4
S. Agarwal et al. / Bioorg. Med. Chem. Lett. xxx (2013) xxx–xxx
the effect of solvent (DMSO) alone. Complete or 100% inhibition
is defined by the absence of virus infected cells (foci).
Our syntheses afforded several potent compounds which
show strong inhibitory effects in the virus replication assay with
IC50 and IC90 values in the low micromolar range (Table 1). It ap-
pears that the combination of the cholestane scaffold with an
amino function either attached to the steroidal A-ring (choleste-
rylamines and derivatives thereof) or incorporated in the A-ring
(azacholestanes or azahomocholestanes) is a structural motif
triggering anti-influenza viral activity. In contrast, cholesteryl
sulfate did not inhibit viral replication when tested at concentra-
tions of up to 50 lM, the MTC for this compound. Also O-3b-
cholesterylglycolic acid, which is negatively charged under assay
conditions, and cholestan-3-one oxime exerted no inhibition.
Furthermore,
4a,5a-dihydroxycholestan-3-one (IC50 = 16.0
lM)
and cholestane-3b,4
a
,5 -triol (IC50 = 16.1 M) (compound 3, Ta-
a
l
ble 1) provided only weak potency. Thus, strong polarity local-
ized at the steroidal A-ring does not suffice for antiviral
activity. The presence of an amino function appears to be neces-
sary to obtain a particularly high activity of the compounds.
Since trans-2-aminomethyl-1-cyclohexanol does not show any
inhibitory effect, an amino or aminoalcohol moiety attached to
a cyclic hydrocarbon motif is not the sole reason for the anti-
influenza activity. Clearly, the cholestane scaffold imparts unique
partitioning behavior. These compounds are obviously also dis-
tinct from the classical anti-influenza drugs, the cyclic hydrocar-
bons amantadine (1-adamantylamine) and rimantadine (1-(1-
adamantyl)ethanamine), that target the viral M2 proton chan-
nel.22 We conclude that the anti-influenza activity of the steroi-
dal amines depends on their remarkable lipophilicity, combined
with an amino function positively charged under physiological
conditions. Their bulky cholestane scaffold is too large and
hydrophobic to occupy the aqueous lumen of the proton channel
like the adamantylamines. A second, intramembrane drug-bind-
ing site of the M2 protein has also been discussed.22
Figure 1a. Correlation of raft modulating activities of the steroids 1–3 detected by a
tripartite structure (N-terminally sterol-linked peptide)21a as tracer with those
detected by perylene as tracer. Quantification of raft modulation (RM): disrafting
[%] = 100 ꢁ {1 ꢀ P(tracer)steroid/P(tracer)DMSO};
raft
augmentation
[%] = ꢀ100 ꢁ {1 ꢀ P(tracer)DMSO/P(tracer)steroid}; partition coefficients P are approx-
imated by the ratio of tracer concentrations in the membrane and the aqueous
phase.
The antiviral activity of the steroidal amines is likely to be
exerted by the compounds incorporated into viral or cellular
membranes into which they partition nearly quantitatively. In
order to examine effects of compounds partitioned into viral
membranes, we set up assays for virucidal activity and the inhi-
bition of virus budding (VA and BA, Table 1). The virucidal activ-
ity was assayed on influenza virus, incubated with the steroids
1–3 for 30 min and titrated by focus reduction assay.
Virus budding (BA, Table 1) was studied by biotinylation of in-
fected cells 6 h after infection and, after quenching excess biotin,
harvesting the virus budded during the next hour. Biotinylated
virus particles were captured on streptavidin-coated microtiter
plates and detected by ELISA, as described above. Generally, the ef-
fects in the functional assays do not correlate with the IC50 or IC90
values. However, there is a reasonable correlation between raft
modulating activity detected with the N-terminally sterol-linked
peptide as tracer and the effect in the budding assay (R2 = 0.496,
Fig. 1b). Budding was inhibited by raft augmentation, whereas dis-
rafting had no effect, except for compound 2b.
Despite the broad structural homogeneity of this group of com-
pounds, several derivatives, distinct from each other by a single
CH2 or CH3 moiety or by the configuration at C3, exhibit striking
differences in one or more assay results. For example, viruses were
completely inactivated (virucidal assay) by the secondary and ter-
tiary amines 1c, 1d, and 2b, whereas the quaternary ammonium
salt 1e was somewhat less active. In contrast, compound 2a, an
azacyclic steroid with a seven-membered A-ring but otherwise
similar to 2b, has a much weaker virucidal effect. Based on thera-
peutic indices two compounds stand out, 1g and 1k. Their
improved TI results from the lower toxicity of these compounds.
The set of functional assays employed does not cover all raft-
Figure 1b. Correlation of effect on virus budding with raft modulating activity for
the steroids 1–3. Legend for Figures 1a and 1b: s 1a 1b j 1c ꢀ 1d 1e N1f s1g
h1h ꢁ1i
1j d 1k 4 2a } 2b – 3.
than the traditional plaque reduction assay. The antiviral assay was
carried out on microtiter plates and developed as a cell ELISA (en-
zyme-linked immunosorbent assay) using an antibody to virus
nucleoprotein HB65. Cells are pre-incubated for 5 min with serial
dilutions of the test compounds and then infected with the serially
diluted virus. Potency in the virus reproduction and infectivity as-
say (characterized by IC50 and IC90 values, i.e., the concentrations at
which 50% or 90% of viral reproduction is inhibited) was evaluated
and compared to the maximum tolerated concentration (MTC)
determined under the conditions of the focus reduction assay.
The ratio of MTC and IC50 provides the therapeutic index (TI) as
shown in Table 1. Assays were performed at least twice. Where a
clear dose-effect relation was seen, we calculated the IC50 and
IC90 values by regression analysis. Zero inhibition is defined as
Agarwal, Sameer
