HTRF-based identification of small molecules targeting SARS-CoV-2 E protein interaction with ZO-1 PDZ2 Authors: Flavio Alvarez, Florence Larrous, Ariel Mechaly, Benjamin Bardiaux, Quentin Goor, Ahmed Haouz, Blandine Séon-Méniel, Rodolphe Alves de Sousa, Stéphane Bourg, Didier Desmaële, Bruno Figadère, Nicolas Wolff, Hélène Munier-Lehmann, Célia Caillet-Saguy Published: 2025 ABSTRACT The SARS-CoV-2 E protein through its C-terminal PDZ-binding motif (PBM) interacts with several host PDZ-containing proteins, including Zonula occludens-1 (ZO-1) protein via its PDZ2 domain, thereby contributing to viral pathogenesis. Targeting this interaction represents a potential therapeutic strategy. In this study, we determined the X-ray structure of the E PBM peptide in complex with the ZO-1 PDZ2 domain at 1.7 Å resolution. The structure revealed a domain-swapped dimer conformation of ZO-1 PDZ2, with the E PBM peptide conventionally bound within the PDZ domain’s canonical binding groove, exhibiting key interactions characteristic of type II PBM/PDZ interactions. To identify potential inhibitors of the E PBM/ZO-1 PDZ2 interaction, we performed a Homogeneous Time-Resolved Fluorescence (HTRF) screening using a protein-protein interaction-focused library of 1000 compounds. This led to the identification of 36 hits that disrupted this interaction. Subsequent cytotoxicity and dose-response assays narrowed the selection to 14 promising compounds. Docking simulations showed that some compounds bind within or near the PBM-binding pocket, supporting a competitive mechanism of interaction inhibition, while others bind at a central interface between the two PDZ monomers, suggesting an inhibition of dimerization, which in turn prevents PBM binding. Thus, the E PBM/ZO-1 PDZ2 interaction can be inhibited through both direct and indirect mechanisms. Finally, antiviral assays using a NanoLuciferase-expressing recombinant SARS-CoV-2 demonstrated that one compound, C19, significantly reduced viral replication, highlighting its potential as a candidate for further therapeutic development.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-025-31755-y. Introduction The SARS-CoV-2 virus continues to pose a significant public health challenge due to its ongoing mutations and the emergence of new variants, which contribute to periodic outbreaks worldwide1,2. SARS-CoV-2 can lead to severe respiratory illness in humans, commonly referred to as COVID-19, often characterized by a cytokine storm leading to acute respiratory distress syndrome (ARDS) and high mortality rates3–5. Current FDA-approved treatments primarily target viral proteins essential for replication, including antiviral drugs such as Paxlovid, Remdesivir, and Molnupiravir6. These treatments remain effective, targeting highly conserved viral regions such as the SARS-CoV-2 main protease (Mpro) and the viral RNA-dependent RNA polymerase (RdRp). However, monoclonal antibodies targeting the spike protein, another widely used therapeutic strategy, have shown limitations7–9. Their effectiveness decreases as spike protein mutations emerge, highlighting the urgent need for alternative therapeutic targets. Expanding antiviral targets beyond RdRp and Mpro can help prevent cross-resistance, reduce the risk of drug-drug interactions, and enable combination therapies, which are critical for durable antiviral control and enhanced pandemic preparedness. SARS-CoV-2 is an enveloped, positive-sense single-stranded RNA virus that encodes several structural and non-structural proteins crucial for its viral life cycle. Among the structural proteins, the envelope (E) protein stands out for its pivotal role in viral assembly, release, and host-pathogen interactions10,11. This small and highly conserved integral membrane protein acts as a viroporin, enabling ion transport across host cell membranes, and contributes to viral morphogenesis. Notably, the E protein has an exceptionally low mutation rate, with a conservation rate exceeding 99.98%, suggesting its critical role in viral fitness12,13. Given its conserved nature and functional importance in viral assembly and pathogenesis, the SARS-CoV-2 E protein has emerged as an attractive therapeutic target. Recent studies, mainly through computational approaches, have identified natural and synthetic compounds capable of inhibiting its viroporin activity or interactions with host proteins. For instance, compounds derived from flavonoids, such as rutin, and glycyrrhizic acid, a triterpene glycoside, have demonstrated the ability to bind and destabilize the E protein12. Therapeutic approaches targeting protein-protein interactions (PPIs) involving the E protein remain largely underexplored, especially as its interactome discoveries continue to expand14–18. The E protein features a short C-terminal interaction motif, the PBM, capable of interacting with host cellular PDZ domain-containing proteins, leading to the disruption of endogenous PPIs involved in various cellular functions. The PBM of the SARS-CoV-2 E protein and its interactions with host proteins have been studied15,16,19–21. We and others have shown that the PBM plays a significant role in viral pathogenicity22,23. Our previous work showed that the PBM is essential for SARS-CoV-2 virulence through its interaction with ZO-1, a key determinant of COVID-19 pathogenesis22,24. In addition to ZO-1, the E protein engages other PDZ-containing tight junction proteins, such as the Polarity protein associated with Lin-7 1 (PALS-1), disrupting cell polarity and epithelial integrity15,20,24–26. These interactions weaken tight junctions and epithelial barriers, increasing the permeability of lung tissue and other organs, eventually leading to fluid leakage27,28. This contributes to severe complications such as pulmonary edema and ARDS, which are commonly observed in critical COVID-19 cases. Additionally, the breakdown of tight junctions facilitates the uncontrolled migration of immune cells and inflammatory mediators into the affected tissues, amplifying the immune response and potentially triggering a cytokine storm, a major factor in the severity of COVID-1929,30. In this context, targeting the interaction between the E protein and ZO-1 may represent a novel therapeutic approach to limit viral propagation and mitigate the pathogenesis and severity of the disease. In this study, we aimed to identify small molecules capable of inhibiting the interaction between the SARS-CoV-2 E protein and the PDZ2 domain of the ZO-1 protein. Using a Homogeneous Time-Resolved Fluorescence (HTRF) assay, we screened a PPI-focused library of 1,000 small compounds31 and identified 36 candidates for further validation. Inhibition and cytotoxicity dose-response assays allowed us to select 14 promising candidates. Their antiviral activities were assessed in vitro using a recombinant SARS-CoV-2 containing a NanoLuciferase reporter gene. One of these molecules demonstrated significant antiviral activity. We solved the X-ray structure of the complex between the PDZ2 of ZO-1 and the PBM of the E protein and performed docking studies with selected inhibitors to elucidate their mode of action. Our findings provide insights into novel therapeutic strategies targeting the interaction between the SARS-CoV-2 E protein and the PDZ-containing ZO-1 protein. This approach could help maintain barrier integrity, alleviate inflammation, reduce the severity of COVID-19, and potentially provide a groundbreaking therapeutic strategy to combat other viruses employing similar mechanisms. Results The crystallographic structure reveals key interaction determinants between the ZO-1 PDZ2 swapped dimer and the C-terminal region of the E protein To gain structural insights into the interaction between ZO-1 and the C-terminal PBM of the SARS-CoV-2 E protein, we solved the X-ray structure of the ZO-1 PDZ2 domain (residues K186–R262) in complex with the last 12‐residue peptide of the E protein (acNLNSSRVPDLLVCOOH). The crystal structure of ZO1-PDZ2 in complex with the SARS-CoV-2-E-PBM peptide was resolved at 1.72 Å using the molecular replacement method (Fig. 1, Table 1). The final refined model contains two PDZ domains per asymmetric unit, adopting a swapped dimer conformation. Each PDZ domain features a fold comprising six β strands and two α helices. Notably, the N-terminal region, encompassing strands β1 and β2, does not contribute to the core PDZ fold of its own molecule. Instead, these segments complete the PDZ fold of the adjacent molecule in a reciprocal manner. Two primary regions of intermolecular interactions between the PDZ domains significantly contribute to the extensive dimer interface: the inter-domain β1/β6 pairing and the extended antiparallel β2–β3/β2′–β3′ assembly (Fig. 1A). Fig. 1X-ray crystal structure of the PDZ2 domain of ZO-1 bound to the SARS-CoV-2 E protein PBM. (A) Ribbon diagram of the final crystal structure model of the ZO-1 PDZ2 domain in complex with the SARS-CoV-2 E protein PBM peptide. The structure forms a swapped dimer, with one monomer colored light blue and the other dark blue. The bound peptides are shown in orange. Secondary structure elements are labeled according to the canonical PDZ domain scheme. (B) Close-up view of the SARS-CoV-2 E protein PBM bound within the PDZ domain. Key interacting residues are shown as sticks and labeled. Intermolecular hydrogen bonds and polar interactions are indicated by cyan dashed lines. X-ray crystal structure of the PDZ2 domain of ZO-1 bound to the SARS-CoV-2 E protein PBM. (A) Ribbon diagram of the final crystal structure model of the ZO-1 PDZ2 domain in complex with the SARS-CoV-2 E protein PBM peptide. The structure forms a swapped dimer, with one monomer colored light blue and the other dark blue. The bound peptides are shown in orange. Secondary structure elements are labeled according to the canonical PDZ domain scheme. (B) Close-up view of the SARS-CoV-2 E protein PBM bound within the PDZ domain. Key interacting residues are shown as sticks and labeled. Intermolecular hydrogen bonds and polar interactions are indicated by cyan dashed lines. Table 1Data collection and refinement statistics. Statistics for the highest-resolution shell are shown in parentheses.ZO-1 PDZ2 / SARS-CoV-2-E-PBM peptide complexWavelength0.9786Resolution range54.120−1.721 (1.918−1.721)Space groupP 21 21 21Unit cell30.127 76.157 76.922 90 90 90Total reflections180,485 (7430)Unique reflections13,687 (684)Multiplicity13.2 (10.9)Completeness (%)52.9 (99.7)Mean I/sigma(I)11.6 (1.7)Wilson B-factor30.54R-merge0.121 (1.612)R-meas0.126 (1.686)R-pim0.034 (0.476)CC1/20.998 (0.617)Reflections used in refinement13,183 (77)Reflections used for R-free684 (5)R-work0.2030 (0.4132)R-free0.2727 (0.5860)Number of non-hydrogen atoms1492Macromolecules1396Ligands5Solvent91Protein residues183RMS (bonds)0.006RMS (angles)0.82Ramachandran favored (%)99.43Ramachandran allowed (%)0.57Ramachandran outliers (%)0.00Rotamer outliers (%)0.64Clashscore2.83Average B-factor44.02Macromolecules44.00Ligands36.16Solvent44.82 Table 1Data collection and refinement statistics. Statistics for the highest-resolution shell are shown in parentheses.ZO-1 PDZ2 / SARS-CoV-2-E-PBM peptide complexWavelength0.9786Resolution range54.120−1.721 (1.918−1.721)Space groupP 21 21 21Unit cell30.127 76.157 76.922 90 90 90Total reflections180,485 (7430)Unique reflections13,687 (684)Multiplicity13.2 (10.9)Completeness (%)52.9 (99.7)Mean I/sigma(I)11.6 (1.7)Wilson B-factor30.54R-merge0.121 (1.612)R-meas0.126 (1.686)R-pim0.034 (0.476)CC1/20.998 (0.617)Reflections used in refinement13,183 (77)Reflections used for R-free684 (5)R-work0.2030 (0.4132)R-free0.2727 (0.5860)Number of non-hydrogen atoms1492Macromolecules1396Ligands5Solvent91Protein residues183RMS (bonds)0.006RMS (angles)0.82Ramachandran favored (%)99.43Ramachandran allowed (%)0.57Ramachandran outliers (%)0.00Rotamer outliers (%)0.64Clashscore2.83Average B-factor44.02Macromolecules44.00Ligands36.16Solvent44.82 Data collection and refinement statistics. Statistics for the highest-resolution shell are shown in parentheses. The swapped dimer assembly observed in the ZO-1 PDZ2 complex was similarly identified for ZO‐1 and ZO‐2 PDZ232 either in their ligand‐free forms or in complex with the C-terminal peptide of connexin-4333 (Supplementary Fig. 1). Two peptides are bound to each PDZ dimer (Fig. 1A). Electron density maps suggest a well-defined conformation of the last eight residues of the peptide containing the PBM. The E PBM binds to the PDZ in a conventional manner, inserting into a binding groove formed by the β2 strand, the α2 helix, and the “GLGF” motif, and extending the antiparallel β-sheet. The root-mean-square deviation (RMSD) between the structures of ZO‐1 PDZ2 in complex with connexin-43 and the E protein peptides is 0.54 Å, indicating a high degree of structural similarity (Supplementary Fig. 1). The last six residues of the SARS-CoV-2-E-PBM peptide are in contact with the PDZ domain, whereas the upstream residues are distant from the PDZ surface, as observed in the connexin-43 peptide complex33. The interaction between the SARS-CoV-2-E-PBM peptide and the binding pocket of ZO-1 PDZ2 exemplifies a type II PBM-PDZ binding32. Three hydrogen bonds are formed between the C-terminal carboxylate of Val at position 0 (Val0) of the PBM and the backbone amides of Tyr198, Gly199, and Leu200 of the “GLGF” loop in ZO1-PDZ2 (Fig. 1B). A hydrogen bond is also formed between the backbone amide of Val0 and the carbonyl of Leu200 at the beginning of the β2 strand. Additionally, the carbonyl and the amide group of Leu at position-2 (Leu-2) interact through hydrogen bonds with the amide and carbonyl groups of Leu200, respectively. Also, the side chain of Leu-2 engages in a hydrophobic interaction, packing tightly against the side chain of Leu202 located in the β2 strand. An additional H-bond is observed between the carbonyl of Val-5 and the side-chain amide of Asn239 (Fig. 1B). The structural insights into the ZO-1 PDZ2 domain in complex with the SARS-CoV-2 E C-terminal provide valuable clues for therapeutic targeting of this interaction. High-throughput HTRF screening uncovers 36 compounds targeting E/ZO-1 protein-protein interaction To identify potential inhibitors, we employed a homogeneous time-resolved fluorescence resonance energy transfer (HTRF) approach to screen compounds that could disrupt the E/ZO-1 PPI. This variant of FRET uses a persistent fluorescence donor, enabling measurements after a defined delay following excitation. For the assay, we used an N-terminal His-tagged construct encompassing the cytoplasmic region of the E protein (residues 45–75) and a GST-tagged construct of the ZO-1 PDZ2 domain. Various in vitro binding parameters were tested in a 96-well low-volume plate, including incubation time and temperature, as well as the concentrations of purified proteins and antibodies, to determine the optimal conditions for complex formation. Our experimental optimizations ensured robust and reliable FRET efficiency, assessed using the HTRF ratio and Z′ factor to evaluate signal-to-background ratio34. With 3 nM of GST-ZO1-PDZ2 incubated with 60 nM of His-Ecyto (His-tagged E protein C-terminal 12-residue peptide) for 1 h at room temperature, followed by an additional 1-hour incubation with labeled antibodies, we confirmed a strong signal indicative of complex formation. GST-ZO1-PDZ2 alone with buffer was used as a negative control, whereas dimethyl sulfoxide (DMSO), used as an organic solvent for compound solubilization, did not impair PPIs or reduce the reaction system stability at its final concentration of 5%. Using a subset of the CNRS Chimiothèque Nationale, a focused library of 1000 putative PPI inhibitors solubilized in pure DMSO31, we conducted a primary screening at a concentration of 20 µM per compound, tested in duplicate. Our screening achieved a Z′-factor above 0.5 for all the plates, confirming the robustness and high-quality performance of our assay. We identified 36 promising hits with a residual HTRF ratio (% residual HTRF ratio) below 30% (Table 2). Histograms illustrating the percentage of inhibition of the normalized HTRF ratio signal are presented in Fig. 2. Table 2Activity and toxicity metrics of the 36 primary identified hits. This table summarizes the % residual HTRF ratio from the primary screening, along with IC50 values determined from HTRF-based serial dilution assays and CC50 values obtained from cytotoxicity dose-response assays. The CC50/IC50 ratio was calculated to determine the selectivity index (SI). For compounds that did not exhibit cytotoxicity, an SI threshold was estimated by using the highest tested concentration (20 µM) as the CC50 limit. Dashes indicate data that were either not detected or not quantifiable. Values highlighted in bold indicate compounds that fall outside the predefined selection thresholds, set at 25 µM for IC50 (inhibitory potency) and 10 µM for CC50 (cytotoxicity), and were therefore excluded from further consideration.CompoundHTRF assayCytotoxicity CC50 (µM) (n = 1)Selectivity indexPrimary screening percentage residual ratio (%) (n = 2)IC50 (µM) (n = 1)C125.5 (± 24.7)16.4–> 1.2C224 (± 0)–––C323.5 (± 7.8) 25 –> 0.8C422.5 (± 2.1) 39.4 9.1 0.2C520.5 (± 10.6)–26.5–C620 (± 2.8)0.4–> 50C720 (± 0)–––C818.5 (± 7.8)–––C918 (± 1.4)3.4–> 5.9C1017.5 (± 14.8)21.8–> 0.9C1115 (± 12.7)7.7–> 2.6C1214 (± 2.8)–––C1314 (± 4.2)16.3–> 1.2C1412.5 (± 10.6)–––C1512.5 (± 6.4)–––C1612 (± 11.3)6.6 2.8 0.4C1711.5 (± 7.8)–––C1811.5 (± 9.2)–––C1910.5 (± 6.4)6.8–> 2.9C209.5 (± 2.1)1.521.614.4C218.5 (± 10.6)–––C228.5 (± 7.8) 87.2 –0.2C238 (± 1.4)–––C247 (± 7.1) 51.6 –> 0.38C256 (± 5.7)21 5.7 0.27C266 (± 4.2)–––C274.5 (± 0.7)8.7–> 2.3C284 (± 2.8)20.3–> 1C293.5 (± 4.9)1.110.49.45C303.5 (± 3.5)–––C312.5 (± 2.1)–38–C322 (± 2.8)1.916.78.8C331.5 (± 0.7) 483.1 17.50.04C341.5 (± 2.1)2.9 7.4 2.6C351 (± 1.4)1.920.710.9C360.5 (± 0)8.1–> 2.5 Table 2Activity and toxicity metrics of the 36 primary identified hits. This table summarizes the % residual HTRF ratio from the primary screening, along with IC50 values determined from HTRF-based serial dilution assays and CC50 values obtained from cytotoxicity dose-response assays. The CC50/IC50 ratio was calculated to determine the selectivity index (SI). For compounds that did not exhibit cytotoxicity, an SI threshold was estimated by using the highest tested concentration (20 µM) as the CC50 limit. Dashes indicate data that were either not detected or not quantifiable. Values highlighted in bold indicate compounds that fall outside the predefined selection thresholds, set at 25 µM for IC50 (inhibitory potency) and 10 µM for CC50 (cytotoxicity), and were therefore excluded from further consideration.CompoundHTRF assayCytotoxicity CC50 (µM) (n = 1)Selectivity indexPrimary screening percentage residual ratio (%) (n = 2)IC50 (µM) (n = 1)C125.5 (± 24.7)16.4–> 1.2C224 (± 0)–––C323.5 (± 7.8) 25 –> 0.8C422.5 (± 2.1) 39.4 9.1 0.2C520.5 (± 10.6)–26.5–C620 (± 2.8)0.4–> 50C720 (± 0)–––C818.5 (± 7.8)–––C918 (± 1.4)3.4–> 5.9C1017.5 (± 14.8)21.8–> 0.9C1115 (± 12.7)7.7–> 2.6C1214 (± 2.8)–––C1314 (± 4.2)16.3–> 1.2C1412.5 (± 10.6)–––C1512.5 (± 6.4)–––C1612 (± 11.3)6.6 2.8 0.4C1711.5 (± 7.8)–––C1811.5 (± 9.2)–––C1910.5 (± 6.4)6.8–> 2.9C209.5 (± 2.1)1.521.614.4C218.5 (± 10.6)–––C228.5 (± 7.8) 87.2 –0.2C238 (± 1.4)–––C247 (± 7.1) 51.6 –> 0.38C256 (± 5.7)21 5.7 0.27C266 (± 4.2)–––C274.5 (± 0.7)8.7–> 2.3C284 (± 2.8)20.3–> 1C293.5 (± 4.9)1.110.49.45C303.5 (± 3.5)–––C312.5 (± 2.1)–38–C322 (± 2.8)1.916.78.8C331.5 (± 0.7) 483.1 17.50.04C341.5 (± 2.1)2.9 7.4 2.6C351 (± 1.4)1.920.710.9C360.5 (± 0)8.1–> 2.5 Activity and toxicity metrics of the 36 primary identified hits. This table summarizes the % residual HTRF ratio from the primary screening, along with IC50 values determined from HTRF-based serial dilution assays and CC50 values obtained from cytotoxicity dose-response assays. The CC50/IC50 ratio was calculated to determine the selectivity index (SI). For compounds that did not exhibit cytotoxicity, an SI threshold was estimated by using the highest tested concentration (20 µM) as the CC50 limit. Dashes indicate data that were either not detected or not quantifiable. Values highlighted in bold indicate compounds that fall outside the predefined selection thresholds, set at 25 µM for IC50 (inhibitory potency) and 10 µM for CC50 (cytotoxicity), and were therefore excluded from further consideration. Fig. 2HTRF-based screening of compounds targeting the interaction between ZO-1 PDZ2 and the SARS-CoV-2 E protein PBM. The assay was conducted in 96-well, white, low-volume HTRF plates using 5 µL of purified GST-ZO-1-PDZ2 at a final concentration of 3 nM, 5 µL of purified His6-Ecyto at a final concentration of 60 nM, and 1 µL of each of the 1000 compounds from a focused-PPI chemical library. The histogram shows the % residual HTRF ratio. Among the 1,000 compounds screened, only the 36 compounds with % residual HTRF ratio below 30% were classified as primary hits (C1 to C36). Horizontal lines indicate the mean with SD (n = 2). HTRF-based screening of compounds targeting the interaction between ZO-1 PDZ2 and the SARS-CoV-2 E protein PBM. The assay was conducted in 96-well, white, low-volume HTRF plates using 5 µL of purified GST-ZO-1-PDZ2 at a final concentration of 3 nM, 5 µL of purified His6-Ecyto at a final concentration of 60 nM, and 1 µL of each of the 1000 compounds from a focused-PPI chemical library. The histogram shows the % residual HTRF ratio. Among the 1,000 compounds screened, only the 36 compounds with % residual HTRF ratio below 30% were classified as primary hits (C1 to C36). Horizontal lines indicate the mean with SD (n = 2). Upon further analysis of the screening results, we noted a spectrum of inhibitory effectiveness among the compounds. Compound C36, the most effective, exhibited a % residual HTRF ratio of just 0.5 ± 0.0%, highlighting its strong potency as an inhibitor of the E/ZO-1 interaction. Other notable compounds include C35 and C36, with % residual HTRF ratios of 1 and 0.5, respectively. It is important to note that a reduction in HTRF signal may result from compound interference with the fluorescence readout, such as quenching or aggregation. To ensure that the observed signal reduction reflects genuine disruption of the E/ZO-1 interaction, follow-up assays were performed to confirm the inhibitory activity. Overall, the compounds demonstrated varying levels of potency, contributing to a diverse set of candidates for further investigation. Fourteen inhibitors of the E/ZO-1 interaction emerge from the dose-response and cytotoxicity evaluations To confirm reproducible inhibition, we first performed HTRF dose-response assays for all 36 initial hits using a 10-point, two-fold serial dilution, with concentrations ranging from 100 µM to 195 nM (Supplementary Fig. 3, Table 2). This approach aimed to assess both reproducibility and dose-dependent inhibitory effects, enabling the calculation of IC50 values (Table 2). Compounds with IC50 values exceeding 25 µM were excluded based on our selection criteria. Next, we evaluated the cytotoxicity of the 36 compounds on Vero-E6 cells using a 9-point, two-fold serial dilution with inhibitor concentrations ranging from 20 µM to 78 nM (Supplementary Fig. 2, Table 2). Cell viability was measured using the XTT assay, allowing us to determine the CC50 values of the compounds (Table 2). Compounds with a CC50 lower than 10 µM were excluded from further analysis. Notably, compounds such as C16 (2.8 µM), C25 (5.7 µM), and C4 (9.1 µM) posed higher risks of cytotoxicity and were discarded. Among the inhibitors, C6 emerged as the most potent with an IC50 of 0.4 µM, while C20 and C35 displayed low IC50 values of 1.5 µM and 1.9 µM, respectively, combined with high CC50 values (21.6 µM and 20.7 µM), indicating both strong inhibition and low cytotoxicity. C29 and C32 also showed notable IC50 values of 1.1 µM and 1.9 µM, alongside reasonable CC50 thresholds, making them good candidates. Conversely, C22 and C33 exhibited elevated IC50 values, placing them out of the preferred range for effective interaction inhibition. Selectivity indices (SI) were calculated for all compounds, with several showing SI values greater than 1 and some exceeding 10, underscoring the identification of multiple selective candidates (Table 2). Chemical characteristics of selected compounds highlight potential determinants for ZO-1 PDZ2 interaction Based on cytotoxicity and dose-response inhibition assays, 14 promising compounds were selected, with HTRF IC₅₀ values ranging from 0.4 to 21.8 µM. The chemical structures of these selected compounds are shown in Fig. 3. The compounds exhibit complex organic structures, predominantly characterized by fused aromatic rings and distinct functional groups. Many of these compounds incorporate fused aromatic or heteroaromatic rings, forming polycyclic structures. Several molecules possess functional groups such as hydroxyl (–OH), methoxy (–OCH3), amide (–CO–NH–), and isocyanate (–NCO), which are characteristic of highly functionalized PPI compounds designed to promote interactions such as hydrogen bonding, thereby influencing their chemical and biological properties. Compounds such as C20, C32, and C35 exhibit structural symmetry, indicating a regular arrangement of functional groups and rings, which is characteristic of ditercalinium-type compounds with only slight variations in their substituents. Three compounds possess potentially reactive functional groups: compound C10 contains an isocyanate group, compound C19 features an aromatic nitro group, and compound C36 includes a catechol moiety with potential redox activity. Compound C29 presents a linear core structure flanked by fused rings at both ends, whereas other compounds, like C32, adopt a more complex, polycyclic configuration. Compound C28 is identified as a metal chelator, suggesting its ability to bind metal ions, a feature not common to all compounds. Among the most active compounds, compound C19 is constructed around a quinoline substituted atposition 2 by a catechol masked in the form of methyl ether and in position 7 by a nitroaromatic group which can be the origin of redox cycling reactions. Fig. 3Chemical structures of the 14 selected compounds. The 14 compounds shown were selected based on results from cytotoxicity and inhibitory dose-response assays. All chemical structures were generated using ChemDraw software. Chemical structures of the 14 selected compounds. The 14 compounds shown were selected based on results from cytotoxicity and inhibitory dose-response assays. All chemical structures were generated using ChemDraw software. Due to their polycyclic structures and specific functional groups, the compounds may interfere with the interaction between the SARS-CoV-2-E-PBM peptide and the binding pocket of ZO-1 PDZ2. For example, fused aromatic rings and functional groups like amides or isocyanates could potentially bind to the PDZ2 domain of ZO-1, inhibiting its interaction with the E protein PBM. Additionally, the presence of positive charges in at least eight compounds containing more than two charges may enhance electrostatic interactions with charged residues in the PDZ2 domain, thereby increasing their affinity for this target. In summary, despite the structural diversity among these compounds, their ability to interact with the PDZ2 domain of ZO-1 through π-stacking, electrostatic interactions, or metal ion coordination could render them effective in inhibiting the E protein PBM/ZO-1 PDZ2 interaction. Docking analysis identifies key binding sites for selected compounds on the E PBM/ZO-1 PDZ2 complex To further investigate the binding modes of the 14 selected compounds, we performed docking simulations using the DOCK6 program35. As a template, we employed the X-ray structure of the ZO-1 PDZ2 domain alone (PDB ID: 2RCZ). All compounds were docked successfully, with the molecules adopting various orientations (Fig. 4, Supplementary Table 1). The docking poses demonstrated a tendency to cluster in a central region at the interface of the swapped dimer, positioned between the two PDZ monomers. Moreover, among the top poses for the five compounds C1, C10, C19, C27, and C36, some were found in proximity to the PBM peptide-binding site close to the α2 helix (Fig. 4). These findings suggest that while certain compounds may directly compete or interfere with the PBM for binding, some might also disrupt the interaction by targeting adjacent regions crucial for the structural stability of the dimer. The preferential localization of the compounds within this central pocket and near the PBM peptide-binding site highlights a conserved hotspot for ligand binding, which could be exploited for therapeutic targeting of the ZO-1 PDZ2/SARS-CoV-2 E interaction. Furthermore, control docking simulations on the structure of ZO-1 PDZ2 dimer bound to the PBM peptide that we obtained revealed no compounds located near the PBM-binding pocket (Supplementary Fig. 4), consistent with a competitive inhibition mechanism. Fig. 4Docking poses of the top 14 compounds targeting the ZO-1 PDZ2 dimer. The figure shows the best binding poses of the 14 compounds targeting the swapped dimer of ZO-1 PDZ2 using DOCK6 docking simulations. The PDZ is depicted in cartoon. Each compound’s docking pose is represented, positioned within key binding regions of the ZO-1 PDZ2 dimer. The compounds are primarily located at the dimer interface or near the PBM binding site such as C10, C19 or C27. Docking poses of the top 14 compounds targeting the ZO-1 PDZ2 dimer. The figure shows the best binding poses of the 14 compounds targeting the swapped dimer of ZO-1 PDZ2 using DOCK6 docking simulations. The PDZ is depicted in cartoon. Each compound’s docking pose is represented, positioned within key binding regions of the ZO-1 PDZ2 dimer. The compounds are primarily located at the dimer interface or near the PBM binding site such as C10, C19 or C27. A standout molecule among the hits emerges as promising for its antiviral activity To investigate the potential antiviral effects of the 14 selected compounds directly extracted from the library without further chemical optimization, we used a recombinant SARS-CoV-2 virus engineered to express NanoLuciferase as a reporter gene. This system allowed us to quantify viral replication by measuring luminescence from the NanoLuciferase secreted into the infected cell culture supernatant, which directly correlates with viral load (Supplementary Fig. 5). Among the 14 compounds tested, only C19 demonstrated a pronounced antiviral effect in Vero-E6 cells. Indeed, at concentrations above 10 µM, C19 induced a greater than two-log reduction in luminescence, indicating a strong inhibition of viral replication. In contrast, the remaining 13 compounds showed no significant antiviral activity under these experimental conditions. This highlights that, while several compounds effectively inhibit the E/ZO-1 interaction in vitro, their antiviral effects in this particular cell-based assay are limited, with C19 being the only compound showing a clear and robust impact on viral replication. This compound (Fig. 5A), for which all specific data are summarized in Fig. 5, was predicted in 8 out of 9 top docking poses to occupy the PBM interaction pocket, suggesting a potential competitive binding mechanism (Fig. 5B). Supporting this hypothesis, when docking was performed using our structure of the ZO-1 PDZ2 domain in complex with the E PBM peptide, all binding poses of C19 were redirected toward the interface between the two monomers rather than the PBM-binding site (Fig. 5C). Fig. 5Characterization of C19, a promising antiviral compound. (A) Chemical structure of C19, generated using ChemDraw software. (B,C) Docking poses of C19 targeting the ZO-1 PDZ2 swapped dimer. Panels show the best predicted binding poses of C19 when docked to the ZO-1 PDZ2 dimer either in its free form (B) or in complex with the E protein PBM (C), using DOCK6 simulations. The PDZ domains are shown in cartoon representation. The compound is positioned within key binding regions of the ZO-1 PDZ2 dimer, preferentially occupying the PBM-binding site. (D) HTRF inhibition dose-response curve of C19. The assay was performed in 96-well, white, low-volume HTRF plates with 5 µL of purified GST-ZO-1-PDZ2 (final concentration: 3 nM), 5 µL of purified His₆-Ecyto (final concentration: 60 nM), and 1 µL of C19, tested in a 10-point, twofold serial dilution (100 µM to 195 nM). The resulting curve represents the percentage of residual interaction (%R) based on the normalized FRET signal. A sigmoidal curve was generated using nonlinear regression in Prism software, plotting C19 concentration against the normalized response (n = 1). (E,F) Antiviral activity of C19. The assay was conducted in 96-well culture plates using Vero-E6 cells (40,000 cells/well) infected with a NanoLuciferase-expressing recombinant SARS-CoV-2 virus at an MOI of 0.001. Cells were incubated for 72 h in the presence of C19, applied as a 9-point, twofold serial dilution ranging from 20 µM to 78 nM. Viral replication was assessed by measuring luminescence in the cell culture supernatant using the Nano-Glo assay kit (E). Data were normalized to both infected untreated and uninfected untreated control conditions. Sigmoidal curve was generated using nonlinear regression in Prism software, plotting C19 concentration against the normalized luminescence signal (n=3). Viral quantification was performed using a plaque assay on Vero-E6 cells (F). Plaques were observed at 3 days post-infection (dpi) following infection with serial dilutions of supernatants collected from infected non-treated (INT) or infected treated (IT) cells, treated for 72 h with 20 µM of compound C19. Viral titers are reported as plaque-forming units per milliliter (PFU/mL). Horizontal lines represent the mean ± SEM, and dashed lines indicate the limit of detection (n=3; nd=not detected). Statistical significance was determined by a t-test (p < 0.005). Characterization of C19, a promising antiviral compound. (A) Chemical structure of C19, generated using ChemDraw software. (B,C) Docking poses of C19 targeting the ZO-1 PDZ2 swapped dimer. Panels show the best predicted binding poses of C19 when docked to the ZO-1 PDZ2 dimer either in its free form (B) or in complex with the E protein PBM (C), using DOCK6 simulations. The PDZ domains are shown in cartoon representation. The compound is positioned within key binding regions of the ZO-1 PDZ2 dimer, preferentially occupying the PBM-binding site. (D) HTRF inhibition dose-response curve of C19. The assay was performed in 96-well, white, low-volume HTRF plates with 5 µL of purified GST-ZO-1-PDZ2 (final concentration: 3 nM), 5 µL of purified His₆-Ecyto (final concentration: 60 nM), and 1 µL of C19, tested in a 10-point, twofold serial dilution (100 µM to 195 nM). The resulting curve represents the percentage of residual interaction (%R) based on the normalized FRET signal. A sigmoidal curve was generated using nonlinear regression in Prism software, plotting C19 concentration against the normalized response (n = 1). (E,F) Antiviral activity of C19. The assay was conducted in 96-well culture plates using Vero-E6 cells (40,000 cells/well) infected with a NanoLuciferase-expressing recombinant SARS-CoV-2 virus at an MOI of 0.001. Cells were incubated for 72 h in the presence of C19, applied as a 9-point, twofold serial dilution ranging from 20 µM to 78 nM. Viral replication was assessed by measuring luminescence in the cell culture supernatant using the Nano-Glo assay kit (E). Data were normalized to both infected untreated and uninfected untreated control conditions. Sigmoidal curve was generated using nonlinear regression in Prism software, plotting C19 concentration against the normalized luminescence signal (n=3). Viral quantification was performed using a plaque assay on Vero-E6 cells (F). Plaques were observed at 3 days post-infection (dpi) following infection with serial dilutions of supernatants collected from infected non-treated (INT) or infected treated (IT) cells, treated for 72 h with 20 µM of compound C19. Viral titers are reported as plaque-forming units per milliliter (PFU/mL). Horizontal lines represent the mean ± SEM, and dashed lines indicate the limit of detection (n=3; nd=not detected). Statistical significance was determined by a t-test (p < 0.005). Compound C19 exhibits an IC50 of 6.8 µM for inhibition of the E/ZO-1 interaction (Fig. 5D), as measured by HTRF, and an IC50 of 4.8 ± 1.5 µM for inhibition of viral replication (Fig. 5E), determined via luminescence assays in infected NanoLuciferase-expressing virus-infected cells. Viral titration by plaque lysis assay confirmed the antiviral effect, showing a viral titer more than 2 log units lower compared to untreated infected cells (Fig. 5F). Discussion The SARS-CoV-2 E protein plays a multifaceted role in viral pathogenesis. Various molecules and strategies have been investigated to inhibit the SARS-CoV-2 E protein, although research in this area is still ongoing12. These include direct inhibitors, peptide antagonists, antibody-based approaches, repurposed drugs, and computationally designed molecules. As understanding of the E protein’s role in viral pathogenesis and its interactions with host proteins continues to evolve, new and more effective inhibitors are likely to emerge. While monoclonal antibodies and ion channel blockers have been developed, approaches targeting the PDZ-binding motif remain unexplored12. We previously identified ZO-1 PDZ2 as a binding partner of the SARS-CoV-2 E protein and demonstrated that its PBM preferentially binds to ZO-1 PDZ2 in vitro with a dissociation constant (Kd) of 15 µM15,24. ZO-1 is a critical component of tight junctions, which are essential for maintaining epithelial and endothelial barrier integrity. During SARS-CoV-2 infection, ZO-1 has been implicated as a key host factor targeted by the viral E protein, particularly through its C-terminal PBM. We previously highlighted that this PBM acts as an important factor in pathogenesis, demonstrating how disruption of the ZO-1 network contributes to tight junction and epithelial barrier compromise22. This breakdown of barrier integrity may play a significant role in the development of ARDS observed in severe COVID-19 cases. The crystal structure of the ZO-1 PDZ2 domain in complex with the SARS-CoV-2 E PBM reveals key insights into the molecular determinants of the interaction and provides a foundation for the development of novel therapeutic strategies. The adoption of a swapped dimer conformation stabilized by the reciprocal β1/β6 and β2/β3 interactions provides a robust framework for peptide binding. The small RMSD between the structures of the PDZ in complex with the SARS-CoV-2 E PBM peptide and the connexin-43 peptide, which also harbors a PBM (DLEICOOH), further validates the conservation of the interaction mode. This emphasizes the structural similarity and adaptability of the ZO-1 PDZ2 domain in engaging diverse PBM ligands. The high-throughput HTRF screening identified 36 compounds capable of disrupting the E protein/ZO-1 interaction, confirming the effectiveness of this method for discovering PPI inhibitors. Cytotoxicity and dose-response assessments reduced these down to 14 promising candidates. Docking simulations further revealed that some compounds preferentially bind within or near the PBM-binding groove, while others engage adjacent regions at the interface between the two PDZ monomers, suggesting multiple potential mechanisms for disrupting the interaction. While docking poses are presented for all selected compounds to illustrate potential binding modes and inform structure-activity relationships, most of these compounds do not show measurable antiviral activity in the cell-based assays. This distinction is critical, as the purpose of the screening pipeline is to identify compounds with true antiviral potential. Therefore, the docking poses of inactive molecules should be interpreted cautiously, as they are speculative and should be considered as starting points for future investigation. Among these, compound C19 emerged as a standout due to its potent antiviral activity, demonstrated by a significant reduction in viral replication observed using a recombinant SARS-CoV-2 NanoLuciferase reporter system. Importantly, C19 was the only compound among the 14 tested that displayed a clear and measurable antiviral effect in Vero-E6 cells. Although the other compounds did not show significant antiviral activity in this specific context, this may be attributed to the cellular model or assay conditions used. These compounds still hold potential, as their confirmed biochemical inhibition of the E/ZO-1 interaction suggests that they could exhibit antiviral effects under different experimental conditions. Further studies in alternative models could reveal additional insights into their antiviral potential. With these considerations in mind, the next steps should involve the optimization of compound C19 to enhance its specificity, potency, and pharmacokinetic properties. To extend our in-cellulo findings, future studies assessing the antiviral effects of these compounds in more relevant human respiratory epithelial cell models, such as Calu-3 or Human Airway Epithelial cells (HAE), would provide additional insight into their potential therapeutic relevance. The structural insights gained from the E PBM/ZO-1 PDZ2 complex and docking studies provide a framework for designing next-generation inhibitors targeting this critical host-virus interaction. Such inhibitors could not only mitigate the direct effects of SARS-CoV-2 infection but also provide a blueprint for addressing other pathogens that exploit similar host interactions. In addition to ZO-1, the C-terminal PBM of the SARS-CoV-2 E protein has been reported to interact with several other host PDZ domain-containing proteins, including PARD3, MLLT4 (Afadin), LNX2, and MPP5 (PALS1)24. These interactions are thought to contribute to the disruption of cell polarity and tight junctions during infection. Taking into account their critical role, it would be valuable to extend the antiviral inhibitor screening strategy to these additional PBM-interacting proteins. While the compounds were specifically selected to interfere with the E/ZO-1 interaction, it cannot be excluded that they might also affect additional E/PDZ interactions sharing structural and binding similarities, as these binding interfaces often encompass partially conserved surface areas. Although achieving maximal selectivity for the E/ZO-1 interaction remains a key goal, such multi-target activity could also contribute to the overall antiviral effect observed. Future studies assessing the selectivity profiles of these inhibitors across multiple E/PDZ interactions will be important to better understand their mechanisms of action and to optimize their therapeutic potential. Disrupting this interaction could preserve barrier integrity, reduce inflammation, and alleviate COVID-19 severity. Since ZO-1 is expressed in multiple organs, such as the gastrointestinal tract, kidneys, and brain29,36,37, the E protein’s interaction with ZO-1 may contribute to systemic complications like gastrointestinal symptoms, acute kidney injury, and neurological issues observed in some COVID-19 patients. Additionally, the E/ZO-1 interaction may compromise the endothelium, leading to vascular complications, including blood clots, strokes, and multiorgan failure38,39. In conclusion, targeting this interaction may reduce viral fitness and help mitigate both pulmonary and systemic complications of COVID-19, underscoring the importance of continued efforts to develop inhibitors of key host–virus interactions and paving the way for novel antiviral strategies. Materials and methods Expression constructs The pGEX-2T plasmid encoding amino acids 186–262 of human ZO-1 was obtained from Addgene (59735, Addgene)32. The construct features a thrombin cleavage site inserted between the GST tag and the N-terminal PDZ sequence. The cDNA sequence encoding amino acids 45–75 of the SARS-CoV-2 E protein was synthesized and then cloned into the pET-15b vector by GenScript. In this construct, the E protein sequence is appended with a N-terminal His tag, which is linked by a TEV protease cleavage site. Protein expression and purification Vectors encoding His-Ecyto or GST-ZO-1-PDZ2 constructs were transformed into E. coli BL21 Star (DE3) cells (EC0114, Invitrogen™). The bacteria were cultured in LB medium supplemented with 100 mg/L ampicillin at 37 °C. Protein expression was induced by adding 0.2 mM IPTG when the OD600nm reached 0.8-1.0, and cultures were incubated overnight at 18 °C. Following induction, the bacteria were harvested, and the PDZ2 domain of ZO-1 or Ecyto was resuspended in a lysis buffer composed of 50 mM Tris (pH 7.5), 250 mM NaCl, and 2 mM β-mercaptoethanol. The buffer was supplemented with one EDTA-free protease inhibitor tablet per 50 mL (COEDTAF-RO, Roche Diagnostics) and more than 250 units per liter of culture of benzonase (E1014-25KU, Sigma-Aldrich). The cells were lysed under high pressure using a CELLD homogenizer (1.3 kBar, 4 °C). Cell debris was removed by centrifugation at 30,000 × g for 1 h at 4 °C. Proteins fused to GST tag were purified using a GSTrap HP affinity chromatography column (17528201, GE Healthcare), while proteins fused to His tag were isolated using nickel-chelated HiTrap Chelating HP columns (17524801, GE Healthcare). For crystallographic purposes, an additional step was applied to the GST-ZO-1-PDZ2 protein bound to the column. An overnight cleavage at 16 °C was performed with Thrombin protease (SAE0147, Sigma-Aldrich) to remove the GST tag. Cleaved and uncleaved proteins were then processed through a final step of size-exclusion chromatography using a Sephacryl S-100 h 16/60 column (17116501, GE Healthcare). The chromatography buffer consisted of 50 mM Tris (pH 7.5), 150 mM NaCl, 0.5 mM TCEP, supplemented with one EDTA 2% protease inhibitors tablet per 100 mL of buffer (CO-RO, Roche Diagnostics). Peptide synthesis The acetylated peptides containing the C-terminal PBM sequence of E protein (12 residues long) of SARS-CoV-2 (acNLNSSRVPDLLVCOOH) were synthesized in solid phase using the Fmoc strategy (Proteogenix, Schiltigheim, France). The peptides were resuspended in pure water to prepare desired stock solutions. Crystallization, data collection, and structure determination The E PBM/ZO-1 PDZ2 complexes for crystallization were generated by mixing ZO-1PDZ domains with SARS-CoV-2 E protein PBM peptide at a ratio of 1:2, ensuring that at least 92% of complexes were formed. Initial screening of crystallization conditions was carried out by the vapor diffusion method using a Mosquito™ nanoliter-dispensing system (TTP Labtech) following the established protocols40. Sitting drops were set up using 400 nL of a 1:1 mixture of each sample protein and crystallization solutions (672 different commercially available conditions) equilibrated against a 150 µL reservoir in multiwell plates (Greiner Bio-One). The crystallization plates were stored at 4 °C in a RockImager1000® (Formulatrix) automated imaging system to monitor crystal growth. The best crystals were obtained in crystallization conditions containing 20% PEG 3350 and 0.2 M NH4I. Crystals were then flash cooled in liquid nitrogen using the condition of crystallization supplemented with 30% (v/v) of glycerol as cryoprotectant. X-ray diffraction data were collected on the beamlines Proxima-1 Synchrotron SOLEIL (St. Aubin, France). The data were processed by XDS41, and the structures were solved by molecular replacement with PHASER42 using the search atomic models PDB id 2RCZ for ZO-1-PDZ2 as template. The positions of the bound peptides were determined from a Fo–Fc difference electron density maps. Models were rebuilt using COOT43, and refinement was done with phenix.refine of the PHENIX suite44. The crystal parameters, data collection statistics, and final refinement statistics are shown in Table 1. The structure factors and coordinates have been deposited in the Protein Data Bank under accession code 9HM2. All structural figures were generated with the PyMOL Molecular Graphics System, Version 2.7.0 (Schrödinger). Design of a PPI-focused screening library The French national chemical library (Chimiothèque Nationale) was used to design a focused screening library toward PPI. A chemoinformatic study was carried out to select 1004 PPI-like compounds. First, the electronic version of the 62,281 compounds of the “Chimiothèque Nationale” was submitted to VSPrep, a workflow dedicated to preparing molecules for virtual screening45,46. During the preparation of chemical libraries, VSPrep implements several metrics based on the distribution of molecular properties to characterize specific subsets such as PPI-like47. It also labels PAINS compounds (Pan Assay Interference Compounds)48 using substructure search methods implemented in RDKit, a software package dedicated to chemoinformatics. 3350 compounds were identified as PPI-like of which 117 labeled PAINS were removed, yielding 3223 compounds. The structural similarity of the compounds was assessed using FCFP6 fingerprints49, calculated with RDKit. The compounds were then grouped according to their structural similarity using the Butina algorithm50. A Tanimoto coefficient threshold of 0.46 yielded 1,004 clusters from whose centroids were selected to build a library of 1,000 compounds that were solubilized in pure DMSO. HTRF binding assay between GST-ZO-1-PDZ2 and His6-Ecyto The HTRF assay was conducted in 96-well, white, low-volume HTRF plate (66PL96025, Revvity) with a final reaction volume of 20 µL per well. The anti-GST antibody labeled with Terbium (Tb) cryptate as a donor (61GSTTLA, Revvity) was used to detect the GST-ZO-1-PDZ2 protein, while the anti-His6 antibody labeled with the d2 acceptor fluorophore (61HISDLA, Revvity) was employed to detect the His6-Ecyto protein. Experimental conditions were optimized to achieve the maximum HTRF ratio signal. 5 µL of purified GST-ZO-1-PDZ2 at a final concentration of 3 nM and 5 µL of purified His6-Ecyto at a final concentration of 60 nM, both diluted in assay buffer (50 mM Tris-HCl pH 7.5, 150 mM NaCl, 0.5 mM TCEP, 0.05% Tween-20, 1% BSA and one EDTA-free protease inhibitor tablet per 100 mL of buffer), were dispensed into the plate and incubated at room temperature for 1 hour. For the chemical library screening, 1 µL of compound (2 mM in DMSO) was spiked into dry wells. For each plate, columns 1 and 12 contained only 1 µL of DMSO to act as controls. Then, 4.5 µL of anti-GST mAb Tb-conjugate and 4.5 µL of anti-His6 mAb d2-conjugate, both diluted to a final concentration of 1× according to the manufacturer’s protocol, were added to the plate. The mixture was incubated for an additional 1 h at room temperature. The fluorescence signal was measured with an Infinite M1000 Pro plate reader (TECAN) with the following settings: excitation at 340 nm, emission at 665 nm (for acceptor) and 620 nm (for donor) with a delay of 60 µs. The HTRF ratio is calculated for each well using the following formula: Signal at 665 nm/ Signal at 620 nm × 104. and the Z′-factor34 was calculated to assess the robustness of the test, using the formula: Z′ = \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:1-\frac{(3\mathrm{*}{\upsigma\:}\left(\mathrm{X}\mathrm{Y}\right)+3\mathrm{*}{\upsigma\:}\left(\mathrm{Y}\right))}{\mathrm{m}\left(\mathrm{X}\mathrm{Y}\right)-\mathrm{m}\left(\mathrm{Y}\right)}$$\end{document} where “(XY)” refers to the mean (m) or standard deviation (s) of the signal from wells containing both E and ZO-1 (positive controls), and “(Y)” refers to the mean or standard deviation of the signal from wells containing only ZO-1 (negative controls). The data from a plate are further analyzed when the Z’-factor is greater than 0.5. For each compound, the percentage of residual HTRF ratio was calculated via the following equation: % residual HTRF ratio = \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{\mathrm{H}\mathrm{T}\mathrm{R}\mathrm{F}\:\mathrm{r}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\:\mathrm{c}\mathrm{o}\mathrm{m}\mathrm{p}\mathrm{o}\mathrm{u}\mathrm{n}\mathrm{d}\:-\:\mathrm{H}\mathrm{T}\mathrm{R}\mathrm{F}\:\mathrm{r}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\:\mathrm{n}\mathrm{e}\mathrm{g}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{v}\mathrm{e}\:\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}\mathrm{s}}{\mathrm{H}\mathrm{T}\mathrm{R}\mathrm{F}\:\mathrm{r}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\:\mathrm{p}\mathrm{o}\mathrm{s}\mathrm{i}\mathrm{t}\mathrm{i}\mathrm{v}\mathrm{e}\:\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}\mathrm{s}\:-\:\mathrm{H}\mathrm{T}\mathrm{R}\mathrm{F}\:\mathrm{r}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\:\mathrm{n}\mathrm{e}\mathrm{g}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{v}\mathrm{e}\:\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}\mathrm{s}}X100$$\end{document}. Compounds with a value of the % residual HTRF ratio below 30 were defined as primary hits. IC50 HTRF determination The primary hits were validated using HTRF under the same conditions described for the initial compound screening. Purified proteins were added to a 96-well plate containing 1 µL of each selected compound, diluted in DMSO and tested in a 10-point, 2-fold serial dilution ranging from a final well concentration of 100 µM to 195 nM, to determine the IC50 values. For each compound, the interaction percentage, referred to as the HTRF residual interaction percentage, was calculated using the formula: % residual HTRF ratio = \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{\mathrm{H}\mathrm{T}\mathrm{R}\mathrm{F}\:\mathrm{r}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\:\mathrm{c}\mathrm{o}\mathrm{m}\mathrm{p}\mathrm{o}\mathrm{u}\mathrm{n}\mathrm{d}\:-\:\mathrm{H}\mathrm{T}\mathrm{R}\mathrm{F}\:\mathrm{r}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\:\mathrm{n}\mathrm{e}\mathrm{g}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{v}\mathrm{e}\:\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}\mathrm{s}}{\mathrm{H}\mathrm{T}\mathrm{R}\mathrm{F}\:\mathrm{r}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\:\mathrm{p}\mathrm{o}\mathrm{s}\mathrm{i}\mathrm{t}\mathrm{i}\mathrm{v}\mathrm{e}\:\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}\mathrm{s}\:-\:\mathrm{H}\mathrm{T}\mathrm{R}\mathrm{F}\:\mathrm{r}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\:\mathrm{n}\mathrm{e}\mathrm{g}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{v}\mathrm{e}\:\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}\mathrm{s}}X100$$\end{document}. IC50 values were calculated from the experimental data (% residual HTRF ratio) using Prism Graphpad software and compounds with IC50 values below 25 µM were classified as secondary hits. Cell viability assay Cell viability was assessed using the CyQUANT™ XTT Cell Viability Assay (X12223, Invitrogen™). Vero-E6 cells were seeded into 96-well plates at a density of 4 × 104 cells per well and incubated overnight. The following day, compounds were added to the cells in a 9-point, 2-fold serial dilution, ranging from a final concentration in the well of 20 µM down to 78 nM, to determine the CC50. Plates were cultured for 72 h. To assess cell viability, the CyQUANT™ XTT Cell Viability Kit was used according to the manufacturer’s protocol. Briefly, the XTT reagent was prepared by mixing components A and B, and 70 µL of the prepared reagent was added to each well containing 100 µL of culture medium. After 4 h of incubation at 37 °C, absorbance was measured using an Infinit M1000 pro plate reader (TECAN) at 450 nm and 660 nm. Negative controls consisted of culture medium only, while positive controls involved a 1-h treatment with 5% Triton X-100 diluted in culture medium to obtain 100% cell lysis. These controls enabled the calculation of the percentage of cell viability as a function of compound concentration and the subsequent determination of the CC50. Compounds with CC50 values above 10 µM were classified as secondary hits. Docking simulations We performed blind molecular docking simulations using the UCSF DOCK software (https://dock.compbio.ucsf.edu/DOCK_6/). The docking studies were conducted using the X-ray structure of the ZO-1 PDZ2 domain (PDB ID 2RCZ). The receptor structure included two PDZ2 domains of ZO-1 adopting a swapped dimer conformation. The receptor was prepared in MOL2 format with Chimera51 (removal of solvent molecules, addition of hydrogens and charges). The grid was generated to encompass all spheres detected with sphgen at the molecular surface of the receptor. The structures of the 14 selected compounds as ligands were prepared in MOL2 format with OpenBabel52 (addition of hydrogens, charges and generation of 3D conformation). Docking was carried out as flexible with DOCK6.9 using the grid-based score (van der Waals and electrostatic interactions). For each compound, at most 10 poses were ranked after clustering (2 Å RMSD threshold). The docking results were analyzed to identify potential binding sites and interactions between the compounds and the ZO-1 PDZ2 swapped dimer. Visualization of the docking poses and the analysis of intermolecular interactions were performed using PyMOL (Schrödinger, LLC). As a control, the same docking procedure was applied using the structure of the ZO1-PDZ2/E-PBM complex (PDB ID 9HM2) as receptor. Generation of the complete genome of Wuhan Nanoluciferase virus Recombinant viruses expressing Nanoluciferase (nLuc) as a reporter gene were obtained by reverse genetics as previously described53. Wuhan_nLuc was derived from the clone synSARS-CoV-2-GFP-P2A-ORF7a (clone #41) generated as described where the GFP reporter gene was replaced by the nLuc one. Briefly, the plasmid was linearized by EagI-HF (R3505, NEB) digestion, purified by a phenol chloroform process and transcribed in vitro into RNA using the T7 RiboMaxTM Large Scale RNA Production System (P1300, Promega) and the m7G(5’)ppp(5’)G RNA Cap Structure Analog (#S1411, NEB) kits. The synthesized RNA was finally purified using a classical phenol chloroform method and precipitated. Twelve µg of complete viral mRNA and 4 µg of plasmid encoding the viral nucleoprotein gene were electroporated on 8 × 106 Vero-E6 cells resuspended in 0.8 mL of Electroporation Solution (MIR50114, Mirus Bio™ Ingenio™) using the Gene Pulser Xcell Electroporation System (1652660, Bio-Rad) with a pulse of 270 V and 950 µF. The cells were then transferred to a T75 culture flask with 15 mL of DMEM supplemented with 2% FCS (v/v). The electroporated cells were incubated at 37 °C, 5% CO2 for several days until the cytopathic effect was observed. Then, the supernatant was harvested, aliquoted and frozen at − 80 °C until titration. The viral sequence was controlled by Next Generation Sequencing. NanoLuciferase viral inhibition To assess the impact of hit compounds on viral fitness, we used a recombinant virus expressing nLuc as a reporter gene. One day prior to infection, 40,000 Vero-E6 cells per well were seeded in a 96-well plate and incubated at 37 °C with 5% CO2 in 100 µL of medium supplemented with 5% fetal bovine serum (FBS). The following day, the cell culture medium was removed, and the cells were infected at a multiplicity of infection (MOI) of 0.001. After a 1-h incubation period, the viral inoculum was removed. Hit compounds dissolved in 100% DMSO were diluted in cell culture medium without FBS, using a 9-point, 2-fold serial dilution, with final concentrations in wells ranging from 20 µM to 39 nM. Cells were cultured for 72 h post-infection, after which luminescence was measured following the manufacturer’s protocol (N1120, Promega). Briefly, 10 µL of cell culture supernatant was diluted 10-fold into a final reaction volume of 100 µL in a flat-bottom white polystyrene 96-well plate (MSSWNFX40, Millipore®). The mixture was incubated for 10 min with the reconstituted reagent. Luminescence measurements were taken using a VICTOR® Nivo™ Plate Reader (PerkinElmer). Data are expressed as normalized luminescence using untreated infected cells and non-infected cells as controls. Lysis plaque viral titration Viruses’ titers were determined by lysis plaque assay. 1. 106 Vero-E6 cells were seeded to each well of 6-well plates and cultured overnight at 37 °C, 5% CO2. Viruses were serially diluted in DMEM without FCS and 400 µl diluted viruses were transferred onto the monolayers. The viruses were incubated with the cells at 37 °C with 5% CO2 for 1 h. After the incubation, inoculum was removed, and an overlay medium was added to the infected cells per well. The overlay medium per well contained MEM 1× (prepared from MEM 10 × (21430020, Gibco™), distilled water (15230162, Gibco™), L-glutamine (25030123, Gibco™) and gentamicin 10 mg/ml (15710064, Gibco™)), 0.25% sodium phosphate (25080094, Gibco™), and 1X AVICEL (RC-581, Dupont). After 3 days of incubation, the plates were stained with crystal violet (V5265, Merck) and lysis plaques were counted to assess virus titer. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1