Antiviral Screen against Canine Distemper Virus-Induced Membrane Fusion Activity

Canine distemper virus (CDV), a close relative of the human pathogen measles virus (MeV), is an enveloped, negative sense RNA virus that belongs to the genus Morbillivirus and causes severe diseases in dogs and other carnivores. Although the vaccination is available as a preventive measure against the disease, the occasional vaccination failure highlights the importance of therapeutic alternatives such as antivirals against CDV. The morbilliviral cell entry system relies on two interacting envelope glycoproteins: the attachment (H) and fusion (F) proteins. Here, to potentially discover novel entry inhibitors targeting CDV H, F and/or the cognate receptor: signaling lymphocyte activation molecule (SLAM) proteins, we designed a quantitative cell-based fusion assay that matched high-throughput screening (HTS) settings. By screening two libraries of small molecule compounds, we successfully identified two membrane fusion inhibitors (F2736-3056 and F2261-0043). Although both inhibitors exhibited similarities in structure and potency with the small molecule compound 3G (an AS-48 class morbilliviral F-protein inhibitor), F2736-3056 displayed improved efficacy in blocking fusion activity when a 3G-escape variant was employed. Altogether, we present a cell-based fusion assay that can be utilized not only to discover antiviral agents against CDV but also to dissect the mechanism of morbilliviral-mediated cell-binding and cell-to-cell fusion activity.


Introduction
Canine distemper is a highly infectious disease caused by the canine distemper virus (CDV), a member of the family Paramyxoviridae and closely related to the measles virus (MeV) human pathogen. CDV is a negative sense single stranded RNA virus that is present worldwide and has a wide host range within the order Carnivora. While domesticated dogs are common hosts, CDV also infects various wild species within various families, namely, Canidae, Hyaenidae, Phocidae, Felidae, Procyonidae, Mustelidae, Ursidae, and Viverridae, thereby threatening the lives of many endangered animals [1][2][3][4][5].
Primarily, CDV infection in dogs is transmitted by inhalation of infectious aerosol droplets but transmission by direct contact with body fluids or fomites is also possible [6]. Post inhalation, CDV establishes the infection by binding a receptor: signaling lymphocyte activation molecule family member 1 (SLAM/F1, or CD150), expressed by dendritic cells (DCs), subsets of thymocytes, macrophages, and T-and B-lymphocytes [7,8]. This facilitates the systemic dissemination whereby the virus interacts with a second receptor: nectin cell Vero-cSLAM cells were cultured in absence of pen-strep). The NanoLuc ® Binary Technology (NanoBiT; Promega, Madison, WI, USA) is a structural complementation reporter system composed of a Large BiT (LgBiT; 18 kDa) subunit and a small complimentary peptide (e.g., HiBiT; high affinity to LgBiT). Vero-Fs-sH-RFP/HiBiT cells were cultured in presence of 10 µM of Asunaprevir.

Stable Transfection
CDV surface glycoprotein (F and H; Onderstepoort strain)-expressing genes, fused (N-terminal: H protein, C-terminal: F protein) to the Small Molecule Assisted Shut-off (SMASh) tag referred to as Fs and sH, respectively [25], were synthetically generated by Eurofin Genomics (Ebersberg, Germany). CDV surface glycoproteins as well as the GFP-LgBiT and RFP-HiBiT [26] constructs were stably expressed in either Vero or Vero-cSLAM cells using the lentivirus technology, as previously described [27]. Vero-cSLAM-GFP/LgBiT and Vero-Fs-sH-RFP/HiBiT cells were cloned by limiting dilution technique. Vero-cSLAM were transfected with TransIT-LT1 (Mirus) by following the manufacturer's instructions.

Compound Library
Two different commercially available libraries, (A) protein-protein interaction (PPI) library of around 5400 compounds and (B) chemical diverse collection of around 8000 compounds, were tested. Molecules were selected by the Biomolecular Screening Facility (BSF, EPFL, Lausanne, Switzerland) and were purchased from three different suppliers: Enamine, Life chemicals and ChemDiv. The entire process of screening was done at the biomolecular screening facility (BSF): a multidisciplinary technological platform at the EPFL for performing the high-throughput screening in life sciences-related projects.

Nanoluciferase (nLuc)-Based Assay for High-Throughput Screening
Using an ECHO550 liquid handler (Beckman Coulter), 40 nL of the stock compound (10 mM in 100% DMSO) was added to the Greiner bio-one 384-wells plate (40 µL final volume, final concentration 10 µM). Then, 100 µL (containing about 17,000 cells) of each cell suspension (Vero/cSLAM-GFP/LgBiT and Vero-Fs-sH-RFP/HiBiT) was seeded to the wells using a Multidrop Combi Reagent Dispenser (Thermo Scientific, Waltham, MA, USA) and incubated at 37 • C for 24 h. After the incubation, corresponding substrate (Nano-Glo ® Live Cell Assay System, Promega) was added to the plate using Biotek Microflo Select Dispenser (Agilent) and the total amount of luminescence was measured by Synergy NEO HTS Multi-Mode Microplate Reader (Biotek). Afterwards, all the statistical analysis was done by the BSF using the internal Laboratory Information Management System (LIMS).

CDV Inhibition Assay
Using a reverse genetic system for CDV (Onderstepoort (OP) strain; manuscript in preparation), the nLucP gene (encoding for the nanoluciferase fused to the "Pest" degradation motif; Promega) was inserted as an additional transcription unit in between the P and M genes of CDV-OP, which additionally contained the mNeonGreen gene (Neon) N-terminally attached to the CDV-OP N gene via a P2A motif (manuscript in preparation). Both viruses were successfully rescued and referred to as OP neon and OP neon/nLucP , respectively. OP neon/nLucP was used in all viral inhibition assays. A wild-type CDV virus (strain A75/17) was rescued, as previously described [28]. The latter recombinant virus harbors the Neon-P2A-nLucP cassette as an additional transcription unit inserted in between the P and M genes, and referred to as A75/17 neon/nLuP . Desired inhibitors, dissolved in DMSO, were added in 96 wells plate (Greiner bio-one) starting at 100 nM with increasing concentration of half a log up to 100 µM. OP neon/nLucP was then added to the plates at a multiplicity of infection (MOI) of 0.04 and incubated at 4 • C for 1 h. Mixtures were then added to Vero-cSLAM cells pre-seeded in a separate 96 wells plate and incubated at 37 • C for 24 h. The effect of the inhibitors was evaluated by measuring the amount of luminescence emitted by OP neon/nLucP -infected Vero-cSLAM cells using a multiplate reader Viruses 2021, 13, 128 4 of 13 (Cytation 5, Biotek, Winooski, VT, USA). After the measurement, all statistical analysis was carried out using the GraphPad Prism 8 package.

Viability Assay
In order to investigate the viability of cells in presence of hit compound, desired compounds were added in 96 wells plate (Greiner bio-one), similar to inhibition assay, starting at 100 nM with increasing concentration of half a log up to 100 µM. MT cell viability substrate and NanoLuc ® enzyme (RealTime-Glo™ MT Cell Viability Assay; Promega) were added to the Vero-cSLAM cell suspension, according to the manufacturer's instruction. After, 15,000 Vero-cSLAM cells were added per well in the plate containing compounds and incubated at 37 • C for 24 h. The total amount of luminescence was measured using a multiplate reader (Cytation 5, Biotek). After the measurement, all statistical analysis was carried out using the GraphPad Prism 8 package.

In Silico Studies-Molecular Docking with the F-Protein
The F-protein was prepared for the docking studies using the molecular operating environment (MOE) via the QuickPrep option applying the default values for temperature 300 K, pH 7 and salt 0.1. The ligands F2261-0043 and F2736-3056 were imported and 3D coordinates for the ligands were generated directly in MOE using energy minimize with a root mean square deviation (RMSD) gradient of 0.1. The Amber10: EHT force field was applied for the minimizing and docking experiments. We used the "triangle matcher" as placement methods with "London dG" scoring and the "rigid receptor" mode for the refinement with the "GBVI/WSA dG" scoring. Water molecules were not included in the calculation.

Establishment of a Quantitative Cell-Based Fusion Assay
Provided that previous screens involving reporter protein-expressing recombinant live-viruses resulted in the discovery of anti-F or anti-L inhibitors [24], we investigated the possibility to increase the chances to identify novel antivirals by designing a more viral protein-targeted bioassay. We therefore established a quantitative cell-based fusion assay (potentially targeting viral cell entry and spread), which relied on the stable expression of the H, F and SLAM glycoproteins. Two cell lines were generated, namely effector and target cells. While effector cells (Vero) expressed CDV F-and H-proteins (Onderstepoort strain), target cells expressed the receptor only (Vero-cSLAM). Next, the split nanoluciferase (nLuc) system (Promega) was employed to generate a quantitative assay. Indeed, each part of split nLuc (HiBiT and LgBiT) were introduced into effector and target cells, respectively ( Figure 1A). Moreover, in order to rationally control for CDV F and H expression, both glycoproteins' cytosolic tails were fused to a tag, which allowed for protein expression-control (small molecule-assisted shutoff (SMASh) technology [25]). Briefly, by treating proteinexpressing cells with the small molecule Asunaprevir (ASV), target protein-degradation is induced [25]. In summary, the assay relied on mixing effector and target cells together in absence of ASV, and 24 h later, to record the nanoluciferase (nLuc) activity. Thus, nLuc activity relied on productive F/H/SLAM interactions. Such glycoprotein engagements ultimately led to membrane fusion promotion ( Figure 1B), which in turn triggered cytosolic content-mixing and, finally, nLuc reconstruction.
Addition of the nLuc substrate to such a system resulted in emission of luminescence, which indirectly corresponded to the total amount of membrane fusion. When membrane fusion activity was intervened using either 3G (a known fusion inhibitor; [22]) or ASV, luminescence signal was significantly diminished ( Figure 1C), which verified our quantitative cell-based fusion assay as a very sensitive and convenient system to initiate screens for fusion inhibitors. luminescence signal was significantly diminished ( Figure 1C), which verified our quantitative cell-based fusion assay as a very sensitive and convenient system to initiate screens for fusion inhibitors.

Screening of Small Molecules Using a Quantitative Cell-Based Fusion Assay
Firstly, the newly developed quantitative cell-based fusion assay was validated for its coherence with the high throughput screening (HTS) settings. In order to do that, reference controls (3G; positive control and DMSO; negative control) were assayed first. Using the control data, respective means and standard deviations (SD) were calculated, which in turn were utilized to calculate Z' score. Z' score is a characteristic parameter that evaluates the quality of an assay [29]. In our case, the measured Z' score was about 0.5, which qualified the reliability of the assay. Next, approximately 5400 compounds from the PPI library and 8000 compounds from the CDC library were screened at a concentration of 10 µM. Every compound was tested in duplicates. After the screening, the data was analyzed using LIMS software package. The criteria for "hit" declaration were (a) Effector and target cells were mixed either in presence or absence of either 3G or Asunaprevir (ASV). The asterisk highlights syncytia expressing both the red and green fluorescent proteins. (C) Measurement of membrane fusion between effector and the target cells either in presence or absence of either ASV or 3G by recording luminescence emission. Each column represents replicate values from 3 independent experiments. Asterisks indicate statistical significance between the control and each group, as determined using one way ANOVA with Dunnett's test (**** p < 0.0001). ns: not significant. Scale bars: 200 µm.

Screening of Small Molecules Using a Quantitative Cell-Based Fusion Assay
Firstly, the newly developed quantitative cell-based fusion assay was validated for its coherence with the high throughput screening (HTS) settings. In order to do that, reference controls (3G; positive control and DMSO; negative control) were assayed first. Using the control data, respective means and standard deviations (SD) were calculated, which in turn were utilized to calculate Z' score. Z' score is a characteristic parameter that evaluates the quality of an assay [29]. In our case, the measured Z' score was about 0.5, which qualified the reliability of the assay. Next, approximately 5400 compounds from the PPI library and 8000 compounds from the CDC library were screened at a concentration of 10 µM. Every compound was tested in duplicates. After the screening, the data was analyzed using LIMS software package. The criteria for "hit" declaration were (a) measured value of the compound is above "average + 3X SD" of value of negative control and (b) both duplicates are active.
Overall, using the abovementioned screening pipeline, we identified 60 compounds as "hits" (Figure 2A,B). Out of those compounds, F2736-3056 (identified from CDC library) and F2261-0043 (identified from PPI library) exhibited the highest inhibition profile in the fusion assay (Figure 2A,B). All confirmed hits were finally counterscreened (each small molecule at 10 µM) against a recombinant attenuated Onderstepoort (OP)-CDV that additionally expressed mNeonGreen (OP neon ). Post 24 h of infection of Vero-cSLAM with OP neon , GFP expression was qualitatively analyzed. Out of 60 compounds, F2736-3056 and F2261-0043 successfully inhibited the GFP expression, which indirectly corresponded to Overall, using the abovementioned screening pipeline, we identified 60 compounds as "hits" (Figure 2A,B). Out of those compounds, F2736-3056 (identified from CDC library) and F2261-0043 (identified from PPI library) exhibited the highest inhibition profile in the fusion assay (Figure 2A,B). All confirmed hits were finally counterscreened (each small molecule at 10 µM) against a recombinant attenuated Onderstepoort (OP)-CDV that additionally expressed mNeonGreen (OP neon ). Post 24 h of infection of Vero-cSLAM with OP neon , GFP expression was qualitatively analyzed. Out of 60 compounds, F2736-3056 and F2261-0043 successfully inhibited the GFP expression, which indirectly corresponded to CDV replication ( Figure 2C). Upon comparison, the scaffold of both inhibitors resembled the membrane fusion inhibitor AS-48 class compounds ( Figure 2D).

Potency of the Newly Discovered Fusion Inhibitors
Among the 60 tested compounds, F2736-3056 and F2261-0043 emerged as the most potent inhibitors of CDV infection. In order to accurately determine the efficiency of both small molecule compounds, we compared the bioactivity of these two inhibitors with 3G, which represents one of the most active derivatives of the AS-48 class compounds against CDV [22]. The compounds' inhibition and cytotoxic impacts were assessed in Vero-cSLAM cells infected with OP neon/nLucP (a recombinant virus expressing both the mNeonGreen and the nanoluciferase (nLucP) reporter proteins). While the 50% inhibition concentration (IC 50 ) of compound F2736-3056 and F2261-0043 returned values of about 5 µM and 12 µM ( Figure 3A), respectively, the 50% cytotoxic concentration (CC 50 ) was about 170 µM and 140 µM (Figure 3C), respectively. Overall, while the selectivity index (SI) of compound F2736-3056 and F2261-0043 was about 34 and 12, the SI value for 3G was about 21. Indeed, 3G returned IC 50 values of 7 µM and CC 50 values of 150 µM ( Figure 3A,C). Additionally, the inhibitory activity of the compounds was confirmed against a prototype wild-type CDV strains, namely A75/17, which is genetically distant from the OP-CDV strain. The compounds were also active against A75/17 neon/nLucP , a recombinant wild-type virus expressing both, the mNeonGreen and nLucP reporter proteins ( Figure 3B)     In order to exclude the possibility that the compounds were directly inhibiting the nLuc enzymatic activity, we investigated whether the inhibitors were also affecting the cytopathic effect (CPE) induced by OP neon/nLucP , namely syncytia formation. Strikingly, a clear concentration-dependent decrease in number and size of the syncytia was observed ( Figure 3D).

F2736-3056 Partly Preserved Fusion-Inhibition Activity against a 3G-Escape CDV F-Protein Variant
Next, we tested the activity of F2736-3056 and F2261-0043 against a previously described 3G-escape, hyperfusogenic F-mutant (V447T; valine-to-threonine substitution at CDV F position 447) [27]. Of note, the 3G-escape profile of F-V447T precluded the synchronization of fusion activity necessary to run a quantitative cell-cell fusion assay. To tackle this limitation, we conducted a transient qualitative cell-cell fusion assay and recorded the extent of membrane fusion by light microscopy 24 h post-transfection. For this assay, H and F glycoproteins of Onderstepoort (OP) strain were used. While all three compounds efficiently inhibited cell-to-cell fusion induced by the standard F OP -protein (Figure 4), 3G and F2261-0043 lost their inhibitory activity in presence of F OP -V447T. In contrast, F2736-3056 preserved substantial inhibitory activity towards the hyperfusogenic F OP -V447T mutant, since membrane fusion remained partly inhibited (Figure 4). tackle this limitation, we conducted a transient qualitative cell-cell fusion assay and recorded the extent of membrane fusion by light microscopy 24 h post-transfection. For this assay, H and F glycoproteins of Onderstepoort (OP) strain were used. While all three compounds efficiently inhibited cell-to-cell fusion induced by the standard FOP-protein ( Figure  4), 3G and F2261-0043 lost their inhibitory activity in presence of FOP-V447T. In contrast, F2736-3056 preserved substantial inhibitory activity towards the hyperfusogenic FOP-V447T mutant, since membrane fusion remained partly inhibited (Figure 4). . Impact of the inhibitors was assessed using a previously described cell-to-cell fusion assay [30]. HOP and either FOP or FOP-V447T glycoproteins were co-expressed in Vero-cSLAM cells that were individually treated with DMSO (no compound), 3G, F2261-0043 and F2736-3056. Pictures of representative fields of view illustrating the extent of cell-to-cell fusion are shown for each condition.

In Silico Analysis Suggests a CDV F-Protein Hydrophobic Pocket as a Common Binding Site for Diverse Inhibitory Compounds
The binding mode of two fusion inhibitor families (FIP and AS-48) in MeV F was confirmed by X-ray crystallography [31]. The crystal structure of MeV F and the cryogenic electron microscopy (cryo-EM) structure of CDV F [32], both in the prefusion state, revealed high structural similarities.
The co-crystal structures of MeV F with AS-48 or FIP indicated that both binders dock into the similar hydrophobic region at the contact interface between two F-protomers, which locates at the junction of the head and stalk domains of the prefusion state. More specifically, the high-resolution structures revealed the key binding of a phenyl moiety of both inhibitors in a deep and narrow hydrophobic pocket (formed in proximity to Leu351 and Ile340 in MeV F [31]). In addition, a hydrophobic groove (formed in proximity to Pro241 and Pro339 in MeV F [31]), which extends from the deep pocket, is occupied by hydrophobic groups of AS-48 and FIP. Of note, is that the L-Phe side chain of FIP (absent Figure 4. Enhanced inhibition efficiency of compound F2736-3056 against 3G-resistant F OP -mutant (V447T). Impact of the inhibitors was assessed using a previously described cell-to-cell fusion assay [30]. H OP and either F OP or F OP -V447T glycoproteins were co-expressed in Vero-cSLAM cells that were individually treated with DMSO (no compound), 3G, F2261-0043 and F2736-3056. Pictures of representative fields of view illustrating the extent of cell-to-cell fusion are shown for each condition.

In Silico Analysis Suggests a CDV F-Protein Hydrophobic Pocket as a Common Binding Site for Diverse Inhibitory Compounds
The binding mode of two fusion inhibitor families (FIP and AS-48) in MeV F was confirmed by X-ray crystallography [31]. The crystal structure of MeV F and the cryogenic electron microscopy (cryo-EM) structure of CDV F [32], both in the prefusion state, revealed high structural similarities.
The co-crystal structures of MeV F with AS-48 or FIP indicated that both binders dock into the similar hydrophobic region at the contact interface between two F-protomers, which locates at the junction of the head and stalk domains of the prefusion state. More specifically, the high-resolution structures revealed the key binding of a phenyl moiety of both inhibitors in a deep and narrow hydrophobic pocket (formed in proximity to Leu351 and Ile340 in MeV F [31]). In addition, a hydrophobic groove (formed in proximity to Pro241 and Pro339 in MeV F [31]), which extends from the deep pocket, is occupied by hydrophobic groups of AS-48 and FIP. Of note, is that the L-Phe side chain of FIP (absent in AS-48 and 3G) reaches an additional hydrophobic pocket (near Leu358 and Ala354 in MeV F [31]). This pocket is formed between two helices in the stalk domain.
Due to the lack of CDV F-structures in complex with ligands, we used the co-crystal structure of MeV F with AS-48 (PDB-ID: 5YZC) to investigate a potential docking mode of the two newly identified compounds. Interestingly, the docking models illustrate that F2736-3056 and F2261-0043 bind to the same hydrophobic region within the prefusion F-structure ( Figure 5A,B). The 2-phenylacetanilide scaffold has been investigated in the past with AS-48 [19] and 3G [22] as prominent representatives. The structure activity relationship (SAR) of this scaffold showed a low tolerance for substitutions on the phenylacetic acid moiety. According to this SAR, the para fluoro substitution is the only tolerated substitution on this phenyl ring, which is consistent with the para fluoro substitutions of the newly identified hits F2261-0043 and F2736-3056. The seven-membered ring of F2261-0043 is placed in the shallow groove, which offers enough space for the additional fused ring ( Figure 5B). The benzothiazole moiety of F2736-3056 was placed in the hydrophobic groove as the aniline moiety of AS-48 and the Z-group of FIP. The morpholine moiety is pointing towards the pocket occupied by the L-Phe side chain of FIP. The more polar morpholine is not entering into the hydrophobic pocket but the ether oxygen might form a hydrogen bond to the lysine residue at position 469 (the homologous amino acid in CDV F corresponds to the lysine residue at position 581) ( Figure 5A). Collectively, our in silico approach with the two new hits indicated a potential overlap of the binding region with FIP and AS-48 in the previously experimentally determined co-crystal structures [31].
F2736-3056 and F2261-0043 bind to the same hydrophobic region within the prefusion Fstructure ( Figure 5A,B). The 2-phenylacetanilide scaffold has been investigated in the past with AS-48 [19] and 3G [22] as prominent representatives. The structure activity relationship (SAR) of this scaffold showed a low tolerance for substitutions on the phenylacetic acid moiety. According to this SAR, the para fluoro substitution is the only tolerated substitution on this phenyl ring, which is consistent with the para fluoro substitutions of the newly identified hits F2261-0043 and F2736-3056. The seven-membered ring of F2261-0043 is placed in the shallow groove, which offers enough space for the additional fused ring ( Figure 5B). The benzothiazole moiety of F2736-3056 was placed in the hydrophobic groove as the aniline moiety of AS-48 and the Z-group of FIP. The morpholine moiety is pointing towards the pocket occupied by the L-Phe side chain of FIP. The more polar morpholine is not entering into the hydrophobic pocket but the ether oxygen might form a hydrogen bond to the lysine residue at position 469 (the homologous amino acid in CDV F corresponds to the lysine residue at position 581) ( Figure 5A). Collectively, our in silico approach with the two new hits indicated a potential overlap of the binding region with FIP and AS-48 in the previously experimentally determined co-crystal structures [31].

Discussion
Canine distemper virus is a negative sense ssRNA virus that is well-known to infect and cause fatal diseases in various animals. While initially recognized for causing distemper in domestic dogs, CDV is now well known for infecting wide range of hosts, including mostly but not confined to carnivore [1][2][3][4][5]. CDV is increasingly recognized as a major cause of decline in wild carnivore population [33], which inevitably raises concerns about the extinction threat of many endangered species [34]. In addition, lethal CDV outbreaks have been described in unprecedented species such as non-human primates demonstrating the remarkable ability of the pathogen to cross species barriers [35][36][37].
In 1960s, two Modified Live Virus (MLV) vaccines were developed as a control measure against CDV infection [38,39]. However, reports of post vaccinal canine distemper [40][41][42][43] led to the introduction of the recombinant CDV vaccines, which have been tested and shown to be safe to many susceptible species [44][45][46][47]. Albeit the suitability in many

Discussion
Canine distemper virus is a negative sense ssRNA virus that is well-known to infect and cause fatal diseases in various animals. While initially recognized for causing distemper in domestic dogs, CDV is now well known for infecting wide range of hosts, including mostly but not confined to carnivore [1][2][3][4][5]. CDV is increasingly recognized as a major cause of decline in wild carnivore population [33], which inevitably raises concerns about the extinction threat of many endangered species [34]. In addition, lethal CDV outbreaks have been described in unprecedented species such as non-human primates demonstrating the remarkable ability of the pathogen to cross species barriers [35][36][37].
In 1960s, two Modified Live Virus (MLV) vaccines were developed as a control measure against CDV infection [38,39]. However, reports of post vaccinal canine distemper [40][41][42][43] led to the introduction of the recombinant CDV vaccines, which have been tested and shown to be safe to many susceptible species [44][45][46][47]. Albeit the suitability in many species, these recombinant vaccines have a limitation such as milder immunological response as compared to MLV vaccines [48]. As a therapy, there exists no specific antiviral drug against CDV infection in any species, including domestic dogs. Some promising candidates such as AS-48 class molecules (entry inhibitors) and pan-morbillivirus replication inhibitor were, however, recently reported [24]. Nevertheless, simultaneous discovering of more antivirals targeting different viral life cycle stages might mitigate issues such as generation of drug-resistant mutants.
In this study, we present a quantitative cell-based fusion assay, which enabled us to screen small molecule compounds potentially inhibiting CDV-mediated cell entry and spread. The split nature of the nanoluciferase indeed facilitated the indirect reading of the amount of cell-to-cell fusion ( Figure 1C). The high sensitivity of the system and its optimization in 384-wells format offered an attractive framework to initiate HTS. Out of two independent libraries screened, we indeed identified two entry inhibitors, F2736-3056 and F2261-0043 (Figure 2), one from each library, which validated the power of the assay to identify inhibitors against CDV.
Interestingly, both inhibitors resembled the AS-48 class compounds, which inferred a similar mode of binding and action. Indeed, high resolution structural studies of prefusion MeV F-protein in complex with FIP and AS-48 revealed a common binding pocket located at the junction of the head and stalk domains [26]. Moreover, functional studies suggested that such inhibitors may increase the energy threshold required for F-activation in turn efficiently preventing F-refolding and ensuing membrane fusion activity (reviewed in [49]). F2736-3056 and F2261-0043 may hence bind to the same pocket microdomain of the CDV F prefusion state and block its activity. Supporting this hypothesis, our in silico analyses indicated the possibility that both identified compounds may indeed dock into the similar microdomain in CDV F. Of note, the shallow groove, which extends towards the solvent from the deep hydrophobic pocket, enables accommodation of compounds carrying additional fused groups to the 2-phenylacetanilide scaffold ( Figure 5B). Interestingly, F2736-3056 contains a supplementary morpholine moiety, which points outside of the pocket ( Figure 5A). This moiety may support an additional hydrogen bond interaction with a lysine residue (Lys581 in CDV F and Lys469 in MeV F) locating in the F-stalk domain, which resembles the mode of interaction of FIP in the MeV F-protein ( Figure 5A). This additional interaction may thus further stabilize the prefusion state of the CDV F-protein; an observation which is in good agreement with our functional studies ( Figure 4). Indeed, compared to 3G, F2736-3056 preserved substantial fusion-inhibitory activity against a hyperfusogenic, 3G-escape, F-mutant.
Using this assay, we aim to screen additional libraries to discover compounds with novel scaffolds and better potency that would be necessary to increase our arsenal towards the development of multiple antiviral drugs against CDV. Simultaneously, by employing this assay, we also intend to identify compounds that could improve our basic understanding of the molecular mechanisms underlying morbilliviral cell-entry and spread. Furthermore, such molecular tools may enhance the quality of cryo-EM-based prefusion CDV F-structures. In turn, high-resolution CDV F-structures may set the stage to rationally refine the existing fusion inhibitors and/or to act as templates for the design of novel binders that can eventually be utilized synergistically. Indeed, compared to a monotherapy approach, a combined therapy (e.g., targeting multiple players of the viral-induced membrane fusion process) is likely to match with more realistic clinical requirements.
In summary, in this study we present an HTS-compatible, highly-sensitive quantitative cell-based fusion assay that can be employed to further refine our knowledge of the morbilliviral entry and spread mechanisms, as well as to discover molecules able to inhibit such processes.

Data Availability Statement:
The data presented in this study are available in this article and on request from the corresponding author.