From target to drug

Traditionally, antiviral drug discovery has started from detailed knowledge of a specific viral target. Proteins that are critical for a virus’ life cycle (e.g., polymerase, protease, and envelope) are first characterised, ideally by crystallography, and the compounds that interfere with that specific target are then identified. When the viral target is known, the first step is to develop an assay to reveal whether a given compound can block target activity. Depending on the information available on the target itself, different assays can be developed, and ideally, more than one assay should be available to independently verify the effectiveness of any hit. For instance, when the target can be recombinantly expressed and purified, crystallography or NMR can identify compounds that bind to the target. Then, additional assays (e.g., thermal shifts or enzymatic assays) should be run to confirm that interaction, and indeed the compound inhibitory activity. For viral enzymes, this can be fairly straightforward, as different readouts can be used to test for enzymatic activity. However, in such cases, it is also important to validate the results using at least one other approach, which will confirm that the inhibition is specific and not targeted to other assay components. It is not uncommon, for instance, that a given compound might simply inhibit the luciferase enzyme that was used for detection!

From drug to target

A more recently developed approach does not begin with a specific target, but instead with the effect that a compound might have on the ability of a virus to replicate. This approach is particularly advantageous when the information on the virus of interest is limited (as it is often the case for newly emerging viruses), or when the most classical options have been explored with limited success. This approach only requires a phenotypic readout (i.e., an assay able to detect viral replication and/or spread in cell culture), and has the advantage of immediately selecting for compounds that can penetrate the cell membrane, are active antivirals in a cellular environment, and have limited toxicity in the tissue culture settings; all parameters that need to be determined down the line when starting from a more targeted in vitro approach. The downside is that the exact target of any hit compound is unknown at this stage, and not always easy to pin down.

Importantly, hits identified in a phenotypic screening can target viral proteins or cellular proteins that are important for the virus life cycle. The main advantage of hitting a viral target is that the drug will act selectively on the virus and limit side effects (supposing that it does not also interfere with cellular proteins); the main disadvantage is that viral targets are more prone to developing resistance. The exact opposite is true for cellular targets, which also have higher potential for broad-spectrum activity when the same cell target is used by multiple viruses. One of our first ever blog articles addresses these two approaches in more detail.

Between targeted and untargeted approaches sit the so-called nucleoside analogues. These are nucleosides that mimic one of the four bases that make up the genome but with a twist: they generally stop replication by preventing the attachment of the next base or, sometimes more specifically, by creating steric hindrance in the viral polymerase. While highly effective and in some cases having broad-spectrum activity against viruses using similar replication mechanisms, the likelihood of resistance arising can be high.

High-throughput screening: some tips

Whether the target is known or unknown, viral or cellular, the development of an assay that can identify potent inhibitors is critical. The more compounds that are tested, the higher the likelihood of success, and this is why any assay should be designed for high-throughput readouts. High-throughput formats range from 96- to 1536-wells per plate. The smaller the format, the faster the screening, but the robustness of the format should always be assessed early on using the appropriate controls, especially for cell-based assays. The variation of positive and negative signals can be conveniently assessed using the Z’-score, which should always be >0.5. The smaller formats are also more sensitive to evaporation and edge effects, and as such are generally recommended for shorter assays.

Two to three replicates (depending on the robustness of the assay) should always be included, even if further validation steps will follow. Testing of control compounds, if available, should also be included to confirm that the inhibition observed is in the expected range. It is common to only test one or two concentrations of each compound (depending on the number of compounds tested); in vitro assays and fragment screening are generally run at higher concentrations than phenotypic assays.

As large screenings are likely to be run over several days or weeks, small variations can be expected, and because of this, the inclusion of negative (uninfected) and positive (infected without compounds) controls is critical for normalisation. If available, a known inhibitor should also be included in each plate.

Depending on the number of hits obtained and that it is reasonable to carry forward, different activity thresholds can be defined to prioritise the most potent hits, at least in the first instance.

When developing an in vitro assay, it is important to keep it as simple as possible, thereby minimising the risk that a compound interferes with several assay components, resulting in a high number of false positives. When this is not possible (for instance, when multiple steps are required to increased assay sensitivity), it is even more important that further assays are deployed to eliminate false positives as early as possible. For example, if an assay uses luciferase, a counter-assay could test the hit compound against the luciferase only. However, an independent assay that confirms the activity of the compound on the viral target would still be necessary.

Validation and cytotoxicity

Before moving on to further studies, including mode-of-action studies or alternative validations, if the initial screening was performed using an enzymatic or phenotypic assay, it is recommended to confirm the effectiveness of the hits identified by re-testing them in a dose-response manner, to obtain additional information on their potency. For cell-based assays, a cytotoxicity test should be run in parallel to confirm that the antiviral activity occurs at a lower concentration than any cytotoxic effect. A measure of this separation is the Selectivity Index (SI, TC50/IC50): the higher the number, the safer the compound. This is only provisional information on toxicity, as further tests will be needed to assess the toxicity and chemical properties of each compound: this is one of the most critical steps in drug discovery.

Mode-of-action studies

When the target is unknown, the first step in identifying the mode-of-action of hit compounds is to narrow down which stage of the virus life cycle is inhibited. This can be done using time-of-addition studies and a combination of the assays that we have described in our previous blogs (MAY, JUNE, JULY). When this aspect becomes clearer, more specific and targeted assays can be designed using a variety of approaches, including affinity-based techniques and expression cloning (see our blog).

While not always strictly necessary, it can be helpful to know the target of an antiviral compound, which might make it easier to improve potency or pharmacokinetics control, estimate the risk of toxicity, and rationally introduce modifications.

This aspect, as well as how to transition promising candidates to pre-clinical studies, will be covered in the next blog.