A full cycle anti-viral drug screen identifies a clinical 1 compound against adenovirus infection

Human adenoviruses (HAdVs) are fatal to immune-suppressed people, but no effective anti-HAdV therapy is available. Here, we present a novel image-based high-throughput screening (HTS) platform, which scores the full viral replication cycle from virus entry to dissemination of progeny. We analysed 1,280 small molecular compounds of the Prestwick Chemical Library (PCL) for interference with HAdV-C2 infection in a quadruplicate blinded format, followed by robust image analyses, and hit identification. We present the entire set of image-based screening data including all the images, and the image analysis and data processing pipelines, as deposited at the Image Data repository (IDR) 1, accession number idr0081. We identified Nelfinavir mesylate as an inhibitor of HAdV plaque formation, in agreement with the previous notion that Nelfinavir is ineffective in single round HAdV infection assays. Nelfinavir has been FDA-approved for anti-retroviral therapy in humans. Our results underscore the power of image-based multi-round infection assays in identifying viral inhibitors with clinical potential.


14 15
Human adenoviruses (HAdVs) are fatal to immune-suppressed people, but no effective 16 anti-HAdV therapy is available. Here, we present a novel image-based high-throughput 17 screening (HTS) platform, which scores the full viral replication cycle from virus entry to 18 dissemination of progeny. We analysed 1,280 small molecular compounds of the Prestwick 19 Chemical Library (PCL) for interference with HAdV-C2 infection in a quadruplicate blinded 20 format, followed by robust image analyses, and hit identification. We present the entire set 21 of image-based screening data including all the images, and the image analysis and data 22 processing pipelines, as deposited at the Image Data repository (IDR) 1 , accession number 23 idr0081. We identified Nelfinavir mesylate as an inhibitor of HAdV plaque formation, in 24 agreement with the previous notion that Nelfinavir is ineffective in single round HAdV 25 infection assays. Nelfinavir has been FDA-approved for anti-retroviral therapy in humans.

26
Our results underscore the power of image-based multi-round infection assays in 27 identifying viral inhibitors with clinical potential.

29
Background & Summary: 30 Human adenoviruses (HAdV) predominantly cause diseases of the respiratory and gastrointes-31 tinal tracts. They are a significant cause of acute human disease with morbidity and mortality, 32 especially for immuno-compromised patients 2,3 as indicated by a recent outbreak in the USA 33 killing 12 children. Surprisingly, a recent case of HAdV-C2 caused meningoencephalitis was also 34 reported in a middle-aged woman in the US 4 . HAdV have a high prevalence 5-8 and are broadly 35 used as gene therapy vectors 9 and oncolytic viruses 10,11 . The high seroprevalence of HAdV-C2/5 36 12 underlines that HAdV infections are asymptomatic in healthy individuals, but HAdV persist in 37 mucosal lymphocytes, and thereby pose a severe risk for immunosuppressed patients undergoing 38 stem cell transplantation 13 . 39 40 More than 100 HAdV genotypes have been formally approved 14 and are grouped into seven 41 species based on hemagglutination assay and genome sequences 15 . They exhibit a broad range 42 of tissue tropism, including the respiratory and gastrointestinal tracts, the eye, the kidney, the 43 urogenital tracts and blood cells. While species A, F and G target the gastrointestinal tract, HAdV-44 B, C and E cause infections of the respiratory tract, and conjunctivitis is mostly associated with 45 species B and D, but also C types. HAdV-B show the broadest spectrum of tropisms, also infecting 46 the kidney and hematopoietic system 7,13 .

48
HAdV is a non-enveloped virus with a double-stranded DNA genome of ~36 kbp tightly packaged 49 into an icosahedral capsid of about 90 nm in diameter 16,17 . The best-studied HAdV are HAdV-50 C2/5 (species C, type 2 and 5), which are very closely related to each other. HAdV enter cells by 51 receptor mediated endocytosis, penetrate the endosomal membrane by the activation of a viral 52 lytic machinery, and shed virion proteins in a stepwise manner, until they arrive at the nuclear 53 membrane, where they uncoat and release their genome to the nucleus 18,19 . In the nucleus the 54 viral genome gives rise to the immediate early viral mRNA encoding the E1A protein which then 55 transactivates all the subviral promoters and is key to give rise to lytic infection and maintains viral 56 persistence in presence of the innate immune regulator interferon 20 . Mature HAdV progeny is 57 known to be released by cell lysis upon rupture of the nuclear envelope and the plasma 58 membrane, giving rise to cell-free virions 21 been successfully used to identify many compounds for repurposing applications ranging from 72 antimicrobial agents 26 to anticancer candidates 27 . For a full list of publications, see 28 . We 73 performed a phenotypic screen for HAdV-C2 infection, as outlined in ( Figure 1A, 1B). We took 74 advantage of automated fluorescence microscopy and image-based scoring of the progression 75 of multi-round infections using Plaque2.0 software 29 . This high-throughput screening (HTS) 76 modality was carried out at a 384-well plate format. For representative images, see Figure 1C. 77 78 We demonstrate robust imaging methodology, image analysis and data processing routines as 79 concluded from parallel procedures in two teams at independent institutions, the Biomolecular 80 Screening Facility at Ecole Polytechnique Fédérale de Lausanne (EPFL) and the Department of 81 Molecular Life Sciences at University of Zurich (UZH). To score the infection phenotypes, we used 82 five infection assay features obtained from microscopy: the number of nuclei, the number of 83 infected nuclei, the infection index as calculated from the total nuclei and the infected cells, the 84 number of plaques (areas of multi-round infection foci originating from a single infected host cell) 85 and the integrated viral infection marker, in this case the green fluorescence protein (GFP) 86 intensity. All data is available at the Image Data repository (IDR) 1 (IDR accession number 87 idr0081). The structure of the repository is outlined in Figure 2. Raw and scored infection 88 phenotype features are shown for UZH and EPFL analyses (Supplementary Tables 2 and 3, and  89  Supplementary Tables 4 and 5, respectively). Rigorous assay development ensured a high assay 90 quality as indicated in Figure 3 and by mean Z'-factors of 0.52 for the number of plaques (Table  91 1). The screening was performed in four biological replicates at high reproducibility, see Figure 4  92 and Table 2. We further excluded those PCL compounds that showed significant toxicity in the 93 absence of infection (Table 3 and Figure 5). Imaging, image analysis and scoring by the two 94 independent teams yielded well correlated scores, as depicted in Figure 6. 95 96 Our data indicate a high significance of the identified top hit, Nelfinavir mesylate ( Figure 1D and 97 6). We confirmed the efficacy of Nelfinavir as an inhibitor of HAdV infection by biological follow-98 up studies (submission in preparation). 99 100 Methods: 101 102 Virus 103 HAdV-C2-dE3B virus was produced as described 21  Presto-blue toxicity assay 141 Toxicity of the PCL chemical compounds on A549 in absence of infection was tested using 142 compound concentrations, treatment timing and seeding cell numbers corresponding to the 143 screening protocol, and using the Presto Blue Cell Viability reagent (Thermo Fisher Scientific, 144 Waltham, USA). Briefly, following 3.5-day continuous treatment of A549 cells, 10% final 145 PrestoBlue was added to each well and incubated for 1 h at standard cell incubation conditions. 146 Fluorescence intensity (bottom-read) was then measured using a multi-well plate reader (Tecan 147 Infinite F500, Tecan, Männedorf, Switzerland) with excitation at 560/10 nm, emission at 590/10 148 nm at a fixed gain. Doxorubicin hydrochloride (Prestw-438, Prestwick Chemical, Illkirch, France) 149 was used as a positive control for cytotoxicity, at a final concentration of 10 µM, and the 150 corresponding volume of DMSO was used as a negative control. The full PCL library was tested 151 on duplicated plates. The EPFL-BSF in-house Laboratory Information Management System 152 (LIMS) was used for data processing and statistical validation. First, raw PrestoBlue readings 153 were normalized per plate to negative control values at 0 and positive controls at 1. Then, the 154 normalized values of the duplicates were averaged. Assay quality was assessed for each plate 155 through the Z'-factor calculation. Compounds were considered toxic hits when the normalized 156 value for all replicates was higher than the average + 3σ (standard deviation, SD) of the DMSO 157 negative control for the corresponding plate. Scores and score SD were then calculated for hit 158 compounds by averaging normalized value for all replicates. Each screening plate set consisted of 4 plates A to D. Each screening plate consisted of 32 167 technical replicates of positive and negative control, each, and 320 single technical replicate PCL 168 compounds.

170
Wet-lab screening pipeline 171 The screening was performed in four independent biological replicates. Wet-lab liquid handling 172 was performed using a Matrix WellMate dispenser and Matrix WellMate tubing cartridges (Thermo 173 Fisher Scientific, Waltham, USA). Prior to usage, tubings were rinsed with 125 ml autoclaved 174 ddH2O followed by 125 ml autoclaved PBS. Pre-spotted compound plates were thawed at room 175 temperature (RT) for 30 min, briefly centrifuged before compounds were dissolved in 10 µl/ well 176 PBS. 4,000 A549 cells/ well in 60 µl full medium were seeded onto the compounds using standard 177 bore tubing cartridges. Following cell adhesion over night, the cells are inoculated with 1.77*10 5 178 genome equivalents per well of HAdV-C2-dE3B in 10 µl full media using bovine serum albumin 179 (BSA, cell-culture grade, Sigma-Aldrich, St. Louis, USA)-blocked small bore tubing cartridges. The Z'-factor was computed using R version 3.3.2 31 according to Equation (1) 212 where σ+ is the SD of the positive control, σ-is the SD of the negative control, μ+ the mean of the 214 positive control and μthe mean of the negative control.

216
Screening data processing 217 Plaque2.0 results were further independently processed and filtered. Data structure and repository 232 The HAdV screening data comprise the information collected during assay development, 233 including stability, quality and the PCL screening itself. The latter two have been imaged on two 234 different microscopes. We provide the parameters used for Plaque2.0 image analysis, and the 235 code for the subsequent hit filtering in R. The data structure as available at the IDR 1 , accession 236 number idr0081, is outlined in Figure 2. 237 238 Data sets and file types 239 The provided data consists of four data sets 1 to 4.

243
- Assay stability 256 The wet-lab screening pipeline was optimized regarding liquid handling, cell seeding, virus 257 inoculum, positive and negative control, time line, imaging and image analysis to ensure high 258 assay stability and reproducibility. Furthermore, all compounds, especially media and 259 supplements, the BSA for tubing saturation, PFA-and Hoechst-supplemented fixative were 260 prepared as large batch from a single LOT and stored as single-use aliquots. Assay stability with 261 respect to cell and infection phenotype was tested following the established wet-lab, imaging and 262 image analysis pipeline prior to every experiment on pre plates. Since the solvent control had 263 already been spotted in 10 µl PBS, no further PBS was added prior to cell seeding. If infection 264 scores were found to be low due to limited stability of viral stocks, the virus stock dilution in the 265 subsequent experiment was decreased.

267
Independent analysis 268 Imaging, image analysis and screening data processing was performed by two independent 269 research teams from two independent institutions at UZH and EPFL. Both dry-lab pipelines 270 confirmed the high assay quality (Table 1). As summarized in Figure 6  The assay's effect size was assessed following the established wet-lab, imaging and image 276 analysis pipeline for two independently performed Z' plates (Table 1 and Figure 2 Table 3) are found to be toxic.

419
The quality of the screening platform is assessed prior to screening the PCL by two independent Z' plates