High-throughput screening for identification of Influenza A inhibitors using a cell-based immunofluorescence assay

Abstract

A highly sensitive cell-based imaging assay has been used to screen a 200,000 compounds library for potential influenza antivirals. Compounds were screened at a concentration of 10 μM against the influenza strain A/Udorn/72 (H3N2) and the duplex capability of the assay was taken advantage of to select compounds with no or low cytotoxicity. The assay was also used for confirmation in concentration-response of the active compounds. In the set of confirmed hits, three major structural clusters, 23 minor clusters and 86 singletons were identified. Further evaluation of anti-influenza activity was performed for two additional influenza strains of the H1N1 subtype, A/WSN/33 and A/California/07/09. Three compounds from cluster A, SRI-44211, SRI-44215 and SRI-44221 showed selectivity indices for the pandemic strain A/California/07/09 in the range of 54.3 to 252, providing evidence that the core structure of cluster A could play a relevant role in their inhibitory potency and may serve as a starting point for future hit-to-lead optimization efforts.

1. Introduction

The availability of a methodology for detection of small molecule inhibitors of influenza virus that is high-throughput compatible is of crucial importance when rapid identification of novel influenza inhibitors is needed, as in the case of influenza outbreaks that can potentially evolve to a pandemic.

Cell-based assays involving multiple viral infections and replication cycles are the preferred methodology to apply to primary screens for antivirals when a program is agnostic to the molecular target. They allow the identification of inhibitors of any stage of the virus life cycle, which could lead to the identification of novel viral targets. In addition, because the inhibition occurs in a cellular environment, cell-based antiviral assays allow the identification of inhibitors of host factors relevant for viral replication that can be inhibited without significantly affecting the cellular function. The latter is ensured by the in parallel evaluation of the compound’s cytotoxicity.

The cytopathic effect has been widely used as readout in cell-based high-throughput assays to screen for influenza inhibitors, mostly restricted to the use of Madin Darby canine kidney cells (1-4). Virus-inducible reporter assays and reporter-encoding influenza virus assays have also been used in high-throughput screening (HTS) campaigns for detection of broad classes of inhibitors against influenza (5-8) and more stable reporter influenza virus that can be use in high-throughput screens have been generated (9-11). The use of virus-inducible reporter cell lines has shown sensitivity to external factors, rendering a high number of false positive in a screen, and the use of a reporter virus, although preferable over the activation of an exogenous reporter and open to the use of other susceptible cell lines, is limited to the specific recombinant virus strain.

We have previously reported a highly sensitive imaging cell-based assay that directly measures influenza virus that has duplex capability (12). A549 cells are infected at a low multiplicity of infection (MOI) and the virus infection is evaluated by staining the cells with an influenza antibody directly conjugated to a fluorophore. After an infection length of 48 hours, the resulting signal:baseline and variability meet assay performance requirements for screening. Once the fluorescence read is finalized, the same assay plate is used to measure the viability to select only compounds producing a decrease in the fluorescence signal due to virus inhibition and not to a decrease in the number of live cells due to compound cytotoxicity (Fig. 1).

Figure 1.

Duplex capability of the influenza immunofluorescence assay. Cells plus virus are added to plates containing test compounds (steps 1 and 2). After 48 hours incubation at 37 °C / 5 % CO₂, detection antibody is added (step 3) and plates are incubated for the appropriate period of time before the fluorescence is read (step 4). The cytotoxicity of compounds is then assessed by adding CellTiter-Glo, which measures cell viability by ATP quantitation (step 5).

We have employed this assay to screen 200,000 compounds from a library that contained 98,034 Murcko scaffolds with molecular weight and predicted solubility (LogP) distribution ranging from 100 to 600 and −2 to 6, respectively. This screen led to the identification of 1,735 compounds with potential anti-influenza activity, from which 161 were confirmed to possess both anti-influenza activity and low or no cytotoxicity. The identified active compounds were classified according to structural classes in 3 major clusters, 23 minor clusters and 86 singletons, considering the compound substructures of the major clusters as potentially relevant for development of orally bioavailable drug candidates for influenza infections therapy.

2. Materials and Methods

2.1. Cell culture

Human Epithelial Lung Carcinoma cells (A549, ATCC Cat # CCL-185) were adapted to and cultured in OptiMEM I Reduced-Serum Medium with 2 % FBS (12).

2.2. Compound library, control and reference compounds

A library of 200,000 diverse compounds was purchased from ChemDiv. Compounds were screened at a single concentration of 10 μM. Pimodivir / VX-787 (MedChemExpress Cat # HY-12353A) was used as positive control at a concentration of 0.1 μM, where 100 % inhibition was achieved, and in a range of 0.39-200 nM when used in concentration-response.

After confirmation, experimental compounds SRI-00044211 (Cat # L063-0493), SRI-00044215 (Cat # F794-0866), SRI-00044218 (Cat # E859-0367) and SRI-00044221 (Cat # L066-0517) were purchased from ChemDiv.

Baloxavir marboxil (Cat # HY-109025), oseltamivir (Cat # HY-13318) and ribavirin (Cat # HY-B0434) were all acquired from MedChemExpress.

2.3. Compound preparation

Compounds were prepared as previously described (12). For the screen, to 384-well, collagen coated plates (J. G. Finneran Cat # 312003), 5 μL of media (columns 1, 2 and 23 and 24 from A to L) or media containing test compounds diluted to achieve a 10 μM final assay concentration (columns 3-22) were added. Rows from M to P of columns 23 and 24 were used for the positive control compound. Since the test compound stock solutions and the control were in 100 % DMSO, the final assay concentration of DMSO in each well was maintained at 0.4 % or lower. The 32 wells in columns 1 and 2 (cells only) and wells in columns 23 and 24 from row A to L (cells plus virus, no compound) were used for low and high signal controls, respectively.

For the antiviral and cytotoxicity concentration-response confirmation of the compound selection, a stacked plate concentration-response design in which the 10 individual concentrations that make up the curve are contained in different plates, was used (13). A volume of 5 μL of the positive compounds was cherry picked from the source plates of the compound library to create plates containing up to 320 compounds. A small volume of each compound from each newly created source plate was then transferred to a plate containing assay media from which compounds were serially diluted 1:2 into nine more plates. Each of the ten resulting dilution plates contained the same compounds at different concentrations and equal DMSO percentage, and from them 5 μL were transferred to the assay plates.

2.4. Virus strains

A/Udorn/72 (H3N2) and A/Wisconsin/33 (H1N1) were provided by Dr Robert Krug’s laboratory at the University of Texas in Austin. A/California/07/2009 (H1N1) pdm09 was obtained from ATCC (Cat # VR-1894).

2.5. High Throughput Screen

The immunofluorescence assay used to screen for inhibitors of influenza A virus has been previously described (12). Briefly, A549 cells were harvested, suspended in OptiMEM 1 % PS / 0.5 % BSA / 1 μg/mL TPCK-Trypsin, infected with Influenza A strain A/Udorn/72 (H3N2) at an MOI of approximately 0.001 and seeded in the 384-well assay plates at 5000 cells/well in 25 μL. Plates were incubated for 48 h at 37 °C / 5 % CO₂, after which the supernatant was removed and 20 μL of influenza A M2 antibody conjugated to Alexa Fluor 647 (Santa Cruz Biotechnology Cat # sc-32238) at a 1:1000 dilution in OptiMEM / 1 % PS / 2 % BSA were added to each well. The plates were returned to the incubator and the total fluorescence read the next day using a laser scanning cytometer (Mirrorball, SPTLabtech).

2.6. Cytotoxicity assessment for compound selection

To discard cytotoxic compounds, after concluding the fluorescence read, 20 μL of CellTiter-Glo (Promega Cat # G7573) were added to the assay plates to assess the cell viability. The percentage of cell viability in each test well was calculated using the formula:

$$\%\text{Viability}=100\frac{\text{test compound signal}}{\text{Median of cell control signal}}$$

2.7. Confirmatory assays

For confirmation of the ability of compounds to inhibit influenza virus and for determination of the compound concentration that causes a 50 % virus inhibition (inhibitory concentration 50; IC₅₀), positive compounds were assessed in the antiviral assay in a 10-point concentration-response at a concentration range of 0.078 μM-40 μM. Each compound diluted to a concentration 6 times the final assay concentration was added to columns 3-12 and 13-22 of the 384 well assay plate in 5 μL. The inhibitory capability was measured against the screening strain A/Udorn/72 and the H1N1 strain A/WSN/33. MOI and detection antibody used for the WSN strain were both the same as for Udorn.

Selected compounds with confirmed activity in both A/Udorn/72 and A/WSN/33 were tested for their ability to inhibit the pdm09 strain A/California/07/09. For this assay, the MOI was 0.05 and the detection was performed by adding 20 μL of influenza A H1N1 HA mAb (Novus Biologicals Cat # NBP3-06578) at a 1:1500 dilution in OptiMEM / 1 % PS / 2 % BSA.

Compounds were also analyzed in concentration-response in a cytotoxicity assay as previously described (12). Cytotoxicity assay plates were treated the same as antiviral plates and cells only were added to all wells. Wells in columns 1 and 2 received no treatment (high signal control) and wells in columns 23 and 24 were treated with 100 μM hyamine hydroxide (low signal control). After 72 h incubation at 37 °C / 5 % CO₂, 30 μL of CellTiter-Glo were added and the luminescence signal read using a PHERAstar FSX microplate reader. From these data, the concentration of compounds that is toxic to 50 % of the cells (cytotoxic concentration 50; CC₅₀) was established.

2.8. Data processing

Data was analyzed using IDBS ActivityBase and ActivityBase XE (both version 9.7) run on Windows OS, and percentage of inhibition and percentage of viability were determined for assayed compound as previously described (12), as well as $\mathrm{Z}’$, percentage of variation ($\%\text{CV}$), signal-to-background ($\mathrm{S}∕\mathrm{B}$), and signal-to-noise ($\mathrm{S}∕\mathrm{N}$) for each assayed plate (14).

$\mathrm{Z}’$, $\%\text{CV}$, $\mathrm{S}∕\mathrm{B}$ and $\mathrm{S}∕\mathrm{N}$ calculations were made as follows:

$${\mathrm{Z}}^{\prime }=1-\frac{3(\text{StdDevHighControl}+\text{StdDevLowControl})}{\mid \text{MeanHighControl}-\text{MeanLowControl}\mid }$$

$$\%\text{CV}=100\frac{\text{StdDevControl}}{\text{MeanControl}}$$

$$\mathrm{S}∕\mathrm{B}=\frac{\text{MeanHighControl}}{\text{MeanLowControl}}$$

$$\mathrm{S}∕\mathrm{N}=\frac{(\text{MeanHighControl}-\text{MeanLowControl})}{\sqrt{({\text{StdDevHighControl}}^2+{\text{StdDevLowControl}}^2)}}$$

2.9. Cheminformatics analysis

Cheminformatics analysis was performed using BIOVIA Pipeline Pilot (version 7.5, BIOVIA, Dassault Systèmes). The substructure search component in BIOVIA Pipeline Pilot was used to filter the pan-assay interference structures (PAINS). A list of PAINS structures was assembled from the literature (15-18). The clustering component in BIOVIA Pipeline Pilot was used for the structure cluster analysis of the HTS hits. Tanimoto distance of 0.4 was used to distinguish structure clusters. Minor cluster refers to cluster containing less than 4 compounds. Major cluster contains ≥ 4 compounds while singleton refers to 1 compound. Murcko scaffolds and other molecular properties of the compound library were also analyzed using BIOVIA Pipeline Pilot.

3. Results and Discussion

To identify novel compounds that inhibit influenza virus, a high-throughput cell-based assay that directly measures influenza virus by staining the cells with an anti-influenza antibody directly conjugated to a fluorophore was employed (12). The duplex feature of the assay was exploited to evaluate the cell viability in each test well in comparison to the cell control, which enabled the exclusion of a significant number of cytotoxic compounds. The reduction in the number of live infected cells that results from the presence of a cytotoxic compound produces a reduction in the intensity of the fluorescence signal, which is expected from a compound with antiviral activity. In the absence of a viability assessment, a high number of cytotoxic compounds would be considered positives, causing a decrease in the confirmation rate.

We screened a library of 200,000 compounds in singlet at a concentration of 10 μM, obtaining 191,384 validated results of which 175,508 were validated results of inhibition and viability (Fig. 2). Validated results were defined as data coming from assay plates that complied with the assay quality criteria of $\mathrm{Z}’$ values higher than 0.5 and a 100 % inhibition of the positive control VX-787. The presence of sections of low fluorescence signal in a pattern across all plates in two of the screen’s batches resulted in the rejection of 8,616 results during review, reducing the number of available data points for selection of positive compounds to 191,384.

Figure 2.

A library of 200,000 compounds was screened using the influenza immunofluorescence assay, from which 191,384 valid antiviral results were obtained. From them, 1,735 met the cut-off criteria of 54 % inhibition or higher, and viability data was available for 1,545. The top 640 compounds based on percentage of inhibition were tested in the concentration-response confirmation antiviral and cytotoxicity assays. The selected set of 640 included a subset of 184 compounds with no cytotoxicity data. One-hundred and sixty-one compounds with low or no cytotoxicity were confirmed as active against the screen strain.

The viability data of 15,876 compounds (one batch of 50 plates) was not measured due to technical difficulties with the cytometer. The batch of plates was kept at 4 °C sealed and protected from light until the issue was resolved. When stored under these conditions, the fluorescence signal of the antiviral assay remains stable for at least 7 days, which allowed collection of the antiviral data but compromised the viability data as the cells do not remain viable under this condition. Therefore, for this batch of plates, only the antiviral data was evaluated.

Of the 191,384 validated results, 1,735 compounds met the statistical cutoff criteria of 54 % for virus inhibition, which was defined as the mean of all compound inhibition results (7.26) plus 3 times the standard deviation (15.58). The number of compounds inside this set for which the viability data was available and that, at the same time, complied with the viability criteria for selection of 50 % or higher was 1,545. The top 640 compounds based on percentage of virus inhibition, which included 184 compounds with no cytotoxicity data, were tested in the concentration-response antiviral and cytotoxicity confirmatory assays. Six-hundred and forty corresponds to 2 sets of 10 assay plates in which the activity of 320 compounds per set of plates can be evaluated in a concentration-response manner. It is a common practice in our laboratory to initially select 640 compounds for confirmation for a screen of this size (200,000). In the case that the number of positives obtained is less than 640, this approach allows the confirmation of less potent compounds. When the number of positives is greater than 640, an additional round of confirmation may be performed, if deemed necessary. Consultation with our medicinal chemists for these selections may be carried out to ensure the prioritization of compounds without known liabilities.

A total of 161 compounds were confirmed in independent concentration-response antiviral and cytotoxicity experiments under the criteria of 50 % inhibition for at least one concentration of the concentration range tested of 0.078 μM-40 μM and CC₅₀ values higher than 40 μM. Table 1 shows the analysis of the concentration-response data in two separate sets of compounds selected from the primary screen, depending on the availability of the viability data. While 30.7 % of compounds were confirmed from the set with viability data, only 11.4 % were confirmed in the set with no viability data, which contain a higher rate of compounds with CC₅₀ values lower than 40 μM.

Table 1.

Analysis of the confirmation results when single concentration data for inhibition and viability were available (A) and when no viability data was obtained (B).

A	N	Rate (%)
Total	456
IC50 < 40 μM; CC50 > 40 μM	140	30.7
IC50 < 40 μM; CC50 < 40 μM	23	5
IC50 > 40 μM; CC50 < 40 μM	7	1.5
IC50 > 40 μM; CC50 > 40 μM	286	62.7
B	N	Rate (%)
Total	184
IC50 < 40 μM; CC50 > 40 μM	21	11.4
IC50 < 40 μM; CC50 < 40 μM	48	26.1
IC50 > 40 μM; CC50 < 40 μM	20	10.9
IC50 > 40 μM; CC50 > 40 μM	95	51.6

The high-throughput suitability of the influenza immunofluorescence assay was previously suggested (12) and is demonstrated here, with statistical values for assay performance during the screen that show its robustness (Table 2, Fig. 3). The compound selection can be narrowed if a more stringent criteria of viability is considered. Table 3 shows how the positive compounds rate of the single concentration screen decreases with the increase of the viability criteria for selection, which results in an increase of the confirmation rate. Criteria for positive compounds selection must be carefully established in a way that relevant scaffolds are not missed, and, at the same time, most inactive compounds are discarded. We chose a more conservative criteria of inhibition of virus infection of 54 % or greater with a corresponding percentage of viability of 50 % or higher that, although resulting in more compounds being confirmed as cytotoxic, increased the chance of identifying those with lower potency values.

Table 2.

Summary of statistical HTS parameters. $\mathrm{Z}’$, variation coefficients, $\mathrm{S}∕\mathrm{B}$ and $\mathrm{S}∕\mathrm{N}$ were calculated for each assay plate and averaged.

Control Summary
average $\mathbf{Z}’$	0.66
average cell control CV	122
average virus control CV	6.53
average positive control CV	84.4
average $\mathbf{S}∕\mathbf{B}$	73.2
average $\mathbf{S}∕\mathbf{N}$	13

Figure 3.

Control graphs of the screen of 200,000 compounds using the influenza immunofluorescence assay. Graphs show relative fluorescence average values of control wells for each plate. Square points represent wells containing A549 cells only (low signal control, Graph A) or wells containing cells infected with influenza virus in presence of 100 nM of VX-787 (positive control, Graph B). Triangle points in A and B represent wells containing cells infected with influenza virus in absence of compound (high signal control).

Table 3.

Number of compounds meeting screen criteria for selection, screen hit rate and hit confirmation rate as the viability criteria cut off increases. Determinations were made considering the set of 1,545 screened compounds for which single concentration inhibition and viability screening data were available.

Criteria	Results
Inhibition ≥ 54 %; Viability ≥ 50 %	Number of compounds	308
	Positive compounds rate	0.18 %
	Confirmation rate	37.7 %
Inhibition ≥ 54 %; Viability ≥ 60 %	Number of compounds	216
	Positive compounds rate	0.12 %
	Confirmation rate	42.6 %
Inhibition ≥ 54 %; Viability ≥ 80 %	Number of compounds	123
	Positive compounds rate	0.07 %
	Confirmation rate	49.6 %

After discarding pan-assay interference structures or PAINS, 3 major clusters of compounds with anti-influenza inhibitory activity were identified within the confirmed compounds and grouped as clusters A (8 compounds), B (7 compounds) and C (4 compounds) (Fig. 4). The residual 142 compounds contained 23 small clusters of two or three compounds and 86 singletons. Each major cluster of active compounds came from clusters present in the same relative proportion in the compound collection. More specifically, cluster A is represented with 61 compounds in the collection, cluster B with 32 compounds and cluster C with 13 compounds. This proportionate identification of positive compounds from these clusters provides further confidence in the approach used in this HTS campaign.

Figure 4.

Core structure of the 3 major clusters of compounds identified within the 161 confirmed hits.

Based on the data from the confirmatory assays and commercial availability, twenty-six compounds were purchased for further analysis by rerunning the antiviral assay using the screening strain A/Udorn/72 and the cytotoxicity assay, but also by profiling their antiviral activity against an H1N1 strain, for which A/WSN/33 was used. This selection contained 7 compounds from cluster A, 1 from cluster B, 3 from cluster C, 6 compounds from minor clusters and 9 singletons. The data revealed cytotoxicity not detected during confirmation using the cherry-picked library compounds and non-toxic compounds active against A/Udorn/72 but inactive against A/WSN/33. From these results and with the purpose of publishing the screen data, the attention was focused on compounds from cluster A SRI-44211, SRI-44215 and SRI-44221 and singleton SRI-44218. Fresh DMSO stocks were prepared from powder obtained from Southern Research repository to run another confirmatory test of the compounds against the screen strain A/Udorn/72, A/WSN/33 and the pdm09 strain A/California/07/09. Reference compounds baloxavir marboxil (19-21), oseltamivir (22) and ribavirin (23) were added to this test using high concentrations of 0.5, 1 and 100 μM, respectively. VX-787 (24, 25) concentration range and single inhibitory concentration added to rows M-P of columns 23-24 of the 384 wells assay plate were adjusted to 0.078-40 nM and 20 nM, respectively. All experimental compounds showed to be slightly active against A/WSN/33 and A/Udorn/72 but highly potent against the pdm09 strain, with IC₅₀ values lower than 1 μM and selectivity indices (SI=CC/IC) ranging from 54.3 to 252 (Table 4 & 5, Fig 5 & 6) despite the notable shift in the CC₅₀ values of the compounds in comparison with the confirmation data and retest after purchasing. This may be explained by the occurrence of structural changes due to chemical degradation over the storage time in between the retest of the confirmed hits and the test to obtain the results showed here. Still, with the probable degradation introducing the cytotoxic effect observed, the compounds preserved their inhibitory activity. Results suggest a relevant role of the core structure of cluster A in the inhibitory properties of compounds SRI-44211, SRI-44215 and SRI-44221. The structural scaffold of singleton SRI-44218 may be of interest as well (SI for A/Ca/07/09 = 64.5), due to the concentration-response results and the fact that the single concentration screen detected two structurally related compounds in the library with percentages of inhibition of 31 and 25. These findings could represent a starting point for chemical optimization.

Table 4.

Structure and inhibitory activity of confirmed compounds from cluster A and singleton SRI-44218.

Compound ID	Structure	A/WSN/33		A/Udorn/72		A/Ca/07/09		CC50 μM
		IC50 μM	SI	IC50 μM	SI	IC50 μM	SI
SRI-44211 (Cluster A)		3.61	3.98	5.62	2.56	0.26	54.3	14.4
SRI-44215 (Cluster A)		1.94	17	2.83	11.6	0.13	252	33
SRI-44221 (Cluster A)		1.82	6.76	1.36	9.05	0.11	109	12.3
SRI-44218 (Singleton)		5.85	5.29	7.37	4.20	0.48	64.5	31

Table 5.

Inhibitory concentration 50 (IC₅₀) expressed in nM of reference compounds (A) included in retesting experiments, (B) interval obtained through multiple determinations.

A	A/WSN/33	A/Udorn/72	A/Ca/07/09
Baloxavir marboxil	19	44	2.2
VX-787	7.9	12	2.3
Oseltamivir	133	49	122
Ribavirin	39980	41710	20880
B	A/WSN/33	A/Udorn/72	A/Ca/07/09
Baloxavir marboxil	16.4 - 57.5	3.6 - 45	1.6 - 3
VX-787	1 - 9.2	1 – 12.5	2.3 - 5
Oseltamivir	12 - 320	11 – 170	59 - 129
Ribavirin	25960 - 49540	20280 - 44300	9940 - 20880

Note: Inhibitory concentration 50 (IC₅₀) interval of VX-787, ribavirin and oseltamivir have been obtained through multiple determinations for A/WSN/33 and A/Udorn/72. Baloxavir marboxil IC₅₀ range for A/WSN/33 and A/Udorn/72 comes from 5 independent determinations. A/Ca/07/09 data for VX-787, baloxavir marboxil, oseltamivir and ribavirin come from 3 independent determinations.

Figure 5.

Antiviral concentration-response curves of compounds from cluster A (SRI-44211, SRI-44215 and SRI-44221) and singleton SRI-44218 against influenza strains A/WSN/33 (green circles), A/Udorn/72 (magenta squares) and A/Ca/07/09 (blue rhombus). Cytotoxicity response is also included (red inverted triangles). Each point represents the average of duplicate determinations, and the error bars represent the standard deviation. IC₅₀, CC₅₀ and SI values of each experimental compound are shown in Table 4.

Figure 6.

Antiviral concentration-response curves of reference compounds baloxavir marboxil, VX-787, oseltamivir and ribavirin tested alongside the experimental compounds of interest: A/WSN/33 (green circles), A/Udorn/72 (magenta squares) and A/Ca/07/09 (blue rhombus). Cytotoxicity response is also included (red inverted triangles). Each point represents the average of duplicate determinations, and the error bars represent the standard deviation. IC₅₀ values are shown in Table 5.

Interestingly, the response of the influenza strains tested in this work against these four experimental compounds showed a pattern similar to baloxavir marboxil, the prodrug of baloxavir acid, with higher IC₅₀ values for A/WSN/33 and A/Udorn/72 compared to the clinical isolate of the pandemic strain A/Ca/07/09. The susceptibility of influenza viruses to different antivirals has been shown to vary depending on the viral strain or subtype (19, 25-27). Therefore, although the differences we observe are notable, they are not unusual. Future studies to support these molecules’ potential as candidates for chemical optimization are to determine the susceptibility of other more recently isolated clinical and zoonotic strains to these compounds, as well as conducting other in vitro studies to probe potential mechanism of action.

The cell-based immunofluorescence assay has proven to be a powerful high-throughput research tool to identify compounds with anti-influenza activity with the potential for therapeutic development. The assay combines high sensitivity, high-throughput and duplex compatibility, the latter allowing the evaluation of the effect of compounds on virus levels and cell viability in the same well, for a highly efficient screening process. Any influenza strain can be used as screening strain when the antibody for detection is available, making the assay a potent drug discovery tool to respond to new emerging influenza strains presenting a high risk to the public. The assay has been employed here to perform a primary screen of 200,000 compounds as well as confirmation in multiple influenza strains, resulting in the identification of a structural moiety potentially relevant for the development of compounds with improved potency and selectivity.

Article highlights.

• The influenza immunofluorescence assay has high-throughput capability.

• The duplex capability of the screening assay enriches the identified active compounds and increases the efficiency of the screen.

• The immunofluorescence assay is compatible with any influenza strain, allowing the evaluation of active compounds for pan-influenza activity.

• A structural moiety potentially relevant for the development of compounds with improved potency and selectivity was identified.