Screening of potential antiviral molecules against equid herpesvirus-1 using cellular impedance measurement: Dataset of 2,891 compounds.

Data presented in this article are associated with the research article “Identification of antiviral compounds against equid herpesvirus-1 using real-time cell assay screening: efficacy of decitabine and valganciclovir alone and in combination” [1]. These data correspond to the in vitro screening of 2,891 potential antiviral compounds against equid herpesvirus-1 (EHV-1) based on impedance measurements using the xCELLigence® RTCA MP System. This dataset includes compounds from three different libraries: i) 1,199 compounds from the Prestwick® Chemical Library, which contains mostly US Food and Drug Administration approved drugs (Prestwick® Chemical, Illkirch, France); ii) 1,651 compounds from the Centre d'Etudes et de Recherche sur le Médicament de Normandie (CERMN, Caen, France); iii) 41 compounds (called herein in-house antiviral library) selected for their effects against different human viruses. Compounds effective against EHV-1 were selected using the area under normalised curves (AUCn) and the time required for the Cell Index to decrease by 50% after virus infection (CIT50). The full dataset from the screen is made publicly available for further analyses.


706
Values determined using MacSynergy II software (Prichard and Shipman, 1990)        of VGCV and DTB at 1:1 ratio using qPCR assay. Combination Index (CI) was calculated 770 using the Chou and Talalay equation (Chou & Talalay, 1984). CI < 1, CI = 1 and CI > 1 771 indicate synergism, additive and antagonism, respectively. The weighted CI is calculated as 772  In the paper 'Identification of antiviral compounds against equid herpesvirus-1 (EHV-1) using real-time cell assay screening: efficacy of decitabine and valganciclovir alone or in combination', Thieulent at al. have tested a huge battery of compounds that may have an antiviral effect against EHV-1 in three cell types (RK-13, ED, EEK). Since EHV-1 is a champion in evading the immunity, resulting in reproductive and respiratory poblems, antivirals are urgently needed. Therefore, this paper is a very important piece of research.
The work has been performed in an excellent way, with the necessary controls (controlled on cytotoxicity and three different EHV-1 strains). However, there are issues that have to be addressed before it can be published: 1° Introduction -p3, lines 41-50: 'While EHV-1 induced abortion storms have been prevented since the introduction of vaccination three decades ago…' This is not correct. Abortions still occur in correctly vaccinated mares. This has to be adapted.
We agree with the reviewer and "prevented" has now been replaced by "reduced".
2° Introduction -p4, lines 58-60: '…the use of antiviral treatment is sometimes considered to prevent severe forms of EHV-1 induced disease…' This is difficult to follow. What do the authors mean with prevent severe disease? An antiviral will not be used in a preventive way. I think you should mention that one will use it for a treatment.
We agree with the reviewer and "to prevent severe forms of EHV-1 induced disease" has now been replaced by "for the treatment of EHV-1 induced disease". Indeed, we have to be more critical on results obtained on continuous cell lines. This is now clearly discussed lines 377-384 (Nevertheless, our results should be considered only as a first step in this direction as the predictive value of in vitro models based on continuous cell lines is questionable. For example, the antiviral effect of aciclovir observed in vitro on cell lines infected by EHV-1 was not validated ex vivo when using respiratory mucosa explants as a model (Glorieux et al., 2012) and was not transposable in vivo using valaciclovir, the prodrug of aciclovir (Garre et al., 2009). Although data observed in this study are promising, further investigations are necessary on ex vivo models to confirm the antiviral effect of our lead compounds before in vivo experiments implementation.).

Reviewer #2:
The manuscript by Thieulent et al presents data on antiviral effect of 8 compounds from preselected effective 22 compounds that were selected from a screening of 2,897 compounds against Equid herpesvirus 1 (strain Kentucky D). Efficiency of antiviral activity was determined by real time cell analysis (RTCA) using the measurement of cellular "impedance" in cell culture for 96h, compared to a non-infected control. The manuscript focuses on decitabine as a novel anti-viral compound that could be used in synergy with valgancyclovir against EHV-1 infection. The authors present interesting though not fully novel data, and needs major revision before publication.
The manuscript relies on a pre-screening using RTCA methodology but there is no data showing any result from that screening of 2,897 compounds.
We agree with the reviewer remark. The data showing the results of the screening of 2,891 compounds are now presented graphically in supplementary figure 1 and detail of this screening is presented in the associated "Data in Brief" article (Screening of potential antiviral molecules against equid herpesvirus-1 using cellular impedance measurement: dataset of 2,891 compounds, C. Thieulent) as suggested by the editor.

2.
How robust is the RTCA IC50 measurements compared to IC50 measured by qPCR? The authors should show the RTCA results for the 8 selected compounds and qPCR actual measurements rather than just IC50 values.
We have realised a Spearman correlation between EC50 values obtained by RTCA and qPCR for the 16 compounds presented in Table 1 with EC50s<50M ( Figure 1). The R factor of 0.841 and P-value <0.001 now presented lines 221-223 show a good correlation between EC50 values obtained by qPCR and RTCA. These data show the robustness of RTCA EC50 values. In addition, dose-responses curves obtained by RTCA and qPCR measurements for the 8 selected compounds were added in Figure 2 (Figure 1 in the previous version).
3. Figure 2 shows results of potential synergism between valganciclovir and decitabine, using two algorithm approaches. However, experiment in Fig2C is not clear. It seems cells were incubated with both VGCV and DTB compounds but no condition of single treatment was performed as control. If so, no conclusion can be made on any synergistic effect. This is not clear. The authors should provide in one experiment single-treatments and dual treatments.
Previous figure 2 is now figure 4. The figure 4C is obtained after Chou-Talalay method analysis using CompuSyn software. This method requires all the single treatment values associated with dual-treatments in order to calculate combination indexes (CIs) presented on the figure. We agree with the reviewer that the Chou-Talalay analysis and other procedures to assess synergistic effects were not detailed. This is due to space limitations and because these methods were previously described in literature. However and to address Reviewer's comment, please find attached the reports of the 3 experiments of combination using CompuSyn software (Experiment_01.pdf, Experiment_02.pdf and Experiment_03.pdf). Results allowing to obtained data in figure 4C are highlighted in yellow at the bottom of page 5 in the reports. Figure 4 compares DTB with RG108 (a methyl transferase inhibitor) by RTCA and qPCR. This experiment does not include any control (mock-infected and mock-treated). They should be included. Antiviral effect of DTB and RG108 are calculated using the percentage of inhibition based on results of mock-infected and mock-treated cells. The formula is now clearly presented in part 2.4 (lines 157-161; Percentage of inhibition was calculated using the following formula:

4.
, where a corresponds to value of infected cells treated with different concentration of compounds, whereas b and c correspond to values obtained for mock-infected and mock-treated cells, respectively). Figure 5 demonstrates that decitabine is a deoxycytidine analog and that the antiviral effect of this compounds is antagonized by deoxycytidine in excess. Although this information is important, it is an obvious hypothesis. Moreover, the abstract states that decitabine needs to be phosphorylated by DCK in order to be active against EHV-1. This is an overstatement. Although the authors refer to a published study showing the requirement of DCK-mediated phosphorylation of decitabine to be incorporated in cellular DNA, there is no data in the current manuscript supporting that such mechanism is required for antiviral effect against EHV-1. In the same line, the graphical abstract is not appropriately reflecting the results of the study. There is no result supporting that dC directly inhibits DCK-mediated phosphorylation of decitabine. The only section where a 'competition' mechanism is mentioned is in the discussion P16. This should be clarified to avoid misinterpretation of the data.

5.
We agree with the reviewer that information presented in previous figure 5 (now Figure 7) is important. To our knowledge, this is the first demonstration that deoxycytidine can antagonize the antiviral effect of DTB and considering that DTB could act through off-target mechanisms, this result is not obvious. Our observation leads to the conclusion that dC is competing with DTB either at the phosphorylation step by DCK which is required for both dC and DTB incorporation in DNA, or a the level of DNA incorporation per se.
This was clarified by modifying several sentences to avoid misinterpretation: -In abstract, "this study demonstrated that decitabine needs to be phosphorylated by deoxycytidine kinase in order to be active against EHV-1" has now been replaced by "this study suggests that decitabine needs to be phosphorylated by deoxycytidine kinase in order to be active against EHV-1". -In graphical abstract, we have removed the indication that deoxyxytidine (dC) inhibited DTB effect. We suggest now a competition effect between dC and DTB for DCK as mentioned in discussion P16.

6.
P12: the authors conclude that decitabine has a 'cell-protective' effect against EHV-1; but they show that deoxycytidine in excess reverses the antiviral effect. Thus, the effect is only potentially cell protective because EHV-1 replication is impaired and does not protect the cells on its own.
We agree with the reviewer and "cell-protective" has now been replaced by "antiviral effect" (line 299).

7.
The authors do not discuss hypotheses related to the efficacy of decitabine as an analog of dC to inhibit EHV-1 replication in culture, and why dC analogs have apparently not been used against other herpesviruses. This is now discussed in the discussion section.
8. There is no discussion related to the problematic of the absence of effect against latencyassociated virus. It is difficult to discusses the problematic of latency and antiviral treatment because of the absence of pertinent model for EHV-1.

Minor comments: -
Tables: the Excel format is not easily accessible, with footnotes being in the main text and structural information of compounds barely readable.
Structural conformations of compounds in table 1 were re-drawing with Chemdraw and are now more readable.
-P10: "RK" in "RK13" cells stands for "rabbit kidney". These cells are thus not rodent cells as rabbits are not rodents.
This has been corrected.
- Table 2, 3 and 4 would be much more accessible if graphed as bars +/-SD.
We agree with the reviewer and table 2 and 3 are now graphed as bars +/-SD in the new figure 3A and 3B. However, we proposed to maintain in a table form the previous presentation of "Combination analysis of compounds against EHV-1 KyD strain on E. Derm cells" (now table 2).
-P10: how do the authors explain the difference of efficacy in EEK and RK13 compared to E.Derm cells?
The difference of efficacy in different cell models is difficult to access for the different drugs. We proposed lines 355-358 an explanation for DTB and GTB. This point is also discussed in line 377-384.
- Table 4: Additive rather than Additif This has been corrected.

-
The discussion is of poor quality with successive repetitions of the results rather than actual discussion, interpretation and discussion of them.
In concordance with reviewers' suggestions, the discussion has been carefully revised.

Highlights
-Real-time cell assay screening of 2,8917 compounds was conducted against EHV-1.
-Decitabine phosphorylation by deoxycytidine kinase is required to be active. 1

Title:
Identification of antiviral compounds against equid herpesvirus-1 using real-time cell assay screening: efficacy of decitabine and valganciclovir alone or in combination.

List of authors:
Côme Thieulent 1,2 , Erika Hue 1,2,3 , Gabrielle Sutton 1,2 , Christine Fortier 1,2,3 , Patrick Despite preventive treatments with vaccines, resurgence of EHV-1 infection still constitutes a major threat to equine industry. However, no antiviral compound is available to treat infected horses. In this study, 2,891 compounds were screened against EHV-1 using impedance measurement. 22 compounds have been found to be effective in vitro against EHV-1.
Valganciclovir, ganciclovir, decitabine, aphidicolin, idoxuridine and pritelivir (BAY 57-1293) are the most effective compounds identified, and their antiviral potency was further assessed on E. Derm, RK13 and EEK cells and against 3 different field strains of EHV-1 (ORF30 2254A/G/C). We also provide evidences of synergistic interactions between valganciclovir and decitabine in our in vitro antiviral assay as determined by MacSynergy II, isobologramm and Chou-Talalay methods. Finally, we showed that deoxycytidine reverts the antiviral effect of decitabine, thus supporting some competition at the level of nucleoside phosphorylation by deoxycytidine kinase and/or DNA synthesis. Deoxycitidine analogues, like decitabine, is a

Introduction
Herpesviruses (order Herpesvirales, family Herpesviridae) are enveloped viruses with a linear, double-stranded DNA genome of 125-290 kb. Among the five equid herpesviruses (EHV-1 to 5) frequently isolated in horses, EHV-1 is the most pathogenic and is endemic worldwide. EHV-1 infection in horses is associated with several clinical signs of disease, from usually mild respiratory distress, cough and discharge, to more severe secondary forms of diseases such as abortion, neonatal foal death and equine herpes myeloencephalopathy (EHM) (Allen, 2002). The prevalence of latent EHV-1 is estimated to be greater than 60% in horse population (Lunn et al., 2009).
Several vaccines are available against EHV-1. Their use reduces clinical signs of respiratory disease and virus shedding, which limits the extent of outbreaks. However, the protection provided against the secondary forms of the disease presents some limitations. While EHV-1 induced abortion storms have been reduced since the introduction of vaccination three decades ago, none of the commercially available EHV-1 vaccines have demonstrated its efficacy to prevent EHM. EHV-1 vaccine coverage is often too low to provide effective herd immunity. In this context, outbreaks still occur worldwide in horse populations. A recent outbreak reported in France in 2018 (Sutton et al., 2019), led to the cancellation of more than To complement prevention measures, such as vaccination and biosecurity, the use of antiviral treatment is sometimes considered for the treatment of severe EHV-1 diseases, especially EHM. The occasional use of aciclovir during EHV-1 outbreaks has been reported (Friday et al., 2000;Henninger et al., 2007;Murray et al., 1998) but the therapeutic efficacy of this compound is difficult to assess in the absence of untreated animals as a control. Two experimental infections in horses treated with valaciclovir, an aciclovir pro-drug, have also shown divergent results (Garre et al., 2009;Maxwell et al., 2008).
EHV-1 is an alphaherpesvirus genetically closely related to herpes simplex virus type 1 (HHV-1) and varicella zoster virus (HHV-3) for which antiviral therapies are available.
However, the emergence of human herpesvirus strains resistant to antiviral treatments, such as aciclovir, has motivated researches for new antiviral therapies (Jiang et al., 2016) and helicase primase inhibitors seem to be good candidates (James et al., 2015;Kleymann et al., 2002).
Over the last two decades, drug repositioning has proven to be an effective strategy to meet therapeutic needs with nearly a hundred drugs repositioned since (Jourdan et al., 2020). Even in absence of approved EHV-1 antiviral treatment for practitioners, few antiviral molecules have been studied against EHV-1 in vitro and correspond to those already used in human medicine against herpesviruses such as aciclovir, ganciclovir, cidofovir and penciclovir (Maxwell, 2017;Vissani et al., 2016). Other compounds such as aphidicolin (Goodman et al., 2007), A-5021 (Glorieux et al., 2012), quercetin (Ferreira et al., 2018;Gravina et al., 2011) and the histone demethylase inhibitor OG-L002 (Tallmadge et al., 2018) have been studied against EHV-1 in different cell culture models. However, these molecules have never been tested in a standardised cellular model allowing proper comparisons.
We have recently developed a standardised Real-Time Cell Analysis (RTCA) model for evaluating the effect of antiviral compounds against EHV-1. This system relies on the 5 measurement of cellular impedance in culture wells, which reflects cellular adhesion and proliferation. Results are expressed as Cell Index (CI) that enables a standardised and accurate analysis of EHV-1 cytopathic effects. This system has proven successful to determine the efficacy of molecules against EHV-1 such as spironolactone (Thieulent et al., 2019). In the present work, a chemical library of 2,891 compounds comprising new chemical entities and FDA-approved drugs has been screened by impedancemetry to identify compounds against the EHV-1 Kentucky D (KyD) reference strain. As some associations between a DNA polymerase (ORF30) genotype (G/A at position 2254) and the type of disease have been reported by several studies (Goodman et al., 2007;Lunn et al., 2009;Nugent et al., 2006;Pronost et al., 2010), active molecules identified were subsequently tested against a panel of EHV-1 strains (A2254 or G2254), including the newly identified EHV-1 ORF30 variant (C2254) (Paillot et al., 2020). Decitabine was one of the most effective molecules identified, and the mode of action of this cytidine analogue has been further investigated.

EHV-1 strains
The EHV-1 Kentucky D (KyD) strain (ATCC ® VR700™) was used as the EHV-1 reference strain for compound screening and subsequently to confirm the antiviral effect of selected hits in the different cell lines. In addition, three French EHV-1 strains were also used in this study,  Table 1). RG108 (MedChemExpress) was dissolved at 20 mM in DMSO. All compounds were stored at -20°C before used.

Screening of compound libraries using the RTCA system
The screening by impedancemetry was performed with EHV-1 KyD-infected E. Derm cells using the RTCA MP system (ACEA Biosciences Inc., San Diego, CA, USA) as previously described (Thieulent et al., 2019). Control cells were treated with 0.5% DMSO in presence or absence of the virus. The screening was performed under blind conditions and 80 compounds were tested by plate at a final concentration of 10 µg/mL (Preswitck ® Chemical Library), 10 µM (CERMN library) or 50, 10, 2 and 0.4 µM (in-house antiviral library) in 0.5% DMSO.
Each plate includes the controls required for calculation of the Z'-factor (Zhang et al., 1999).
Only plates with a Z' factor upper than 0.5 were considered for further analysis as previously described by Thieulent et al. (2019). For each compound, the area under normalised Cell Index (CI) curves was calculated from 0 to 96 hours post-infection (hpi) (AUCn; (Pan et al., 2013). The time required for the CI to decrease by 50% after virus infection was also determined (CIT50; (Fang et al., 2011), and compared with controls. All the details are presented in Data in Brief (Thieulent et al., submitted). Any increase in these two parameters reflects some protection of E. Derm cells from EHV-1 induced cytopathic effects. The cut-off determined for a molecule to be considered with an antiviral potential were (i) the AUCn increasing by 25%, and (ii) the CIT50 being delayed by >8 h as compared to non-treated cells (Thieulent et al., 2019).
After the screening, dose-response curves were obtained for each selected compound by using percentage of inhibition calculation. The following formula was used: Inhibition (%) = 100 ×

Toxicity measurement
Cells were seeded in white opaque 96-well plates and after 24 h of culture, were treated with compounds. Cell viability was measured at 48 h post-treatment (hpt) by impedancemetry and ATP measurement using the CellTiter Glo ® Luminescent Cell Viability Kit (CTG; Promega, Charbonnière-les-bains, France), according to the manufacturer's instructions. Luminescence signal was acquired using an Infinite ® M200 luminometer (Tecan, Lyon, France).

Research of synergistic effects between compounds against EHV-1
Drug combinations were tested on EHV-1 KyD-infected E. Derm cells using impedancemetry as a read out. For each combination, the two selected drugs were prepared separately by 2fold serial dilution and mixed in 96-well plates to create an 8 by 10 matrix of single and combined diluted drugs. For each compound, the dilution range was designed to have the EC50 in the middle of the range, and the highest concentration inferior to the EC90. In each plate, infected and non-infected cells with 1% DMSO were used as positive and negative controls, respectively. Synergistic or antagonistic effects were determined with the MacSynergy II program using first the Bliss independence model (Prichard and Shipman, 1990) applied on AUCn values. This software calculated the volume of synergy/antagonism produced by the drug combination in a 95% confidence interval. Volumes were given as the area under a dose-response curve in the two dimensional situation (µM² %) and interpretation was made as previously described by Prichard et al. (1990). Values of 0-25, 25-50, 50-100, and >100 µM² % in either a positive or negative direction were defined as additive, minor synergy or antagonism, moderate synergy or antagonism, and strong synergy or antagonism, respectively .   1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60  61  62  63  64  65 Isobologram analysis and the Chou-Talalay method using the Loewe additivity model were used to confirm synergistic effect firstly observed by MacSynergy II program. Isobolograms were built as previously described by Feng et al. (2009) from EC50 values obtained by impedance measurement. The Chou-Talalay method is based on the median-effect equation and computed by CompuSyn software version 1.0 (ComboSyn, Inc., Paramus, New Jersey) (Chou and Talalay, 1984). The software extrapolated a combination index representing the interaction between two drugs from the percentage of inhibition of log10 viral genome copies number produce in presence of each drug alone and in combination. The weighted average combination index (CIwt) value was calculated as previously described (Drouot et al., 2016).

Statistical analysis
EC50 and CC50 values were calculated using a non-linear regression dose response inhibition curve (GraphPad Prism ® software 6.0; La Jolla, CA, USA). The Selectivity Index (SI) was determined for each compound using the following formula: SI = CC50/EC50. Spearman's correlation test and values comparison using ANOVA with Tukey post hoc test were both statistically evaluated using GraphPad Prism ® software.
These molecules were then evaluated in dose-response assays on E. Derm cells by two-fold serial dilution (50 to 0.1 µM) by qPCR assay and impedance measurement. Antiviral properties were confirmed for 22 out of the 25 compounds (Table 1), and EC50 values were precisely determined by both qPCR and RTCA for 16 of them (EC50s<50M). The R factor of 0.841 and P-value <0.001 obtained by Spearman's correlation showed a good correlation between EC50 values obtained by qPCR and RTCA for these 16 molecules (Figure 1). Among these compounds, eight were selected as they complied with more stringent criteria: (i) the absence of toxicity on E. Derm cells at all the concentrations tested (CC50 > 50µM) and (ii) EC50 values below 50 µM at all time points between 48 and 120 hpi (Figure 2A). Doseresponses curves obtained by qPCR and RTCA at 48 hpi for the eight compounds are presented in Figure 2B. Three of these compounds were acyclic guanosine analogues (aciclovir, ACV; ganciclovir, GCV; and ganciclovir prodrug valganciclovir, VGCV) inhibiting the viral DNA polymerase, two were deoxycytidine analogues (decitabine, DTB; gemcitabine, GTB) used for tumor growth inhibition by incorporation in cellular DNA, and one was a deoxyuridine analogue (idoxuridine, IDU) targeting viral DNA synthesis.

Research of synergistic effect and antiviral activity of the valganciclovir/decitabine
combination Dual-combinations were tested between valganciclovir, one of the best candidates, and four other compounds (aphidicolin, pritelivir, decitabine and idoxuridine) for synergistic or antagonistic effects against EHV-1. Using MacSynergy II analysis, only the valganciclovir/decitabine combination showed a synergistic effect that is illustrated by the strong signal above additive effects in the matrix of drug interactions ( Figure 4A). The synergy volume of 63.24 µM² % obtained supports a moderate synergy ( in combination at a 1:1 ratio. A synergistic effect was observed for the valganciclovir/decitabine combination as assessed by a weighted average combination index (CIwt) of 0.20 ( Figure 4C). The three other combinations (valganciclovir/aphidicolin, valganciclovir/ pritelivir, valganciclovir/ idoxuridine) tested were additive when measured by MacSynergy II method (Table 2) and were not tested by isobologram nor median-effect analysis. No cytotoxicity was observed at the maximal drug combinations tested for the four different combinations (Supplementary Figure 2).

Decitabine pre-treatment did not confer cell resistance to EHV-1 replication.
Although valganciclovir was developed as an antiviral against herpesviruses in the first place, this is not the case of decitabine. Indeed, decitabine is an anticancer agent which induces hypomethylation after integration in cellular DNA (Liu et al., 2007). To evaluate whether decitabine integration in target cell DNA provides protection from EHV-1 infection, cells were treated overnight with decitabine before infection and/or just after infection. Both results obtained by cell impedance measurement ( Figure 5A) and EHV-1 viral load measurement ( Figure 5B) showed that decitabine pre-treatment did not protect cells from CPE formation and virus replication. A post-infection treatment with decitabine was required to observe some significant inhibition of EHV-1 replication. The effect of RG108, another well know DNA methyltransferase inhibitor, was then tested against EHV-1 on E. Derm cells. RG108 did not show any antiviral effect when assessed by impedance measurement ( Figure 6A) or virus load quantitation ( Figure 6B). Altogether, this suggests that cellular DNA hypomethylation does not account for the inhibition of EHV-1 by decitabine.

Deoxycytidine competitively inhibits the antiviral effect of decitabine
Decitabine is a deoxycytidine analogue and a pro-drug that must be successively phosphorylated by components of the deoxyribonucleoside salvage pathway involving deoxycytidine kinase (DCK), CMP monophosphate kinase (in particular CMPK1) and 13 nucleotide diphosphate kinases (Momparler, 2005;Stresemann and Lyko, 2008). It has been shown that high levels of deoxycytidine (dC) can reverse the anticancer activity of gemcitabine, another dC analogue, by competition for DCK-mediated phosphorylation, the rate-limiting step in dC phosphorylation to dCTP (Halbrook et al., 2019). We thus tested if dC could similarly reverse the inhibitory effects of decitabine on EHV-1. Infected E. Derm cells were treated with decitabine in the presence of high concentrations of dC or other nucleosides including cytidine, uridine, adenosine, guanosine ( Figure 7A). Of all tested nucleosides, only dC blocked the antiviral activity of decitabine. This result was confirmed by microscopic observations and impedancemetry as dC reversed the antiviral effect of decitabine against EHV-1 ( Figure 7B and 7C). Altogether, these results demonstrate a competition of decitabine and dC for the same metabolic pathway in our in vitro infection model.

Discussion
In this study, 2,891 compounds were screened against EHV-1 by impedancemetry as previously described (Thieulent et al., 2019), and 22 compounds were identified for their antiviral properties against this virus. AUCn values coupled to CIT50 calculation were the two major criteria for filtering raw data and identify hits. The antiviral effect of selected compounds was confirmed by dose-response assay using both impedancemetry and viral load quantitation. As the readouts were not the same, EC50 values obtained with these two methods differed as previously reported Thieulent et al., 2019). However, the good correlation observed between EC50 values obtained by RTCA and qPCR demonstrates the pertinence of RTCA for antiviral evaluation. Among the 22 compounds inhibiting EHV-1, eight molecules were selected for further evaluations using stringent criteria, including EC50 values below 50 µM over time and lack of cytotoxicity when used at 50 µM.
Ganciclovir and aciclovir are approved medications to treat herpesviruses and were previously shown to be effective against EHV-1 in vitro (Garre et al., 2007;Thieulent et al., 2019). Valganciclovir, the pro-drug and valine ester of ganciclovir, presents here an antiviral activity against EHV-1 similar to ganciclovir. Pritelivir is another antiviral drug developed to treat herpesviruses. It is an inhibitor of the helicase-primase complex of herpesviruses discovered in 2002 . The antiviral effect of pritelivir was previously reported against HHV-1 and HHV-2 . However, this study is the first demonstrating the antiviral effect of this molecule against equid herpesviruses and in particular against EHV-1. It would be interesting to evaluate the antiviral effect of these compounds against other equid herpesviruses, such as EHV-3. Idoxuridine is also a well-known antiviral compound against human herpesviruses such as HHV-1 and HHV-2, and is also active against different animal herpesviruses such as feline herpesvirus type-1 (De Clercq and Li, 2016;Maggs and Clarke, 2004). Interestingly, idoxuridine is one of the three deoxyuridine analogues, together with brivudine and trifluridine, which have been used for decades against herpes simplex viruses (De Clercq and Li, 2016). Idoxuridine and trifluridine have showed a good efficacy against EHV-1 without toxicity in our cellular model, whereas brivudine was inactive (data not shown). Brivudine is the only one that needs to be specifically phosphorylated by viral thymidine kinase (TK) to become active (De Clercq and Li, 2016), suggesting that EHV-1 TK is unable to phosphorylate brivudine, which is in line with previous reports (De Clercq, 1984;Kit et al., 1987). Maribavir is also a new nucleoside analogue in development against human cytomegalovirus (HHV-5, a betaherpesvirus) that was tested in our screen and was inactive against EHV-1 (data not show) (Price and Prichard, 2011). This result is in line with the lack of activity against the alphaherpesviruses HHV-1, HHV-2 and HHV-3 (Williams et al., 2003).
The antiviral activity of the 8 most efficient molecules was also validated in three cell lines and against different strains of EHV-1. E. Derm cells and EEK cells are both equine cell lines and most adapted to identify new antiviral compounds in equid species, especially EEK that was derived from a horse foetus that is one of the target of EHV-1 (Léon et al., 2008;Smith et al., 2010). Even though RK13 cells are not equine cells, they have been most frequently used in EHV1 antiviral studies (Azab et al., 2010;de la Fuente et al., 1992;Gibson, 1992;Rollinson, 1987). All compounds except aciclovir and gemcitabine showed some consistent antiviral activity in the three different cell lines. Aciclovir is the least active of the eight selected compounds, and was inactive when used with EEK cells. More surprisingly, although gemcitabine is very effective on E. Derm cells, it has no antiviral activity on RK13 and EEK cells. This suggests that gemcitabine is not properly phosphorylated by RK13 and EEK cells kinases. In line with this hypothesis, decitabine, which is structurally very close to gemcitabine and also needs to be phosphorylated, is less active on EEK and RK13 cells.  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60  61  62  63  64  65 previous reports comparing the susceptibility of A2254 and G2254 strains (Garre et al., 2007;Thieulent et al., 2019). Only aciclovir, pritelivir and idoxuridine were slightly less efficient on the FR-38991 (A2254) strain. No difference of susceptibility was observed between the three strains for aphidicolin treatment. This result differs from a previous report showing that a strain with the G2254 genotype is more sensitive to aphidicolin than a strain with the A2254 genotype (Goodman et al., 2007).
In this study, ganciclovir and its prodrug valganciclovir are the most effective compounds in vitro against EHV-1. Interestingly, the pharmacokinetic of valganciclovir was previously studied in horse (Carmichael et al., 2013), showing 40% bioavailability after oral administration. This positions valganciclovir as the best candidate in our short list of active molecules for treating horses infected by EHV-1 even if the cost of the molecule could be a limitation. Nevertheless, our results should be considered only as a first step in this direction as the predictive value of in vitro models based on continuous cell lines is questionable. For example, the antiviral effect of aciclovir observed in vitro on cell lines infected by EHV-1 was not validated ex vivo when using respiratory mucosa explants as a model (Glorieux et al., 2012) and was not transposable in vivo using valaciclovir, the prodrug of aciclovir (Garre et al., 2009). Although data observed in this study are promising, further investigations are necessary on ex vivo models to confirm the antiviral effect of our lead compounds before in vivo experiments implementation.
This report is also the first analysis of drug combinations against EHV-1. Of all the combinations tested, only valganciclovir plus decitabine showed a synergic effect. It was previously demonstrated that gemcitabine in association with ganciclovir enhanced the antitumoral effects of HSV-TK suicide gene in a synergistic manner (Wang et al., 2016). This result is in agreement with our study .   1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60  61  62  63  64  65 To our knowledge, decitabine has never before been reported to inhibit cellular infections by a herpesvirus. This led us to further investigate the mode of action of this compound. The pretreatment of cells with decitabine did not provide antiviral effects against EHV-1, suggesting that decitabine incorporation in cellular DNA did not mediate the antiviral effect of decitabine. This rather suggests that the antiviral effect of decitabine depend on its incorporation into viral DNA and/or some interference with the viral polymerase. Decitabine is well-known for preventing DNA methylation and this account for its antitumoral properties (Atallah et al., 2007;Schmelz et al., 2005). We thus tested the antiviral effect of RG108, which is a hypomethylation agent acting differently through DNA methyltransferase inhibition. The absence of RG108 activity against EHV-1 suggests that the inhibitory effect of decitabine against EHV-1 is not mediated by viral or cellular DNA hypomethylation. Finally, we showed that the addition of dC to culture medium inhibits decitabine antiviral activity.
This strongly suggests that dC and decitabine channel through the same metabolic pathway, including phosphorylation-dependent activation and incorporation into cellular and viral DNA. Based on collected observations, we propose that decitabine is integrated into the EHV-1 DNA and/or jams the viral polymerase, thus leading to the inhibition of viral growth.
Interestingly, and as opposed to ganciclovir that needs activation by viral TK (Sullivan et al., 1992), decitabine phosphorylation only relies on cellular kinases to be activated and may represent a good candidate against viral strains resistant to ganciclovir.

Supplementary Figure 2:
Cytotoxicity assay of combinations on equine dermal cells.