The influenza virus endonuclease: an innovative target for antiviral therapy.

Influenza viruses cause seasonal epidemics as well as pandemic outbreaks, associated with huge economic costs and significant morbidity and mortality, particularly in elderly or frail individuals. The widely recommended influenza vaccines require annual updates and provide inadequate protection, especially in immunocompromised and elderly persons, which are both steadily growing populations. In addition, vaccination cannot counteract the threat of a suddenly emerging influenza pandemic. Hence, antiviral drugs are an indispensable component of the broad approach to treat and prevent influenza infections, including for pandemic preparedness planning and response. Mutant viruses that are resistant to currently available anti-influenza drugs are widely distributed, even among untreated patients, and hence more potent drugs with a different and directly suppressive mode of action are urgently required. The influenza polymerase complex is widely recognized as a key drug target, given its critical role in virus replication and high degree of conservation among influenza A (of human or zoonotic origin) and B viruses. Its PA (polymerase acidic protein) subunit performs the ‘capsnatching’ endonuclease reaction in which cellular pre-mRNAs are cleaved to yield capped primers for transcription of viral RNA to mRNA. This PA endonuclease activity and the inhibition hereof by PA inhibitors (PAIs) to halt virus replication, represent the subject of this thesis project.

The foundations of our project were laid when we investigated the binding mode of one of the lead PAIs (L-742,001) in the PA active site. This β-diketo acid (DKA) compound was already discovered by researchers at Merck two decades ago, but at that time, the complete lack of knowledge regarding its target protein substantially hampered further PAI development. The crystal structure of the endonuclease catalytic site in the N-terminal part of PA (PA-Nter) was revealed in 2009, and this enabled us to perform computer-assisted docking of L-742,001 in the PA active site. A comprehensive mutational analysis was performed to reveal the binding mode of L-742,001 and determine which amino acid changes within the catalytic centre of PA or its surrounding hydrophobic pockets, alter the antiviral sensitivity to L-742,001 in cell culture. We explicitly chose to use cell culture-based methods, to obtain an accurate insight into the compound’s binding mode in a relevant virus infection model, instead of a system using isolated PA enzyme. A high level of resistance (up to 20-fold) to L-742,001 was noted for mutations I120T and H41A within the catalytic core of PA-Nter, consistent with the notion that the metalchelating effect of the bidentate DKA moiety is a crucial factor in endonuclease inhibition by L-742,001. Our observation that a virus carrying the H41A mutation in PA was able to replicate was totally unexpected, since it contradicts the assumption (which was based on PA crystal structures and enzymatic assays with isolated polymerases) that His41 is indispensable for metal-chelation and integrity of PA, and even of the entire polymerase complex. Mutations of the Gly81 residue (to Val, Phe or Thr) resulted in at least 9-fold resistance, suggesting that Gly81 may form a hydrophobic interaction with the benzyl moiety of L-742,001 or may be critical to shape the binding pocket for L-742,001. For PA mutant viruses and viral ribonucleoproteins (vRNPs) bearing mutations within the predicted hydrophobic binding pockets for L-742,001, only a modest (maximum 5-fold) level of resistance to L-742,001 was observed, which may be explained by suboptimal fitting of the compound’s aromatic wings in the proposed binding cavities, thus lacking tight binding interactions. Furthermore, our resistance data implicated a role for Arg124, Val122 and Tyr130, which surround the proposed RNA substrate binding site in PA; this was not acknowledged in previous cocrystal experiments performed by others.
In addition to the binding mode of L-742,001, our mutational analysis provided important information concerning potential resistance sites, of high relevance to anticipate on possible resistance development upon future clinical application of PAIs. The changes which totally destroyed the activity of the reconstituted vRNPs are all located in the catalytic centre of PA. This heavily compromised polymerase activity implies that it is highly unlikely that prolonged exposure to L-742,001 or another PAI would select viruses carrying these single mutations in their PA protein. On the other hand, several mutations situated at non-catalytic sites in PA had no or only marginal impact on the enzymatic functionality of viral ribonucleoprotein complexes, consistent with the less conserved nature of these PA residues. Thus, an endonuclease inhibitor optimized for tight binding to any of these residues might select for escape mutants at the corresponding sites. It is impossible to speculate on the consequences for the antiviral efficacy of such an inhibitor. Nevertheless, our biological data demonstrate that structure-based design of PA inhibitors should be accompanied by cell culture evaluation against specific PA mutant viruses, to verify the proposed mode of action and anticipate on any potential resistance sites that might be encountered during future clinical use.

Next, we focused on a series of DKA-based compounds encompassing different scaffolds and analysed their structure-activity relationship to build a plausible five-points 3D pharmacophore model, defining strict structural requirements along with chemical features. Several DKA derivatives were found to cause potent inhibition of the PA-Nter enzyme with IC50 values comparable to that of the prototype L-742,001. Three compounds (DKA-10, with a pyrrole scaffold, and DKA-40 and DKA-41, with an indole scaffold) exhibited moderate antiviral activity in cell culture, and were proven to affect viral RNA synthesis. In addition, we developed a novel real-time enzymatic assay based on a molecular beacon (MB) substrate, which is amenable to high-throughput screening and allowed to measure the enzyme kinetics of PA-Nter. Using this MB assay, we convincingly demonstrated that the setup of the enzymatic assay (i.e. substrate, metal cofactor and type of readout) should be carefully chosen during PAI evaluation. Whereas most enzymatic studies with isolated PA-Nter have indicated that the enzyme is considerably more active in the presence of Mn2+ compared to Mg2+, our MB assay works equally well with Mg2+ as with Mn2+. Since the intracellular concentration of free Mg2+ is at least 1000-fold higher than that of Mn2+, magnesium may be more biologically relevant, and evaluation of potential PAIs against both metals (as possible with our MB assay) seems recommended. For most of the DKA inhibitors tested here, the inhibitory activity against PA-Nter was far less with Mg2+ than with Mn2+, yet the lead compound L-742,001 appeared unique in having equal activity against either Mg2+ or Mn2+. This may implicate that PA-Nter assays using Mg2+ are more stringent for evaluating potential PAIs. Furthermore, while our enzymatic studies identified several PA inhibitors, only a few of these molecules were subsequently proven to have activity in influenza virus-infected cell cultures, which emphasizes the importance of proof-of-concept cell culture evaluation during early hit discovery of potential PAIs. Hence, we propose a comprehensive approach to guide PAI development, by integrating complementary enzymatic, cellular and mechanistic assays.

In the last part of our project, we aimed at implementing our gathered experience and assessed diverse scaffolds of potential metal-chelating PAIs, namely 2-hydroxybenzamides, thiosemicarbazones, N-acylhydrazones and dihydroxyindole-2-carboxamides. We combined our enzymatic and cellular assays to evaluate their inhibitory potency, and several compounds were found to produce marked inhibition in PA enzymatic assays (IC50 values < 10 μM), with promising antiviral activity in cell culture and favourable selectivity. However, although conceived as PAIs, for most of these inhibitors, the antiviral target in cell culture was unrelated to PA, but rather associated with an early or late event in the viral life cycle. We embarked on the further unravelling of their mechanism of action with time-of-addition studies and confocal microscopy, and identified at least one original compound which most likely blocks viral replication by interfering with PA endonuclease activity. In conclusion, we made an important contribution to the development and optimization of potential PAIs and achieved unique insights concerning the precise conformation and metal requirements of the PA enzyme. In addition, we revealed pitfalls in currently used enzymatic assays and convincingly demonstrated that, in order to be successful, the hit-to-lead process on novel PAIs requires rapid progression of potential hit compounds to relevant cell-based mechanistic assays.