level of PrPSc
started to decrease. These results mean that at the ratio 1:1, almost all molecules of 1 bind the PrP,
causing just a small inhibition of prions aggregation. In Figure 9C, results from the RT-QuIC of PrP
25

treated with the solution obtained after the pre-incubation and centrifugation in the Nanosep tubes
of the mix PrP:1 = 1:5 is also represented. The delayed aggregation curves (starting around 50
hours) and the low ThT fluorescence (4000 AU) could be due to a low presence of prions, that
corresponds exactly to the strong inhibition of the PK-resistant PrPSc seen in WB. This again could
mean that even if 1 binds PrP it is in a large excess and can cross the 3kDa membrane. So, when the
solution in the bottom of the Nanosep tube is added to the 96-well plate, there is enough of 1, which
is able to bind the PrPC and partially inhibit the aggregation.
Control experiments were also performed, in order to confirm that the vehicle (EtOH-DMSO) does
not influence the results (Figure S2).
Taking together the WBs and RT-QuIC experiments, we conclude that the 1 is able to bind the PrP
(23-231) and so the cellular prion protein.
3.3 Effect of chronic treatment with 1 on ScN2a cell line
A possible cause of the common clinical failure of several small molecules that had been previously
shown to decrease the level of mouse PrPSc and/or inhibit the prion aggregation [30,47–50] is the
drug-prion resistance developed after a continuous chronic treatment and the prion strain selection.
This is what happened for quinacrine treatment in prion-inoculated MDR0/0 mice [51].
In order to assess if ScN2a-RML cells treated with 1 can be walking into a strain resistant and to
check if the therapeutic potential of 1, a chronic treatment was performed. We treated ScN2a-RML
cells with 4 µM of 1 for 5 passages that correspond to 20 days of treatment. After this period, the
incubation with 1 was stopped and the cells were maintained in culture without any treatment for
other 5 passages to check if this kind of treatment with 1 for 20 days is able to clear completely
cells from prions (Figure 10A).
After 40 days, 10 passages of ScN2a-RML cells were obtained. This experiment was performed 3
times and each time in double, because we assessed the activity of 1 in chronic treatment using both
WB analysis and RT-QuIC assay.
For the WBs, we used the lysates of uninfected N2a cells and the ScN2a-RML before the treatment
(#0) as controls. So, looking at the WB, it is possible to notice that the PrPSc in N2a cells is absent,
as expected. Indeed, after PK-digestion no signal was detected. At #0, ScN2a-RML cells showed
the typical pattern of the RML prion strains: the three bands with the mono-glycosylated one more
evident. But immediately after the first passage of the 4 µM treatments with 1, PK-resistant PrPSc is
almost absent. In the following passages (from #2 to #10) the PrPSc is not even detectable. It is very
important to focus the attention on the fact that from the #6 to #10 the treatment with 4 µM of 1 is
stopped. So, the absence of PrPSc signal in these passages shows that the 1 is able to “cure” the
26

ScN2a-RML cells from prions and that it does not undergo a drug-prion resistance. The level of
total PrP was evaluated, observing the samples not treated with PK-digestion. The total PrP does
not reflect the results obtained for the PrPSc, because in this case the #0 and #1 have the exactly
same pattern, while in the PK-treated samples the level of PrP was strongly reduced already in the
#1. Conversely, the total PrP signal from #2 until #10 is unchanged, but is exactly the same of the
PrPC signal from the uninfected N2a cells (line 1). Thus, from the WB it seems that in the samples
of ScN2a-RML since the #2 the only prion protein present is the cellular prion protein.
The same samples were checked also in the RT-QuIC assay. After cell collection, the pellet was
dissolved in 50 µl of PBS and then sonicated. The sonicated proteins were quantified, as already
mentioned in the previous paragraph, and 1 µg of them was used as seed. In Figure 10C, it is
possible to observe the RT-QuIC experiments that we performed with the same type of samples
used for WBs, to compare the two different approaches. In the control (Figure 10C, orange line),
without treatment with 4 µM of 1, prion aggregation started after 1050 minutes (17,5 hours). The
same result was obtained in the #1, where the ThT fluorescence intensity and the lag phase were
identical as in #0. At #1 the aggregation curve obtained in RT-QuIC (Figure 10C, gray line) reflects
the WB result without the digestion with the proteinase K (Figure 10B), but not the result of the
PK-digested sample.
While in WB, at #2, in both PK and non-PK treated samples, the PrPSc seemed completely cleared,
in RT-QuIC experiments, the lag phase was a little elongated (1260 minutes, 21 hours) and the ThT
fluorescence was decreased, for some peaks to 7000 AU in #0 and #1 to 5000 AU in #2 (Figure
10C, yellow line). The same kind of incongruence was noticed in #3, because in the WB the PrPSc
signal is completely absent, while in the RT-QuIC the presence of prions is evident by the
aggregation curve (Figure 10C, light blue line). At passage #4, in two out of three experiments, the
PrP is not able to aggregate; just one sample started the aggregation around 45 hours (2730
minutes) (Figure 10C, light green line). We did not consider it as a technical problem, because all
the 4 wells of that experiment aggregate at the same time and with the same intensity, rather we
believed that after 4 passages of treatment with 1, the prions in almost all the experiments were
cleared but it is possible that sometimes they were not completely eliminated. Importantly, the fifth
treatment with 1 is incontrovertibly able to remove all the prions present in the ScN2a-RML cells
and the proof is the absence of an aggregation curve in all the passages after the #5.
The fact that even after the #5 that corresponds to the treatment interruption none of the samples
demonstrated to have prions using both the RT-QuIC assay and WB analysis, confirmed the ability
of 1 at 4 µM of concentration to cure the ScN2a from the RML prion strain.
27

Looking at the results, this reduction is mainly characterized with a decrease in the ThT
fluorescence, rather than in an increase in the lag phase.
Figure 10. Chronic treatment with 1 on ScN2a-RML cells. (A) Schematic representation of the chronic treatment
with 1. ScN2a-RML cells were incubated with 4 µM of compound 1 for 5 passages, after this period the treatment was
stopped, and the cells were passaged for others 5 times. (B) Representative WB of ScN2a-RML cell lysates. The upper
part corresponds to the PK-resistant PrPSc detected using the W226 primary antibody. The same was used in the cell
lysates not subjected to PK-digestion. In the bottom part the actin is represented, detected using the anti β-actin
antibody. (C) Aggregation curves of PrP after the addition of a seed, obtained from a chronic treatment on ScN2a-RML
cells. RT-QuIC performed with 0.2 mg of PrP and using 1 µg of sonicated protein from ScN2a-RML cells. The RT-
QuIC experiment using only the PrP is shown in blue. As it is presented in the legend the different colors represented
curve aggregation using as a seed the different passages obtained from the N2a-RML treated with 1.
28