Second generation of diazachrysenes: Protection of Ebola virus infected mice and mechanism of action

Article abstract of DOI:10.1016/j.ejmech.2018.10.061

Ebola virus (EBOV) causes a deadly hemorrhagic fever in humans and non-human primates. There is currently no FDA-approved vaccine or medication to counter this disease. Here, we report on the design, synthesis and anti-viral activities of two classes of compounds which show high potency against EBOV in both in vitro cell culture assays and in vivo mouse models Ebola viral disease. These compounds incorporate the structural features of cationic amphiphilic drugs (CAD), i.e they possess both a hydrophobic domain and a hydrophilic domain consisting of an ionizable amine functional group. These structural features enable easily diffusion into cells but once inside an acidic compartment their amine groups became protonated, ionized and remain trapped inside the acidic compartments such as late endosomes and lysosomes. These compounds, by virtue of their lysomotrophic functions, blocked EBOV entry. However, unlike other drugs containing a CAD moiety including chloroquine and amodiaquine, compounds reported in this study display faster kinetics of accumulation in the lysosomes, robust expansion of late endosome/lysosomes, relatively more potent suppression of lysosome fusion with other vesicular compartments and inhibition of cathepsins activities, all of which play a vital role in anti-EBOV activity. Furthermore, the diazachrysene 2 (ZSML08) that showed most potent activity against EBOV in in vitro cell culture assays also showed significant survival benefit with 100% protection in mouse models of Ebola virus disease, at a low dose of 10 mg/kg/day. Lastly, toxicity studies in vivo using zebrafish models suggest no developmental defects or toxicity associated with these compounds. Overall, these studies describe two new pharmacophores that by virtue of being potent lysosomotrophs, display potent anti-EBOV activities both in vitro and in vivo animal models of EBOV disease.

Full text of DOI:10.1016/j.ejmech.2018.10.061

Accepted Manuscript
Second generation of diazachrysenes: Protection of Ebola virus infected mice and
mechanism of action
Života Selaković, Julie P. Tran, Krishna P. Kota, Marija Lazić, Cary Retterer, Robert
Besh, Rekha G. Panchal, Veronica Soloveva, Vantongreen A. Sean, Wells B.
Jay, Aleksandar Pavić, Tatjana Verbić, Branka Vasiljević, Kathleen Kuehl, Allen J.
Duplantier, Sina Bavari, Rajini Mudhasani, Bogdan A. Šolaja
PII:
S0223-5234(18)30938-3
DOI:
https://doi.org/10.1016/j.ejmech.2018.10.061
EJMECH 10847
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 6 September 2018
Revised Date: 17 October 2018
Accepted Date: 27 October 2018
Please cite this article as: Ž. Selaković, J.P. Tran, K.P. Kota, M. Lazić, C. Retterer, R. Besh, R.G.
Panchal, V. Soloveva, V.A. Sean, W.B. Jay, A. Pavić, T. Verbić, B. Vasiljević, K. Kuehl, A.J. Duplantier,
S. Bavari, R. Mudhasani, B.A. Šolaja, Second generation of diazachrysenes: Protection of Ebola virus
infected mice and mechanism of action, European Journal of Medicinal Chemistry (2018), doi: https://
doi.org/10.1016/j.ejmech.2018.10.061.
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. 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.

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT
Second generation of diazachrysenes: protection of Ebola virus infected mice and
mechanism of action
#
‡
‡
#
‡
Života Selaković, Julie P. Tran, Krishna P. Kota, Marija Lazić, Cary Retterer, Robert
‡
‡
‡
‡
‡
Besh, Rekha G. Panchal, Veronica Soloveva, Vantongreen A. Sean, Wells B. Jay, Aleksandar
¶
#
¶
‡
‡
‡
Pavić, Tatjana Verbić, Branka Vasiljević, Kathleen Kuehl, Allen J. Duplantier, Sina Bavari,
‡
,∇
, *
#,§,*
Rajini Mudhasani,
Bogdan A. Šolaja
#
University of Belgrade, Faculty of Chemistry, Studentski trg 12-16, P.O. Box 51, 11158,
Belgrade, Serbia
‡
Molecular and Translational Sciences Division, United States Army Medical Research Institute
of Infectious Diseases, 1425 Porter Street, Frederick, Maryland 21702, United States
¶
Institute of Molecular Genetics and Genetic Engineering, University of Belgrade, Belgrade,
Serbia
∇
Department of Pathology and Microbiology, University of Nebraska Medical Center, 985900
Nebraska Medical Center, Omaha, NE 68198-5900, United States
§
Serbian Academy of Sciences and Arts, Knez Mihailova 35, 11158 Belgrade, Serbia
*
Corresponding authors
Keywords: diazachrysene filovirus inhibitors, naphthyridines, Ebola virus entry inhibitors, late
endosomes, lysosomotroph
1

ACCEPTED MANUSCRIPT
ABSTRACT
Ebola virus (EBOV) causes a deadly hemorrhagic fever in humans and non-human primates.
There is currently no FDA-approved vaccine or medication to counter this disease. Here, we
report on the design, synthesis and anti-viral activities of two classes of compounds which show
high potency against EBOV in both in vitro cell culture assays and in vivo mouse models Ebola
viral disease. These compounds incorporate the structural features of cationic amphiphilic drugs
(CAD), i.e they possess both a hydrophobic domain and a hydrophilic domain consisting of an
ionizable amine functional group. These structural features enable easily diffusion into cells but
once inside an acidic compartment their amine groups became protonated, ionized and remain
trapped inside the acidic compartments such as late endosomes and lysosomes. These
compounds, by virtue of their lysomotrophic functions, blocked EBOV entry. However, unlike
other drugs containing a CAD moiety including chloroquine and amodiaquine, compounds
reported in this study display faster kinetics of accumulation in the lysosomes, robust expansion
of late endosome/lysosomes, relatively more potent suppression of lysosome fusion with other
vesicular compartments and inhibition of cathepsins activities, all of which play a vital role in
anti-EBOV activity. Furthermore, the diazachrysene 2 (ZSML08) that showed most potent
activity against EBOV in in vitro cell culture assays also showed significant survival benefit with
1
00 percent protection in mouse models of Ebola virus disease, at a low dose of 10 mg/kg/day.
Lastly, toxicity studies in vivo using zebrafish models suggest no developmental defects or
toxicity associated with these compounds. Overall, these studies describe two new
pharmacophores that by virtue of being potent lysosomotrophs, display potent anti-EBOV
activities both in vitro and in vivo animal models of EBOV disease.
2

ACCEPTED MANUSCRIPT
INTRODUCTION
The latest Ebola virus disease (EVD) epidemic hit West Africa from late 2013 to mid-2016,
resulting in more than 11,000 deaths.[1] The average fatality rate (about 40%) was lower than in
many of the previous outbreaks (up to 90%). However, the number of cases exceeded all
previous outbreaks combined by more than tenfold and the number of deaths by more than
sevenfold.[2] A more recent outbreak has been reported in West Africa with a total of 78 cases of
hemorrhagic fever, including 44 deaths so far.[3]
EBOV is a single-stranded, negative sense RNA virus and is a member of the genus Ebolavirus
in the family Filoviridae. It is closely related to Marburg virus, which, like Ebola Virus (EBOV),
causes a hemorrhagic fever in humans and non-human primates.[4] The natural reservoir of
filoviruses are possibly bats, albeit transmission to humans can also occur via non-human
primates and possibly other animals as well.[5]
Currently there is no FDA licensed drug, vaccine or any other therapeutic to treat EVD.[6] At
least three vaccine candidates have shown promising results and have entered clinical trials to
7
,8,9,10
various degrees.[7-10]
The unpredictable nature of occurrence of new outbreaks, followed
by a prolonged period usually taken to attribute the patients’ symptoms to EVD, often results in
rapid spread of the contagion. This underlines the need for a readily available therapeutic, with a
long shelf life and one that could be easily stored. Research conducted by different teams,[11-
1
1,12,13,14
15,16,17
1
4]
as well as our own,[15-17]
has provided a number of active small molecule
EBOV inhibitors, that could meet the aforementioned criteria (Figure 1). Favipiravir, BCX4430
and GS-5734, act as prodrugs that are metabolized into selective EBOV RNA-polymerase
inhibitors.[11,12,15,16]
3

ACCEPTED MANUSCRIPT
More recently, repurposing drug screens identified small molecules belonging to cationic
1
8,19,20
amphiphilic drugs (CADs) family, as potent inhibitors of EBOV entry.[18-20]
CADs are a
wide group of chemicals that typically have both a hydrophobic domain containing an aromatic
ring system and a hydrophilic domain containing an ionizable amine functional group.[21] These
structural features, enables easily diffusion into cells but once inside an acidic compartment their
amine groups become protonated, ionized and remain trapped. As a result, CADs are
lysosomotropic, i.e., accumulate to high concentrations in the acidic compartments such as Late
Endosomes and Lysosomes (LE/Ly) thereby altering their morphology and functions. Regardless
of their intended target, many FDA approved drugs with a CAD moiety in their backbone,
blocked EBOV entry in in vitro assays by altering LE/Ly functions.[13,20]
EBOV entry is exclusively dependent on EBOV Glycoprotein (GP) trimers that project as long
spikes on the lipid bilayer envelope. The viral entry begins with the viral attachment to the host
cell surface followed by virus internalization through macropinocytosis. The virus is then
trafficked sequentially from the early endosomes to the LE and Ly. In the LE/Ly, the cysteine
proteases; Cathepsin B and/or Cathepsin L under acidic pH and reducing conditions, cleave GP
2
2,23,24
to generate a 17-19 kDa fusogenic version of GP.[22-24]
The cleaved GP also unmasks its
binding site to the LE/Ly resident receptor, the Niemann-Pick C1 (NPC1). GP-NPC1 interaction
promotes the filovirus/cell membrane fusion, that eventually leads to the release of viral lipid
membrane enclosed ribonucleoprotein complex (RNP) into the cytoplasm for
transcription.[25,26] For successful entry, trafficking of EBOV virions to specific NPC1
expressing LE/Ly compartments is crucial to promote EBOV fusion with the limiting membrane
to release the enclosed nucleocapsid into cytoplasm. CADs appear to block these interactions by
modulating the LE/Ly, however the exact mechanism is not known.
4

ACCEPTED MANUSCRIPT
In spite of many CADs being effective in vitro in cell culture assays, only few of them were
effective in vivo in mouse models of Ebola virus disease. Among the successful ones,
chloroquine (CQ) and amiodarone at 90 mg/kg, protected mice in one study but failed in
1
8,27,28,29
another.[18,27-29]
Amodiaquine (AQ) did not protect mice,[29] but its usage in humans
during 2014 EBOV outbreak, in place of an antimalarial containing lumefantrine at Médecins
Sans Frontières-led treatment units, was associated with an observed decrease in human fatality
rates.[30] Toremifene (Figure 1) at 60 mg/kg protected mice but the doses used were higher than
what was normally tested for clinical usage and hence is potentially associated with serious side
effects.[13] Sertraline offered 70% and Bepridil offered 100% protection of mice at a lower
dosage of 20 mg/kg/day and 24 mg/kg/day, respectively.[20] Potency differences may correlate
with interactions with lipid and protein moieties within the LE/Ly compartments that ultimately
impact EBOV entry. However, the primary intended target of the approved drugs was not to
function as CADs and their anti-EBOV activity required several 10-100 higher fold usage of the
drug in in vitro studies when compared to their activity against their original intended target.
Their application in animal models would require re-testing tolerability at higher doses and are
compounded by the complications of their effects on their primary intended targets. Therefore, a
more efficient CAD that disrupts EBOV entry without causing any serious side effects would be
highly desirable.
5

ACCEPTED MANUSCRIPT
NH2
NH2
N
H
N
O
O
N
O
P
N
H
N
HN
O
N
HO
N
O
O
CN
OH OH
GS-5734
OH OH
BCX4430
EC50 = 0.06-0.14 µM
in vivo (mice): not active
EC50 = 3.4-11.8 µM
in vivo (C. macaques): 100%, 3.3 mg/kg/d
in vivo (mice): 100%, 300 mg/kg/d
O
N
N
NH
N
O
O
N
F
N
N
NH2
Cl
OH
HN
N
O
toremifene
EC50 = 0.97-1.73 µM
in vivo (mice): 50%, 60 mg/kg/d in vivo (mice): 100%, 30 mg/kg/d in vivo (mice): 90%, 10 mg/kg/d
favipiravir
EC50 = 67 µM
1 (ZS48)
EC50 = 0.70 µM
Figure 1. Selected examples of small molecule EBOV inhibitors.
Here, we report on the development of a novel anti-EBOV pharmacophore with a 1,4-
naphthyridine core (Chart 1). Furthermore, for the first time the synthesis of diazachrysene
derivatives substituted by two different aminoalkyl side chains has been achieved, diverging
from 1 (ZS48) and alike.[17,31] The synthesis, pharmacokinetic analysis, in vitro and in vivo
antiviral screening efforts, and detailed mechanism of action (MOA) studies suggest that the
studied compounds are CADs with unique features and represent the most potent inhibitors of
EBOV within the CAD structural class.
6

ACCEPTED MANUSCRIPT
Chart 1. General structures of investigated derivatives
1
R
R
NH
N
R
HN
N
HN
N
N
N
N
NH
R
HN
2
R
aminoalkyl derivatives of 4,10-diazachrysene
with two different substituents
aminoalkyl derivatives of 1,4-naphthyridine
RESULTS AND DISCUSSION
Synthesis
Being aware that increased number of aromatic rings may disfavor our drug development efforts
based on 4,10-diazachrysene core,[32] beside extensive toxicity evaluation thereof, we entered
the core ring re-grouping to two central aromatic rings aiming at naphthyridine core instead
(
Figure 2). Furthermore, the envisioned naphthyridine derivatives have a structural similarity to
the FDA approved antimalarial drug CQ, which was found to have moderate anti-EBOV
activity.[18,27] The planned naphthyridine compounds would preserve the 4-alkylaminopyridine
constitutional motif, akin to alkylamino substituted 4,10-diazachrysenes, which appears to
decrease the susceptibility to metabolic activation and the toxicity proceeding thereof.[31,33]
7

ACCEPTED MANUSCRIPT
O
O
N
N
O
NH
N
NH
N
NH
N
N
b)
N
a)
N
N
O
HN
HN
HN
N
N
N
O
2
(ZSML08)
O
1 (ZS48)
3
N
HN
N
N
N
HN
N
HN
N
N
N
c)
c)
H
H
Cl
HN
N
5
chloroquine, CQ
4
new structures
starting structures
new structures
Figure 2. Planned transformations: a) Desymmetrization 1→2; b) Core reduction 1→3; c) CQ
surrogates.
3
6,37,38
Firstly, two naphthyridine core compounds, 6[34,35] and 7[36-38]
were synthesized using
slightly modified approaches than those already known (see Supporting Information 1). These
two compounds were then subjected to microwave-assisted nucleophilic substitution reactions
with appropriate amines. This gave rise to 12 alkylamino substituted compounds (3-5 and 8-16,
as shown in Scheme 1, thus affording for the first time the 4,8-alkylamino substituted 1,5-
naphthyridines.
8

ACCEPTED MANUSCRIPT
Scheme 1. The synthesis of alkylamino substituted naphthyridines.
R' = H
R
R'
N
5: R'' = -CH(CH )(CH ) N(CH CH ) , 78%
3 2 3 2 3 2
a)
N
8
: R'' = -(CH ) N(CH CH )O, 91%
2 3 2 2
N
9: R'' = -(CH ) N(CH CH ) , 59%
2 3 2 3 2
N
Cl
1
0: R'' = -CH CH(OH)CH N(CH CH ) , 54%
NHR''
2 2 2 3 2
6
: R = H
11: R'' = -(CH ) NHR''', 38%
2 4
7
: R = Cl
R' = -NHR''
o
a) RNH , MW, NMR, 180 C, 2h;
2
3
4
1
1
1
: R'' = -(CH ) N(CH CH ) O, 71%
2 2 2 2 2
: R'' = -CH(CH )(CH ) N(CH CH ) , 51%
3
2 3
2
3 2
Cl
N
2: R'' = -(CH ) N(CH CH ) O, 33%
2
3
2
2 2
3: R'' = -(CH ) N(CH CH ) , 41%
2
3
2
3 2
4: R'' = -CH CH(OH)CH N(CH CH ) , 67%
2
2
2
3 2
R'''
15: R'' = -(CH ) N(CH CH ) , 46%
2 2 2 3 2
1
6: R'' = -(CH ) N(CH CH ) O, 46%
2 4 2 2 2
Our initial diazachrysene structures were substituted with two identical aminoalkyl
sidechains.[17] Therefore, for the synthesis of unsymmetrically substituted diazachrysenes, the
experience in tuning up the conditions for microwave-assisted nucleophilic aromatic substitution
on the diazachrysene core[31] helped us optimize the procedure that resulted in monosubstitution
reaction keeping one chloro substituent intact (Scheme 2). The remaining chlorine in 18 was then
substituted with different aminoalkyl side-chains. In addition, we synthesized three uniformly
substituted derivatives, 21, 22 and 23.
9

ACCEPTED MANUSCRIPT
Scheme 2. The synthesis of alkylamino-substituted diazachrysenes.
O
O
Cl
N
N
N
NH
N
NH
a)
b)
N
N
N
N
x4HCl
Cl
17
18: 63%
Cl
NHR
2
: R = -(CH ) N(CH CH ) O, 56%
b)
2 4 2 2 2
NHR
19: R = -(CH ) N(CH CH ) , 40%
2 3 2 3 2
2
0: R = -CH CH(OH)CH N(CH CH ) O, 73%
2 2 2 2 2
N
N
2
1: R = -(CH ) O(CH ) N(CH ) , 89%
2 2 2 2 3 2
x4HCl
except 22: x2HCl
NHR
22: R = -(CH ) NH , 72%
2 4 2
2
3: R = -(CH ) NH , 99%
2 3 2
o
a) NH (CH ) N(CH CH ) O, NMP, MW, 150 C, 20 min;
2
2
2
2
2 2
o
b) 1. RNH , NMP, MW, 180 C, 5 h, 2. HCl, EtOH
2
Inhibitory Activity Against EBOV
The compounds were screened for their antiviral activity in cell based High throughput, high
Content Imaging (HCI) assay, and their EC and/or EC and CC were assessed by running a
5
0
90
50
dose response curve analysis (Table 1).[39,40] Briefly, HeLa or human foreskin fibroblast (HFF)
cells were pretreated with different doses of the compounds for 2 h prior to incubating with
EBOV. Forty-eight hours post infection (following multiple cycles of viral infections), cells were
subjected to immunofluorescence analysis (IFA) followed by HCI assay to evaluate the
percentage of EBOV-GP expressing cells. Diazachrysene derivatives were tested as water
soluble hydrochloride salts, while naphthyridines were tested as free base forms (1.0 %
DMSO/water).
10

ACCEPTED MANUSCRIPT
Table 1. Inhibitory activity of naphthyridine derivatives against EBOV and their cytotoxicity in
HeLa cells.
Compound
EBOV
EBOV
CC (µM)
SI
5
0
EC (µM) EC (µM)
5
0
90
0
0
0
0
.70
.55
.26
.21
<5
>20
>28.5
>36.5
30
1
2
(ZS48)[31]
a
a
a
a
1.36
>20
8
0.85
(ZSML08)
a
a
a
a
0.78
2.4
11.2
3
4
5
8
9
1
1
1
1
1
1
1
1
2
2
2
2
not active
0.78
-
>25
>10
-
-
4.73
>13
~10
-
-
not active
~10
-
-
-
-
-
-
0
not active
2.87
-
-
-
1
2
3
4
5
6
9
0
1
2
3
13
-
>25
-
>8.7
-
~10
4.82
37.43
-
>10
-
>2
not active
14.11
2.24
-
62.6
9.9
>25
>25
>10
>10
7
>1.8
>11.2
>12.6
>16.7
2.0
>1
0.79
4.85
3.50
6.74
23.5
20.5
-
0.60
3.53
~10
>10
>10
-
~10
>1
2
1,b
CQ
16
-
11

ACCEPTED MANUSCRIPT
Table 1: EC50 or EC90, which is the effective concentration to achieve 50% or 90% infection
inhibition respectively, and CC50 (50% cytotoxicity) was accomplished by running the dose
response curve analysis in four replicates (n = 4). Selectivity index (SI) =CC /EC . HeLa cells
50
50
a
b
were used for assays, unless noted otherwise. HFF cells. Vero cells.
Several bis-aminoalkyl naphthyridine derivatives (4, 13, and 16) were found active in this assay
Table 1) with compound 16 with morpholino substituents connected with four carbons to
aromatic moiety among the most active anti-infectives with EC = 2.24 µM. Other bis-
(
5
0
aminoalkyl derivatives, such as 4 and 13, were also considerably active, with EC = 0.78 µM
50
and 4.82 µM, respectively. Moreover, the low EC90 = 4.73 µM emphasized the possible high
potency of 4 as anti-EBOV inhibitor. Interestingly, 11, which possesses 7-chloroquinoline
moiety as its side-chain (a feature of CQ) also has acceptably good in vitro activity with EC50
=
2
.87 µM, contrary to the mono-aminoalkyl substituted naphthyridines (5, 8, 9 and 10) which
appeared either completely inactive or possessed borderline activity in our screen.
Diazachrysene derivatives with dissimilar aminoalkyl substituents, 2, 19 and 20 showed
improved anti-EBOV activity (Table 1) as compared to their symmetrically substituted analogue
and related compounds[31] with EC = 0.26-0.79 µM, and they also retained the low level of
1
5
0
cytotoxicity. The inhibitor 2, EC = 0.26 µM, EC = 0.85 µM, appeared as one of the most
5
0
90
potent EBOV inhibitors known.[16,41,42] The most active new compound 2, as well as our
preceding derivative 1, were further tested in a primary cell-based assay, in HFF cells. The
activity proved to be consistent with HeLa based assay results.
Symmetrically substituted diazachrysene derivatives 21, 22 and 23 showed no improved activity
and were rather in-line with our previous results.[31]
12

ACCEPTED MANUSCRIPT
Pharmacokinetic analysis and physicochemical properties
The in vitro ADME profile for compounds 2, 4, and 13 (Table 2) was encouraging. Both
naphthyridines 4 and 13, as well as the diazachrysene 2, showed good plasma stability (60 or
>
85% remaining after 1 h), adequate solubility at pH 7.4 (>50 � g/mL), low to moderate CYP
inhibition, and high mouse and human microsomal stability. Compound 2, the most potent
inhibitor, was further shown to be highly permeable in an MDCK permeability assay.
Table 2. ADMET parameters for 2, 4, and 13.[43]
Parameter tested
2
4
13
a
Plasma stability, 1h, 37 °C (%)
60
>85
>50
>10
>10
>10
8.0
>85
>50
>10
>10
>10
>10
>10
<23/<23
b
Solubility, pH 7.4 (µg/mL)
>50
>10
>10
>10
>10
>10
22.3/32.9
c
CYP3A4 / BQ inhibition (µM)
c
CYP3A4 / DBF inhibition (µM)
c
CYP3A4 / BFC inhibition (µM)
c
CYP2D6 / AMMC inhibition (µM)
c
CYP2C19 / CEC inhibition (µM)
3.6
d
Clearance mouse/human (µL/min/mg)
32.9/<23
13

ACCEPTED MANUSCRIPT
d
Microsomal t1/2 mouse / human (min)
62.1/42.1
39.3
42/>60
>60/>60
-6
e
Permeability A→B (10 cm/s)
-
-
-
-
-6
e
Permeability B→A (10 cm/s)
43.2
a
Plasma stability (percent remaining after 1 hour), determined following incubation for 1 hour
b
at 37 °C. Compound solubility, determined using laser nephelometry to measure light scattering,
performed in triplicate. The IC values for CYP inhibition, determined from the net fluorescent
c
5
0
signal from incubation at 37°C for a set time with an active cytochrome P450 enzyme and a
d
fluorescent probe substrate. Clearance in mouse and human models, substrate depletion
experiments were performed by incubating test compound with liver microsomes for 1 hour at 37
e
°
C; in this test t1/2>30 min is considered good. Apparent permeability (PAPP) of compounds
determined with Madin Darby Canine Kidney (MDCK) Cell line; A→B apical to basolateral
membrane movement; B→A basolateral to apical membrane movement; three biological
replicates.
In order to assess the efficacy of diazachrysene transport, the interactions of 2 and 1 with HSA
and AGP proteins, respectively, have been evaluated utilizing changes in fluorescence spectra of
proteins upon addition of increasing amounts of each compound (see Supporting Information 1).
4
-1
4
-1
The binding constants (K ) were (6.12±0.20) × 10 M for AGP and (8.85±0.30) × 10 M for
SV
4
-1
5
-1
HSA (for 2); and (8.12±0.07) × 10 M for AGP and (1.13±0.04) × 10 M for HSA (for
1
).[44,45]
Furthermore, during in vivo tolerability evaluation of 2 in healthy mice (vide infra) mouse serum
was collected and analyzed for total compound levels, and possible metabolites thereof.[46]
Total concentration of compound 2 was detected in serum (0.35-1.20 µM) 24 h after last 20
mg/kg dose was administered, as presented in Figure 3. The detected concentration is
conveniently in the range of the compound’s EC value (0.85 µM), exceeding the EC value
9
0
50
(
0.26 µM HeLa; 0.21 µM HFF) and is significantly lower than the CC value in HeLa, and HFF
50
14

ACCEPTED MANUSCRIPT
cell assays and the LC50 value of 18.9 µM (zebrafish assay, see Supporting Information 1).
Moreover, this serum concentration relative to EC is similar to that for our diazachrysene
5
0
compound 1 (0.46-1.03 µM; Supporting Information 1) which was efficacious in a mouse model
of EBOV infection.
Figure 3. Chromatograms of quantification of free 2 (ZSML08) levels in mouse serum. Samples
from two mice treated with 20 mg/kg/day of 2 (ZSML08) were taken 24 hours after the last dose
and prepared as described in Supporting Information 1. Chromatograms were obtained using
UHPLC/MS. No products of reduction (M+2), oxygenation (M+16), dehydrogenation (M-2),
bis-oxygenation (M+32) or of any combination thereof were detected.
The ability of a compound to be protonated inside of the endosome is critical for the CAD
mechanism. Therefore, acidity constants for three compounds selected for in vivo screening (2,
15

ACCEPTED MANUSCRIPT
4
, 13) were experimentally determined by potentiometric titration in four replicates as presented
in Table 3 and Figure 4 (conjugated acids, Supporting Information 1), as well as for CQ and 1,
for comparison. This allowed the calculation of the net charge for the compounds at
physiological and lysosomal pH (Table 3). Furthermore, the distribution coefficients at
pH=7.3±0.1 were calculated using ACD/Labs software.
Table 3. Experimental acidity constants of selected compounds and calculated distribution
coefficients.
d
Compound
pKa side
chain 1
pKa side
chain 2
pKa
core 1
pKa
core 2
Net
charge at
pH 7.35
Net
charge at
pH 5.0
clogDpH=7.3
a
a
b
b
c
c
1
7
1
5.96
6.65
5.24
5.60
8.41
8.50
/
7.35
7.70
5.72
5.51
4.37
0.54
0.87
1.94
1.95
1.54
2.72
3.04
2.84
2.76
1.99
2.71
2.53
1.36
-0.18
1.53
2
4
4.56
10.81
10.97
9.18
/
/
/
1
3
CQ
7.45
a
b
pKa values of alkylamino side-chains. pKa values of one or two aromatic nitrogens at the
c
core. Total degree of protonation (net charge) at (physiological) pH = 7.35 and pH = 5.0
lysosome), respectively, calculated from pKa values (approximating the ionic strength of
solution to zero). Calculated distribution coefficient at pH 7.3 using ACD/Labs Release
2.0.[47]
(
d
1
16

ACCEPTED MANUSCRIPT
pKa sidechain 1
O
+pKa sidechain 1
N
H
H
N
+
NH
NH
pKa core 1
H
N
+
H
N
+
+
N
H
pKa core 1
N
+
HN
N
H
pKa core 2
H
N
HN
+
O
pKa sidechain 2
4
pKa sidechain 2
2
Figure 4. Sites of protonation from pKa determination experiment.
The pKa measurement results showed that naphthyridine derivatives 4 and 13 are more basic
than diazachrysenes 2 and 1, having two times higher net charge at physiological pH. This fully
corresponds to their computed clogD values which are lower than the values of diazachrysene
derivatives. The charge at physiological pH deserves a few comments: diazachrysenes, 2 and 1,
are less than monoprotonated at pH 7.35 and charge distribution sits ca. 79-93 % at core nitrogen
and the rest stays on morpholine nitrogens (Supporting Information 1).[48,49] Therefore, the
positive charge delocalization across the aromatic diazachrysene core should ease the plasma
transport and passage across membranes of 1 and 2 over 4 and 13 (both possessing +2 localized
charge at side-chain nitrogens) as in vivo results would suggest. On the other hand, the charge at
pH 5.0 (lysosomal compartments) is very similar for all new compounds, being ~3.0; therefore,
the higher degree of protonation is expected to prevent the compound from leaving the
endosomal compartment once inside (vide infra).[50]
17

ACCEPTED MANUSCRIPT
Toxicity profiling
The most promising EBOV inhibitor 2, and 1 and CQ for comparison, were thoroughly
investigated for toxicity in vivo (mortality, teratogenicity, cardiotoxicity, hepatotoxicity and
myelosuppression) using zebrafish embryos (Figure 5).
Figure 5. A. The results of toxicity assay on zebrafish embryos at 120 h hours post fertilization
(
hpf) exposed to antiviral drugs 2 (ZSML08), 1 (ZS48) and CQ, expressed as the LC (µM). B.
50
Morphology of wild type (wt) zebrafish embryos after the treatments with 30 µM of 2
ZSML08), 40 µM of 1 (ZS48) and 40 µM of CQ. The embryonic liver is indicated with a red
(
curved area. C. In vivo imaging of the liver morphology of Tg(l-fabp:EGFP) zebrafish larvae
treated with 2 (ZSML08) and 1 (ZS48) confirming the lack of their hepatotoxicity upon applied
doses contrary to CQ. For details see Supporting Information 1.
18

ACCEPTED MANUSCRIPT
The antivirals 1 and 2 exhibited a dose-dependent mortality and did not induce any appearance
of teratogenic (developmental) malformations in the surviving embryos (Figure 5; Supporting
Information 1). As the cardiotoxicity and the liver toxicity are the most common drawbacks of
the drugs approved for human use, diazachrysenes 1 and 2 were checked for the above toxic
effects every day within 72-120 hours post fertilization (hpf). The obtained results showed no
cardiotoxic effects in any of the surviving embryos, such as the appearance of pericardial edema
(
Figure 5B) and the disturbed heartbeat rate (data not shown).
To address the possible hepatotoxicity of selected diazachrysenes, we explored their effect on the
liver development (applying compounds at 6 hpf when the liver has not yet been formed) and
function (applying compounds at 72 hpf when the liver is fully functional) using the transgenic
Tg(l-fabp:EGFP) zebrafish embryos with fluorescently labeled liver. The hepatotoxicity was
evaluated at 120 hpf old embryos, using the liver area index (the ratio of liver area and lateral
body area) that proved to accurately represents the liver damages.[51] As shown in Figure 5C,
the antivirals 2 and 1 did not change the liver area index nor its fluorescence at any tested dose
up to 30 µM and 40 µM, respectively, irrespective to the time of addition (Figure 5B, 5C,
Supporting Information 1), thus clearly demonstrating no liver toxicity even at concentrations 73
times higher than cell based activity EC50 (for 2) and 103 times higher (for 1). On the contrary,
CQ significantly reduced both the liver area index (P < 0.05) and fluorescence at ≥ 30 µM if
administered at 6 hpf (Figure 5B, C), and at ≥ 80 µM when applied at 72 hpf (Supporting
Information 1).
Since myelosuppression is also one of the adverse side effects of many clinically used drugs, the
antivirals 1 and 2 were examined for toxicity in vivo towards neutrophils. To address this
potential issue, we used the transgenic Tg(mpx:GFP) zebrafish embryos with fluorescently
19

ACCEPTED MANUSCRIPT
labeled neutrophils, thus enabling direct visualization of applied compounds on the neutrophil
genesis and occurrence. Obtained results revealed that 1 and 2 exerted no myelotoxic effect at
any tested concentration up to 20 µM and 40 µM respectively, as determined by the whole
embryos fluorescence (Figure 6). CQ appeared to be myelotoxic and markedly reduced the
neutrophil occurrence at the doses ≥ 80 µM (Figure 6).
20

ACCEPTED MANUSCRIPT
Figure 6. The effect of 2 (ZSML08) and 1 (ZS48) on the neutrophil population assessed in
Tg(mpx:GFP) zebrafish embryos with GFP-labeled neutrophils at 72 hpf. Neutrophil
fluorescence (A) and occurrence (B) upon the treatment with 2 (ZSML08), 1 (ZS48), and CQ.
To address the therapeutic potential of the explored compound, the toxicity was evaluated by
determination of the lethal concentration (LC50, Supporting Information 1), defined as the
treatment concentration resulting in 50% mortality of embryos over a period of 120 hpf. The
obtained results reveal a very good therapeutic potential of both diazachrysenes that showed a
low overall toxicity with the LC50(2 (ZSML08)) = 18.9 µM and LC50(1 (ZS48)) = 72.1 µM. The
observed toxicity is ~70-100 times lower than their respective inhibitory activity against EBOV
observed in HeLa (HFF) cells, Table 1.
Tolerability studies in mice
Six healthy mice were subjected to seven daily doses of 2 (20 mg/kg, a dose twice as high as the
one used in the EBOV challenge test; i.p. administration, c.f. Supporting Information 1). The
mice were then observed twice daily for behavior and appearance for a total of 28 days after the
last administered dose. No overt manifestations of acute toxicity were observed.
Compound 2 Protected Mice from EBOV
Based on the above potency, cellular toxicity, ADME, tolerability and PK properties, results, the
two most potent inhibitors with minimal cytotoxicity, the compounds 2 and 4, were further
21

ACCEPTED MANUSCRIPT
evaluated for in vivo efficacy in a mouse model of EBOV infection[52] using three different
doses of 10, 5 or 1 mg/kg (Figure 7). This mouse model has been often used for evaluating
5
3,54,55
different antivirals.[20,27,41,53-55]
A third compound, 13, a less active homologue of 4,
was also chosen for comparison. Compound 2 showed a dose dependent protection with 100%
protection at a dose of 10 mg/kg and 50% protection at lower doses of 5 mg/kg and 1 mg/kg
(
Figure 7A). Although uniform mice lethality was not observed in the vehicle treated control
group infected with mouse adapted Ebola virus, a statistically significant (p = 0.0013) increase in
survival was observed in the 2 (ZSML08) (10 mg/kg) treated group compared to control group.
Furthermore, minimal weight loss of the surviving mice was observed (Figure 7B). There was
also 40% survival with 4 at its highest dose of 10 mg/kg (Figure 7C-D). These in vivo protection
data corroborated with the compound’s anti-viral activity derived from cell based assay (Table 1)
with compounds potency following 2 (ZSML08) > 4 (ZS103) > 13 (ZS102).
22

ACCEPTED MANUSCRIPT
Figure 7. The compound 2 (ZSML08) conferred most protection in mice from EBOV challenge.
Mice (n = 10 mice/group) were treated with indicated concentrations of the compounds at 2 h
prior to challenge with 1000 plaque forming units (PFU) of mouse-adapted EBOV. Treatment
was continued daily for six days. The percent survival curves and the corresponding percent
mean body weight change relative to day 0 (just prior to treatment or challenge) with 2
(
ZSML08) (A-B), 13 (ZS102) (C-D) and 4 (ZS103) (E-F) during the 14 days period of study.
The P values were evaluated using log-rank test to compare survival rate with the control group.
(ZSML08) p = 0.0013 (10 mg/kg); 19 (ZS103), p = 0.017 (10 mg/kg); 13 (ZS102) p = 0.03 (1
mg/kg).
2
23

ACCEPTED MANUSCRIPT
Investigation on the mechanism of action of diazachrysenes
Compounds 1 and 2 Blocked Viral Entry
Figure 8. 1 (ZS48) and 2 (ZSML08) target viral entry. (A) Time of Compound Addition assay
(TCA) in which cells were untreated (mock) or treated with 2 (ZSML08) (at 2 µM) at 2 h prior
to infection (-2 h), or concurrent with virus infection (0 h), or at various time points post-
incubation with EBOV (MOI=10 for 24 h), followed by immunofluorescence analysis (IFA) and
HCI assay to determine the percentage of GP expressing cells. The infections at each time point
were normalized with mock treated cells to determine the infection inhibition. (B) GFP
expression by HeLa cells, which were pre-treated at 2 h with the indicated compounds followed
by infection with rVSV-EBOV-GP-GFP or VSV-GFP viruses for 8 h and 6 h respectively. Cells
were visualized by staining with Cell MaskDeep Red. (C-D) Same as in B but cells were either
untreated (mock) or treated with indicated concentrations of the various compounds and infected
24

ACCEPTED MANUSCRIPT
with rVSV-EBOV-GP-GFP (C) or VSV-GFP (D). Relative infection inhibition for each
compound treatment was determined by normalizing percentage of GFP expressing cells with
mock treated cells and subtracting the result with 1. Data is represented as mean ±SD of three
replicates and is representative of three independent experiments. P values were evaluated by
multiple t tests. N.S is not significant and ***, P<0.0001
CADs are known to target EBOV entry by modifying the internal milieu of LE/Ly, such that
viral membrane is unable to fuse with host lysosomal limiting membrane.[19,56] Thus, we first
examined if 1 (ZS48) and 2 (ZSML08) targeted viral entry by Time of Compound Addition
(TCA) assay. As shown in Figure 8A, compound treatment prior to infection (-2 h) or at the time
of infection (0 h) was necessary for its antiviral activity. Treating cells with the compound at 2 h
after incubating with virus, by when the virus already entered cells, had no effect on virus
infection.[57]
We next checked if defects in viral entry were indeed specifically related to changes in the
LE/Ly without influencing early endosomes or other early entry events. To address this question,
we used recombinant vesicular stomatitis virus (rVSV) that expressed green florescent protein
(GFP) and either expressed its own GP (VSV-GP) or pseudotyped to express EBOV-GP (VSV-
EBOV-GP) in the place of VSV-GP.[58] EBOV-GP trafficking to functional LE/Ly
compartments is critical for its fusion competency, whereas VSV-GP trafficking to the early
endosomal compartments is sufficient to make it fusion-competent.[59] Thus, if the compounds
were modulating LE/Ly then their antiviral effects would be more pronounced in VSV infection
regulated by EBOV-GP rather than VSV-GP. As shown in Figure 8 (B-D) even at 0.1 µM
25

ACCEPTED MANUSCRIPT
concentration, 1 and 2 inhibited rVSV-EBOV-GP but had no effect on rVSV-GFP infection. At
higher concentrations compounds inhibited VSV infection irrespective of the entry mediated by
VSV-GP or EBOV-GP but were much more sensitive to entry by latter. Two controls, AQ and
CQ, were also shown to inhibit EBOV entry albeit to a lesser extent.[18,28,60] At higher
concentrations (50 µM) at which compounds were all cytopathic (Figure 8C-D), all compounds
inhibited VSV infection. Collectively, these data confirmed that 1 and 2 targeted viral entry
mostly by modulating LE/Ly functions and were significantly more potent when compared to the
other structurally related CADs such as CQ and AQ.
Antivirals 1 and 2 are Lysosomotropic and Modify Acidic Compartments
Figure 9. Compounds 1 (ZS48)/ 2 (ZSML08) are lysosomotropic. HeLa cells were untreated
(mock) or treated with Bafilomycin A1 (BafA1) at 0.01 µM, 1 (ZS48) at 2 µM, 2 (ZSML08) at
2
1
µM, AQ at 10 µM and CQ at 10 µM for an hour followed by treatment with Lysosensor DND
89 for 10 min. (A) Cells were washed and then imaged at an excitation/emission of 443/505 nm
and 405/450 nm for detecting Lysosensor dye and the tested compounds, respectively. (B) The
intensity/area of the fluorescence emitted by the Lysosensor dye from the collected images as in
A was further evaluated using image analysis software (Columbus). Each data point is mean ±
SD of a replicate of 8 wells of a 96 well plate with approximately 3000 cells per well and is
26

ACCEPTED MANUSCRIPT
representative of three independent experiments. P value was evaluated by One Way Anova
analysis. ****, P < 0.0001.
CADs freely diffuse into cells but once inside an acidic compartment they become protonated
and remain trapped. The ensuing physiological and morphological changes in the LE/Ly can
make the virus fusion incompetent. The acidic pH of LE/Ly is also necessary for converting
EBOV-GP to the cleaved fusogenic form that is essential for viral entry.[61] Thus, we examined
1
(ZS48) and 2 (ZSML08) effects on the morphology and pH changes of LE/Ly by live cell
microscopy, using commercially available fluorescent reporter, LysoSensor DND 189, which
localizes to acidic compartments, and exhibits pH dependent increase in fluorescence intensity
upon acidification. HeLa cells were pre-treated with the compounds for 1 h prior to imaging cells
with LysoSensor. Compounds 1 and 2 were also fluorescent that allowed their visualization at an
excitation (Ex. 405) and emission (Em. 450). There was no fluorescence detected by the
compound alone under the Ex/Em conditions used for visualizing LysoSensor (data not shown).
As shown in Figure 9A, antivirals 1 and 2 strongly co-localized with LysoSensor suggesting that
these compounds freely diffused into cells, but once inside the acid compartments, they became
3+
heavily protonated (~ZSH , Table 3, Figure 4) and trapped. Furthermore, 1 and 2 induced the
3
clustering of LE/Ly at a high density, unlike in mock treated cells, in which they were fewer and
showed dispersed distribution throughout the cytoplasm (Figure 9, A-B). Surprisingly, there was
also sharp increase in florescence intensity of the acid compartments, thus ruling out the
possibility of neutralizing or alkalizing of LE/Ly due to compound protonation. LysoSensor 189
sharply drops its florescence intensity by both, decreasing pH from 4 to 0.5 or by increasing pH
from 4 to 9.[62] The plausible explanation for the increase in intensity appears to be due to an
increase in uptake of LysoSensor by the expanding LE/Ly (vide infra). The additional CAD
27

ACCEPTED MANUSCRIPT
controls, CQ and AQ showed a minor increase in fluorescence intensity when compared to mock
treated cells (Figure 10B) but no other changes were noted. As expected, control Bafilomycin A1
+
(
BafA1) treatment, an inhibitor of vacuolar H ATPase that regulates low acidic pH in the
lysosomal lumen, reduced the fluorescence intensity.[63]
We next examined if 1 (ZS48) and 2 (ZSML08) expanded the acidic compartments using
lysosomotropic weak base Acridine Orange (AO). AO binds to DNA and RNA as a monomer
and emits fluorescence in green channel. However, uptake of AO and its retention in protonated
+
(
AOH ) form in the acid compartments, increases its local concentration several fold causing
+
AOH oligomerization and shifting its florescence spectra from green to red. Furthermore,
increase in the intensity of red fluorescence is proportional to the increase in concentration of
+
AOH in LE/Ly.[64] No fluorescence was detected by the compound alone under the Ex/Em
conditions used for visualizing AO (data not shown).
28