Combined siRNA and Small-Molecule Phenotypic Screening Identifies Targets Regulating Rhinovirus Replication in Primary Human Bronchial Epithelial Cells

Abstract

Human rhinovirus (RV) is the most common cause of acute upper respiratory tract infections and has recently been shown to play a significant role in exacerbations of asthma and chronic obstructive pulmonary disease (COPD). There is a significant unmet medical need for agents for the prevention and/or treatment of exacerbations triggered by human RV infection. Phenotypic drug discovery programs using different perturbation modalities, for example, siRNA, small-molecule compounds, and CRISPR, hold significant value for identifying novel drug targets. We have previously reported the identification of lanosterol synthase as a novel regulator of RV2 replication through a phenotypic screen of a library of siRNAs against druggable genes in normal human bronchial epithelial (NHBE) cells. Here, we describe a follow-up phenotypic screen of small-molecule compounds that are annotated to be pharmacological regulators of target genes that were identified to significantly affect RV2 replication in the siRNA primary screen of 10,500 druggable genes. Two hundred seventy small-molecule compounds selected for interacting with 122 target gene hits were screened in the primary RV2 assay in NHBE cells by quantifying viral replication via in situ hybridization followed by secondary quantitative PCR-based assays for RV2, RV14, and RV16. The described follow-up phenotypic screening allowed us to identify Fms-related tyrosine kinase 4 (FLT4) as a novel target regulating RV replication. We demonstrate that a combination of siRNA and small-molecule compound screening models is a useful phenotypic drug discovery approach for the identification of novel drug targets.

Keywords

phenotypic screen rhinovirus COPD NHBE FLT4

Introduction

Human rhinovirus (RV) is responsible for more than 50% of cold-like illnesses and is the most common cause of acute upper respiratory tract infection.^1,2 Recent studies have demonstrated that RV is also a lower respiratory pathogen and plays a significant role in asthma exacerbations and chronic obstructive pulmonary disease (COPD) exacerbations.^3–5 Approximately two-thirds of respiratory virus infections associated with asthma or COPD exacerbations are due to RV,^4,5 and exacerbations have been viewed as one major cause of morbidity and mortality in asthma and COPD.⁶ Despite the recognized importance of RV infection in asthma and COPD exacerbations, the mechanism by which RV causes asthma or COPD exacerbations is not well understood, and there are currently no approved anti-RV agents. Therefore, identifying novel targets regulating RV replication and infection could offer new approaches for the treatment and/or prevention of asthma and COPD exacerbations.

RV is a single positive-strand RNA virus belonging to the family of Picornaviridae. There are more than 160 different RV serotypes grouped into three species (RV-A, RV-B, or RV-C) based on phylogenetic sequence identity. RV-A and RV-B can also be classified into a major and a minor group based on their binding to intercellular adhesion molecule (ICAM)-1 or to members of the low-density lipoprotein (LDL) receptor family, respectively.² RV-C was recently reported to bind to cadherin-related family member (CDHR)-3 as a potential entry receptor.⁷ The large number of RV serotypes makes it challenging to develop anti-RV agents with broad-spectrum activity against the many RV serotypes.

Phenotypic screening approaches using different perturbation modalities, for example, small-molecule compounds, siRNA, and CRISPR, have been shown to hold significant value for identifying novel targets for drug discovery.^8–11 Each perturbation modality has its own advantages and drawbacks. For siRNA screening, both pooled and arrayed siRNA libraries have been used since 2000 for functional genomics screens to elucidate novel gene functions and pathways.^12–14 We have recently reported the identification of lanosterol synthase (LSS) as a novel regulator of RV2 replication through a phenotypic screening of a library of arrayed siRNAs against 10,500 druggable genes in normal human bronchial epithelial (NHBE) cells.¹⁵ One major challenge in the siRNA screening field is that this approach is abundant in false positives mediated mainly by off-target effects,^9,16–19 necessitating thorough validation of first-pass hits through follow-up experiments. In order to investigate for the possibility to identify more potential targets regulating RV2 replication from the siRNA screen, we performed a follow-up phenotypic screen using small-molecule compounds that are annotated to be pharmacological regulators of the target genes that were identified to significantly affect RV2 replication in the previously described siRNA screen.¹⁵

Here, we describe the combined siRNA and small-molecule screening approach, including the compound selection strategy. We screened 270 small-molecule compounds selected for interacting with 122 target gene hits using the same high-content imaging RV2 assay applied in the previous siRNA screen, followed by the secondary quantitative PCR (qPCR) assays for RV2, RV14, and RV16. This follow-up screen allowed us to identify an additional novel target, Fms-related tyrosine kinase 4 (FLT4), that inhibits RV replication in human primary NHBE cells. We demonstrate that a combination of siRNA and small-molecule compound screening is a useful phenotypic drug discovery approach for the identification of novel drug targets.

Materials and Methods

Preparation of NHBE Cells

Cryopreserved NHBE cells were from Lonza (CC2540, Lot 7F4363; Lonza, Basel, Switzerland). Before screening, NHBE cells were thawed and expanded in complete bronchial epithelial growth medium (BEGM) (CC3171, CC4175; Lonza) according to Lonza’s instructions, and frozen at passage 4 with 0.45 million cells/vial and stored at −150 °C until use. Cryopreserved expanded NHBE cells were thawed, seeded in 175 cm² cell culture flasks, and precultured for 7 days in complete BEGM. Cells were detached with TrypLe Express (12604-013; ThermoFisher, Waltham, MA) and used in the screen as described in the following sections.

RV2, RV14, and RV16 Viral Stocks

RV serotypes 2, 14, and 16 were obtained from ATCC (Manassas, VA; VR-482, VR-284, VR-283). Bulk virus stocks were prepared by inoculation of HeLa OHIO cells in cell culture flasks and incubating at 37 °C/5% CO₂ until 100% cell death could be visually observed. The flask was then transferred to a −80 °C freezer overnight. The contents of the flask were then thawed and centrifuged at 600g for 10 min. The supernatant was then aliquoted and stored at −80 °C. To determine stock concentration, RV stocks were titrated on HeLa OHIO cells to determine the TCID₅₀/mL using the Spearman Karber method.^20,21 The concentrations of RV stocks are RV2 stock (4 × 10⁸ plaque-forming units [PFU]/mL), RV14 stock (1.4 × 10⁵ PFU/mL), and RV16 (3.3 × 10⁷ PFU/mL).

siRNA Library Screening Using In Situ Hybridization

The siRNA library was the SilencerSelect Druggable Genome library (10,415 genes; Ambion, ThermoFisher, Waltham, MA), plus siRNA targeting 82 additional genes (Ambion). siRNA screening was performed as described previously.¹⁵ Briefly, NHBE cells were seeded at 900 cells/well in 384-well plates (356663; BD Bioscience, San Jose, CA) and incubated for 24 h at 37 °C/5% CO₂. siRNA transfections were then performed using 30 nM siRNA with 0.03 µL/well RNAiMax (13778; Invitrogen/ThermoFisher). After 72 h of siRNA transfection, cells were infected with 63,000 PFU/well of RV2 (reaching a multiplicity of infection [MOI] of 70 with respect to seeded cells) for 24 h at 33 °C/5% CO₂. Cells were then fixed and stained with MitoTracker Deep Red (M22426; ThermoFisher), Hoechst 33342 (H3570; ThermoFisher), and a ViewRNA RV2-specific probe (VF1-10249-03; Panomics, ThermoFisher, Waltham, MA). The plates were sealed and imaged on an ImageXpress Micro automated microscope system (Molecular Devices, Berkshire, UK) using 20× objectives. Image analysis was performed using automated eCognition (Definiens) software, using algorithms for segmentation and classification of nuclei customized for NHBE cells.

Small-Molecule Compound Follow-Up Screening Using In Situ Hybridization

NHBE cells were seeded at a density of 5000 cells/well in 30 µL of complete BEGM in 384-well plates (356663; BD Bioscience, San Jose, CA) and incubated at 37 °C/5% CO₂. After 24 h of incubation, DMSO-negative control, positive control ZD9720 (a LSS inhibitor),²² and compounds dissolved in DMSO were prediluted in growth medium and then dosed in 5 µL/well to the appropriate wells using CyBio pipetting and incubated at 37 °C/5% CO₂. The final DMSO concentration was 0.1% and the compounds were screened at eight concentrations ranging from 0.01 to 30 µM in triplicate. Twenty-four hours post-compound addition, cells were infected with 150,000 PFU/well of RV2 (MOI of 30). Wells for noninfected control received 5 µL of growth medium. Cells were infected at 33 °C/5% CO₂ for 30 h until fixation. Cells were then stained, imaged, and analyzed as described under “siRNA Library Screening Using In Situ Hybridization.”

Small-Molecule Compound Secondary Screening Using Quantitative RT-PCR

NHBE cells were seeded at a density of 2500 cells/well in 30 µL of complete BEGM in 384-well plates (781098; Greiner, Kremsmünster, Austria) and incubated at 37 °C/5% CO₂. After 24 h of incubation, DMSO-negative control, positive control pirodavir (an RV VP1 inhibitor),²³ and compounds dissolved in DMSO were directly dosed in 90 nL/well using an Echo liquid handler (Labcyte, Sunnyvale, CA). The final DMSO concentration was 0.3% and the compounds were screened at 10 concentrations ranging from 0.002 to 30 µM in duplicate. Twenty-four hours post-compound addition, cells were infected with 8745 PFU/well of RV2 (MOI of 3.5), 1.0 PFU/well of RV14 (MOI of 0.0004), or 15,000 PFU/well of RV16 (MOI of 6) for 24–26 h at 33 °C/5% CO₂. Cells were then rinsed with cold DPBS (14190250; ThermoFisher), lysed with cell lysis buffer (5943523001; Roche Diagnostics, Basel, Switzerland), sealed, and stored at 4 °C for a maximum of 2 days, and then subjected to quantitative real-time reverse transcription polymerase chain reaction (RT-PCR). RT-PCR was conducted with One-Step RT-PCR assay kits (4991885001/4692136001/05189268001; Roche Diagnostics) to assess the RNA level of RV2 (forward primer 5′-3′ AATGTTGGTTACAACTACACACACC, reverse primer 5′-3′ GGATTCAATCTCGTCTCAGCTT; Sigma), RV14 (forward primer 5′-3′ CCTGAATGCGGCTAACCTT, reverse primer 5′-3′ GACCTTCAACCACTGGATCG; Sigma), and RV16 (forward primer 5′-3′ AGCCCAATGTGTGCTGAAT, reverse primer 5′-3′ ACTGGTAGACCTTGCACAATG; Sigma). Peptidylprolyl isomerase A (PPIA) (human PPIA gene assay, 05189268001; Roche Diagnostics) was used as an endogenous control. For performing RT-PCR analysis, 10× concentrated RT-PCR mix containing LightCycler 480 RNA Master Hydrolysis probes, RV2, RV14, or RV16 specific primers, and UPL probes (probe 96 for RV2, probe 1 for RV14, probe 50 for RV16; Roche Diagnostics) was prepared on ice according to the manufacturer’s instructions (4991885001; Roche Diagnostics). Concentrated RT-PCR mix and cell lysate were added to qPCR plates (4729749001; Roche Diagnostics) with 9 and 1 µL/well, respectively. Fifty cycles of reaction were performed on a LightCycler480 instrument (Roche Diagnostics). Only samples reaching threshold values before cycle 36 were included. The relative expression levels of specific genes were calculated using delta delta Cp. The compound inhibition effect for the tested compounds was calculated using GeneData Screener software using the following calculation method: compound % activity = 100*[(0% control – X)/(0% control – 100% control)], where X represents the normalized value for the compound based on the DMSO (100% control) and positive control pirodavir (0% control).

Small-Molecule Compounds Selected for Follow-Up Screening

All compounds and biological data were selected using ChemistryConnect, and filtered and analyzed with Pipeline Pilot^24–26 (https://www.3dsbiovia.com/products/collaborative-science/biovia-pipeline-pilot). Initially, 270 compounds were selected to cover 122 targets identified by the RV siRNA experiment. These are annotated as antagonists and retrieved from internal and external sources. Clustering was performed within Pipeline Pilot using substructure definitions. The nearest-neighbor search was performed in the multifingerprint tool selecting from the available screening compound collection.^24,25

Statistical Analyses

Redundant siRNA activity (RSA) analysis was used to aid the selection of hit mRNA targets from the siRNA library screen.²⁷ Data are presented as means ± SEM unless otherwise indicated in the figure legends. Compound concentration response (CR) data analysis using nonlinear regression fitted to a sigmoidal four-parameter equation, and statistical analysis were performed in GraphPad Prism 7 (La Jolla, CA). Statistical tests applied are indicated in the figure and table legends.

Results

Combined siRNA and Small-Molecule Compound Screening Approach for Target Identification

To identify putative targets that affect RV replication, we performed a combined siRNA and small-molecule compound phenotypic screening approach to first screen a library of siRNAs and, subsequently, screen a set of small-molecule compounds that are annotated to be pharmacological regulators of target genes that were identified to significantly affect RV2 replication in the siRNA primary screening. The siRNA library consists of 10,500 druggable genes with three siRNAs per gene, and the small-molecule compound set consists of 270 compounds. This set of 270 small-molecule compounds was selected using literature and AstraZeneca internal biological activity data via ChemistryConnect based on the criteria shown in Figure 1 , to cover 122 targets that were identified from the primary siRNA screen. In the secondary qPCR assays for RV2, RV14, and RV16, the active compounds from the initial screen were validated, together with their nearest neighbors (NNs) and identified target-enriched compounds. The annotated biological activity of these compounds was investigated to generate hypotheses for targets that may be driving the perturbation of RV replication.

Figure 1.

Overview of a combined siRNA and small-molecule compound screening cascade for identifying novel targets that inhibit RV replication in NHBE cells, number of siRNA and small molecules screened at each step, and criteria used for selecting compounds for follow-up screening.

Primary Screening of siRNA Library for RV2 Replication Regulators

A primary phenotypic screening of a library of siRNAs against druggable genes (three siRNAs per gene) in NHBE cells was performed to identify siRNAs that affect RV2 replication. RV2 replication was quantified via in situ hybridization with RV2 area index (the proportion of the cell cytoplasm positive for RV2 RNA) as an indication of viral replication, and number of nuclei (absolute number of nuclei per image field) as an indication of cell proliferative activity ( Fig. 2A ). Data for all readouts were normalized to the median of all negative controls on the same plate and, subsequently, results for interplate triplicates were averaged for each well. RSA analysis was used to aid the selection of hit mRNA targets from the siRNA primary screen.²⁷ In order to be as inclusive as possible in selecting targets for the small-molecule follow-up screen, and thus reduce the risk of false negatives, a low threshold for defining a hit was used whereby one of the following three criteria needed to be met: (1) an RV2 area index for at least two of three siRNAs ≥50% lower than the control or ≥30% higher than the control), (2) an RSA p value of <0.05 for inhibition or enhancement effect, or (3) a false discovery rate adjusted t test based on the mean RV2 area index value of the three siRNAs per gene compared with the control (q < 0.1). A gene was defined as a hit if it met one of these criteria and had at least two of three siRNAs with a normalized mean nuclear number value ≥50% of the control. Using these criteria, 1128 siRNAs were selected as candidate regulators of RV2 replication ( Fig. 2B ). siRNA knockdown of these 1128 genes either inhibited or enhanced RV2 replication.

Figure 2.

Overall results of RV2 siRNA screening using in situ hybridization in NHBE cells. NHBE cells were transfected with 30 nM siRNAs for 72 h, followed by infection with RV2 at an MOI of 70 for 24 h. Cells were then fixed and stained with MitoTracker Deep Red, Hoechst 33342, and a ViewRNA RV2-specific probe, and imaged with 20× objectives. (A) In situ hybridization RV2 ViewRNA images. (B) Overall results of the 1128 active genes based on the p value derived from the RSA method (see Materials and Methods). Dark circles: 109 active genes showing inhibition effect on RV2 replication with available small molecules. (C) Dark squares: 13 active genes showing enhancement effect on RV2 replication with available small molecules. (D) Coverage of target classes in the 270 selected compounds covering 122 active genes (109 genes inhibiting and 13 genes enhancing RV2 replication).

Follow-Up Screening of Small-Molecule Compounds for RV Replication Regulators

Searching within the AstraZeneca available compound collection for compounds that are annotated to be pharmacological regulators of the 1128 target genes revealed that only 122 targets have available compounds in the compound collection.^24,25 Among the 122 targets, siRNA knockdown of 109 targets inhibited RV2 replication in NHBE cells ( Fig. 2B ), and knockdown of 13 genes resulted in an enhancement effect on RV2 replication ( Fig. 2C ). These 122 targets cover different target classes, including G-protein-coupled receptors (GPCRs), ion channels, nuclear hormone receptors (NHRs), and kinase and nonkinase enzymes, of which kinases constituted a major part ( Fig. 2D ). Two hundred seventy structurally diverse compounds known to interact with the 122 targets were selected and screened in a follow-up RV2 screen on 27 plates at an 8-point CR in NHBE cells. The small-molecule RV2 in situ hybridization assay was robust with a signal-to-background ratio of >10 on all the plates, and Z′ values of 0.3–0.57 on 23 of the 27 plates, demonstrating good reproducibility ( Fig. 3A ). From this follow-up screen, we identified 81 compounds that reduced RV2 replication with an IC50 of <10 µM ( Fig. 3B ). Among the 81 active compounds, 13 actives strongly inhibited RV2 replication with an IC50 of <3 µM, and 10 compounds inhibited RV2 replication with an IC50 of <10 µM, without measurable cytotoxicity, as determined by counting the number of nuclei and assessing nuclear morphology. Figure 3C,D shows compound ZD9720 (PubChem CID 18697995) and one AstraZeneca proprietary compound, compound-A, which targets the RV directly induced CR effect on RV2 replication. A target activity profile was created to perform a target enrichment analysis. This analysis showed no statistically significantly enriched target due to the low number of strong hits (13 top actives inhibited RV2 replication with an IC50 of <3 µM) and overlapping targets. Instead, the correlation between the respective target activities and the RV2 screen readout was used. Targets with more than one active compound, with a positive slope, passing manual inspection, or identified as a hit in the siRNA screen were selected for further studies. For these targets all stock available compounds with measured antagonist activity were abstracted from ChemistryConnect. A diversity selection was carried out, leading to a total set of 39 compounds. These were enriched by 31 NNs to the top 13 actives. Subsequently, all three sets (83 compounds in total) were combined and evaluated at a 7-point CR in NHBE cells in a secondary RV2 qPCR assay for confirmation. The RV2 qPCR assay showed good effect separation between DMSO and the positive control pirodavir ( Fig. 4A ). Pirodavir and compound-A induced CR effects on RV2 replication in the RV2 qPCR assay ( Fig. 4B,C ). The RV2 qPCR secondary screen generated 38 actives displaying an IC50 of <1 µM on RV2 with 20 of the 38 active compounds strongly inhibiting RV2 replication without measurable cytotoxicity as determined by assessing the effect on the housekeeping gene PPIA. A target enrichment analysis revealed a number of potential targets, including INSR, FLT4, PRKAA1, MAOA, and PHKG2 (see the full list in Fig. 4 ).

Figure 3.

Overall results of the RV2 small-molecule follow-up screening using in situ hybridization in NHBE cells. NHBE cells were treated with small molecules for 24 h, followed by infection with RV2 at an MOI of 30 for 30 h. Cells were then fixed and stained with MitoTracker Deep Red, Hoechst 33342, and a ViewRNA RV2-specific probe, and imaged with 20× objectives. RV2 replication was quantified with RV2 area index and data were normalized as percent activity based on the on-plate DMSO (100% control) and positive control ZD9720 (0% control). (A) Normalized RV2 area index from one representative screening plate (384 wells) treated with DMSO (dark circles), positive control ZD9720 (empty squares), and 34 compounds at an 8-point CR (gray circles). (B) Potency of 270 small-molecule compounds on inhibition of RV2 replication. (C,D) Two positive controls, ZD9720 and compound-A, induced concentration-dependent inhibition of RV2 replication. Data presented in C and D are mean ± SEM from n = 4 individual experiments.

Figure 4.

Overall results of the RV2 secondary screening using qRT-PCR in NHBE cells. NHBE cells were treated with small molecules for 24 h, followed by infection with RV2 at an MOI of 3.5 for 24–26 h. Cells were then lysed and cellular viral RNA analyzed by qRT-PCR for RV2 and a housekeeping gene PPIA. RV2 replication was quantified with Cq of RV2 and data were normalized to delta Cq of RV2, delta delta Cq, and then as percent activity based on the on-plate DMSO control (100% control) and positive control pirodavir (0% control). (A) Delta delta Cp (RV2) from one representative screening plate (384 wells) treated with DMSO (dark circles), positive control pirodavir (empty squares), and 15 compounds at a 10-point CR in duplicates. (B,C) Concentration-dependent inhibition of RV2 replication induced by pirodavir (B) and compound-A (C). Table: A list of potential targets emerging from the RV2 qPCR screening of the ViewRNA top hits and NN screening. Data presented in B and C are mean ± SEM from n = 2 individual experiments.

Identification of FLT4 as a Regulator of RV2 Replication in NHBE Cells

Biochemical validation of the potential targets was performed by correlating the potency of each compound in the RV2 assay with the potency of that compound in biochemical assays of the target retrieved from literature. From this analysis, FLT4 was found to be the most promising target. Overall, 14 compounds (CHEMBL347086, CHEMBL29197, CHEMBL3544937, CHEMBL357815, CHEMBL150187, CHEMBL443962, CHEMBL329201, CHEMBL276711, CHEMBL535, CHEMBL226471, CHEMBL21156, CHEMBL2105708, CHEMBL1230541, and CHEMBL1336) (see Suppl. Figs. S1–S3 and Suppl. Table S1 for more information about these compounds) with an activity range of 3–4 log units and the biochemical potencies within three small clusters (clusters 1–3) correlated reasonably well with RV2 inhibition ( Fig. 5A,D ). These 14 compounds are also reported to show activities against other targets (Suppl. Figs. S2 and S3). However, our activity correlation analysis and manual inspection only support the hypothesis that FLT4 is a potential responsible target. This observation does not exclude the possibility that the observed effect is driven by an unknown target or is due to cross-target activity. The observed activity correlation between any targets and RV replication needs to be supported by the siRNA screen data. Furthermore, the combined set of 83 compounds was also evaluated in RV14 and RV16 qPCR assays for specificity. The FLT4 activity versus RV14 or RV16 inhibition plots ( Fig. 5B,C ) also suggests a potential role of FLT4 in inhibiting RV14 and RV16 replication. FLT4 was a weak hit in the primary siRNA screen in that only one of the three different criteria used to identify the 1128 hits from the siRNA primary screen were met (the t test). It was not significant by RSA, and it did not meet the fold change cutoff criterion. The siRNA screening results show that knockdown of FLT4 gene using siRNA in NHBE cells resulted in an inhibition of RV2 replication as shown by the reduced mean RV2 area index of 36% and 55%, respectively, by treating NHBE cells with two of the three FLT4 siRNA variants (variants 1 and 3) ( Fig. 6 ). Taken together, from the results of the siRNA primary screening and follow-up small-molecule compound screening, we hypothesize that FLT4 is a potential target for the inhibition of RV replication.

Figure 5.

Potency correlation between compound effect on RV2, RV14, or RV16 and effect on FLT4. NHBE cells were treated with small molecules for 24 h, followed by infection with RV2 at an MOI of 3.5, RV14 at an MOI of 0.0004, or RV16 at an MOI of 6, respectively, for 24–26 h. Cells were then lysed and cellular viral RNA analyzed by qRT-PCR for RV2, RV14, or RV16 and the housekeeping gene PPIA. Anti-RV2 (A), anti-RV14 (B), and anti-RV16 (C) IC50 values of 14 compounds generated from the RV2, RV14, and RV16 qPCR screens were plotted against mean FLT4 IC50 values retrieved from the literature and originating from different biochemical assays. (D) Core structures for the corresponding clusters of the 14 FLT4 and RV active compounds. These 14 compounds are CHEMBL347086, CHEMBL29197, CHEMBL3544937, CHEMBL357815, CHEMBL150187, CHEMBL443962, CHEMBL329201, CHEMBL276711, CHEMBL535, CHEMBL226471, CHEMBL21156, CHEMBL2105708, CHEMBL1230541, and CHEMBL1336.

Figure 6.

FLT4 results from the primary RV2 siRNA screening in NHBE cells. NHBE cells were transfected with three independent FLT4-specific siRNA variants (30 nM) for 72 h, followed by infection with RV2 at 70 MOI for 24 h. The proportion of the cytoplasmic area staining positive for RV2 RNA, “RV area index,” was calculated as described in Materials and Methods and normalized to the median of the wells treated with the negative control siRNA. FLT4 and negative control gene knockdown results are presented as mean ± SEM from three independent experiments.

Discussion

In this study, we report a combined siRNA and small-molecule phenotypic screening approach to identify targets regulating RV replication in human bronchial epithelial cells. This combination screening approach uses an in situ hybridization assay to first screen a library of siRNAs consisting of 10,500 druggable genes with 1128 active siRNA hits identified to either inhibit or enhance RV2 replication and, subsequently, using the same RV2 in situ hybridization assay, screen 270 small-molecule compounds that are annotated to be pharmacological regulators of 122 target genes that were identified in the siRNA primary screen. The active small-molecule compounds identified from the screen of 270 compounds, and chemically similar compounds, were further tested in the secondary qPCR assays for RV2, RV14, and RV16. With this combined siRNA and small-molecule screening approach, we have successfully identified FLT4 as a novel target regulating RV replication.

Phenotypic screens where physiologically relevant cells and biologically relevant in vitro assays are used to study perturbance-induced phenotypic changes have been shown to be very useful for identifying novel targets for treating human disease.^10,28,29 Various genome-wide siRNA phenotypic screens studying the effects of siRNA-mediated gene knockdown on biological phenotypes, including viral replication, apoptosis, cell adhesion, and cell proliferation, have been reported.^30–33 Lin and coworkers reported that the top-scoring siRNA hits, in a screen to identify novel regulators of the HIF-α transcription pathway, did not act through specific knockdown of the intended targets, but rather by targeting the HIF1-α mRNA through a microRNA-like mechanism.³⁴ Recent studies by Lin et al. and Smith et al. demonstrated that RNAi is more susceptible to off-target effects than CRISPR.^19,35 One approach to reduce the number of false positives is to use a higher cutoff. However, at the same time as minimizing the false-positive rate, this approach may also increase the number of false negatives.

Small-molecule compound libraries, especially pharmacologically annotated tool compound collections, have been successfully used in various phenotypic screens for the identification of novel targets for biology of interest. Phenotypic effects induced by small-molecule inhibition of a target in a cell-based assay should in many cases be similar to the gene knockdown-induced effect of the same target. Using siRNA to knock down the gene of interest is a commonly used approach for the confirmation of any possible targets generated from small-molecule phenotypic screens as well as for the validation of cellular assays in the case that tool compounds for regulating the target of interest are not available.^36,37 Following a rigorous process of technical and biological validation, our primary siRNA screen identified eight “high-confidence” hit mRNA targets, including LSS, which significantly reduced RV replication.¹⁵ To investigate whether there is any potential for identifying more drug targets regulating RV replication from the siRNA screen, we applied a low threshold in the siRNA screen in order to reduce the number of false negatives, and this produced 1128 hits, a number that is not practical for detailed target validation studies. Instead, we integrated a follow-up small-molecule screen using the same high-content imaging RV2 assay applied in the primary siRNA screen. We did this by screening small-molecule compounds that are annotated to be pharmacological regulators of target genes that were identified in the siRNA primary screen. Our rationale was that a siRNA screen in combination with an annotated small-molecule compound screen might help to reduce false negatives by using a low threshold and facilitate the identification of additional drug targets that are not picked up as top-scoring primary siRNA hits. A similar approach has recently been reported in the search for host targets against influenza virus infection, whereby experimental or clinically approved drugs were selected based on the hits from an siRNA screen, enabling the identification of a host signaling pathway for targeting influenza virus.³⁸ Based on this hypothesis, we searched within the AstraZeneca compound collection for compounds that are annotated to be pharmacological regulators of the 1128 target genes. Only 122 targets had available compounds in the AstraZeneca compound collection. Two hundred seventy compounds were initially selected to cover these 122 targets. We used the same high-content imaging RV2 assay by quantifying viral replication via in situ hybridization as in the siRNA screen, to ensure comparability between the two screens. The confirmed small-molecule RV2 hits, and NNs of the confirmed hits, were further studied in a secondary RV2 assay as well as in RV14 and RV16 assays in which viral replication was quantified by qPCR. This combination screening approach allowed us to identify a novel target, FLT4, which had not been originally identified as a strong hit in the primary siRNA screen, that inhibits RV2 replication as well as shows a potential role in RV14 and RV16 inhibition. Thus, we were able to identify an additional target that would have been overlooked, had we not performed the follow-up small-molecule screen. This approach of combining genome-wide siRNA and small-molecule screens for the identification of novel drug targets could have wide applications. However, in our study, among the 1128 primary siRNA hits, only 122 siRNA hits had available compounds in the AstraZeneca compound collection, which limits our opportunity of identifying more additional targets. Interestingly, Lesch et al. identified available compounds for a similar proportion of the hits from their influenza virus siRNA screen (43 compounds covering 14 of the total 133 hit genes).³⁸ We might have been able to identify more targets using this combination screening approach if more compounds covering more primary siRNA hits had been available. Therefore, the success of this combination screening approach highly depends on the availability of small-molecule compounds.

FLT4 (VEGFR-3) is a receptor tyrosine kinase thta is primarily expressed on the lymphatic endothelium. The ligands for FLT4 are VEGF-C and VEGF-D, and receptor engagement and activation mediates lymph angiogenesis and maintenance of the lymphatic endothelium. Corresponding with this key function, mutations in the FLT4 gene cause hereditary lymphedema. FLT4 has also been linked to tumor metastasis.³⁹ Several drugs with activity against multiple receptor tyrosine kinases, including FLT4, have proven to be effective antitumor therapies, for example, erdafitinib, telatinib, sunitinib, and axitinib.^40–43 However, to our knowledge, our finding of a link between FLT4 and RV infection is novel. In a study of natural RV infections in human subjects, VEGF levels were increased in nasal secretions.⁴⁴ However, it should be noted that the enzyme-linked immunosorbent assay (ELISA) used in this study detects VEGF-A, which does not bind to FLT4, and whether it can also detect VEGF-C or VEGF-D is unknown. Little is known about the role of FLT4 in viral infections in general. However, intriguingly, this is not the first time that FLT4 has emerged as a hit from an siRNA functional genomics screen for viral infection. For example, in the influenza virus siRNA screen performed by Lesch et al., not only was FLT4 one of the top 3 hit genes, but also 3 of the 14 selected compounds in their follow-up testing were active against FLT4.³⁸ FLT4 was also one of 5 genes that survived multiple rounds of testing and data analysis methods in an siRNA screen for hepatitis C infection.⁴⁵ In another study, siRNA-mediated knockdown of FLT4 inhibited chikungunya virus infection.⁴⁶ In a limited screen of siRNA against 65 genes against Ebola, virus-like particles showed that silencing of the FLT4 gene suppressed viral replication.⁴⁷ The authors went on to show by co-immunoprecipitation and Ch-IP that FLT4 protein interacts with both viral glycoprotein and viral RNA. The mechanism was suggested to be related to inhibition of cellular secretory functions.

The mechanism by which silencing of FLT4 suppresses viral infection is poorly understood. The FLT4 inhibitors tested by Lesch et al. in their influenza studies were shown to inhibit fusion of the viral and endosomal membranes.³⁸ FLT4 was shown to be induced via Ets 1 in Kaposi’s sarcoma virus-infected cells.⁴⁸ Ma et al. demonstrated that respiratory syncytial virus-, influenza-, or dsRNA-induced type I interferon (IFN) responses inhibit VEGF responses in human VEGF-transgenic mice, suggesting a potential link between VEGF and innate antiviral immunity.⁴⁹ However, it should be noted that the mechanism was shown to be via suppression of VEGFR-1 expression but not FLT4. A recent study showed a link between stimulation of VEGF-C expression and suppression of type I IFN responses.⁵⁰ Further studies would be required to investigate whether our findings could be explained by enhanced type I IFN responses. To our knowledge, there are no published studies investigating FLT4 in asthma. However, VEGF levels have been shown to be increased in the asthmatic airway, a finding that may relate to airway remodeling.^51,52 As noted above, the ELISA used in these studies is most likely to detect VEGF-A, and it is not known whether these studies were also detecting VEGF-C or VEGF-D. In one study, FLT4 was shown to be expressed by cultured distal lung fibroblasts from COPD patients, and these cells produced VEGF in response to iloprost or TGF-β, implicating VEGF signaling in vascular remodeling in COPD.⁵³ However, to our knowledge, there are no published data linking FLT4 with susceptibility to viral exacerbations in COPD.

In summary, we performed a small-molecule phenotypic screen to follow up hits identified in a primary siRNA screen for modulators of RV replication. This combination screening approach allowed us to identify FLT4 as a novel target regulating RV replication. Our study demonstrates that a combination of siRNA and small-molecule compound screening models is a useful approach for the identification of novel drug targets.

Supplemental Material

Supplemental_Info_Combined_siRNA_and_SMs_HRV_screen_by_Ding_et_al – Supplemental material for Combined siRNA and Small-Molecule Phenotypic Screening Identifies Targets Regulating Rhinovirus Replication in Primary Human Bronchial Epithelial Cells

Supplemental material, Supplemental_Info_Combined_siRNA_and_SMs_HRV_screen_by_Ding_et_al for Combined siRNA and Small-Molecule Phenotypic Screening Identifies Targets Regulating Rhinovirus Replication in Primary Human Bronchial Epithelial Cells by Mei Ding, Christian Tyrchan, Elisabeth Bäck, Jörgen Östling, Steffen Schubert and Christopher McCrae in SLAS Discovery

Supplemental Material

Supplemental_Table_S1_Combined_siRNA_and_SMs_HRV_screen_by_Ding_et_al – Supplemental material for Combined siRNA and Small-Molecule Phenotypic Screening Identifies Targets Regulating Rhinovirus Replication in Primary Human Bronchial Epithelial Cells

Supplemental material, Supplemental_Table_S1_Combined_siRNA_and_SMs_HRV_screen_by_Ding_et_al for Combined siRNA and Small-Molecule Phenotypic Screening Identifies Targets Regulating Rhinovirus Replication in Primary Human Bronchial Epithelial Cells by Mei Ding, Christian Tyrchan, Elisabeth Bäck, Jörgen Östling, Steffen Schubert and Christopher McCrae in SLAS Discovery

Footnotes

Acknowledgements

The authors would like to thank Niklas Blomberg for initiating the work to select the small-molecule compounds for the follow-up screening, scientists at Cenix for supporting the in situ hybridization experiments, and Per-Erik Strömstedt and Ryan Hicks for valuable discussions.

Supplemental material is available online with this article.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Mei Ding

References

Uncapher

C. R.

DeWitt

C. M.

Colonno

R. J.

The Major and Minor Group Receptor Families Contain All but One Human Rhinovirus Serotype. Virology 1991, 180, 814–817

Jacobs

S. E.

Lamson

D. M.

St. George

; et al. Human Rhinoviruses. Clin. Microbiol. Rev. 2013, 26, 135–162.

Friedlander

S. L.

Busse

W. W.

The Role of Rhinovirus in Asthma Exacerbations. J. Allergy Clin. Immunol. 2005, 116, 267–273.

Papadopoulos

N. G.

Christodoulou

Rohde

; et al. Viruses and Bacteria in Acute Asthma Exacerbations—A GA(2) LEN-DARE Systematic Review. Allergy 2011, 66, 458–468.

Kurai

Saraya

Ishii

; et al. Virus-Induced Exacerbations in Asthma and COPD. Front. Microbiol. 2013, 4, 293.

Hershenson

M. B.

Rhinovirus-Induced Exacerbations of Asthma and COPD. Scientifica (Cairo) 2013, 2013, 405876.

Bochkov

Y. A.

Watters

Ashraf

; et al. Cadherin-Related Family Member 3, a Childhood Asthma Susceptibility Gene Product, Mediates Rhinovirus C Binding and Replication. Proc. Natl. Acad. Sci. U.S.A. 2015, 112, 5485–5490.

Deans

R. M.

Morgens

D. W.

Okesli

; et al. Parallel shRNA and CRISPR-Cas9 Screens Enable Antiviral Drug Target Identification. Nat. Chem. Biol. 2016, 12, 361–366.

Bhinder

Antczak

Ramirez

C. N.

; et al. An Arrayed Genome-Scale Lentiviral-Enabled Short Hairpin RNA Screen Identifies Lethal and Rescuer Gene Candidates. Assay Drug Dev. Technol. 2013, 11, 173–190.

10.

Moffat

J. G.

Vincent

Lee

J. A.

; et al. Opportunities and Challenges in Phenotypic Drug Discovery: An Industry Perspective. Nat. Rev. Drug Discov. 2017, 16, 531–543.

11.

Kurata

Wolf

N. K.

Lahr

W. S.

; et al. Highly Multiplexed Genome Engineering Using CRISPR/Cas9 gRNA Arrays. PLoS One 2018, 13, e0198714.

12.

Cole

K. A.

Huggins

Laquaglia

; et al. RNAi Screen of the Protein Kinome Identifies Checkpoint Kinase 1 (CHK1) as a Therapeutic Target in Neuroblastoma. Proc. Natl. Acad. Sci. U.S.A. 2011, 108, 3336–3341.

13.

Quon

Kassner

P. D.

RNA Interference Screening for the Discovery of Oncology Targets. Expert Opin. Ther. Targets 2009, 13, 1027–1035.

14.

Zhou

Huang

; et al. Genome-Scale RNAi Screen for Host Factors Required for HIV Replication. Cell Host Microbe 2008, 4, 495–504.

15.

McCrae

Dzgoev

Stahlman

; et al. Lanosterol Synthase Regulates Human Rhinovirus Replication in Human Bronchial Epithelial Cells. Am. J. Respir. Cell Mol. Biol. 2018, 59, 713–722.

16.

Jackson

A. L.

Linsley

P. S.

Recognizing and Avoiding siRNA Off-Target Effects for Target Identification and Therapeutic Application. Nat. Rev. Drug Discov. 2010, 9, 57–67.

17.

Mohr

Bakal

Perrimon

Genomic Screening with RNAi: Results and Challenges. Annu. Rev. Biochem. 2010, 79, 37–64.

18.

Sigoillot

F. D.

King

R. W.

Vigilance and Validation: Keys to Success in RNAi Screening. ACS Chem. Biol. 2011, 6, 47–60.

19.

Lin

Giuliano

C. J.

Palladino

; et al. Off-Target Toxicity Is a Common Mechanism of Action of Cancer Drugs Undergoing Clinical Trials. Sci. Transl. Med. 2019, 11, eaww8412.

20.

Spearman

The Method of “Right and Wrong Cases” (Constant Stimuli) without Gauss’s Formula. Br. J. Psychol. 1908, 2, 227–242.

21.

Kärber

Beitrag zur kollektiven Behandlung pharmakologischer Reihenversuche. Naunyn Schmiedebergs Arch. Exp. Pathol. Pharmakol. 1931, 162, 480–483.

22.

Pyrah

I. T.

Kalinowski

Jackson

; et al. Toxicologic Lesions Associated with Two Related Inhibitors of Oxidosqualene Cyclase in the Dog and Mouse. Toxicol. Pathol. 2001, 29, 174–179.

23.

Lacroix

Laconi

Angius

; et al. In Vitro Characterisation of a Pleconaril/Pirodavir-Like Compound with Potent Activity against Rhinoviruses. Virol. J. 2015, 12, 106.

24.

Kogej

Engkvist

Blomberg

; et al. Multifingerprint Based Similarity Searches for Targeted Class Compound Selection. J. Chem. Inf. Model. 2006, 46, 1201–1213.

25.

Steffen

Kogej

Tyrchan

; et al. Comparison of Molecular Fingerprint Methods on the Basis of Biological Profile Data. J. Chem. Inf. Model. 2009, 49, 338–347.

26.

Dassault Systèmes BIOVIA. BIOVIA Workbook, Release 2017; BIOVIA Pipeline Pilot, Release 2017; Dassault Systèmes: San Diego, 2017.

27.

Konig

Chiang

C. Y.

B. P.

; et al. A Probability-Based Approach for the Analysis of Large-Scale RNAi Screens. Nat. Methods 2007, 4, 847–849.

28.

Eder

Sedrani

Wiesmann

The Discovery of First-in-Class Drugs: Origins and Evolution. Nat. Rev. Drug Discov. 2014, 13, 577–587.

29.

Swinney

D. C.

Anthony

How Were New Medicines Discovered?

Nat. Rev. Drug Discov. 2011, 10, 507–519.

30.

Huang

Wang

J. Y.

Systems Analysis of Quantitative shRNA-Library Screens Identifies Regulators of Cell Adhesion. BMC Syst. Biol. 2008, 2, 49.

31.

Kimura

Nguyen

S. T.

Liu

; et al. A Functional Genome-Wide RNAi Screen Identifies TAF1 as a Regulator for Apoptosis in Response to Genotoxic Stress. Nucleic Acids Res. 2008, 36, 5250–5259.

32.

Schlabach

M. R.

Luo

Solimini

N. L.

; et al. Cancer Proliferation Gene Discovery through Functional Genomics. Science 2008, 319, 620–624.

33.

Ghosh

Kaushik

Khurana

; et al. An RNAi-Based High-Throughput Screening Assay to Identify Small Molecule Inhibitors of Hepatitis B Virus Replication. J. Biol. Chem. 2017, 292, 12577–12588.

34.

Lin

Ruan

Anderson

M. G.

; et al. siRNA-Mediated Off-Target Gene Silencing Triggered by a 7 nt Complementation. Nucleic Acids Res. 2005, 33, 4527–4535.

35.

Smith

Greenside

P. G.

Natoli

; et al. Evaluation of RNAi and CRISPR Technologies by Large-Scale Gene Expression Profiling in the Connectivity Map. PLoS Biol. 2017, 15, e2003213.

36.

Ding

Brengdahl

Lindqvist

; et al. A Phenotypic Screening Approach Using Human Treg Cells Identified Regulators of Forkhead Box p3 Expression. ACS Chem. Biol. 2019, 14, 543–553.

37.

Rondinone

C. M.

RNAi: For Functional Analysis and Target Validation. Expert Opin. Ther. Targets 2006, 10, 337–340.

38.

Lesch

Luckner

Meyer

; et al. RNAi-Based Small Molecule Repositioning Reveals Clinically Approved Urea-Based Kinase Inhibitors as Broadly Active Antivirals. PLoS Pathog. 2019, 15, e1007601.

39.

Chen

Dang

; et al. FGF19-Mediated Upregulation of SOX18 Promotes Hepatocellular Carcinoma Metastasis by Transactivating FGFR4 and FLT4. Hepatology [Online early access.] DOI: 10.1002/hep.30951. Published Online: Sept 16, 2019.

40.

Hanna

K. S.

Erdafitinib to Treat Urothelial Carcinoma. Drugs Today (Barc.) 2019, 55, 495–501.

41.

Diekstra

M. H.

Fritsch

Kanefendt

; et al. Population Modeling Integrating Pharmacokinetics, Pharmacodynamics, Pharmacogenetics, and Clinical Outcome in Patients with Sunitinib-Treated Cancer. CPT Pharmacometrics Syst. Pharmacol. 2017, 6, 604–613.

42.

van IJzendoorn

D. G. P.

Sleijfer

Gelderblom

; et al. Telatinib Is an Effective Targeted Therapy for Pseudomyogenic Hemangioendothelioma. Clin. Cancer Res. 2018, 24, 2678–2687.

43.

Tzogani

Skibeli

Westgaard

; et al. The European Medicines Agency Approval of Axitinib (Inlyta) for the Treatment of Advanced Renal Cell Carcinoma after Failure of Prior Treatment with Sunitinib or a Cytokine: Summary of the Scientific Assessment of the Committee for Medicinal Products for Human Use. Oncologist 2015, 20, 196–201.

44.

Leigh

Oyelusi

Wiehler

; et al. Human Rhinovirus Infection Enhances Airway Epithelial Cell Production of Growth Factors Involved in Airway Remodeling. J. Allergy Clin. Immunol. 2008, 121, 1238–1245.e4.

45.

Suratanee

Rebhan

Matula

; et al. Detecting Host Factors Involved in Virus Infection by Observing the Clustering of Infected Cells in siRNA Screening Images. Bioinformatics 2010, 26, i653–i658.

46.

Karlas

Berre

Couderc

; et al. A Human Genome-Wide Loss-of-Function Screen Identifies Effective Chikungunya Antiviral Drugs. Nat. Commun. 2016, 7, 11320.

47.

D. S.

Weng

T. H.

C. Y.

; et al. Chaperones, Membrane Trafficking and Signal Transduction Proteins Regulate Zaire Ebola Virus trVLPs and Interact with trVLP Elements. Front. Microbiol. 2018, 9, 2724.

48.

Gutierrez

K. D.

Morris

V. A.

; et al. Ets-1 Is Required for the Activation of VEGFR3 during Latent Kaposi’s Sarcoma-Associated Herpesvirus Infection of Endothelial Cells. J. Virol. 2013, 87, 6758–6768.

49.

Dela Cruz

C. S.

Hartl

; et al. RIG-Like Helicase Innate Immunity Inhibits Vascular Endothelial Growth Factor Tissue Responses via a Type I IFN-Dependent Mechanism. Am. J. Respir. Crit. Care Med. 2011, 183, 1322–1335.

50.

Ning

; et al. Hes1 Attenuates Type I IFN Responses via VEGF-C and WDFY1. J. Exp. Med. 2019, 216, 1396–1410.

51.

Asai

Kanazawa

Kamoi

; et al. Increased Levels of Vascular Endothelial Growth Factor in Induced Sputum in Asthmatic Patients. Clin. Exp. Allergy 2003, 33, 595–599.

52.

Hoshino

Takahashi

Aoike

Expression of Vascular Endothelial Growth Factor, Basic Fibroblast Growth Factor, and Angiogenin Immunoreactivity in Asthmatic Airways and Its Relationship to Angiogenesis. J. Allergy Clin. Immunol. 2001, 107, 295–301.

53.

Westergren-Thorsson

Bagher

Andersson-Sjoland

; et al. VEGF Synthesis Is Induced by Prostacyclin and TGF-beta in Distal Lung Fibroblasts from COPD Patients and Control Subjects: Implications for Pulmonary Vascular Remodelling. Respirology 2018, 23, 68–75.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.55 MB

0.02 MB