K. Shamim et al.
Bioorganic & Medicinal Chemistry Letters 40 (2021) 127906
structure–activity relationship of the modified derivatives, we generated
a binding model of niclosamide with viral protein target NS3 using an
ensemble docking approach.52 Our model as depicted in Fig. 2 predicts
that niclosamide binds to the NS3-NS2B interface in an inner, well-
defined hydrophobic pocket adjacent to a highly polar region. The sal-
icylic acid motif inserts into the pocket by forming extensive hydro-
phobic and aromatic stacking interactions with the surrounding residues
Tyr-23, Phe-46, Leu-58, Leu-65, and Tyr-79. The OH group on C-2 forms
a hydrogen bond with the backbone O atom of Arg-24 while the 5-Cl
points downward and forms halogen-binding interactions inside the
pocket. Both interactions are predicted to be key for the binding affinity
and activity of niclosamide. On the other side of the molecule, the ani-
lide motif points out of the pocket to the solvent exposed region with the
nitro group forming salt bridge with Arg-28 and Arg-59. These binding
interactions compares well to the NS2B-NS3 protein–protein interaction
(PPI) site53 with NS2B forming key interactions with residues Tyr-52,
Ile-53, and Glu-54 at the PPI interface (Also see Supplementary infor-
mation Suppl. S1).
Scheme 1. Synthesis of anilide derivatives.
permeability of 334 X 10ꢀ 6 cm/s. Thus, by addressing the electronic
properties in the anilide region we could identify 25 and 28 exhibiting
similar IC50s in NS-1 assay but much better physicochemical properties
than the parent niclosamide.
We next turned our attention to potential replacements of the nitro
group and other heteroaromatic/ aliphatic ring system replacements of
the anilide moiety. Introduction of the 5-isopropylthiazol-2-amine in 29
and of 5-nitrothiazol-2-amine in 30 as heterocyclic anilide replacements
proved ineffective, as did the 3-chloropyridin-4-amine in 31. A 4′-amino
substitution replacing 4′-NO2 on the aniline in 32 and the aliphatic 1-
methyl-piperidin-4-amine in 33 also produced inactive compounds.
Replacing the 4′-NO2 substituent with the electron-withdrawing 4′-
cyano substituent in 34, however, proved beneficial for potency and
showed an IC50 of 1.7 µM in the NS-1 inhibition assay.
These docking studies of the niclosamide derivatives with NS3 pro-
vided an understanding of the structural basis for the observed SAR. As
shown in Fig. 3, these analogs were predicted to adopt the same binding
conformation as niclosamide at the binding site. As the salicylic acid ring
was accommodated in a deep and restricted pocket, it was postulated
that substituents with bulky groups would pose steric hindrance and
thereby restrict potency. As expected, replacement of the OH group on
the salicylic acid ring significantly impaired potency (compounds 2, 3,
and 4). Similarly, replacement of the Cl with Me or OMe (9 and 10)
resulted in decreased potency, but substituents with 5-F and 5-Br (5 and
6) exhibited comparable potencies as niclosamide. Interestingly, other
substitutions on salicylic acid were tolerated (7, 11 and 16). For the
anilide motif the binding interaction with Arg-28 was predicted to be
key, and in fact replacement of the 4′-NO2 group with 4′-CN and 4′-CF3
(34, 41 and 42) showed comparable potencies as niclosamide, whereas
other substitutions disrupting the H-bond interaction dramatically
decreased activity (44, 45, 51, 55).
The 4′-fluoro replacement was then tried, and 5 analogs were syn-
thesized. Simple replacement of 4′-NO2 with 4′-F in 35 seemed to be
tolerated as was the 2′, 4′-difluoro analog 36, but the other fluorinated
analogs such as 37 and 38 were less potent in both NS-1 and E-protein
assays and compound 39 was inactive. Compound 40 with a 4′-sulfon-
amide replacement of the nitro moiety was also tried but failed to
improve potency compared to niclosamide as measured in the NS-1
assay.
A CF3 replacement of the 4′-NO2 group was then attempted and
compounds 41 and 42 were synthesized. This modification proved su-
perior and produced more potent compounds in both NS-1 and E-protein
assays (41, NS-1 = 0.8 µM and E-protein = 1.4 µM, and 42, NS-1 = 1.6
µM and E-protein = 1.7 µM) thus showing that the potential liability
associated with the nitro group in niclosamide can be removed without
compromising potency. The 2′-CF3 analog 43 had modest NS-1 and E-
protein inhibition values.
With these insights from our docking studies, advantageous modifi-
cations in the salicylamide and anilide region were then combined
(Table 3). To this end, the 5-bromo-2-hydroxybenzoic acid from 6 that
had shown good potency and ADME characteristics was combined with
the 4′-CF3 and 4′-CN replacements of 4′-NO2 on anilide (59–64). We
were encouraged to see that all of these compounds showed NS-1 IC50s
in the low micromolar range. However, these compounds did not show
improved solubility and PAMPA permeability over niclosamide. Bicyclic
naphthalene in the salicylic acid region and the biphenyl compounds 67
and 68 with extra aromatic appendage showed decreases in potency. An
indole replacement in 69 was also not effective in improving potency.
Homologation in the amine motif of niclosamide was also tried and
the results are shown in Table 1 in the Supplementary information.
Finally, we wanted to corroborate the SAR we had generated with a
Zika virus focus-forming assay to demonstrate actual reductions in viral
titer. To test this, we assayed a set of analogs that varied over a wide
range of potencies in the NS-1 and E-protein assays. As shown in Fig. 4
and Table 4, there was a good correlation between relative potencies
observed in the molecular assays compared to those observed in the viral
titer reduction assay. Viral titer IC50 comparison with cell viability in
SNB-19 cells is shown in Table 4 and plotted in Fig 5.
Introduction of a 4′-acetamido electron-withdrawing group to
replace the 4′-NO2 in 44 or a 3′-dimethyl carbamoyl group in 45 pro-
duced inactive compounds. A 4′-chloro substitution in the anilide was
then examined in 46, 47, and 48 but these compounds did not show
improvements in potencies over niclosamide. Compounds 49 and 50
with 2′-6′-dichloro and 2′-6′-difluoro substitution were inactive.
With electron-donating 4′-OMe and 3′-OMe substitutions in 51 and
52 the IC50s for inhibition were adversely affected. The same trend
continued in 53, 54, 55 and 56 where reduced potencies were observed.
Introduction of bicyclic nitro naphthalenes in the anilide region were
also investigated with compounds 57 and 58; these both showed
desirable inhibition values in the NS-1 assay.
Thus, by modifying the anilide region we found 4′-CN and 4′-CF3 as
suitable 4′-NO2 replacements in compounds 34, 41 and 42, all exhibit-
ing similar potencies as niclosamide in the NS-1 inhibition assay.
Modifying the electronic properties in the anilide region in 25 and 28
resulted in similar IC50s for inhibition in the NS-1 assay compared to
niclosamide but provided improved physicochemical characteristics in
terms of RLM stability, solubility, and PAMPA permeability.
To understand further the binding interaction of niclosamide and the
As we were preparing to report these SAR findings around niclosa-
mide against the Zika virus, our labs were also redirecting drug
screening and development resources against the SARS-CoV-2 virus,
5