A bacteria-based search for drugs against avian and swine flu yields a potent and resistance-resilient channel blocker

Abstract

Influenza represents a significant threat with seasonal epidemics that can transition to global pandemics, and cross-species infection presenting a continuous challenge. While vaccines and several antiviral drugs are available, constant genetic changes vitiate these prevention and treatment options. Consequently, we decided to search for inhibitors against one of the virus's validated drug targets, its M2 channel that is blocked by aminoadamantanes. Regrettably, widespread mutations in M2 abolish the antiflu activity of said blockers. Therefore, we devised bacteria-based genetic assays that can screen for drugs against aminoadamantane-sensitive and resistant M2 channels and map the resistance potential of any identifiable blocker. Subsequent in cellulo testing and structure-activity relationship studies yielded a synergistic combination of two compounds, Theobromine and Arainosine, that exhibited remarkable antiviral activity by directly inhibiting the virus's channel. The drug duo was potent against H1N1 pandemic swine flu, H5N1 pandemic avian flu, and aminoadamantane-resistant and sensitive strains alike, exhibiting activity that surpassed oseltamivir, the leading antiflu drug on the market. When this drug duo was tested in an animal model, it once more outperformed oseltamivir, considerably reducing disease symptoms and viral RNA progeny. Importantly, harnessing the bacterial genetic selection, we could demonstrate that the drug duo's potential for eliciting drug resistance is significantly smaller and molecularly distinct from that of aminoadamantanes. In conclusion, the outcome of this study represents a new potential treatment option for influenza alongside an approach that is sufficiently general and readily applicable to other viral targets.

Keywords: channel blockers; influenza; ion channel.

Fig. 1.

Screening results employing the negative assay against the aminoadamantane-sensitive M2 channel (A) and against a channel containing the S31N mutation (B) that renders the channel aminoadamantane resistant (28). Bacteria without the IPTG inducer (green) and untreated bacteria (red) serve as controls. Effective blockers increase bacterial growth. Note that rimantadine exhibits potency against the aminoadamantane-sensitive channel (A) but, as expected, has no activity against a channel with the S31N mutation and is incapable of restoring bacterial growth (B).

Fig. 2.

In cellulo antiviral dose–response studies of active drugs. MDCK cells were infected with an H1N1 aminoadamantane-resistant (in red) or H5N1 aminoadamantane-sensitive (in green) virus at an MOI of 0.3 and their viability was monitored by MTS after 48 h. Results are normalized relative to uninfected cells and untreated cells. K_s, ${\mu }_{\text{max}}$, and corresponding R² values are shown for each drug in the Inset. Note that for the theobromine and arainosine drug combination, equal concentrations were used and the fit employed data until 0.3 μM.

Fig. 3.

Combination antiviral in cellulo studies of asunaprevir, fludarabine, flunisolide grazoprevir, theobromine, and vidarabine. MDCK cells were infected with the H1N1 virus at an MOI of 0.3 and their viability was monitored by MTS after 48 h. All compounds were at 0.1 μM. Diagonal elements (boxed) represent activity of the individual drugs. Results are normalized relative to uninfected cells and untreated cells and represent the average of at least three measurements.

Fig. 4.

In cellulo binding mechanism analyses of theobromine and arainosine. MDCK cells were infected with the H5N1 virus at an MOI of 0.3, and their viability was monitored by MTS after 48 h. All drugs are given at 300 nM except adamantanol, which, on its own or in combination with other drugs, is administered at 5 μM. Results are normalized relative to uninfected cells and untreated cells.

Fig. 5.

Structural activity relationship study of theobromine (A) and vidarabine (B). MDCK cells were infected with the H1N1 virus at an MOI of 0.3 and their viability was monitored by MTS after 48 h. Dashed lines indicate viability of untreated cells, while uninfected cells are set to 100%.

Fig. 6.

In vivo antiviral activity. (A) Scheme of experimental protocol. (B) Mice were infected with the A/Wisconsin/629-D02452/2009 (H1N1)pdm09 strain and viral RNA in the lung was measured after five days (C). Same as (B), but mice were infected with the PR8 H1N1 strain.

Fig. 7.

In vivo antiviral activity. Relative weight changes of mice receiving different treatments, as indicated, during the course of the infection. (A–E) Individual mice weight changes. (F) Average data for comparative purposes whereby values of four days post infection are listed in the graph.

Fig. 8.

Analysis of the resistance potential against the M2 channel blockers. A library of bacteria harboring a randomly mutated M2 channel was grown in liquid media for three hours after which their plasmids were sequenced. The prevalence of every amino acid is tabulated and color-coded as indicated in the bar. The data correspond to difference prevalence of bacteria that are exposed to (A) rimantadine or (B) theobromine relative to untreated bacteria.

Fig. 9.

Proposed mechanism for inhibition of the arainosine-theobromine drug duo. The structure of the M2 protein bound to rimantadine [PDBID 6US9, (40)] is sliced in the Middle to depict the channel’s pore and drug binding site. Two out of the four histidines that are important for channel activity are depicted in ball-and-stick alongside rimantadine. Theobromine (Top) and arainosine (Bottom) are shown on the Right with proposed hydrogen bonding in green.