What are the considerations when selecting a model for influenza drug discovery?

KEYWORDS:

• Influenza virus
• drug discovery
• antivirals
• Drug-resistance mutants
• host factors
• animal models
• 3D cell culture

1. Introduction

There have been several viral outbreaks in the 21^st century, not only through the influenza virus and the ongoing SARS-CoV-2 pandemic but also through SARS-CoV, Ebola, MERS-CoV, Zika, and Nipah virus epidemics [1]. Our particular recent experience underlines the importance of developing efficient vaccines and antivirals against viral infection. Besides the pandemics and epidemics of the past, seasonal flu (influenza) causes severe mortality and morbidity annually, especially in those with weak immune responses. Indeed, a recent modeling study has shown that an estimated 290,000 to 650,000 mortalities occur annually in addition to the 3–5 million severe cases worldwide [2] and this is in spite of both vaccines and antivirals.

Vaccines are the preferred means for controlling influenza virus infection and for reducing the risk of influenza-associated illness. However, these vaccines are not without issues including low effectiveness and vaccination rate [3]. Meanwhile, there are two types of FDA-approved antiviral drugs on the market; firstly, there are the neuraminidase inhibitors (NAIs), oseltamivir and zanamivir, and the polymerase acidic protein (PA) endonuclease inhibitor (ENI), baloxavir. For a time, M2 ion-channel blockers, such as amantadine and rimantadine, were also approved by the FDA for clinical use, but they are not currently used in clinics due to the emergence of drug-resistant mutants. In the case of NAIs, although most of the circulating viruses are sensitive to NAI treatment, there are still several highly resistant mutants to NAIs [4]. Moreover, as for baloxavir, several reports have shown that one or two amino acid mutations are sufficient to confer drug resistance [5], which again highlights the need for us to continue developing novel antivirals against influenza virus infection (several potential antiviral targets are described in reference [6]). The purpose of this editorial is to briefly describe the current drug discovery strategies against influenza and to discuss the considerations when starting an influenza virus drug discovery project.

2. Drug discovery strategies

The discovery and development of antiviral drugs begins with knowledge in virology, and in particular, how the virus infects the host cells and how the virus interacts/utilizes the host protein within the cells [7]. In the case of segmented negative-stranded RNA viruses (sNSVs), viral RNA forms a complex with viral polymerase and nucleoprotein to form a viral ribonucleoprotein (vRNP) complex. Influenza viruses utilize viral neuraminidase, the prime antiviral target for NAIs, that cleaves the host receptor sialic acid to release the newly assembled virions from the infected cells at the later stage of infection [8]. To determine whether the target of interest can be interfered with to cripple viral replication, methods such as reverse genetics [9] or CRISPR screening [10] can be implemented. When selecting a model that is efficient for antiviral drug discovery, several considerations must be made. These considerations, in combination with a knowledge of chemical biology, will accelerate future drug design and discovery. We list below some of the important points that must be considered.

2.1. Consideration point 1: the initial screening method

To start the drug discovery process, it is first necessary to choose the screening methods required for hit identification. These screening methods can be divided into two categories: structure-based virtual screening and high-throughput screening (HTS) [11]. Additionally, in silico pharmacophore-based drug discovery has gained considerable attraction due to its potential to reduce time and cost during the hit identification step [7]. Indeed, the introduction of machine learning techniques in virtual screening [12] has improved its performance and aided the prediction of the ADMET (Absorption, Distribution, Metabolism, Excretion, and Toxicity) properties of antivirals, thereby helping to avoid adverse drug reactions in the clinical stage of drug development. In the case of HTS, it is important to decide whether the goal of the screening campaign is to find a compound that inhibits the overall replication of the virus or a specific viral enzyme or protein–protein interaction [13]. The advantage of the latter approach is that it is possible to select novel scaffolds for optimization prior to the determination of their antiviral activities in cell-culture. Recent advances in fragment-based drug discovery, which involve a modest number of fragments, can represent a large diverse scaffold through limiting molecular size; this approach has gained much interest in the drug discovery field and can be applied to both pharmacophore-based de novo design and HTS [14,15].

2.2. Consideration point 2 – the generation of small molecule library

Successful drug discovery relies on the quality of the small molecular library in use. Although there is extensive information on chemical compounds from a plethora of public sources, it is crucial to 1) select compounds that have drug-likeness properties such as cell permeability, 2) generate a diverse compound library from large chemical library databases such as DrugBank, PubChem, or ChEMBL, and 3) construct a focused library based on the initial hit compounds from the screening campaign or a set of chemical fragments [11]. If the undertaken research is focused on a specific target, it is beneficial to identify the protein function prior to selecting the compound library. For example, it is beneficial to generate a library that interacts with metals when designing the discovery of influenza virus PA endonuclease inhibitors which utilize divalent metal ions, such as magnesium or manganese, as crucial cofactors for the enzymatic function [16].

2.3. Consideration point 3 – which animal model?

The development of antiviral drugs heavily relies on preclinical efficacy to provide answers that need to be addressed prior to human clinical trials. There are numerous animal models available for influenza antiviral research; however, each animal model comes with its own advantages and limitations [17]. Animal models for influenza virus drug discovery include mice, ferrets, Guinea pigs, swine, and non-human primates. The authors believe that ferrets are the ideal animal model as they are permissive to infection and produce similar pathological outcomes to those observed in humans. On the other hand, mice are often used for initial in vivo screening as well as for toxicity evaluation due to their lower cost, ease of handling, and availability [17]. In addition to what is mentioned above, Zebrafish and Drosophila are gaining attention as a unique animal model for developing drugs modulating the host antiviral RNA interference pathway [18,19].

3. Expert opinion

Due to the viral reservoir within animals that allows frequent animal-human transmission, the eradication of influenza virus disease is doubtful, with the virus anticipated to continue to circulate and infect humans [20]. The segmented nature of the RNA genome and the lack of a proof-reading mechanism in virus replication facilitate an antigenic shift and drift that not only lowers vaccine effectiveness but also leads to the emergence of antiviral-resistant strains. Furthermore, concerns over the ‘twin-demic’ situation where the influenza virus and SARS-CoV-2 co-circulate [21] have led to researchers urgently seeking novel antiviral agents against both viruses. So far, several antiviral compounds have been identified that share common targets. Based on mechanism-of-action studies, these antiviral compounds either target viral proteins such as surface glycoproteins and viral polymerase, or host factors involved in viral entry or the posttranslational processing of viral proteins [1]. Importantly, while extensive studies are still needed, these reports provide a conducive plan for the ‘twin-demic’ situation.

As mentioned in the ‘Drug Discovery Strategies’ section, with the aid of in silico pharmacophore-based drug discovery, target-based drug discovery has the potential to reduce time and cost during the hit identification process. The unique ‘cap-snatching’ mechanism of influenza viruses and other sNSVs offers attractive antiviral targets, such as the viral cap-binding domain and viral endonuclease. Currently, there are several investigative drugs that inhibit influenza virus replication and which have the potential to reduce clinical symptoms. Notably, ZSP1273, an anti-influenza small-molecule drug that targets the PB2 cap-binding domain, shows antiviral activity higher than that of pimodivir (a first-in-class influenza PB2 inhibitor) under similar experimental conditions. Furthermore, the report of this drug demonstrates that the efficacy of the delayed treatment of ZSP1273 was higher than that of oseltamivir all the while having no adverse side-effects [22]. We also note that, as an alternative to the replication step, the viral entry step can also be an attractive antiviral target. It is also of interest that type II transmembrane serine proteases (TTSPs) are currently being investigated in a clinical setting for their role in the proteolytic cleavage of the HA protein, which is critical for the fusion of the viral membrane with the cellular endosomal membrane [23]. So far, three TTSPs (HAT, TMPRSS2, TMPRSS4) were reported to be involved in proteolytic cleavage, and TTSP knockout mice have shown a significant reduction in lung pathology as well as body weight loss and mortality.

Recent advances in three-dimensional (3D) cell culture systems have offered the opportunity for linking the gap between in vitro 2D culture and in vivo animal models. For example, Brown and colleagues were able to demonstrate that some mutations in the NA gene acquired from the MDCK cell culture were selected against in differentiated human airway epithelial (HAE) cells [24], which yielded clinical isolate-like NA mutations. Furthermore, recent drug discovery processes often include resistant-profiling while air–liquid interface cultures may also provide clinically relevant information [25]. Additionally, studies on viral dynamics using mathematical models in combination with pharmacokinetic and pharmacodynamic profiling have enabled us to understand and quantify the efficacy of antiviral drugs. One such example was demonstrated that can be found through Rarra-Rojas and their colleagues who investigated the contribution of NAIs and this analysis can be applied to other antiviral drugs based on their mechanism of action [26].

As an alternate to conventional drug discovery, target protein degradation (TPD) technology is emerging as a therapeutic approach that utilizes the ubiquitin-proteasome system to remove the target protein of interest from cells [27]. The key feature of this approach is either proteolysis targeting chimeras (PROTACs) or molecular glues that hijack E3 ligases to induce ubiquitination and the degradation of the protein of interest. However, the main drawback currently for this TPD technology is the duration of the discovery phase, though this will hopefully be addressed/shortened as the technology continues to evolve.

Last but not least, the concerted efforts to increase the novelty and drug-likeness in small-molecule libraries should continue. There is also a plentiful amount of literature on the anti-influenza activity of several novel natural products derived from medicinal plants or marine sources [28,29], which have potent antiviral activity against influenza viruses through targeting hemagglutinin (HA), PA endonuclease, and other host factors [30].

Declaration of Interest

The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

Reviewer disclosures

Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Additional information

Funding

The authors were supported by the VITAL-Korea project from the Ministry of Health and Welfare (MoHW), of the Korean Government (grant number: HV22C0259).