Mechanisms of action of novel influenza A/M2 viroporin inhibitors derived from hexamethylene amiloride

Article abstract of DOI:10.1124/mol.115.102731

The increasing prevalence of influenza viruses with resistance to approved antivirals highlights the need for new anti-influenza therapeutics. Here we describe the functional properties of hexamethylene amiloride (HMA)'derived compounds that inhibit the wildtype and adamantane-resistant forms of the influenza A M2 ion channel. For example, 6-(azepan-1-yl)-N-carbamimidoylnicotinamide (9) inhibits amantadine-sensitive M2 currents with 3- to 6-fold greater potency than amantadine or HMA (IC₅₀ 5 0.2 vs. 0.6 and 1.3 μM, respectively). Compound 9 competes with amantadine for M2 inhibition, and molecular docking simulations suggest that 9 binds at site(s) that overlap with amantadine binding. In addition, tert-butyl 4′-(carbamimidoylcarbamoyl)-2′,3-dinitro- [1,1′-biphenyl]-4-carboxylate (27) acts both on adamantanesensitive and a resistantM2variant encoding a serine to asparagine 31 mutation (S31N) with improved efficacy over amantadine and HMA (IC5050.6 μMand4.4mM, respectively).Whereas 9 inhibited in vitro replication of influenza virus encoding wild-type M2 (EC505 2.3 μM), both 27 and tert-butyl 4′-(carbamimidoylcarbamoyl)-2′,3- dinitro-[1,1′-biphenyl]-4-carboxylate (26) preferentially inhibited viruses encoding M2(S31N) (respective EC50 5 18.0 and 1.5 μM). This finding indicates thatHMAderivatives can be designed to inhibit viruses with resistance to amantadine. Our study highlights the potential of HMA derivatives as inhibitors of drug-resistant influenza M2 ion channels.

Full text of DOI:10.1124/mol.115.102731

Molecular Pharmacology Fast Forward. Published on May 18, 2016 as DOI: 10.1124/mol.115.102731
This article has not been copyedited and formatted. The final version may differ from this version.
Mechanisms of action of novel influenza A/M2 viroporin inhibitors derived from hexamethylene
amiloride
Pouria H. Jalily, Jodene Eldstrom, Scott C. Miller, Daniel C. Kwan, Sheldon S. -H. Tai, Doug Chou,
Masahiro Niikura, Ian Tietjen, David Fedida
Department of Anesthesiology, Pharmacology, and Therapeutics, Faculty of Medicine, University of
British Columbia, Vancouver, British Columbia, Canada (P.H.J, J.E, S.C.M, D.C.K, D.C, I.T, D.F)
Faculty of Health Sciences, Simon Fraser University, Burnaby, British Columbia, Canada (S.S.T,
M.N, I.T)

Molecular Pharmacology Fast Forward. Published on May 18, 2016 as DOI: 10.1124/mol.115.102731
This article has not been copyedited and formatted. The final version may differ from this version.
Mechanisms of action of novel influenza A/M2 viroporin inhibitors
Corresponding Author: Dr. David Fedida
2.301 Life Sciences Institute, 2350 Health Sciences Mall, Vancouver, BC Canada V6T 1Z3
Telephone: +1 (604) 822 5806, Fax: +1 (604) 822 2281, Email: David.Fedida@ubc.ca
Number of Text Pages: 21
Number of Tables: 3
Number of Figures: 7
Number of References: 37
Number of Words in Abstract: 195
Number of Words in Introduction: 745
Number of Words in Methods: 1591
Number of Words in Results: 4424
Number of Words in Discussion: 1162
Abbreviations
Boc; tert-Butyloxycarbonyl, EIPA; 5-(N-ethyl-N-isopropyl)-amiloride, GFP; green fluorescent
protein, HMA; hexamethylene amiloride, I-V; current-voltage relationship, IC50; half maximal
inhibitory concentration, MEM; Modified Eagle’s Medium, MES; 4-morpholineethanesulfonic acid,
NMDG; N-Methyl-D-glucamine, RMSD; root mean square deviation, SARS-CoV; SARS
coronavirus, TEVC; two electrode voltage clamp, WT; wild type

Molecular Pharmacology Fast Forward. Published on May 18, 2016 as DOI: 10.1124/mol.115.102731
This article has not been copyedited and formatted. The final version may differ from this version.
Abstract
The increasing prevalence of influenza viruses with resistance to approved antivirals highlights the
need for new anti-influenza therapeutics. Here we describe the functional properties of hexamethylene
amiloride (HMA)-derived compounds that inhibit the wild-type and adamantane-resistant forms of the
influenza A M2 ion channel. For example, 6-(azepan-1-yl)-N-carbamimidoylnicotinamide (9) inhibits
amantadine-sensitive M2 currents with 3 to 6-fold greater potency than amantadine or HMA (IC50s =
0.2 vs. 0.6 and 1.3 µM, respectively). Compound 9 competes with amantadine for M2 inhibition, and
molecular docking simulations suggest that 9 binds at site(s) that overlap with amantadine binding. In
addition, tert-butyl 4'-(carbamimidoylcarbamoyl)-2',3-dinitro-[1,1'-biphenyl]-4-carboxylate (27) acts
both on adamantane-sensitive and a resistant M2 variant encoding a Serine to Asparagine 31 mutation
(S31N) with improved efficacy over amantadine and HMA (IC50s = 0.6 µM and 4.4 µM,
respectively). While 9 inhibited in vitro replication of influenza virus encoding wild-type M2 (EC50
=
2.3 µM), both 27 and tert-butyl 4'-(carbamimidoylcarbamoyl)-2',3-dinitro-[1,1'-biphenyl]-4-
carboxylate (26) preferentially inhibited viruses encoding M2(S31N) (respective EC50s = 18.0 and 1.5
µM). This indicates that HMA derivatives can be designed to inhibit viruses with resistance to
amantadine. Our study highlights the potential of HMA derivatives as inhibitors of drug-resistant
influenza M2 ion channels.

Molecular Pharmacology Fast Forward. Published on May 18, 2016 as DOI: 10.1124/mol.115.102731
This article has not been copyedited and formatted. The final version may differ from this version.
Introduction
Viroporins are virally-encoded transmembrane proteins that facilitate conduction of ions or
small molecules and are required for efficient viral replication (Nieva et al., 2012). Despite their small
size (frequently < 100 amino acids), viroporins in many cases have evolved to function as highly
regulated ion channels, which makes them attractive minimalist models of ion conductance and ion
channel evolution (Pinto et al., 1992; Stouffer et al., 2008; Thiel et al., 2011; OuYang et al., 2013;
OuYang & Chou, 2014). The M2 viroporin of influenza A is a 97 amino acid, type I transmembrane
domain protein that forms a tetrameric ion channel that is proton-gated and proton-selective (Nieva et
al., 2012). After viral entry into host cells, M2 conducts protons from acidic host cell endosomes to
the virion interior to allow for uncoating and release of viral RNA. M2 on host cell endosomal
membranes is also observed in some cases to conduct protons to elevate secretory vesicle pH, thereby
delaying egress of nascent virion particles and preventing viral hemagglutinin from adopting a non-
functional, low pH conformation (Sugrue et al., 1990; Alvarado-Facundo et al., 2015). The M2 ion
channel of influenza B (B/M2) is a functional homolog of A/M2. It is 109 residues long and forms a
homotetramer in the membrane like A/M2. Furthermore, B/M2 exhibits higher channel activity but
shows a similar pH-dependence in terms of its proton conductance. There however exist major
differences between the two channels; apart from the HXXXW sequence motif crucial for channel
activity, the two proteins share nearly no sequence homology and unlike A/M2, the B/M2 proton
conductance activity is entirely insensitive to amantadine and rimantadine. (Mould et al., 2003; Wang
et al., 2009).
The compounds amantadine and rimantadine (Figure 1A) are potent inhibitors of A/M2
proton conductance and licensed influenza antivirals (Hay et al., 1985; Pinto et al., 1992; Chizhmakov
et al., 1996). However, M2 sequence changes that render resistance to adamantanes are now so
prevalent that these compounds are no longer recommended for use (Fiore et al., 2011). For example,
over 90% of transmissible adamantane-resistant influenza strains encode an M2 serine to asparagine
mutation at position 31 (S31N). This mutation disrupts adamantane interactions within the M2 pore
without adversely affecting ion channel activity (Hay et al., 1986; Belshe et al., 1988; Pinto et al.,
1992; Bright et al., 2005; Stouffer et al., 2008, Balannik et al., 2010). Thus, new small molecules that

Molecular Pharmacology Fast Forward. Published on May 18, 2016 as DOI: 10.1124/mol.115.102731
This article has not been copyedited and formatted. The final version may differ from this version.
inhibit adamantane-resistant M2 are desired for both improved understanding of the chemical space
and mechanisms by which M2 activity can be modified in addition to development of new influenza
antivirals.
To date, few compounds are reported to act on the S31N form of M2 from influenza A.
Furthermore, several proposed M2 inhibitors that are derived from established M2 inhibitors and act
on subsets of drug-resistant influenza viruses in vitro instead confer antiviral activity through
alternative mechanisms (Jizhou Wang et al., 2013; Kolocouris et al., 2014), indicating that direct
screening of M2 ion channel activity is necessary to identify bona fide M2(S31N) inhibitors. One of
the most potent adamantane derivative for which detailed M2 current analysis is available, N-((5-
(thiophen-2-yl)isoxazol-3-yl)methyl)adamantan-1-amine (M2WJ352; Figure 1A), inhibits M2(S31N)
but not wild-type (WT) M2 proton currents as measured by two electrode voltage clamp (TEVC)
electrophysiology, with an IC50 of 14 µM (Jun Wang et al., 2013).
A separate class of compounds that remains poorly explored for anti-M2 activity is the
acylguanidines (Kleyman & Cragoe, 1988; Gazina & Petrou, 2012). Initially identified as potassium-
sparing diuretics, acylguanidine-containing amilorides, and hexamethylene amiloride (HMA) in
particular (Figure 1B), are inhibitors of multiple viroporins including those of HCV, HIV-1, and
SARS-CoV (Kleyman & Cragoe, 1988; Ewart et al., 2002; Premkumar et al., 2004; Pervushin et al.,
2009; Gazina & Petrou, 2012). HMA has also been reported to inhibit M2(WT) (IC50 = 1.1 µM by
TEVC) but not M2(S31N), while a related compound N-(5-(1-methyl-1H-pyrazol-4-yl)naphthalene-
2-carbonyl)guanidine (BIT-225; Figure 1B) inhibits currents from the viroporins of Hepatitis C and
HIV-1 and shows promising activity in early clinical trials (Khoury et al., 2010; Luscombe et al.,
2010; Gazina & Petrou, 2012). However, while additional acylguanidines including ethyl-isopropyl
amiloride (EIPA) and (6-(1-methylpryazol-4-yl)-2-napthoyl)guanidinium (BIT-314; Figure 1B) are
also reported to have antiviral activity against multiple viroporins and viruses, their effects on M2
currents are not yet reported (Ewart et al., 2009; Gazina & Petrou, 2012). Herein we describe an
electrophysiology-driven approach to characterize the mechanism of and pharmacologically evaluate
a series of acylguanidines and HMA-like derivatives on inhibition of wild-type and adamantane-
resistant influenza M2 viroporins.

Molecular Pharmacology Fast Forward. Published on May 18, 2016 as DOI: 10.1124/mol.115.102731
This article has not been copyedited and formatted. The final version may differ from this version.
Materials and Methods
Chemistry. Detailed information for synthesis and characterization of compounds 1-33 can be found
in Supplemental Material. M2WJ352 was synthesized as described previously (Jun Wang et al.,
2013).
Amantadine
hydrochloride,
5-(N,N-Hexamethylene)amiloride,
5-(N-Ethyl-N-
isopropyl)amiloride, and 1-Benzoylguanidine were purchased from Sigma-Aldrich.
Electrophysiology. tsA-201 cells, a derivative of the HEK 293T cell line, or ltk- murine fibroblast
(LM) cells were cultured in Modified Eagle’s Medium plus 10% fetal calf serum, 100 U/mL
penicillin, and 100 g/mL streptomycin (MEM + medium). cDNA sequences encoding full-length M2
were derived from the A/Hong Kong/1073/99(H9N2) or B/Lee/1940 references sequence and
contained an N-terminal FLAG-tag plus 3 (Gly) repeat linker. This tag was used to confirm M2(WT)
expression on the cell surface of transfected cells by immunocytochemistry (Supplemental Figure 1).
M2 sequences were cloned into the pcDNA3 plasmid and transiently co-transfected with a pcDNA3
plasmid encoding eGFP into tsA-201 cells using standard transfection protocols (Lipofectamine 2000,
Life Technologies). 24-48 h after transfection, single GFP-positive cells were perfused continuously
at 3-5 mL/min with external (bath) solution containing (in mM): 150 NMDG, 10 HEPES, 10 D-
glucose, 2 CaCl2, 1 MgCl2 buffered at pH 7.4. For low external pH (pHo 5.9 or 5.5) solution, HEPES
was replaced by MES. Patch electrodes were pulled from thin-walled borosilicate glass (World
Precision Instruments) and fire-polished before filling with standard pipette solution containing (in
mM): 140 NMDG, 10 EGTA, 10 HEPES, and 1 MgCl2 buffered at pH 7.2, or pH 6.0 (10 mM MES).
M2 currents were detected in > 90% of GFP positive cells assayed during the course of the study.
Voltage-clamp experiments were performed with an Axopatch 200B amplifier (Molecular
Devices, CA) connected to a Digidata1322A 16-bit digitizer. Pipettes typically had a resistance of 3-5
M . Data were acquired with pCLAMP9.2 software (Molecular Devices, CA) sampled at 10 kHz and
low-pass filtered at 5 kHz. The standard voltage protocol consisted of holding a cell at -40 mV,
followed by a 100-ms pulse to -80 mV, a 300-ms ramp to +40 mV, and a 200-ms step to 0 mV before
stepping back to -40 mV repeated every 4 s at 20-22 °C.

Molecular Pharmacology Fast Forward. Published on May 18, 2016 as DOI: 10.1124/mol.115.102731
This article has not been copyedited and formatted. The final version may differ from this version.
Compound screening. Compounds were prepared in DMSO at 100 mM and diluted with external pH
5.5 solution to desired concentrations. To measure inhibition of M2 currents by compounds, cells
were exposed repeatedly to pHo 7.4 and pHo 5.5 solutions until stable, pH-dependent inward currents
were reproducibly observed. Cells were then treated with compounds at defined concentrations in pHo
5.5 solution for 2 min. Data are presented as percent current inhibition at a given concentration of
compound or concentration required for 50% inhibition (IC50). All compounds were initially tested at
100 µM. Compounds that showed greater than 50% inhibition at 100 µM were further tested at lower
concentrations to estimate the half maximal inhibitory concentration (IC50). This was calculated from
nonlinear regression fitting of percentage inhibition at minimum of three concentrations, and
experiments were performed at least three times at each concentration.
Current-voltage relations. M2-transfected ltk- murine fibroblast (LM) cells were used 18-48 h post-
transfection. Cells were voltage clamped using the whole-cell patch clamp configuration described
above. To maximize internal pH (pHi) control, high concentrations of pH buffer were used as
impermeant ions: The patch pipette contained 90 mM N-methyl-D-glucamine (NMDG), 10 mM
EGTA and 180 mM N-2-hydroxyethylpiperazine-N'-2-ethanesulphonic acid (HEPES). The bath
contained a similar solution with 2 mM CaCl2 replacing EGTA. The pH of all solutions was lowered
to desired values by addition of 5 M aqueous HCl. The osmolality of both bath solutions (pHo 7.4 and
5.6) as well as the patch pipette solution were precisely adjusted to 300 mOsmol/kg by addition of
glucose and measured by an osmometer (OsmetteIITM - Precision Systems Inc, MA). Membrane
current was recorded at 20-22 °C, digitized at 10 kHz. Current was measured using a modified voltage
ramp protocol that extended the voltage range from -80 to +120 mV. pHi was held constant at 7.2,
while pHo was changed by fast perfusion close to the cell, and washout was by slow perfusion. M2
current-voltage relations were determined by subtracting current at pHo 7.4, obtained during the ramp
phase of the voltage protocol, from the activated M2 currents at pHo 5.6, or from current after
exposure to compound 9 or amantadine at pHo 5.6. Reversal potentials (Erev) were estimated by linear
interpolation of data points of the I-V relation on each side of zero current.
Molecular Docking. Blind docking was carried out using AutoDock4.2 software (Morris et al., 1998)
using the default parameters, the Lamarckian genetic algorithm with local search, and 25 million

Molecular Pharmacology Fast Forward. Published on May 18, 2016 as DOI: 10.1124/mol.115.102731
This article has not been copyedited and formatted. The final version may differ from this version.
energy evaluations (Long evals) per run. 150 runs were performed for 9, 120 runs for 26 and 100 for
27. The solution NMR structure of M2 (A/Hong Kong/156/97; PDB 2LY0) (Jun Wang et al., 2013),
with M2WJ332 removed, was used as the receptor. Autodock4.2 allows for 32 flexible bonds to be
modeled; two were used for compound 9, six for compound 26 and seven for compound 27. The
remaining bonds were used to allow flexibility of the side chains of Val27, Ser31, and His37. The
remaining residues of the channel were held rigid during the docking process. The grid box size in
initial runs were 20.25 x 30 x 20.25 angstroms in the x, y and z dimensions, which covered Ser23 to
Arg45 of the crystal structure, and was expanded to 17.62 x 45 x 18.75 angstroms in subsequent runs
to include Asp21 to Lys49, both centered around the inner pore. Figures were generated using
PyMOL with the PyMOL Autodock Plugin (Seeliger & De Groot, 2010).
Cytotoxicity. tsA-201 cells were assessed for cell viability using the MTT reagent kit (Life
Technologies). Cells were plated in 96-well plates (2.5 104 cells / well) and incubated for 24 h,
followed by treatment with drug for 24 h. Cells were then incubated with MTT reagent for 4 h and
treated with an equal volume of 10% SDS + 0.01 M HCl, Cell cultures were read by
spectrophotometry at an absorbance wavelength of 570 nm (A570). A570 background values from wells
without cells were subtracted from the A570 cell culture values and normalized to the average A570
value of cell cultures in the absence of compounds. Data were obtained from at least 3 experiments
and at least 5 concentrations.
Generation of amantadine-sensitive and resistant influenza viruses. Recombinant influenza A
viruses were generated using the reverse genetic system based on the A/Puerto Rico/8/34 (PR8) strain
(Neumann et al., 1999) provided by Dr. Y. Kawaoka (University of Wisconsin, Madison). The
endogenous M2 sequence of PR8 encodes both a threonine at position 27 (T27) and asparagine at
position 31 (N31) that render amantadine resistance. Amantadine-sensitive PR8 viruses were
therefore generated by mutating T27 and N31 to valine (V27) and serine (S31), respectively.
Amantadine-resistant M2 encoding V27 and N31 was also generated. Nucleotide substitutions were
introduced by modifying the sequence of pPol-PR8-MG-M (one of the 12 plasmids that make up the
reverse genetics system) using two-step overlap extension PCR and cloning of subsequent DNA
fragments into the pHH21 vector. A DNA fragment containing the S31 point mutation was generated

Molecular Pharmacology Fast Forward. Published on May 18, 2016 as DOI: 10.1124/mol.115.102731
This article has not been copyedited and formatted. The final version may differ from this version.
using the primers TSH289, TSH293, TSH294, and TSH292 (Supplemental Table 1) and pPol-PR8-
HG-M (M2 T27 + N31) as the template. The PCR fragment and pHH21 vector were then treated with
restriction endonuclease BsmBI and ligated to generate pPol-PR8-HG-M-V27-N31 (M2 V27 + N31).
A DNA fragment encoding V27 and S31 was generated using the same primers and pPol-PR8-HG-M-
A27-S31 (M2 A27 + S31) as the template, which in turn was generated using primers TSH289,
TSH290, TSH291, and TSH292 (Supplemental Table 1) and pPol-PR8-HG-M (M2 T27 + N31) as the
template, resulting in generation of pPol-PR8-HG-M-V27-S31 (M2 V27 + S31). PCRs were
performed using the Expand High Fidelity PCR System (Roche), and all constructs were confirmed
by sequencing using primers TSH284 and TSH285 (Supplemental Table 1). Viruses were generated
by transfecting one of the recombinant pPol-PR8-HG-M plasmids with the other 11 plasmids of the
reverse genetics system to 293FT cells (Life Technologies). Cells were plated in a 6-well plate and
transfected with the plasmid mixture (250 ng each) using TransIT LT-1 (Mirus). Culture supernatants
were collected 48 h after transfection and passed through a 0.45 µm filter. After propagating virus in
Madin-Darby canine kidney (MDCK) cells for 3-4 passages, virus stocks were stored at -80 ˚C.
Viral cytopathic assay. MDCK cells were cultured in DMEM (Life Technologies) plus 10% heat-
inactivated fetal bovine serum (Hyclone), 100 U/mL penicillin, and 100 g/mL streptomycin (Life
Technologies) except during generation of virus stocks and plaque assays, when serum was removed
from the media. Plaque reduction assays were carried out in triplicate as described (Song et al., 2005)
with minor modifications. Briefly, approximately 100 plaque forming units of either PR8 recombinant
with M2 V27 + S31 (PR8M2(WT)) or one with M2 V27 + N31 (PR8M2(S31N)) were mixed with different
concentrations of the compounds, ranging from 100 M to 1 M, and inoculated on confluent MDCK
monolayers in six-well plates. After a 1-hour adsorption at 37 °C, the inoculums were removed and
cells were washed twice with phosphate-buffered saline (PBS). The cells were then overlaid with
DMEM containing 1% Noble agar (Affymetrix), 10 mM HEPES buffer (Lonza), 0.00075% DifcoTM
Trypsin 250 (BD Biosciences), 100 U/mL penicillin, 100 g/mL streptomycin, and each compound at
test concentration. Following an incubation in 5% CO2 at 37°C for three days, plaques were
visualized and counted by staining the cells with 0.01% Neutral Red (Sigma-Aldrich). The
concentration that reduces plaque number by 50% compared to the DMSO control (EC50) was
calculated by regression analysis of the dose–response curves.

Molecular Pharmacology Fast Forward. Published on May 18, 2016 as DOI: 10.1124/mol.115.102731
This article has not been copyedited and formatted. The final version may differ from this version.
Results
Detection of M2 currents by whole-cell patch-clamp electrophysiology. We initially investigated
baseline M2 ion channel activity by co-transfecting tsA-201 cells with plasmids encoding GFP and
M2(WT) (A/Hong Kong/1073/99(H9N2)) and recording pH-dependent ion currents in GFP-positive
cells by whole-cell patch clamp electrophysiology (Figure 2; Chizhmakov et al., 1996; Kolocouris et
al., 2014). Cells were held at a constant membrane potential of -40 mV, and currents were recorded
every 4 s by applying 100 ms pulses to -80 mV. When M2-expressing (i.e., GFP-positive) cells were
incubated in solution at pHo 7.4 (designated by the horizontal black lines above each panel in Figure
2), minimal negative or inward current was observed (< 10 pA), which was normalized to 0.0 pA after
all leak and initial current changes had settled down. In contrast, exposure to pHo 5.9 – 5.5 (designated
by the horizontal white rectangles above each panel) resulted in an initially large inward current that
decayed ~50% during continued exposure to low pHo. Return to pHo 7.4 caused the steady inward
current to deactivate. Subsequent, repeated exposures to extracellular acid solutions reversibly
activated the relatively constant level of inward current (Figure 2A), consistent with previous
observations (Kolocouris et al., 2014; Pinto et al., 1992). Currents were also detected in cells
transfected with M2(WT) lacking the N-terminal FLAG epitope (data not shown) but not in cells
transfected with only GFP control vector (Supplemental Figure 2).
Also consistent with previous reports (Pinto et al., 1992; Chizhmakov et al., 1996),
extracellular administration of amantadine inhibited M2(WT) currents at acidic pHo in a dose-
dependent manner (Figure 2B), indicating that this protocol, which differs in cell-type, M2 strain, and
pulse protocol from previous electrophysiology studies (Pinto et al., 1992; Chizhmakov et al., 1996),
could be used to screen test agents for M2 current inhibition. No obvious differences were observed at
pHo 5.5 to 5.9 in efficacy or rate of M2 inhibition by amantadine and subsequent compounds (data not
shown). For all compounds tested in this study, we obtained dose-response relationships if 50%
inhibition of M2 current was observed at 100 µM. For amantadine, we determined an IC50 of 0.6 ± 0.2
µM (Table 1).
Using this protocol, we then screened four acylguanidine containing molecules (Figure 1B;
Table 1). The most potent compound was HMA, which blocked M2(WT) currents with an IC50 of 1.3

Molecular Pharmacology Fast Forward. Published on May 18, 2016 as DOI: 10.1124/mol.115.102731
This article has not been copyedited and formatted. The final version may differ from this version.
± 0.3 µM (Figure 2C), followed by EIPA, which blocked M2(WT) currents with an IC50 of 52 ± 8
µM. In contrast, both BIT-225 and BIT-314 showed only modest activity against M2(WT) (30 ± 5%
and 14 ± 2% inhibition of M2(WT) at 100 µM, respectively) and were not explored further.
Synthesis of a novel HMA derivative with improved activity against M2(WT). As HMA
exhibited the most activity of the acylguanidines tested against M2(WT), we next asked which
functional groups of HMA were responsible for its inhibitory activity. To test initially the role of
central ring substitutions, we began by replacing the central pyrazine with a phenyl ring to generate
compound 1. Despite removal of the ring nitrogens, the exocyclic amino groups and the chloro atom,
1 exhibited improved activity against M2(WT) (IC50 = 0.4 ± 0.1 µM). In contrast, an analog of 1 with
complete removal of the 7-membered azepane ring (1-Benzoylguanidine) had almost no activity (<
10% block; Table 1), indicating that a distal moiety is required. We did find that substitution of the
azepane ring with cyclohexane was somewhat tolerated for anti-M2(WT) activity (e.g., 4; IC50 = 5.0 ±
1.0 µM); however, this activity was sharply reduced when the cyclohexane was replaced with a more
distally-polar morpholine ring (6; 33 ± 3% inhibition at 100 µM). Moreover, substitution with a
pyrrolidine, benzene, or methoxybenzene was also less tolerated (IC50 = 50 ± 10 µM for 5; 33%
inhibition at 100 µM for 2 and 3). Furthermore, the residual activity of 2 was abolished when the
distal ring was shifted to the ortho or meta position of the central ring relative to acylguanidine (i.e.,
<10% block of M2(WT) by 100 µM of compounds 7, or 8), emphasizing the importance of ligand
linearity, and consistent with our observations of limited anti-M2(WT) efficacy of non-linear
molecules BIT-225 and BIT-314 (Table 1).
We next substituted a single phenyl carbon of 1 with nitrogen and synthesized compound 9.
Like 1, compound 9 was also highly potent and inhibited M2(WT) current at pHo 5.5 (Figure 3A),
with an IC50 of 0.2 ± 0.1 µM (Table 1). In contrast, no inhibition of pH-dependent currents from the
M2 ion channel of influenza B (B/Lee/1940) (Mould et al., 2003) was observed with up to 100 µM of
9 (Figure 3B), demonstrating that the inhibitory activity of 9 is specific to M2 encoded by influenza
A.
Further structure activity studies were performed using 9. With respect to the central ring,
replacing the pyridyl nitrogen with an exocyclic nitrogen in the form of a nitro group (11) also largely

Molecular Pharmacology Fast Forward. Published on May 18, 2016 as DOI: 10.1124/mol.115.102731
This article has not been copyedited and formatted. The final version may differ from this version.
retained activity against M2(WT) (IC50 = 1.8 ± 0.1 µM), but shifting the central pyridyl nitrogen to
the ortho position relative to the acylguanidine moiety (12) substantially reduced activity (38 ± 7%
inhibition at 100 µM). Similar to the compound 1 to 1-benzoylguanidine transition, removal of the
distal 7-membered azapane ring of 9 also eliminated activity (10). Moreover, removal of the
guanidinium and/or carbonyl groups of the acylguanidine moiety (13, 14, 15) completely abolished or
strongly reduced M2(WT) current inhibition at 100 µM (e.g., maximum 36 ± 3% inhibition for 15;
Table 1).
Compound 9 exhibited the most activity against M2(WT) currents, with almost an order of
magnitude improved efficacy over the parent compound HMA. Moreover, and consistent with the
docking model described below, molecule linearity combined with the presence of acylguanidine and
distal ring moieties were necessary for M2(WT) activity, while some modifications to the central and
distal rings were tolerated.
Effects of compound 9 on M2(WT) current-voltage relationships. We were interested to
understand the voltage-dependence of 9 effects on pure M2 conductance in mammalian cells, and so
carried out voltage clamp experiments in LM cells expressing M2(WT) protein. H+ whole-cell
currents were measured in the absence of other monovalent cations (Na+ and K+) using pipette and
external media that contained only impermeant organic ions N-methyl D-glucamine (NMDG+) and
HEPES- or MES-. Consistent with results obtained from tsA-201 cells, in LM cells transfected with
M2(WT), a change in pHo from 7.4 to 5.6 induced an inward current at -80 mV that was sensitive to
both 9 (Figure 4A) and amantadine (Supplemental Figure 3). Current-voltage relationships obtained
during the ramp phase of the protocol between -80 and +120 mV in pHo 7.4, 5.6, and at pHo 5.6 plus
100 M 9 are shown in Figure 4B. It can be seen that all three relationships cross close to +80 mV.
Subtraction of the relationship obtained at pHo 7.4, which we assume contains little M2 current, from
those at pHo 5.6 places the reversal potential of the pHo 5.6 relations on the zero current axis, and
gives an Erev of +80 mV (Figure 4C), which is close to the predicted reversal potential of +92 mV for
a pure, M2-dependent H+ current. Similarly, subtraction of the two relationships obtained at pHo 5.6
gave the 9-sensitive current (Figure 4D), which showed an Erev of +85 mV, again close to that
predicted for a pure H+ current. The S.D of the data points around the best-fit line through the current-

Molecular Pharmacology Fast Forward. Published on May 18, 2016 as DOI: 10.1124/mol.115.102731
This article has not been copyedited and formatted. The final version may differ from this version.
voltage relations at their reversal potentials was ± 8 mV, indicating the overall error in estimating Erev.
These results suggest that lowered pHo activates a relatively pure M2 H+ current, and that 9 (Figure 4)
and amantadine (Supplemental Figure 3) both directly inhibit M2 H+ currents in these cells.
Molecular modelling of M2(WT) block. To gain insight into how 9, 26 and 27 (described below)
might bind and inhibit the M2 channel we performed molecular docking analyses (Huey et al., 2007;
Morris et al., 1998; Seeliger & De Groot, 2010) with the M2 transmembrane domain tetramer (PDB
entry 2LY0, the NMR structure of residues 19-49 of M2 of A/Chiba/5/71(H3N2) in
dodecylphosphocholine micelles, computationally modified to include S31 when necessary; Jun
Wang et al., 2013). In addition, we ran a compound that did not block in electrophysiological studies
to observe the predictions made by the program. Due to constraints of the AutoDock 4 software that
limit the number of flexible bonds in the ligand plus the receptor to 32 for accuracy (Trott & Olson,
2009), the M2 structure was largely held rigid during docking simulations with the exception of 3
flexible residues per tetramer subunit (Val27, Ser31 or Asn31 and His37). These residues were
selected based on 9’s interactions during a trial molecular docking simulation where M2(WT) was
held entirely rigid, in addition to a visual scan for residues with side chains that were most obviously
pointing into the inner vestibule and could provide steric hindrance if left rigid.
Using these parameters, we identified several binding sites for 9 (Figure 5A-5D). Many of the
lowest energy binding conformations were in the turret (Figure 5A), and many did not actually block
the channel, suggesting this might not be the block site. This is consistent with the fact that 9 does not
block N31 (Table 2 and described below) and the two share identical sequence in the turret. Autodock
also predicted in some conformations that the hydrophobic 7-membered azepane ring of 9 could pass
through the constriction created by the valines at position 27 while the acyl guanidine portion
maintained polar interactions with Ser22 and Asp24, creating an effective plug of the pore (Figure.
5B). Much of 9 block is readily reversible upon washout (Figure 5F), and this binding configuration
might explain this reversibility since the majority of 9 remains outside the inner vestibule. The 20%
block that is not reversible might then be a result of full entry of the drug into the inner vestibule
(Figure 5C, 5D). Approximately 50% of predicted binding conformations were in the inner vestibule
but all of these fell 1-2 kcal/mol higher in energy than the majority of turret binding conformations (-

Molecular Pharmacology Fast Forward. Published on May 18, 2016 as DOI: 10.1124/mol.115.102731
This article has not been copyedited and formatted. The final version may differ from this version.
13.61 to -11.85 kcal/mol). These could be with either the acyl guanidine pointing towards the valines
(Figure 5C,D) or towards the histidines at position 37 (Supplemental Figure 6) with similar binding
energies. However, if the initial trajectory of the passage involves a hydrophobic interaction between
the 7-membered azepane ring and the valines, this would favour an inner vestibule orientation as
shown in Figure 5C (shown in more detail in Figure 5D). In this conformation the 7-membered
azepane ring is reaching down to interact with Histidine 37 from two subunits as well as the backbone
of Leucine 38. In addition, the azepane ring and the central pyridyl ring make hydrophobic
interactions with Gly34 and Ala30 respectively and the acyl guanidine interacts with valines that
define the roof of the inner vestibule (Figure 5D).
Interestingly, mutations at multiple M2 residues predicted to interact with 9, including Val27,
Ala30, and Gly34, lead to adamantane resistance in vitro and/or in transmissible viruses (Hay et al.,
1985, 1986; Belshe et al., 1988). Facets of the docking approach, such as use of an empirical
hydration force field for the drug binding in the water-filled lumen and use of a micellar protein
structure and homology model thereof, require a conservative interpretation confined to assessment of
the steric fit of the drug in the pore. Nevertheless, taken together, these observations further support
that 9 inhibits M2(WT), but potentially not M2(S31N), by a pore-blocking mechanism. That 9 does
not block M2-N31, leads us to speculate that the extra bulk and hydrophilicity of the asparagine side
chain hinders, or makes energetically unfavourable, the entry of the bulky hydrophobic 7-membered
azepane ring of 9 into the inner vestibule of the mutant channel.
As 9 and amantadine are predicted to interact with many of the same M2 pore-lining residues,
we next tested whether 9 and amantadine compete for M2(WT) inhibition (Figure 5E-H). Treatment
of M2(WT)-expressing cells with 10 µM amantadine at pHo 5.5 reduced pH-dependent currents by
63.5% (Figure 5E). However, following removal of amantadine from extracellular solution, little
recovery of baseline M2(WT) current was observed after either at least 3.3 min at pHo 5.5 (Figure 5E)
or iterative treatment with pHo 7.4 and pHo 5.5 solutions (data not shown), consistent with previous
observations that amantadine does not readily dissociate from blocked M2(WT) channels (Balannik et
al., 2009, 2010). In contrast, while 10 µM of 9 reduced M2(WT)-dependent currents by 80.5% in
Figure 5F, these currents recovered 57.1% within 3.3 min of removal of 9, indicating reversible

Molecular Pharmacology Fast Forward. Published on May 18, 2016 as DOI: 10.1124/mol.115.102731
This article has not been copyedited and formatted. The final version may differ from this version.
inhibition and rapid recovery of M2(WT) currents. The different M2(WT) current recovery rates
observed after removal of amantadine and 9 from pHo 5.5 solution let us test which compound
preferentially blocked M2 (WT) when applied in competition. As shown in Figure 5G, when 10 µM
of 9 was administered at pHo 5.5 to inhibit M2(WT) (90.5% inhibition in this example), no obvious
further inhibition by 10 µM amantadine was observed. However, after removal of both compounds,
M2(WT) current recovered by 58.0% after 5 min, consistent with a model where M2(WT) was
inhibited primarily by 9 and amantadine was largely unable to dislodge it from the pore. These
recovered M2(WT) currents were subsequently inhibited by 10 µM amantadine (data not shown),
showing that removal of 9 from the pore allowed amantadine access. When 5 µM of 9 was used
(Figure 5H), which resulted in 61% block, subsequent exposure to 10 µM amantadine caused a further
reduction in current. This suggests that amantadine can access M2(WT) channels left unblocked by
the lower concentration of 9. Taken together, these experiments suggest that 9 and amantadine likely
compete for the same or highly overlapping binding sites to inhibit M2(WT).
Activity of compound 9 against adamantane-resistant M2 sequence. To assess the inhibitory
potential of 9 against mutant M2 viroporins, we examined inhibition of pHo-dependent currents in
tsA-201 cells expressing an A/Hong Kong/1073/99(H9N2) M2 sequence encoding S31N. pH-
dependent currents from M2(S31N)-expressing cells were identical to those of cells expressing
M2(WT) (Figure 6). Consistent with this mutation conferring adamantane resistance (Balannik et al.,
2010; Williams et al., 2013), we observed that M2(S31N) was inhibited only 24 ± 1% by 100 µM
amantadine and 10 ± 3% by 100 µM HMA in patch clamp experiments (Table 2). Similarly,
sequential exposure to 10 and 100 µM of 9 at pHo 5.5 had almost no activity on M2(S31N) as shown
in Figure 6A and Table 2 (<10% inhibition).
Discovery of HMA derivatives with activity against M2(S31N). As M2(S31N) is the most
prevalent adamantane-resistant mutation observed in transmissible influenza (Bright et al., 2005), we
next asked if derivatives of 9 could inhibit M2(S31N). As S31N affects both the polarity and the size
of the hydrophobic adamantane binding site (Wang et al., 2011; Williams et al., 2013), we
hypothesized that modifications to the hydrophobic terminus of 9 incorporating more polar groups
might improve M2(S31N) inhibition, perhaps by stabilizing proximal and/or distal interactions within

Molecular Pharmacology Fast Forward. Published on May 18, 2016 as DOI: 10.1124/mol.115.102731
This article has not been copyedited and formatted. The final version may differ from this version.
the M2(S31N) pore. We first measured whether M2(S31N) could be inhibited by compounds 6, 16,
18, and 20, which all contain polar substituents in the distal ring; however, these compounds had little
or no activity at 100 µM (e.g., maximum 16 ± 4% inhibition for 16; Table 2). Furthermore, 100 µM of
1 and 11 also had limited or no activity against M2(S31N) (e.g., maximum 10 ± 1% inhibition for 11;
Table 2), indicating that these central ring modifications, on their own, did not substantially improve
activity against M2(S31N).
We then assessed the combined effects of a more polar, Boc-conjugated distal piperazine (i.e.,
16) with a polar nitro group on the central ring (i.e., 11). Interestingly, this compound (28) inhibited
M2(S31N) with slightly more than additive effects relative to 16 and 11 (32 ± 5% at 100µM; Table
2), indicating that both the nitro and terminal Boc group of the piperazine are needed for M2(S31N)
inhibition. However, 28 did not inhibit M2(WT). This observation, combined with activity against
M2(S31N) which is only slightly improved over amantadine, prompted us to continue exploring 28
derivatives with improved M2(S31N) and M2(WT) activities.
Following our identification of 28 as an early lead for M2(S31N) inhibition, we next found
that substitution of the central ring nitro group with either a chlorine (23) or nitrile (22) reduced
activity against M2(S31N) (<10% and 17 ± 8% at 100 µM respectively). Shifting of the nitro group to
ortho position with regard to the acylguanidine moiety (17) also reduced activity (10 ± 5% at 100
µM). Activity against M2(S31N) was also eliminated by increasing the size of the second ring from
piperazine (28) to homopiperazine (21), with inhibition less than 10% at 100 µM. Finally, compounds
with different size carboxylates (29 and 32) and sulfonamide (31) did not exhibit potent activity
against M2(S31N) (maximum 11 ± 8% at 100 µM). However, switching the Boc-conjugated
piperazine (28) with a piperidine ring with a slightly more polar terminal group (25) resulted in a
compound having the same level of efficacy. Replacing the heterocyclic ring with an aromatic ring
(26) substantially improved activity against M2(S31N) (60 ± 7% at 100 µM; IC50 = 42 ± 5 µM;
Figure 6B). Unlike 28, compound 26 also had observable, albeit weak activity against M2(WT) (20 ±
5% inhibition at 100 µM; Figure 6C). Further substitution of a polar nitro group into the distal ring
resulted in compound 27 which had even more activity against both M2(WT) and M2(S31N), with
IC50s of 0.6 µM and 4.4 µM, respectively.

Molecular Pharmacology Fast Forward. Published on May 18, 2016 as DOI: 10.1124/mol.115.102731
This article has not been copyedited and formatted. The final version may differ from this version.
In this way we identified multiple derivatives of 9 that inhibit M2(S31N) currents, with lead
compounds 26 and 27 inhibiting M2(S31N) with greatly improved activity over HMA and
amantadine while retaining submicromolar M2(WT) activity, in the case of 27. As whole-cell patch
clamp electrophysiology and TEVC are not directly comparable, we synthesized M2WJ352 (Jun
Wang et al., 2013), a potent adamantane derivative, and found its IC50 to be 44 µM (Table 2), some
10-fold weaker than 27.
Molecular docking studies of M2(S31N) block by 26 and 27. When molecular docking studies
were performed with 26 within the M2(S31N) pore, two confirmations with inverse orientations were
again observed to interact with Val27, Ala30, Gly34, His37, and Leu38, along with Asn31 (Figure
6D-E). However, similar to 9, the orientation of the acylguanidine differed by interacting with either
one or more of the valines at position 27 (Figure 6D) or by interacting with the histidines of the
proton sensor (Figure 6E). Both the guanidine amines and the oxygens of the nitro-substituted phenyl
ring could form hydrogen bonds as shown in Figure 6D. Also in some conformations (still within 2.0
RMSD), the oxygens of the Boc terminal group contributed to hydrogen bonding with His37 or
Asn31 (data not shown). Taken together, in comparison to 9, the longer molecule of 26 with added
bulk and polarity of both the nitro and Boc group may help stabilize interactions across more of the
pore, thereby improving block against M2(S31N).
While 26 does not bind in the turret the variety of conformations observed for this compound
was no less impressive than for 9 (Supplemental Figure 4). The compound could be found in an
orientation with the acyl guanidine pointing towards the extracellular domain (Figure 6D and
Supplemental Figure 4B) or cytoplasmically (Figure 6E, Supplemental Figure 4A) and at varying
depths within the M2 pore. It is conceivable that since the lowest energy conformations are deeper in
the pore (Figure 6D and E, Supplemental Figure 4A, B far left) that the other conformations are
intermediate states of the compound as it progresses to a preferred binding site, though all of the
conformations lead to block. 26 does not wash out upon solution change (data not shown) leading
one to speculate that once at least the bulk of the drug is past the valine ring it is difficult to reverse
the entry process.

Molecular Pharmacology Fast Forward. Published on May 18, 2016 as DOI: 10.1124/mol.115.102731
This article has not been copyedited and formatted. The final version may differ from this version.
Binding patterns for 27 were fairly similar whether there was a serine at position 31 or an
asparagine. Averaged over many runs for each channel construct, Val27 was occupied ~67% of the
time, Ala30 ~47%, Ser/Asn31 ~27%, and Gly34 ~16%. Again, compound 27 was found to bind the
channel in “Boc up” or “Boc down” conformations. When Serine was present at position 31 the lower
energy cluster was “Boc up” (Figure 7A(i), Supplemental Figure 5A) though 63% of the runs fell into
one of the “Boc down” conformations. When asparagine was present, the lower energy clusters were
“Boc down”, as well 54% of runs fell into these two clusters (Figure 7A(iii, iv) Supplemental Figure
5B). The lowest energy conformation found in M2(N31) was almost a kcal/mol more than the lowest
energy conformation found when serine was present, which is consistent with the electrophysiological
data showing 27 to be almost 10 times more effective in blocking the WT channel (Tables 1 and 2).
The nitro-substituted phenyl rings tended to interact with the hydrophobic wall of the inner
vestibule, Val27, Ala30, Ile33 and Gly34. The nitro groups on these phenyl rings interacted with
serine or asparagine, and even histidine if the drug bound low enough in the pore, and occasionally
hydrogen bonds were predicted (Figure 7A(iv)). When 27 bound residues in the turret, hydrogen
bonds were much more common between the acyl guanidine hydrogens and the oxygens of Asp21,
Ser22, Ser23 and Asp24. The Boc group made hydrophobic interactions, depending on the depth and
orientation, with the carbons in the histidine side chains (Figure7A(ii,iv)), or with Ala30 and Val27
(Figure 7A (i,iii)). Given that the Boc group presents an hydrophobic terminus it is tempting to
speculate that it is this end that would pass through the valine ring more readily and that the “Boc
down” orientation might be favoured. Adding the bulkier, more hydrophilic asparagine just below the
valine might make entry slightly less favourable.
Molecular docking of compound 14 onto M2. Compound 14 is one which failed to block M2(WT)
at 100 µM (<10%, Table 1), and modeling frequently predicted that 14 would bind in the extracellular
turret (Figure 7B) without blocking the channel with very low binding energy (-17.16 kcal/mol).
Interactions were made with Asp21, Ser22, Ser23, Asp24, and sometimes with one of the valines,
often with the amine hydrogens forming hydrogen bonds with the aspartic acid oxygens. It is likely
possible that more than one molecule of 14 could in fact bind in the turret given the four-fold
symmetry of the channel. Even in this situation, given that the compound is off the central axis of the

Molecular Pharmacology Fast Forward. Published on May 18, 2016 as DOI: 10.1124/mol.115.102731
This article has not been copyedited and formatted. The final version may differ from this version.
pore, there would be no block. Where a binding site was found in the internal vestibule, the lower
energy conformations were almost 2 kcal/mol higher in energy than the turret site (-15.41 kcal/mol;
not shown). Migration into the inner vestibule would require the high affinity turret site remain vacant
to allow an additional molecule of 14 to pass. 14 is a smaller molecule than 9, 26 and 27. It lacks the
acyl guanidine of 9, 26 and 27 that provides bulk and length and the nitro groups bulking up the rings
of 26 and 27. It is likely for this reason that when 14 binds the channel it can fail to block the channel
(Figure 7B).
In vitro cytotoxicity. Compounds that inhibited > 50% of M2(WT) or > 30% of M2(S31N) currents
at 100 µM were assessed for cytotoxicity. Using a standard MTT-based cell metabolic assay, no
obvious effects were observed in tsA-201 cells incubated for 24 h with up to 100 µM amantadine,
M2WJ352, or compound 9 (Table 1). In contrast, HMA inhibited growth of tsA-201 cells with a 50%
cytotoxic concentration (CC50) of 4.7 ± 0.3 µM. Thus, while HMA inhibits M2(S31) currents with
comparable efficacy to the established inhibitor amantadine and 9, it also has undesirable cellular
toxicity consistent with previous observations (Balgi et al., 2013). Intermediate levels of cytotoxicity
between HMA and 9 were observed for both inhibitors of M2(S31N) and other 9 derivatives (e.g.,
CC50 = 25 ± 5 µM for 26, 55 ± 17 µM for 27; Table 2).
In vitro activities against influenza viruses in a cytopathic assay. As functional experiments were
primarily performed to this point on heterologously-expressed M2 ion channels, we next determined
the effects of select compounds on replication of influenza A virus in vitro using a viral cytopathic
assay with MDCK cells (Table 3). Here, we used a previously-described reverse genetic system based
on the A/Puerto Rico/8/34 (PR8) strain (Neumann et al., 1999) to generate PR8 strains encoding M2
with exclusively S31 (PR8M2(WT), containing both V27 and S31) or N31 (PR8M2(S31N) containing V27
and N31). As expected, amantadine exhibited antiviral activity against PR8M2(WT) but not PR8M2(S31N)
in these assays (50% effective concentration (EC50) < 1 µM for PR8M2(WT) vs. > 100 µM for
PR8M2(S31N); Table 3). Compound 9 also inhibited PR8M2(WT) (EC50 = 2.3 ± 0.1 µM), consistent with
results from patch clamp studies using M2(WT). Notably, 9 also exhibited ~10-fold weaker activity
against PR8M2(S31N) (EC50 = 23 ± 5 µM). As 9 did not inhibit M2(S31N) currents at up to 100 µM,
these observations suggest that compound 9 may have an additional antiviral mechanism in addition

Molecular Pharmacology Fast Forward. Published on May 18, 2016 as DOI: 10.1124/mol.115.102731
This article has not been copyedited and formatted. The final version may differ from this version.
to M2(WT) blockade. While off-target antiviral effects are frequently observed for multiple M2
inhibitors (Kolocouris et al., 2014), the disproportionate ability of 9 to inhibit PR8M2(WT) relative to
PR8M2(S31N) replication is nevertheless consistent with its ability to exclusively inhibit M2(WT) in
patch clamp studies and further suggests that the additional efficacy of compound 9 against PR8M2(WT)
relative to PR8M2(S31N) is due to M2 blockade.
The previously described M2(S31N) inhibitor M2WJ352 (Jun Wang et al., 2013) inhibited
PR8M2(S31N) with an EC50 = 1.8 ± 0.7 µM, and with a ~19-fold reduced activity against PR8M2(WT)
(EC50 = 34 ± 6 µM). As M2WJ352 did not inhibit M2(WT) currents at up to 100 µM, these
observations suggest that M2WJ352, like 9, may have antiviral mechanisms in addition to M2(S31N)
blockade. Similarly, PR8M2(S31N) replication was blocked by both 26 and 27 (EC50 vs. PR8M2(S31N) = 1.5
± 0.1 µM and 18 ± 1 µM, respectively), but weaker activity against PR8M2(WT) was also observed
(EC50 vs. PR8M2(WT) = 6.9 ± 1.6 µM and 40 ± 1 µM, respectively; Table 3). While the antiviral activity
of 26 is consistent with electrophysiology studies, antiviral and electrophysiology profiles of 27 are
discordant; thus the antiviral activity of 27 observed here is unlikely to be due primarily to M2
blockade. Finally, no obvious cellular toxicities in MDCK cells were observed for any tested
compound with the exception of 26, where clear cell death was observed only at higher concentrations
(100 µM).

Molecular Pharmacology Fast Forward. Published on May 18, 2016 as DOI: 10.1124/mol.115.102731
This article has not been copyedited and formatted. The final version may differ from this version.
Discussion
Here we show that of four acylguanidine compounds previously reported to act against
multiple viroporins and/or viruses (Gazina & Petrou, 2012), HMA had the most potent efficacy
against influenza A M2(WT) currents, as measured by whole-cell patch-clamp electrophysiology. We
then show that the HMA derivative 9 inhibits M2(WT) and has improved efficacy over amantadine
and HMA, while inhibition of adamantane-resistant M2(S31N) can be achieved with other derivatives
of this chemical class such as 26 and 27. Using a combination of electrophysiological, molecular
docking, and structure-activity relationship studies, we identify the transmembrane pore of the M2
tetramer as the likely interaction site of 9, 26 and 27, which leads to M2 current block, in a region
which overlaps the reported adamantane interaction site.
Mechanisms of block of M2(WT). The M2 protein can transfer protons selectively across
membranes with a H+ electrochemical gradient, a property consistent with its role in modifying virion
and trans-Golgi pH during virus infection (Sugrue et al., 1990; Alvarado-Facundo et al., 2015). In LM
cells expressing the M2 protein, with pHi 7.2 and pHo 5.6, the I-V plots were similar in conductance
and shape across all transfected cells. The inhibitory effect of amantadine (Supplemental Figure 3)
and 9 (Figure 4) brought about an identical compression of the current across all voltages, while the
pH- and drug-dependent currents through M2 demonstrated a reversal potential close to the
equilibrium potential for the transmembrane pH gradient at pHo 5.6, suggesting that 9 and amantadine
inhibited proton currents similarly in our experiments. Our initial molecular docking studies predicted
that, in the lowest-energy state, 9 interacts within the M2(WT) pore in two different orientations
(Figure 5, Supplemental Figure 6) in a manner reminiscent of the previously-described orientations of
rimantadine and amantadine. While alternative models of 9-M2(WT) interactions are clearly possible,
our model is supported by the competition of 9 with amantadine for M2(WT) inhibition (Figure 5),
similar effects on current-voltage relations, and the inability of 9 to inhibit currents from adamantane-
resistant M2(S31N) (Figure 6A). Our structure-activity relationship study also identifies distinct
moieties of 9 that are required to maintain anti-M2(WT) activity. For example, derivatives of 9
lacking portions of the acylguanidine group (e.g., 13, 14, and 15) were less effective in inhibiting
M2(WT), while removal of the azepane ring (compounds 10 and 1-benzoylguanidine) or substitution

Molecular Pharmacology Fast Forward. Published on May 18, 2016 as DOI: 10.1124/mol.115.102731
This article has not been copyedited and formatted. The final version may differ from this version.
with smaller rings (compounds 2, 3, 4, 5 and 6) did not improve on the activity of 9 (Table 1). A
linear molecule was also required for activity, as shifting of the distal ring to either ortho or meta
position relative to the acylguanidine also eliminated activity (7, 8). While some, but not all
compounds with substitutions in the central pyridyl ring maintained activity against M2(WT) (e.g., 1
and 11), 9 retained the most potent inhibition of M2(WT) of the novel compounds tested here.
However, 9 was ineffective in blocking the adamantane-resistant M2(S31N) variant. We then
envisioned that the polarity changes brought about by the S31N mutation would require a molecule
with both a more distal polar end as well as a polar substitution on the central ring, and 28 bearing a
Boc distal end and a nitro-substituted central ring appeared to bind and block the S31N mutant.
Mechanisms of block of M2(S31N). When compared to amantadine, 28 and 25 somewhat improved
blockade of currents from adamantane-resistant M2(S31N) (32-33% inhibition at 100 µM). However,
we did identify two other HMA derivatives, 26 and 27, of which 26 showed improved potency
against M2(S31N) with an IC50 of 42 ± 5 µM, and weak activity against M2(WT), while 27 exhibited
a potent dual inhibitory effect against both M2(WT) as well as M2(S31N) with an IC50 of 4.4 ± 0.5
µM and 0.6 ± 0.1 respectively, as measured by whole-cell patch clamp (Table 1 and 2). Autodock
predicts binding of 26 in S31 that would lead to block, and so the failure to see significant block in the
electrophysiological experiments suggests that forces driving the drug into the pore are different
between S31 and N31. The addition of the second set of oxygens on the second ring structure in 27
appears to overcome this. The overall orientation of 26 and 27 with M2(S31N) is, to a first
approximation, similar to that of 9 with M2(WT). Both 26, 27 and 9 are predicted to enter their
binding site in two orientations to interact variably with both the selectivity filter (Val27) and the
proton sensor (His37) (Figure 5A-D and Figure 6D-E). The binding site for 26 and 27 is predicted to
span from Val27 to His37 (Figure 6 and 7, Supplemental Figure 4 and 5) which also overlaps to a
degree with the binding site previously reported by NMR for the less potent inhibitor M2WJ332 (Jun
Wang et al., 2013). Notably, the acylguanidine carbonyl of both 9, 26 and 27 may contribute to an H-
bond with Ser31 in M2(WT) (for 9) or Val27 in M2(S31N) (for 26 and 27), and removal of this
acylguanidine carbonyl, as seen in compounds 14 and 15, significantly reduced activity against
M2(WT). Thus the ability of 26 and 27 to act on M2(S31N) appears to be driven primarily by the
presence of the acylguanidine moiety, the second aromatic ring, an additional distal Boc group and

Molecular Pharmacology Fast Forward. Published on May 18, 2016 as DOI: 10.1124/mol.115.102731
This article has not been copyedited and formatted. The final version may differ from this version.
the central ring nitro that increase bulk and perhaps stabilize deeper interactions within the expanded
M2(S31N) pore. The second aromatic ring of 26, (in comparison to 28 with a piperazine ring) appears
to play a critical role in further stabilizing hydrophobic interactions with the walls of the amphipathic
pore.
Correlation of viral and electrophysiology assays. Interestingly, in viral assays (Table 3), both
amantadine and 9 displayed significant activity against PR8M2(WT) but not PR8M2(S31N) virus,
supporting that 9 targets this virus, at least in part, by inhibiting proton flux through M2. In contrast,
both M2WJ352 and 26 were more potent against PR8M2(S31N), further supporting their action against
M2(S31N). We note that the limited ability of 9 to inhibit PR8M2(S31N), and 26 to inhibit PR8M2(WT)
electrophysiologically, but significant antiviral activity against these isoforms, suggests that effects of
the M2 block may be amplified in vitro, or indeed that these acylguanidine-based compounds may
exhibit additional antiviral mechanisms, as previously suggested for adamantanes (Kolocouris et al.,
2014). Notably, compounds 9, 26, and 27 also exhibited markedly reduced cytotoxicity compared to
the parent compound HMA in MTT assays with tsA-201 cells, further suggesting that these
compounds may be improved starting points for future antivirals compared with HMA.
In summary, we describe novel acylguanidine-containing inhibitors of M2 viroporins with binding
mechanisms similar to adamantanes, with potent activity against wild-type and adamantane-resistant
M2 ion channels and viruses. The novel HMA derivatives, including compounds 26 and 27 as dual
inhibitors of both M2(WT) and M2(S31N), provide scaffolds that may aid in development of further
non-adamantane compounds with improved inhibitory activity against drug-resistant forms of M2.
These promising leads for additional medicinal chemistry optimization will also require further in
vivo studies to assess their antiviral efficacy, stability, and pharmacokinetic parameters in animal
disease models.
.

Molecular Pharmacology Fast Forward. Published on May 18, 2016 as DOI: 10.1124/mol.115.102731
This article has not been copyedited and formatted. The final version may differ from this version.
Acknowledgements
We thank Luping Yan, Hannah E. Boycott, Shunping Lin, Zhuren Wang, and Daniel
Mendonca for superb technical assistance, Raymond Andersen and Luping Yan for assistance with
mass spectrometry, and Zabrina Brumme and Mark Brockman for use of facilities and critical review
of the manuscript.
Authorship Contributions
Participated in research design: Jalily, Tietjen, Miller, Eldstrom, Tai, Niikura, and Fedida.
Conducted experiments: Jalily, Tietjen, Miller, Eldstrom, Tai, Niikura, Chou, and Kwan.
Preformed data analysis: Jalily, Tietjen, Eldstrom, and Fedida.
Wrote or contributed to the writing of the manuscript: Tietjen, Jalily, Eldstrom, Tai, and Fedida.

Molecular Pharmacology Fast Forward. Published on May 18, 2016 as DOI: 10.1124/mol.115.102731
This article has not been copyedited and formatted. The final version may differ from this version.
References
Alvarado-Facundo E, Gao Y, Ribas-Aparicio RM, Jiménez-Alberto A, Weiss CD, and Wang W
(2015) Influenza Virus M2 Protein Ion Channel Activity Helps To Maintain Pandemic 2009
H1N1 Virus Hemagglutinin Fusion Competence during Transport to the Cell Surface. J Virol
89: 1975-1985.
Balannik V, Wang J, Ohigashi Y, Jing X, Magavern E, Lamb RA, DeGrado WF, and Pinto LH (2009)
Design and pharmacological characterization of inhibitors of amantadine-resistant mutants of
the M2 ion channel of influenza A virus. Biochem 48:11872-11882.
Balannik V, Carnevale V, Fiorin G, Levine BG, Lamb RA, Klein ML, DeGrado WF, and Pinto LH
(2010) Functional studies and modeling of pore-lining residue mutants of the influenza A
virus M2 ion channel. Biochem 49: 696-708.
Balgi AD, Wang J, Cheng DY, Ma C, Pfeifer TA, Shimizu Y, Anderson HJ, Pinto LH, Lamb RA,
DeGrado WF, and Roberge M (2013) Inhibitors of the influenza A virus M2 proton channel
discovered using a high-throughput yeast growth restoration assay. PLoS ONE. 8, e55271.
Belshe RB, Smith MH, Hall CB, Betts R, and Hay AJ (1988) Genetic basis of resistance to
rimantadine emerging during treatment of influenza virus infection. J virol 62: 1508-1512.
Bright RA, Medina M, Xu X, Perez-Oronoz G, Wallis TR, Davis XM, Povinelli L, Cox NJ, and
Klimov AI (2005) Incidence of adamantane resistance among influenza A (H3N2) viruses
isolated worldwide from 1994 to 2005: a cause for concern. Lancet 366: 1175-1181.
Chizhmakov IV, Geraghty FM, Ogden DC, Hayhurst A, Antoniou M, and Hay AJ (1996) Selective
proton permeability and pH regulation of the influenza virus M2 channel expressed in mouse
erythroleukaemia cells. J Physiol 494: 329-336.
Ewart GD, Mills K, Cox GB, and Gage PW (2002) Amiloride derivatives block ion channel activity
and enhancement of virus-like particle budding caused by HIV-1 protein Vpu. Eur Biophys J
31: 26-35.

Molecular Pharmacology Fast Forward. Published on May 18, 2016 as DOI: 10.1124/mol.115.102731
This article has not been copyedited and formatted. The final version may differ from this version.
Ewart GD, Luscombe CA, and Miller M (2009) Hepatitis c antiviral compositions and methods. US
Patent WO2009018609A1.
Fiore AE, Fry A, Shay D, Gubareva L, Bresee JS, and Uyeki TM (2011) Antiviral agents for the
treatment and chemoprophylaxis of influenza: recommendations of the Advisory Committee
on Immunization Practices. MMWR Recomm Rep 60: 1-24.
Gazina EV and Petrou S (2012) Viral targets of acylguanidines. Drug Discovery Today 17: 1039-
1043.
Hay AJ, Wolstenholme AJ, Skehel JJ, and Smith MH (1985) The molecular basis of the specific anti-
influenza action of amantadine. EMBO J 4: 3021-4.
Hay AJ, Zambon MC, Wolstenholme AJ, Skehel JJ, and Smith MH (1986) Molecular basis of
resistance of influenza A viruses to amantadine. J Antimicrob Chemoth 18: 19-29.
Huey R, Morris GM, Olson AJ, and Goodsell DS (2007) A semiempirical free energy force field with
charge-based desolvation. J Comput Chem 28: 1145-1152.
Khoury G, Ewart G, Luscombe C, Miller M, and Wilkinson J (2010) Antiviral Efficacy of the Novel
Compound BIT225 against HIV-1 Release from Human Macrophages. Antimicrob Ag
Chemoth 54: 835-845.
Kleyman TR and Cragoe EJ (1988) Amiloride and its analogs as tools in the study of ion transport. J
Membrane Biol 105: 1-21.
Kolocouris A, Tzitzoglaki C, Johnson FB, Zell R, Wright AK, Cross TA, Tietjen I, Fedida D, and
Busath DD (2014) Aminoadamantanes with persistent in vitro efficacy against H1N1 (2009)
influenza A. J Med Chem 57: 4629-4639.
Luscombe CA, Huang Z, Murray MG, Miller M, Wilkinson J, and Ewart GD (2010) A novel
Hepatitis C virus p7 ion channel inhibitor, BIT225, inhibits bovine viral diarrhea virus in vitro
and shows synergism with recombinant interferon- -2b and nucleoside analogues. Antiviral
Res 86: 144-153.

Molecular Pharmacology Fast Forward. Published on May 18, 2016 as DOI: 10.1124/mol.115.102731
This article has not been copyedited and formatted. The final version may differ from this version.
Morris GM, Goodsell DS, Halliday RS, Huey R, Hart WE, Belew RK, and Olson AJ (1998)
Automated docking using a Lamarckian genetic algorithm and an empirical binding free
energy function. J Comput Chem 19: 1639-1662.
Mould JA, Paterson RG, Takeda M, Ohigashi Y, Venkataraman P, Lamb RA, and Pinto LH (2003)
Influenza B virus BM2 protein has ion channel activity that conducts protons across
membranes. Dev Cell 5: 175-184.
Neumann G, Watanabe T, Ito H, Watanabe S, Goto H, Gao P, Hughes M, Perez DR, Donis R,
Hoffmann E, Hobom G, and Kawaoka Y (1999) Generation of Influenza A Viruses Entirely
from Cloned cDNAs. Proc. Natl. Acad. Sci. U. S. A. 96, 9345-9350.
Nieva JL, Madan V, and Carrasco L (2012) Viroporins: structure and biological functions. Nature Rev
Microb 10: 563-574.
OuYang B, Xie S, Berardi MJ, Zhao X, Dev J, Yu W, Sun B, and Chou JJ (2013) Unusual
architecture of the p7 channel from hepatitis C virus. Nature 498: 521-5.
OuYang B, and Chou JJ (2014) The minimalist architectures of viroporins and their therapeutic
implications. Biochim Biophys Acta 1838: 1058-1067
Pervushin K, Tan E, Parthasarathy K, Lin X, Jiang FL, Yu D, Ardcharaporn V, Soong TW, Liu DX,
and Torres J (2009) Structure and inhibition of the SARS coronavirus envelope protein ion
channel. PLoS Pathog 5: e1000511.
Pinto LH, Holsinger LJ, and Lamb RA (1992) Influenza virus M2 protein has ion channel activity.
Cell 69: 517-528.
Premkumar A, Wilson L, Ewart G, and Gage P (2004) Cation-selective ion channels formed by p7 of
hepatitis C virus are blocked by hexamethylene amiloride. FEBS Lett 557: 99-103.
Seeliger D and De Groot BL (2010) Ligand docking and binding site analysis with PyMOL and
Autodock/Vina. J Comput Aid Mol Des 24: 417-422.

Molecular Pharmacology Fast Forward. Published on May 18, 2016 as DOI: 10.1124/mol.115.102731
This article has not been copyedited and formatted. The final version may differ from this version.
Song JM, Lee KH, and Seong BL (2005) Antiviral Effect of Catechins in Green Tea on Influenza
Virus. Antiviral Res. 68, 66-74.
Stouffer AL, Acharya R, Salom D, Levine AS, Di Costanzo L, Soto CS, Tereshko V, Nanda V,
Stayrook S, and DeGrado WF (2008) Structural basis for the function and inhibition of an
influenza virus proton channel. Nature 451, 596-9.
Sugrue RJ, Bahadur G, Zambon MC, Hall-Smith M, Douglas AR, and Hay AJ (1990) Specific
structural alteration of the influenza haemagglutinin by amantadine. EMBO J 9: 3469-3476.
Thiel G, Baumeister D, Schroeder I, Kast SM, Van Etten JL, and Moroni A (2011) Minimal art: or
why small viral K+ channels are good tools for understanding basic structure and function
relations. Biochim Biophys Acta 1808: 580-588.
Trott O, and Olson AJ (2009) AutoDock Vina: Improving the speed and accuracy of docking with a
new scoring function, efficient optimization, and multithreading. J Comput Chem 31: 455-
461.
Wang J, Pielak RM, McClintock MA and Chou JJ (2009) Solution structure and functional analysis of
the influenza B proton channel. Nat Struct Mol Biol 16: 1267-1271.
Wang J, Ma C, Fiorin G, Carnevale V, Wang T, Hu F, and DeGrado WF (2011) Molecular dynamics
simulation directed rational design of inhibitors targeting drug-resistant mutants of influenza
A virus M2. J Am Chem Soc 133: 12834-12841.
Wang Jizhou, Ma C, Wang J, Jo H, Canturk B, Fiorin G, DeGrado WF (2013) Discovery of Novel
Dual Inhibitors of the Wild-Type and the Most Prevalent Drug-Resistant Mutant, S31N, of
the M2 Proton Channel from Influenza A Virus. J Med Chem 56: 2804-2812.
Wang Jun, Wu Y, Ma C, Fiorin G, Wang J, Pinto LH, Lamb RA, Klein ML, and Degrado WF (2013)
Structure and inhibition of the drug-resistant S31N mutant of the M2 ion channel of influenza
A virus. P Natl Acad Sci USA 110: 1315-1320.

Molecular Pharmacology Fast Forward. Published on May 18, 2016 as DOI: 10.1124/mol.115.102731
This article has not been copyedited and formatted. The final version may differ from this version.
Williams JK, Tietze D, Wang J, Wu Y, DeGrado WF, and Hong M (2013) Drug-induced
conformational and dynamical changes of the S31N mutant of the influenza M2 proton
channel investigated by solid-state NMR. J Am Chem Soc 135: 9885–97.

Molecular Pharmacology Fast Forward. Published on May 18, 2016 as DOI: 10.1124/mol.115.102731
This article has not been copyedited and formatted. The final version may differ from this version.
Footnotes
This work was funded by a Canadian Institutes of Health Research Industry-Partnered Collaborative
Research Operating Grant [IPR-124291] which was co-funded by Cardiome Pharma Corp.,
Vancouver, Canada. We also wish to acknowledge the support from Grand Challenges Canada [0487-
01-10]. Grand Challenges Canada is funded by the Government of Canada and is dedicated to
supporting Bold Ideas with Big Impact in global health