DDX3X inhibitors, an effective way to overcome HIV-1 resistance targeting host proteins

Article abstract of DOI:10.1016/j.ejmech.2020.112319

The huge resources that had gone into Human Immunodeficiency virus (HIV) research led to the development of potent antivirals able to suppress viral load in the majority of treated patients, thus dramatically increasing the life expectancy of people living with HIV. However, life-long treatments could result in the emergence of drug-resistant viruses that can progressively reduce the number of therapeutic options, facilitating the progression of the disease. In this scenario, we previously demonstrated that inhibitors of the human DDX3X helicase can represent an innovative approach for the simultaneous treatment of HIV and other viral infections such as Hepatitis c virus (HCV). We reported herein 6b, a novel DDX3X inhibitor that thanks to its distinct target of action is effective against HIV-1 strains resistant to currently approved drugs. Its improved in vitro ADME properties allowed us to perform preliminary in vivo studies in mice, which highlighted optimal biocompatibility and an improved bioavailability. These results represent a significant advancement in the development of DDX3X inhibitors as a novel class of broad spectrum and safe anti-HIV-1 drugs.

Full text of DOI:10.1016/j.ejmech.2020.112319

Journal Pre-proof
DDX3X inhibitors, an effective way to overcome HIV-1 resistance targeting host
proteins
Annalaura Brai, Valentina Riva, Francesco Saladini, Claudio Zamperini, Claudia
Immacolata Trivisani, Anna Garbelli, Carla Pennisi, Alessia Giannini, Adele Boccuto,
Francesca Bugli, Maurizio Martini, Maurizio Sanguinetti, Maurizio Zazzi, Elena
Dreassi, Maurizio Botta, Giovanni Maga
PII:
S0223-5234(20)30288-9
DOI:
https://doi.org/10.1016/j.ejmech.2020.112319
EJMECH 112319
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 29 January 2020
Accepted Date: 6 April 2020
Please cite this article as: A. Brai, V. Riva, F. Saladini, C. Zamperini, C.I. Trivisani, A. Garbelli, C.
Pennisi, A. Giannini, A. Boccuto, F. Bugli, M. Martini, M. Sanguinetti, M. Zazzi, E. Dreassi, M. Botta,
G. Maga, DDX3X inhibitors, an effective way to overcome HIV-1 resistance targeting host proteins,
European Journal of Medicinal Chemistry (2020), doi: https://doi.org/10.1016/j.ejmech.2020.112319.
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.

DDX3X inhibitors, an effective way to overcome HIV-1
resistance targeting host proteins
Annalaura Brai,1,† Valentina Riva,2,† Francesco Saladini,3 Claudio Zamperini,1 Claudia Immacolata
Trivisani,1 Anna Garbelli,2 Carla Pennisi,2 Alessia Giannini,3 Adele Boccuto,3 Francesca Bugli,4,5
Maurizio Martini,6,7 Maurizio Sanguinetti,4,5 Maurizio Zazzi,3 Elena Dreassi, 1, Maurizio Botta*1,8 and
Giovanni Maga*2
1Dipartimento Biotecnologie, Chimica e Farmacia, Università degli Studi di Siena, Via A. De Gasperi 2,
2
I-53100 Siena, Italy. Istituto di Genetica Molecolare "Luigi Luca Cavalli - Sforza", IGM-CNR, Via
3
Abbiategrasso 207, I-27100 Pavia, Italy. Dipartimento di Biotecnologie Mediche, Università degli
Studi di Siena, I-53100 Siena, Italy. 4Fondazione Policlinico Universiterio A. Gemelli IRCCS,
Dipartimento di Scienze di Laboratorio e Infettivologiche, I-00168, Roma, Italy. 5Istituto di
6
Microbiologia, Università Cattolica del SC, L.go F. Vito 1, I-00168, Roma, Italy. Fondazione
Policlinico Universitario A. Gemelli IRCCS, Servizio di Istopatologia e Citodiagnosi, Rome, Italy.
8
7Istituto di Patologia, Università Cattolica del Sacro Cuore, Rome, Italy. Sbarro Institute for Cancer
Research and Molecular Medicine, Center for Biotechnology, College of Science and Technology,
Temple University, BioLife Science Building, Suite 333, 1900 N 12th Street, Philadelphia, PA 19122,
USA.
† These authors equally contributed to this work
KEYWORDS: DDX3X, host factors, HIV-1, antiviral agents, drug resistance, coinfections.

ABSTRACT: The huge resources that had gone into Human Immunodeficiency virus (HIV) research
led to the development of potent antivirals able to suppress viral load in the majority of treated patients,
thus dramatically increasing the life expectancy of people living with HIV. However, life-long
treatments could result in the emergence of drug-resistant viruses that can progressively reduce the
number of therapeutic options, facilitating the progression of the disease. In this scenario, we previously
demonstrated that inhibitors of the human DDX3X helicase can represent an innovative approach for the
simultaneous treatment of HIV and other viral infections such as Hepatitis c virus (HCV). We reported
herein 6b, a novel DDX3X inhibitor that thanks to its distinct target of action is effective against HIV-1
strains resistant to currently approved drugs. Its improved in vitro ADME properties allowed us to
perform preliminary in vivo studies in mice, which highlighted optimal biocompatibility and an
improved bioavailability. These results represent a significant advancement in the development of
DDX3X inhibitors as a novel class of broad spectrum and safe anti-HIV-1 drugs.

INTRODUCTION
The human immunodeficiency virus type 1 (HIV-1) is one of the most devastating viral infections
affecting humans and still represents a serious global health issue. Despite antiretroviral therapy (ART)
has been introduced in 1996, it has been estimated that 36.7 million persons are living with HIV-1
worldwide, with more than 1 million annual deaths per year[1]. Due to the persistence of latently
infected cells and to the high genetic variability, all the efforts to completely eradicate this pathogen or
to develop effective prophylactic or therapeutic vaccines have failed so far. Nowadays, thanks to the
improved potency of currently available antivirals, HIV-1 replication can be successfully halted in most
of treated patients[2]. Unfortunately, ART is life-long and its efficacy can be compromised by the
emergence of toxicities, comorbidities and the selection of resistance mutations, which ultimately led to
the progression of the disease[3]. Moreover, immunocompromised patients are susceptible to bacterial,
fungal, and viral infections[4,5] complicating their therapeutic regimen with increased pill burden,
complex drug interactions and high costs. On these bases, discovery and development of drugs endowed
with a new mechanisms of action, non-inferior tolerability, improved pharmacokinetic profile and
possibly contributing to reduce the burden of latent cellular reservoir and the chronic immune
activation, is of great necessity to improve the quality of existing treatments.
The indirect-acting antiviral agents (IAAs) represent an interesting class of drugs, characterized by a
great genetic barrier that make them less susceptible to resistance[6]. In facts viruses exploit the
components of the host cells in order to complete their replication, making human cofactors useful
targets to develop new antiviral agents[7,8]. The DEAD-box family comprehends a large number of
human proteins known to be recruited by different viruses[9–11]. In 2004 the ATPase/RNA helicase X-
linked DEAD-box polypeptide 3 (DDX3X) has been identified as an indispensable human cofactor for
the Rev/CRM1 dependent nuclear export of the unspliced HIV-1 RNAs[12]. Starting from such
evidence, in the last ten years we thoroughly analyzed DDX3X structure, identifying ligands able to
selectively bind the ATPase or the helicase binding site, increasing their activities and reaching optimal
specificity against our target protein. Accordingly, our small molecule DDX3X inhibitors were able to

inhibit HIV-1 replication[13–17] (Chart 1). In a previous work we identified compound EI01D[15] the
first DDX3X inhibitor designed to selectively block the helicase binding pocket by competing with
RNA. Through optimization studies we discovered the derivative 16d[17] (compound 1 of the present
paper) that represents the first DDX3X inhibitor with broad spectrum antiviral activity against multiple
RNA viruses (Hepatitis C virus, West Nile virus, Dengue virus) including several HIV-1 mutant strains
resistant to the current direct acting antivirals (DAAs). According to these features and to the novel
mechanism of action, we investigated the biocompatibility and biodistribution (BD) in vivo of a selected
DDX3X inhibitor.
Chart 1. Structures of known DDX3X inhibitors with anti-HIV-1 activity
Since DDX3X is well conserved from yeasts to humans, we performed preliminary in vivo studies in
rats that share 99% DDX3X sequence identity with humans[18,19]. Despite the promising antiviral
activities, as anticipated by our preliminary in vitro ADME analysis, the poor aqueous solubility of
compound 1 limited its BD in plasma and organs, preventing us from carrying out in vivo experiments
that require multiple administrations such as sub-chronic toxicity and pharmacodynamics.

In the present study, we designed and synthesized novel derivatives with improved kinetic aqueous
solubility with the purpose to ameliorate the BD and the pharmacokinetic (PK) properties of our hit
compound.
Due to the unavailability of the DDX3X crystal structure in the active closed conformation, we
previously generated a homology model that led us to identify several RNA-competitive
inhibitors[16,17,20]. Herein, through our computational model we rationally designed a series of novel
derivatives, by inserting polar groups respectively on the side chain at triazole (C4) position or by
inserting different hydrogen bond donors on tolyl ring (compounds 15a-e). The molecules able to retain
compound 1 main interactions within the helicase binding site and characterized by increased in silico
predicted aqueous solubility were then synthesized and validated as inhibitors of the DDX3X helicase
activity. 21 novel DDX3X inhibitors were found, and their ADME properties were experimentally
confirmed in vitro. 11 compounds were then evaluated against HIV-1 infected cells, revealing
promising antiviral activities and selectivity indexes (SI). Taking into account ADME parameters,
antiviral activity and SI, compound 6b was selected for further studies. According to its host-protein
targeting direct mechanism, the antiviral activity of compound 6b was confirmed against a panel of
HIV-1 resistant strains carrying mutations to currently approved DAAs. Finally, BD, PK and sub-
chronic toxicity experiments were performed in mice, revealing optimal tolerability and improved
biodistribution.
RESULTS AND DISCUSSION
Molecular modelling
Twenty-five novel derivatives of compound 1 were rationally designed using the homology model of
the closed conformation of DDX3X previously built by us[16]. Compounds were selected on the basis
of their docking score and taking into account their predicted absorption, distribution, metabolism, and
excretion (ADME) properties calculated with the QikProp software[21]. DDX3X can assume three
different conformations, called open, pre-RNA binding and closed conformation[22] and only the closed

conformation presents the structural properties competent for the RNA binding. Docking analysis within
the RNA binding site was executed using Gold version 5.2.[23], investigating the region with a distance
of 10Å from Arg276. We decided to use a consensus scoring method, with chemscore that has been
used as docking fitness function and goldscore as rescore fitness function. For each molecule 100 poses
were generated. Visual analysis was performed using Pymol, version 1.8.4.0.[24]
a
b
c

Figure 1 Binding mode of selected compounds. Molecules were
docked in RNA binding site of our in-house built homology model
of the closed conformation of DDX3X. (a) In magenta is showed
the binding pose of 15a; (b) in yellow is reported the binding pose
of 6a, (c) in aquamarine is reported the binding pose of 6b.
As represented in Figure 1, the triazole moiety of 15a forms hydrogen bonds with Arg326, Gly302 and
Thr323. The two urea NH-groups form hydrogen bonds with the backbone of Pro274, while the ether-
oxygen binds Thr498. The aromatic group is accommodated in a hydrophobic pocket, constituted by
Gly473, Val500 and Ala499. Analogously, triazole moiety of 6a binds the guanidine group of Arg276,
while the two urea NH-groups bind the carbonyl group of the backbone of Pro274. The isoquinoline
moiety interacts hydrophobically with Gly473, Val500 and Ala499. Compound 6b takes hydrogen
bonds with the backbone of Pro274, while the triazole ring takes contacts with Arg276. As for the
previous compound, the isoquinoline moiety is accommodated in a hydrophobic pocket constituted by
Gly473, Val500 and Ala499.
Molecular dynamic simulations were executed on the molecules. Amber 16 was used to perform 40 ns
of molecular dynamic simulation at the temperature of 300K. In particular, as showed in figure 2, by the
root-mean-square deviation (rmsd) value (Figure 2), of 6a, 15a and 15b, they are stable in the active site

6a, shows rmsd values stabilized around 2.8Å. The trajectory analysis confirms that the molecule
interacts with the side chain of Arg276 and with the backbone of Pro274. During the simulation, the
triazole ring takes contact with the backbone of Gly303 while the isoquinoline moiety is accommodated
in the hydrophobic region constituted by Gly473, Val500 and Ala499, and the nitrogen atom makes a
hydrogen bond with Arg351.
15a is stable in the RNA binding site, despite the increment of rmsd values due to a rotation of phenyl
and butyl groups of the molecule in the binding site. 15a takes contacts with Arg326, Gly302 and
Thr323 and other hydrogen bonds are also formed between the urea moiety and the backbone of Pro274
and the side chain of Thr498.
Finally, the trajectory analysis of 6b shows a compound able to establish long lasting contacts with the
amino acids previously described, especially between the urea NH-groups and the backbone of Pro274
and between the isoquinoline group and the side chain of Arg326.
a
b

c
Figure 2 (a) Plot of the rmsd for 6a. The plot shows an increasing value for the rmsd, which is
stabilized around 2.8 Å. (b) Plot of the rmsd for 15a. In a first stage, the rmsd oscillates around 2Å,
then, in the red portion, it increases its value. This variation is only caused by the rotation of the phenyl
moiety of 15a and not by the shift of the whole molecule into the binding site. Concerning the blue
region, the increase in the rmsd value is due to the movement of the butyl terminal of 15a into the
solvent. (c) Plot of the rmsd for 6b. Also in this case, the rmsd variation in the first part of the
production is due to the rotation of the isopentyl moiety of 6b, then it stably interacts with the amino
acids that constitute the binding site. This behaviour is related to the rmsd value that oscillates around
2Å during the rest of the simulation.

Chemistry
The general strategy to synthesize the triazole derivatives is shown in Scheme 1. Azide 3 was
synthesized from 4-nitroaniline by diazotization reaction. Huisgen 1,3-dipolar cycloaddiction[25,26]
with selected terminal alkynes 4a-g furnished 1, 2, 3-triazoles 5a-g, subsequent Pd/C mediated
hydrogenation led to desired anilines with excellent yields. Finally, target ureas 6a, 6b, 7c, 7g, 8d, 9e,
9f, 10a were produced by treatment with different commercially available isocyanates or with
isocyanates synthesized by reacting the opportune aromatic amine with triphosgene with 60-86% yields.
Alkyne 4c was synthesized through Jones oxidation[27] of 4-pentyn-1-ol, 4f was synthesized by
reaction with β-D-ribofuranose 1-acetate 2,3-5 and BF3 Et2O as a Lewis acid and 4g by esterification p-
toluensolfonic acid (PTSA) catalyzed of 4c.
Scheme 1. Synthesis of ureas 6a-10a modified on triazole sidechaina
aReagents and conditions: i. a) tert-butyl nitrite (t-BuONO), CH3CN, 20 min. 0°C; b) Trimethylsilyl
azide (TMSN3), CH3CN, 2h, room temperature (r.t.) (99%); ii. PTSA, DCM/ EtOH (4:1), 36h, reflux
(80%); iii. Alkyne 4a-g, CuSO4·5 H2O, sodium ascorbate, H2O t-BuOH (1:1), MW 120°C, 300W, 10
min (70-86%); iv. a) H2, Pd/C, MeOH, 1h (80-99%), b) opportune isocyanate, CH2Cl2, 12h, r.t. (60-
86%)

As depicted in Scheme 2 aniline 11, was converted into phenol 12 by Sandmeyer reaction, using the
synthetic procedure already reported by Taniguchi et al.[28] and obtaining a comparable yield.
Subsequent Mitsunobu reaction with 3-hydroxytetrahydrofuran, or alkylation with cyclopentyl chloride,
or phosphorylation in presence of diethylchlorophosphate, o methoxymethylchloride (MOM-Cl) or with
β-D-ribofuranose 1-acetate 2,3-5 tribenzoate provided ethers 13a-e that were reduced and converted into
desired ureas 15a-e as described above. Acidic deprotection of ether 15e provided phenol 15f.
Scheme 2. Synthesis of O-alkylated ureas modified on tolyl ring 15a-ea
aReagents and conditions: i. (a) H2SO4-H2O (3:1), 100°C, 30 min.; (b)NaNO2(aq) 0°C to rt c) 150°C, 2h
(70%); ii. (for 13a) PPh3, Diisopropyl azodicarboxylate (DIAD), 3-hydroxytetrahydrofuran, THF, r.t. 9h
(84%); iii. (for 13c) titanium (IV) isopropoxyde (TTIP); diethyl chlorophosphate, TEA, DCM, O/N.
r.t.(67%); iv. (for 13d) (a) β-D-ribofuranose 1-acetate 2,3-5 tribenzoate, BF3 Et2O 0°C, CH2Cl2 ,15 min;
(b) K2CO3 15 min (62%); v. (for 13b and 13e) (a) NaH, DMF 0°C to rt (b) cyclopentyl chloride or
MOM-Cl, tetrabutylammonium iodide (TBAI), DMF, r.t., 1h (70-99%); vi. H2, Pd/C, MeOH, r.t., 1h
(99%); vii. opportune isocyanate, CH2Cl2, 12h, r.t. (69-79%) viii. HCl (3N), MeOH, r.t. 12h (92%).
With the aim of increasing aqueous solubility several polar groups were introduced on the butyl
sidechain, such as amino groups, alcohols, ethers and esters. The synthesis of the series bearing an
amino moiety in the aliphatic side chain was performed as depicted in Scheme 3. Alcohols 5h-l were
synthesized by click reaction, activated by tosylation (16a-c), and converted into desired amines by
nucleophilic displacement. Subsequent reduction Pd catalysed and reaction with the opportune
isocyanates provide ureas 20a-c, 21c and 22c. Alcohols intermediates 5h-l were converted into ureas

23a and 23b and subsequently into derivatives 25-27 through Steglich esterification. In order to check
the importance of a phosphate in the sidechain, alcohol 23a was phosphorylated with diethyl
chlorophosphate and titanium (IV) isopropoxyde (TTIP) as a Lewis acid (24a).
Scheme 3. Synthesis of ureas modified on triazole sidechaina
aReagents and conditions: i. alkyne, CuSO4·5 H2O, sodium ascorbate, H2O tBuOH (1:1), MW
300W,10 min, 120°C (84-90%); ii. KOH, TsCl, THF (dry) 24h, r.t. (72-83%); iii. opportune amine,
DCM, 9 h, 80°C (69-95%); iv (a) H2, Pd/C, MeOH, 1h (99%), (b) opportune isocyanate DCM, 5h r.t.
(67-75%); v. TTIP; diethyl chlorophosphate, TEA, DCM o.n. r.t, (75%)vi.opportune acid, DCC, DMAP,
DCM, DMF 9 h, r.t. (68-74%).

Biological evaluation
Compounds were tested for their ability to inhibit the helicase activity of DDX3X. Data are presented
as half-maximal inhibitory concentration (IC50) and were calculated using a fluorescence resonance
energy transfer (FRET)-based assay previously published[17].
As reported in Table 1, the introduction of a planar isoquinolyl ring is well tolerated, being able to
hydrophobically interact with Gly473, Val500 and Ala499. Accordingly, compounds 6a and 6b reach
inhibitory activities of 0.15 and 0.12 µM.
Reverse ester 7g maintains a good inhibitory activity of 2.02 µM, in contrast its corresponding
carboxylic acid 7c precipitated during the assays.
Ether derivatives 8d and 9e retain satisfactory activities of 1.51 and 1.00 µM, in contrast the
introduction of ribose in the sidechain abolish the activity (compound 9f), probably due to the spatial
constrains of the pocket as highlighted in our docking experiment.
In addition, we checked the effect of hydrogen bond acceptor on tolyl ring. Even if the carbonyl group
of 10a establishes an additional interaction with the backbone of Gly473, 10a presents a small reduction
of the activity, probably due to the pocket constrains. Best results were obtained with tetrahydrofuran
derivative 15a that reaches an IC50 of 0.10 µM. As already discussed in molecular modelling section,
15a confirms all the key interactions already seen for compound 1, in addition the ether-oxygen
establishes a hydrogen bond with Thr498. Cyclopentyl derivative 15b, phosphoric ester 15c and para
methoxymethyl-hydroxy derivative 15e show high inhibitory activities of 0.14, 0.94 and 0.82 µM. All
compounds form the key interactions with Gly302, Arg326 and Thr323, in addition the phenyl ring of
15c establishes a cation-π interaction with Arg351, while the ether-oxygen of 15e forms a hydrogen
bond with Thr498. Phenol 15f maintains good inhibitory activity of 0.49 µM, in contrast the
introduction of ribose is detrimental for the activity, probably due to the high steric hindrance
(compound 15d).
Alkyl chain at triazole (C4) position was explored by introducing amines. Three different morpholine
derivatives were synthesized, increasing the chain length from one to three methylenes (derivatives 20a-

c), in addition we replaced morpholine with methylpiperazine 21c and N,N-dimethylamine 22c. As
reported in Table 1, the inhibitory activities are directly proportional to the length of the methylene
spacer and dimethylamino group is preferred respect to the more hindered morpholino or
methylpiperazino moiety. These data confirms our previous published results, which highlighted that
linear and long sidechains are preferred with respect to bulky and hindered groups[17,20]. Accordingly,
alcohols 23a and 23b are moderately active, probably due to their lower occupancy within the pocket,
and maintain the interactions with Gly302 and Pro274. Phosphate ester 24 retains a good inhibitory
value, confirming that polar atoms are well tolerated in presence of linear and flexible substituents. In
particular, the phosphate ester of 24 establishes a hydrogen bond with Arg276, and the phenyl ring
makes a cation-π interaction with Arg480. Among derivatives characterized by the introduction of polar
groups on the triazole sidechain the best results were obtained with ester derivatives 25 and 27, which
make an additional hydrogen bond between the carbonyl ester and Arg276. In fact, 25 has an activity of
1.29 µM, and the corresponding trifluoromethyl derivative 26 shows an IC50 of 0.40µM. Ester 27, due
to the presence of the bulkier cinnamic acid moiety, is less active with an IC50 of 5.09 µM, that
confirmed again our previous findings.
Table 1. Anti-enzymatic activity of the compounds against the DDX3X helicase a.
Cmpd.ID
IC50± SD b(µM)
Cmpd.ID
IC50± SD b(µM)
0.82±0.12
0.30±0.16
1*
15e
15f
0.15±0.09
0.12±0.10
0.49±0.10
6a
6b
na
20a

nd
59.68±3.0
53.80±1.1
7c
20b
CF3
H
H
20c
N
N
O
N
N
N
2.02±0.7
COOEt
7g
1.51±0.8
1.00±0.5
na
40.06±1.7
2.90±0.2
13.00±3.1
8d
9e
21c
22c
23a
9f
2.49±0.7
11.20±1.2
0.89±0.6
10a
23b
24
0.10±0.07
15a
15b
25
0.14±0.05
1.29±2.1

26
27
0.94±0.12
0.40±0.5
5.09±0.8
15c
15d
na
a Data represent means values of at least two experiments each performed in duplicate ± S.D. b IC50
half maximal inhibitory concentration. na: not active. nd: not determined, compound precipitated from
medium
Table 2. Antiviral activity of selected compounds against HIV-1 infected cells.
c
IC50b ± SD
M)
CC50
d
Cpd IDa
SI
(
μ
(μ
M)
1.11 ± 0.5*
2.2 ± 1.5
2.1 ± 1.0
>50
90
81
36
41
-
1*
6a
80
86
6b
125
90
7g
3.8 ± 1.1
2.6 ± 1.0
0.2 ± 0.02
5.5 ± 1.6
>50
24
6
8d
16
15a
15b
15c
15d
23b
24
40
200
2.9
-
16
100
80
>50
-
>50
50
-
>50
95
-
26
aData represent mean ± standard deviation at least two experiments; b IC50: half maximal inhibitory
concentration; c CC50: Half maximal cytotoxic concentration; d Selectivity Index (SI): CC50 to IC50 ratio;
*Previously published data.

The 9 compounds endowed of the most promising DDX3X inhibitory activities, alcohol 23b with
inhibitory activity below average and compound 15d completely inactive against DDX3X, were then
evaluated through the BiCycle assay as previously described[29]. Briefly, human T cell lymphoma
derived clone H9 was infected with HIV-1 wild type NL4-3 strain in presence of serial dilution of
compounds. After 72 hours, supernatants from each well were used to infect the TZM-bl cell line, which
allows the quantitative analysis of HIV infection by measuring the expression of the luciferase gene
integrated in the genome of the cells under the control of HIV-1 LTR promoter. Recently, Van Voss et
al. [30] reported that the small molecule RK-33 caused lower oxygen consumption rates and decreased
intracellular ATP concentrations. Taking into account this potential biological effect, we calculated half
maximal cytotoxic concentrations (CC50) of our compounds using CellTiter-Glo 2.0 assay (Promega),
which evaluates the cell viability by measuring cellular ATP as a marker of the cellular metabolic
activity. According to the reported mechanism of toxicity, the CC50 of compound 1, that was previously
tested by MTT assay in different cell lines was equal to 90 µM.
As reported in Table 2 compounds 6a and 6b and 15b are characterized by the best antiviral activities,
and good selectivity indexes, respectively 36, 41 and 200. Noteworthy, compound 15b shows an IC50 of
0.2 µM, 5-times higher than hit compound 1 already published. Compound 8d retains appreciable
activity and low cytotoxicity, in contrast 15a and 15c were found toxic at the concentration of 16µM.
Compound 15d, which was inserted in the antiviral panel as a negative control, was found inactive at
the maximum concentration tested (50µM). Analogously 23, probably due to its low inhibiting
concentration, does not show any significant activity up to 50 µM. 24 and 26 were surprisingly found
inactive, further ADME analysis (see ADME paragraph) demonstrated their high susceptibility to
hydrolysis.
Activity on HIV-1 resistant strains
Taking into account the low aqueous solubility of 15b (for details see ADME section), lower than hit
compound 1 and outside the recommended range reported for a drug candidate[31], we decided to

perform subsequent studies on compound 6b. Its antiviral activity was evaluated against a panel of HIV-
1 strains carrying mutations that confer resistance to the major classes of currently approved antivirals
such as protease inhibitors (PIs), nucleos(t)ide reverse transcriptase inhibitors (NRTIs), non-nucleoside
reverse transcriptase inhibitors (NNRTIs), integrase inhibitors (INIs) (Table 3). As expected, IC50 values
are comprised between 1.5 to 2.3 µM, resulting in fold change values ranging from 0.7 to 1.1, clearly
confirming that resistant strains are fully susceptible to DDX3X inhibitor 6b. Results are in agreement
with our previously published results[17] and confirm once more the efficacy of IAAs in overcoming
viral resistance.
Table 3. Antiviral activity of compound 6b against HIV-1 resistant strains
c
d
Drug class
resistanceb
Virusa
NL4-3
Resistance mutations
IC50 ± SD (
μ
M)
FC
none
none
2.1 ± 0.8
2.3 ± 0.1
-
Major: V32I, I54V, I84V, L90M
Accessory: L10F, V11I, K20T,
L33F, E35G, A71I, G73S, L89V
PIs
1.1
0.8
11808
7400
M41L, E44D, D67N, T69D,
M184V, L210W, T215Y
NRTIs
1.8 ± 0.8
INIs
G140S, Q148H
1.5 ± 0.1
1.8 ± 0.1
0.7
0.8
11847
12231
NNRTIs
K103N, V179F, Y181C
b
aNIH AIDS Reagent Program catalogue number (www.aidsreagent.org). PIs: protease inhibitors;
NRTIs: nucleos(t)ide reverse transcriptase inhibitors; NNRTIs: non-nucleoside reverse transcriptase
c
inhibitors; INIs: integrase inhibitors. Antiviral activity calculated with a two round of infection assay;
dFold change values indicate the ratio between IC50 values from resistant and wild type strains.

In vitro ADME analysis
Selected compounds were profiled in vitro for aqueous solubility at pH 7.4, liver microsomal stability
and membrane passive permeability.
Kinetic solubility was calculated adding 1 mg of compound into 1 mL of water. After 24 h of stirring
at 27°C, the mixture was filtered and the quantity of solubilized compound determined by LC-MS-MS.
As shown in Table 4, except for ether 15b, accordingly to our in silico prediction all the selected
compounds showed improved aqueous solubility respect to the parent compound 1. In fact, ether 15b,
despite its very promising antiviral activity has an aqueous solubility outside the range recommended
for a drug candidate. This drawback, already prevented us to perform the repeated toxicological studies
with hit 1, due to the compound’s accumulation and to the problems related to its formulation at higher
doses. Since repeated administrations are necessary to perform efficacy tests, 15b was discarded. Except
for esters 7g and 26, which had only modest improvement of solubility (respectively 3 and 6-times
higher), the other compounds were from 10 (isoquinolyl derivatives 6a and 6b) to 100-times (9e and
15c) more soluble than 1.
We then analyzed passive membrane permeability (AppP) using Parallel Artificial Membrane
Permeability Assay (PAMPA). Briefly, compound was dissolved in DMSO, diluted with phosphate
buffer and placed into the donor compartment of a multi-well microtitre plate. After 5 hours of
incubation, the quantity of compound able to diffuse from the donor compartment to an acceptor
compartment, passing through a semipermeable artificial membrane, was calculated by LC-MS-MS. As
reported in Table 4 compounds are characterized by low but non limiting AppP, esters derivatives 7g
and 26 have values comparable or higher than 1.
Finally, microsomal metabolic stability tests were performed. Each compound was solubilized in
DMSO solution and incubated at 37 °C for 60 min in phosphate buffer in presence of human liver
microsomes. The reaction mixtures were then centrifuged, and the parent drug and metabolites were
subsequently determined by LC-UV-MS. As reported in table 4 compounds showed good to excellent

metabolic stability, with exception of esters derivatives 7g and 26 which were respectively converted
into the corresponding alcohol and carboxylic acid.
Table 4. In vitro ADME studies of selected compounds.
LogSb
HLM Stabilityc
Cmpd.ID
AppPa
2.86·10-6
-7.0
99
1
6a
<0.1·10-6
0.18·10-6
1.56·10-6
0.69·10-6
1.96·10-6
0.68·10-6
0.72·10-6
4.24·10-6
-5.8
-5.7
-6.4
-4.3
-7.4
-4.7
-5.4
-6.7
95.2
89.7
16.9
94.2
92.5
85.2
53.7
49.3
6b
7g
9e
15b
15c
24
26
Apparent permeability reported in cm·s-1. Aqueous solubility expressed as Log of molar
a
b
c
concentration. Human Liver Microsomal Metabolic Stability expressed as percentage of unmodified
parent drug.
In vivo BD, PK and toxicity studies
Despite the promising antiviral activity of 6b, its aqueous solubility outside the recommended range
coupled with the previous issues related to hit 1 formulation and accumulation in the organs, 6b was
chosen as preclinical candidate. Its low passive permeability, as reported in ADME section, in absence
of any information about its active permeability, led us to investigate its administration using
intravenous (i.v.) route. In order to choose the best dose of compound 6b to be administered, and to
identify its concentration in plasma and organs, four groups of 18 Balb/c mice each, were treated with
compound 6b, at the dose of 8 mg/kg, 16 mg/kg or 32 mg/kg and the other received vehicle. Mice were
all injected by i.v. administration via tail vein and blood samples were collected at 15 and 30 min and 1,
2, 4, 6, and 24 h. Samples were processed and analyzed by HPLC-MS (For details see methods). The
main pharmacokinetic parameters from single- compartment model analysis are summarized in Figure

3. The half-life elimination and the plasmatic clearance values denoted that 6b was slowly eliminated
after i.v. administration.
Figure 3. Pharmacokinetic analysis. Four groups of 18 Balb/c mice each, were treated with compound 6b, at the
dose of 8 mg/kg, 16 mg/kg or 32 mg/kg. Mice were all injected by intravenous administration via tail vein and
blood samples were collected at 15 and 30 min and 1, 2, 4, 6, and 24 h. Samples were processed and analyzed by
HPLC-MS.
Tissues (kidneys, liver, spleen and brain) were collected at 15 and 30 min and 1, 2, 4, 6, and 24 h after
i.v. administration of 16 mg/kg and 32 mg/kg of 6b. Tissue samples were accurately processed and
analyzed, results are reported in Figure 4 (For details see methods). Compound 6b reached the
maximum concentration after 1h, and was completely eliminated after 24h. No trace of 6b was found
into the brain, demonstrating its inability to cross blood-brain barrier. As reported by Nolan et al.[32]
the spleen may act as a reservoir for HIV thus contributing to the viral persistence. Interestingly, we
found important concentrations of 6b in the spleen, that suggest a potential use of our compound in the
reservoir reduction. According to the ADME data, 6b reached good concentrations, that are in every
case major than its IC50. In fact, after 240 minutes its plasmatic concentration was around 10 µM, 5-

times higher than its IC50. After 24 hours the plasmatic concentration decreased to 2µM, suggesting that
a daily administration will be necessary in a future efficacy test.
Figure 4. Biodistribution of 6b in Balb/c mice. Concentration levels of compound in mice tissues at 5, 30, 60,
120, 240 and 1440 minutes of 6b at the dose of 16 mg/kg, and 32 mg/kg iv.
In order to identify potential target organs and toxicities two groups of 10 Balb/c mice each,
were treated with 6b at the highest dose of 32 mg/kg. Mice were all injected by intravenous
administration via tail vein, three times a week for three consecutive weeks. Observations were
made twice daily for clinical signs of pharmacologic and toxicological effects of the treatment.
The body weights of all mice were recorded before the experiment and once every day after the
treatment showed no significant changes. No animal died before the end point. Two days after
the last injections, liver, kidneys and brain were excised from all animals, extensively washed
with distilled water and stored at 4°C in 4% formaldehyde for subsequent analysis. Depth
histopathological examination was carried out following necropsy. Organ Samples sections were
stained with hematoxylin and eosin (H&E). As highlighted in Figure 5, no histological
alterations were found at the tested dose, demonstrating that 6b is well tolerated.

Figure 5: Representative images of the histological examination of HE-stained sections of
Brains (A, B, C, D), A-B: Treated mice, C: Vehicle group, D: Wilde Type group; Liver (E, F, G,
H), E-F: Treated mice, G: Vehicle group, H: Wilde Type group; Kidneys (I, L, M, N), I-L:
Treated mice, M: Vehicle group, N: Wilde Type group. Treated mice does not exhibit abnormal
histopathological changes compared with control groups. The microphotographs were taken
using digital camera (Nikon SLR-D3000) at original magnification of 200X. kidneys.
23

Conclusions
In the present paper, starting from hit compound 1, 25 novel potential inhibitors of the helicase
activity of the human protein DDX3X were rationally designed using a homology model
previously generated by us. The molecules able to retain compound 1 main interactions within
the helicase binding site and characterized by increased in silico predicted aqueous solubility
were then synthesized and validated against DDX3X enzyme. As a result, 21 novel DDX3X
inhibitors were found, with inhibitory activities ranging from 0.1 to 60µM. Eleven compounds
were then evaluated against HIV-1 infected cells, by using an in vitro phenotypic cell-based
assay, revealing activities ranging from the low micromolar to submicromolar range. A
preliminary ADME analysis was performed, in order to select the most promising compound for
further in vivo assays. Compounds metabolically instable or poorly soluble were abandoned.
Unfortunately, despite the promising antiviral activity of 0.2 µM, compound 15b was discarded
due to its very low aqueous solubility, which already prevented us to perform subchronic in vivo
studies. Taking into account ADME parameters, antiviral activity and SI, compound 6b, 10-
times more soluble than hit compound 1, was selected to perform additional studies. According
to its mechanism, directed against a host target, the antiviral activity was retained against a panel
of HIV-1 resistant strains carrying resistance mutations in the viral proteins targeted by the
currently approved DAAs, with fold changes comprised between 0.8 and 1.1 that demonstrated
the full susceptibility to the drug.
Finally PK, BD and subchronic toxicity in mice were carried-out, revealing optimal tolerability
and improved biodistribution respect to hit compound 1 previously published[17]. In conclusion,
we report herein a novel promising DDX3X inhibitor, characterized by antiviral activity against
24

HIV-1 wt and resistant strains. In vivo results highlighted optimal tolerability and suggest that a
daily dose of 32 mg/kg could be used to perform efficacy studies in HIV-1 infected mice,
reaching concentrations 4-times higher than IC50. The absence of histological alterations,
coupled with the good PK profile and the promising antiviral activity, make 6b a good candidate
for in vivo experiments alone or in combination with other drugs with different mechanism of
action as a novel strategy to overcome the drug resistance associated with other DAAs. In
addition, the high concentration found into the spleen suggested a possible therapeutic use in the
reservoirs’ treatment. Finally, the SAR reported herein will be taken into account to synthesize
novel back-up derivatives with improved ADME properties.
Even if further studies are necessary to prove the efficacy on an animal model of HIV-1
infection, our results pave the way for the use of host targeting antivirals such as DDX3X
inhibitors against viral diseases, in particular in the treatment of resistant strains that are no more
susceptible to the DAAs currently available on the market.
AUTHOR INFORMATION
Corresponding Author
* E-mail *botta.maurizio@gmail.com (BM), giovanni.maga@igm.cnr.it (GM),
Author Contributions
MB and GM had the original idea. BA designed, synthesized compounds and wrote the
manuscript. VR purified the recombinant DDX3X and performed all the enzymatic assays. FS,
AG and AB performed antiviral and cytotoxicity assays. AG and CP helped with protein
purification and enzymatic assays. CZ performed ADME assays and pharmacokinetic analysis.
25

CT designed compounds and performed computational analysis. FB, MM and MS performed in
vivo studies All authors have given approval to the final version of the manuscript.
ACKNOWLEDGMENTS
This work was supported by First Health Pharmaceuticals B. V. (Amsterdam). Financial support
has been also provided by Tuscany Region Grant FAS Salute 2014 (DD 4042/2014) to MB and
MZ. GM and VR were partially supported by the CNR Project "Identification of novel molecules
with therapeutic potential against Zika virus and other emerging viruses"
NOTES
This paper is dedicated to the memory of Prof. Maurizio Botta, pioneer in DDX3X research and
beloved master.
ABBREVIATIONS
Human Immunodeficiency Virus 1 (HIV-1); tert-Butyl nitrite (t-BuONO); trimethylsilylazide
(TMeSiN3); Titanium(IV) isopropoxide (TTIP); N,N'-Dicyclohexylcarbodiimide (DCC), 4-
Dimethylaminopyridine (DMAP); absorption distribution metabolism and excretion (ADME);
(SAR) structure activity relationships; mechanism of action (MOA); Diisopropyl
azodicarboxylate, (DIAD);
standard deviation (SD); hematoxylin and eosin (H&E);
pharmacokinetic (PK), biodistribution (BD); pharmacodynamics (PD); selectivity indexe (SI).
REFERENCES
[1]
[2]
AIDSinfo | UNAIDS, (n.d.). http://aidsinfo.unaids.org/ (accessed September 27, 2018).
F.J. Lee, J. Amin, A. Carr, Efficacy of initial antiretroviral therapy for HIV-1 infection in
adults: a systematic review and meta-analysis of 114 studies with up to 144 weeks’
26

follow-up., PLoS One. 9 (2014) e97482. https://doi.org/10.1371/journal.pone.0097482.
L. Muri, A. Gamell, A.J. Ntamatungiro, T.R. Glass, L.B. Luwanda, M. Battegay, H.
Furrer, C. Hatz, M. Tanner, I. Felger, T. Klimkait, E. Letang, G. Isglobal, Development of
HIV drug resistance and therapeutic failure in children and adolescents in rural Tanzania:
[3]
an
emerging
public
health
concern,
AIDS.
31
(2017)
61–70.
https://doi.org/10.1097/QAD.0000000000001273.
[4]
[5]
K. Naidoo, N. Dookie, K. Naidoo, N. Yende-Zuma, B. Chimukangara, A. Bhushan, D.
Govender, S. Gengiah, N. Padayatchi, Infection and Drug Resistance Dovepress Recurrent
tuberculosis among hIV-coinfected patients: a case series from KwaZulu-natal, Infect.
Drug Resist. (2018) 11–1413. https://doi.org/10.2147/IDR.S150644.
S. Das, J. Opoku, A. Allston, M. Kharfen, Detecting spatial clusters of HIV and hepatitis
coinfections,
PLoS
One.
13
(2018)
e0203674.
https://doi.org/10.1371/journal.pone.0203674.
[6]
[7]
Z. Lou, Y. Sun, Z. Rao, Current progress in antiviral strategies, Trends Pharmacol. Sci. 35
(2014) 86–102. https://doi.org/10.1016/j.tips.2013.11.006.
A. Garbelli, V. Riva, E. Crespan, G. Maga, How to win the HIV-1 drug resistance hurdle
race: running faster or jumping higher?, Biochem. J. 474 (2017) 1559–1577.
https://doi.org/10.1042/BCJ20160772.
[8]
[9]
C. Tintori, A. Brai, A.L. Fallacara, R. Fazi, S. Schenone, M. Botta, Protein-protein
interactions and human cellular cofactors as new targets for HIV therapy, Curr. Opin.
Pharmacol. 18 (2014) 1–8. https://doi.org/10.1016/j.coph.2014.06.005.
W.-T. Wang, T.-Y. Tsai, C.-H. Chao, B.-Y. Lai, Y.-H. Wu Lee, Y-Box Binding Protein 1
Stabilizes Hepatitis C Virus NS5A via Phosphorylation-Mediated Interaction with NS5A
27

To
Regulate
Viral
Propagation,
J.
Virol.
89
(2015)
11584–11602.
https://doi.org/10.1128/JVI.01513-15.
[10] M.E. Loureiro, A.L. Zorzetto-Fernandes, S. Radoshitzky, X. Chi, S. Dallari, N. Marooki,
P. Lèger, S. Foscaldi, V. Harjono, S. Sharma, B.M. Zid, N. López, J.C. de la Torre, S.
Bavari, E. Zúñiga, DDX3 suppresses type I interferons and favors viral replication during
Arenavirus
infection,
PLOS
Pathog.
14
(2018)
e1007125.
https://doi.org/10.1371/journal.ppat.1007125.
[11] V. Meier-Stephenson, T. Mrozowich, M. Pham, T.R. Patel, DEAD-box helicases: the Yin
and Yang roles in viral infections, Biotechnol. Genet. Eng. Rev. 34 (2018) 3–32.
https://doi.org/10.1080/02648725.2018.1467146.
[12] V.S.R.K. Yedavalli, C. Neuveut, Y.-H. Chi, L. Kleiman, K.-T. Jeang, Requirement of
DDX3 DEAD box RNA helicase for HIV-1 Rev-RRE export function., Cell. 119 (2004)
381–92. https://doi.org/10.1016/j.cell.2004.09.029.
[13] A. Garbelli, M. Radi, F. Falchi, S. Beermann, S. Zanoli, F. Manetti, U. Dietrich, M. Botta,
G. Maga, Targeting the human DEAD-box polypeptide 3 (DDX3) RNA helicase as a
novel strategy to inhibit viral replication., Curr. Med. Chem. 18 (2011) 3015–27.
http://www.ncbi.nlm.nih.gov/pubmed/21651478 (accessed August 22, 2018).
[14] G. Maga, F. Falchi, M. Radi, L. Botta, G. Casaluce, M. Bernardini, H. Irannejad, F.
Manetti, A. Garbelli, A. Samuele, S. Zanoli, J.A. Esté, E. Gonzalez, E. Zucca, S. Paolucci,
F. Baldanti, J. De Rijck, Z. Debyser, M. Botta, Toward the Discovery of Novel Anti-HIV
Drugs. Second-Generation Inhibitors of the Cellular ATPase DDX3 with Improved Anti-
HIV Activity: Synthesis, Structure-Activity Relationship Analysis, Cytotoxicity Studies,
and
Target
Validation,
ChemMedChem.
6
(2011)
1371–1389.
28