High throughput screening for SARS-CoV-2 inhibitors targeting 5 helix bundle

Abstract

SARS-CoV-2 and other related viruses enter host cells via receptor recognition and membrane fusion. A crucial part of this is mediated by 5HB which is capable of binding to the viral spike heptad repeats (HR2) making 5HB a potential druggable target of virus entry. Thus, we constructed a 5-Helix Bundle (5HB) pentamer assay for the purpose of identifying potential inhibitors SARS-CoV-2 virus entry. Following implementation and optimization into a 1536 well format, we validated this assay via a pilot HTS and proved we were able to find small molecule inhibitors that appear to compete with the 5HB binding to HR2. This allowed us to push forward and complete the full HTS campaign testing 635,262 compounds. Upon completion of the 5HB pentamer HTS, we also tested and validated a monomer version of the 5HB assay against a pilot screen and then used it to help confirm on-target activity. This allowed for the selection of 130 compounds which were tested in dose titration format against the 5HB pentamer assay. The same compounds were tested in secondary cell-based assays for SARS2 and Machupo virus entry via a dual luciferase transient transfection system. We also incorporated a live/dead cytotoxicity counterscreen. At the conclusion of these screens, 41 compounds were found to be selective inhibitors of the 5HB pentamer assay. From these assays, 31 compounds and analogs were selected which were tested in both the pentamer and monomer assays. 5 compounds emerged which showed good potency in both assays which were then tested in the SARS pseudo virus assay to round out this exercise.

1. Introduction

Although the COVID-19 pandemic has come to an end, we must try to prevent this from happening again. Besides the devasting loss of life, the initial stages of the pandemic had an estimated cost of $13 trillion dollars[1]. To prevent another pandemic, apriori strategies and therapies must be in place to treat any outbreak that occurs. While vaccines and some treatments are already available, none are full effective, so we have leveraged our drug discovery capabilities available through the Midwest AViDD Center, in this case targeting virus entry as the key element. SARS-CoV2 virus enters the host cell by binding its spike protein to the hACE2 binding site of the host cell[2]. If the spike protein can be disrupted, then entry of the virus into the host can be blocked. The spike protein is made up of S1 and S2 regions. S1 protein is responsible for binding to ACE2 of the host cell[3]. The S2 region consists of heptad repeats, formally known as 1 and 2 (HR1 and HR2) [4]. We hypothesized that by inhibiting the binding of the heptad repeats, the virus will no longer be able to get into the host cell. Therefore, we implemented 5HB (5 helix bundle) into a biochemical screen to bind to HR2[5]. Fig. 1 illustrates the study design and implementation of a High-Throughput Screening (HTS) assay to identify inhibitors of SARS-CoV-2 entry targeting the 5HB. The goal is to inhibit this binding with drug like small molecules, which when on target would be effective at blocking virus entry.

Fig. 1.

Design and implementation of a high-throughput screening (HTS) assay to identify inhibitors of SARS-CoV-2 entry targeting the 5-helix bundle (5HB). (A) Schematic representation of the 5HB construct. The 5HB mimics the post-fusion conformation of the SARS-CoV-2 spike protein’s heptad repeat regions. It consists of three HR1 domains (red) and two HR2 domains (green), connected by linkers, forming a stable five-helix bundle. The N-terminus (H2N) and C-terminus (COOH) are indicated. (B) Structural model of the 5HB with synthetic HR2 peptide. Based on PDB entry 7Y9N, this model illustrates the helical arrangement. Top and side views (rotated 90 degrees) are shown, with HR1 helices in red and HR2 helices in green and blue(synthetic HR2 peptide). This highlights the overall structure. (C) Fluorescence polarization (FP) assay design. The assay is designed to identify small molecules that disrupt the interaction between 5HB and HR2. A schematic representation of the 5HB (three red and two green helices, labeled N and C termini) is shown alongside a schematic of the HR2 peptide labeled with the AF546 fluorophore (blue helix, labeled N and C termini). The small molecule "compound" is depicted. The assay principle is that a compound that blocks the binding of 5HB with the free, labeled HR2, will cause a lower polariz signal.

Thus, we developed, optimized, and miniaturized the 5HB -HR2 binding assay into a fluorescence polarization assay (FP) to find inhibitors that prevent the binding. The assay was first developed in 384 well format using the 5HB in a pentamer arrangement. The assay was then miniaturized to 1536 well format, tested in a pilot screen of over 14,000 compounds followed quickly by a full HTS campaign testing over 630,000 compounds. Inhibitors were confirmed in secondary assays and the most potent and selective in those assays were tested in titration assays including multiple counterscreen assays. These assays included a confirmatory SARs COV-2 pseudovirus assay and two assays that would eliminate compounds from further pursuit including the Machupo pseudovirus assay and a cytotoxicity assay.

At the completion of the 5HB pentamers HTS, we also developed and used a 5HB monomer assay to help identify hits that may not be found in the pentamer assay. This resulted in 5 compounds that showed inhibition in both the monomer and pentamer assay. These compounds were finally tested in a SARs COV-2 pseudo virus assay, none of which proved to block infection in that assay.

2. Materials and methods

2.1. 384 well initial assay and 1536-well format assay optimization

The assay was initially developed in 384 well format at the University of Minnesota in the Li Lab and transferred to the UF Scripps Molecular Screening Center[6] where it was then readily miniaturized into a 1536 well format. To establish the 384 well assay, (1) small-molecule compound is incubated with the protein (5 helix-bundle or 5HB); (2) the labeled probe (fluorescently labeled HR2 peptide) is added to the mixture of the compound and the protein; (3) the FP signal is measured for evaluation of whether the compound blocks the formation of the protein/probe complex. In the 384 well assay, after titration of the protein and probe, the final concentrations of the labeled HR2 and 5HB was determined to be 5 nM and 4 μM, respectively which provided for a maximal change in FP signal which was measured after 1 h of room temperature incubation. Finally, the assay was tested for DMSO tolerance up to 2 % DMSO with no effects. These outcomes from these optimizations are illustrated in Fig. 2. The assay was then transferred to UF Scripps and miniaturized into 1536 well format for compound screening. To further reduce the interference of FITC-labeled HR2 in the FP assay, a longer wavelength fluorophore, AF546, was used to conjugate HR2 peptide. The original 384 assay was tested using the original HR2-FITC compared to HR2-AF546 (Fig. 3A). The miniaturized 1536 well format was compared to the 384 well format (Fig. 3B). Under matched conditions in 1536 well format, we then titered the 5HB and HR2 and took reads over time (Fig. 3C-D). We noticed that the assay initiates instantly with no delay. Using final conditions of 25nM 5HB and 2.5nM HR2, we confirmed inhibition with control compound P083 titration [5,7,8] (Fig. 3E). The final assay protocol can be found in Table 1.

Fig. 2. Assay development.

(A) HR2-FITC (HR2-*) concentration titration assay tested. (B) 5HB(pentamer) titration using 5nM HR2-FITC after 1 h incubation. (C) DMSO tolerance comparing assay buffer vs 2 % DMSO assay buffer after 1 h incubation with a titration of 5HB (pentamer).

Fig. 3. HTS assay optimization.

(A) Initial HR2-FITC data vs HR2-AF546 data. 50nM 5HB (pentamer) and 5nM HR2 used. The inhibitor used in the assay was 1uM P083 after 1 h incubation (B) 384 well data vs 1536 well data using AF546 labeled HR2. 50nM 5HB(pentamer) 5nM HR2 tested. (C) 5HB (pentamer) titer using 5nM HR2-AF546. Assay tested at different incubation times. (D) HR2-AF546 titer using 25nM 5HB (pentamer). Assay tested at different incubation times. (E) Dose response of P083 using 3 different pintool volumes. IC50s are shown below the graph.

Table 1.

HTS protocol (pentamer assay).

Order	Step	Condition	Comments
1	5HB dispense	3uL/well	25nM final concentration in Assay buffer (0.01M HEPES pH7.4, 0.15M NaCl, 3mM EDTA, 0.05 % v/v Surfactant P20)
2	Pin compounds	57nL/well	Kalypsys Pintool dispense 0.75 % DMSO final
3	Incubate	1 h	Room temperature
4	AF546-HR2 dispense	3uL/well	2.5nM final concentration in assay buffer
5	Incubate	30 min	Room Temperature
6	Read	FP	540nM, 590nM, 590nM Pherastar

2.2. Machupo (FLUC) and SARS2 (NLUC) cell-based luciferase assays

The MLV-based pseudoviruses (PVs) were kindly provided by Dr. Hyeryun Choe (Boston Children’s Hospital, Boston, MA, USA). To produce SARS2 GP and NLUC expressing PV, HEK293T cells (ATCC) were co-transfected with the MLV-gag/pol [9], luciferase expressing plasmid, pQ-F-luc-deltaIRES, or pQ-N-luc-deltaIRES, and pCAGGS-SARS2 (BA.5)-dCT19 into HEK293T cells at ratio 5:5:1, and the supernatant containing PVs was harvested at 43 hs post-transfection. For Machupo GP and FLUC expressing PV, MLV-gag/pol, pQ-F-luc-deltaIRES and pcDNA3.1-MACV-GPC plasmids were co-transfected into HEK293T cells at 1:1:1 ratio and the supernatant was harvested at 48 hs post-transfection. All PVs were clarified by 0.45 um filter and aliquoted for storage at −80°C.

The 1536 well plate PV cell-based assay protocol starts with thawing the hACE-2-H1299 cells (kindly provided by Dr. Choe, Boston Children’s Hospital, Boston, MA, USA) which are dispensed at 250 cells / 2 μL/well into a 1536 well assay plate (Aurora EWB0-42000A) followed by addition of 10 nL of test compounds or vehicle (DMSO). After 24 hs incubation at 37°C and 5 % CO₂, 2 μL / well of assay media (RPMI-1640 (Gibco 61870) supplemented with 10 % FBS (Sigma F2442), 1X Antibiotic-Antimycotic (Gibco 15240-062) and 1 μg/ml puromycin (InvivoGen ant-pr-1) was dispensed for column 1-3 and 2 μL / well of Machupo FLUC PV (1:50 final dilution) and SARS NLUC PV (1:10 final dilution) in assay media for column 4-48. The plates were further incubated for 48 hs at 37°C and 5 % CO2 and then removed from that environment and incubated at room temperature for 10 min followed by addition of 2.5 μL / well of One Glo reagents (Promega N1650). After 10 min incubation at room temperature, firefly luciferase expression was measured using a PHERAstar multimodal plate reader (BGM LABTECH). Subsequently, 2.5 μL / well of NanoDLR Stop & Glo reagents (Promega N1650) was added. After 10 min incubation at room temperature, nano luciferase activity was measured using the PHERAstar. Plate Statistics were calculated using high control wells: hACE-2-H1299 cells + assay media + vehicle (DMSO); low control and data wells: hACE-2-H1299 cells + Machupo FLUC and SARS NLUC PVs + test compound or vehicle. These details of these assays are also described in further detail in these publications [10,11].

2.3. Cytotoxicity assay

After thawing the hACE-2-H1299 cells, the cells seeded at 250 cells / 2 μL / well for column 4-48 of the assay plates (Aurora EWB0-42000A). For column 1-3, 2 μL/well of assay media was dispensed. After addition of 10 nl of test compounds or vehicle (DMSO), plates were incubated 24 hs at 37°C and 5 % CO2. Following that incubation, 2 μL / well of assay media was added to all wells. The plates were further incubated for 48 hs at 37°C and 5 % CO2. The plates were then removed and incubated at room temperature for 10 min followed by the addition of 4 μL/well of CellTiter-Glo reagents (Promega G7573). After 10 min incubation at room temperature, luminescence was measured using the PHERAstar.

2.4. Screening Libraries

The UF Scripps Drug Discovery Library (UF SDDL) currently consists of 666,120 unique compounds representing a diversity of drug-like compound scaffolds targeted to traditional and non-traditional drug-discovery biological targets. It serves as the foundation and functional screening library of the Midwest AViDD HTS Core B facility and has been successfully tested against 14 targets from that program over the past 3 years [11-14]. UF SDDL compounds are selected based on scaffold novelty, physical properties and spatial connectivity. The UF SDDL contains focused sub-libraries targeted to popular drug-discovery targets (i.e. Bioactive lipid collection; FDA-approved drug collection, NCI oncology set), drug-discovery target classes (i.e. kinases/transferases, GPCRs, ion channels, NRs, hydrolases, transporters), diverse chemistries (i.e. click-chemistry, PAINS-free, Fsp3 enriched, and natural product collections), and desirable physical properties (“rule-of-five”, “rule--of-three”, polar surface area, etc.)[15-20]. For pilot scale HTS and assay validation we typically rely on either the LOPAC^®1280 collection of drug-like small molecules or the Maybridge HitFinder (14,400 compounds). In this case we opted for the Maybridge library because it consists of not only a slightly larger library but is representative of drug like molecules that span the chemical diversity and space of a much larger library.

2.5. Data analysis

All data files were uploaded into the UF Scripps institutional HTS database (Symyx Technologies, Santa Clara, CA) for plate QC and hit identification. Activity for each well was normalized on a per-plate basis using the following equation:

$$\%activity=1000\times \left(\frac{Testwell-MedianLowControl}{MedianHighControl-MedianLowControl}\right)$$ (1)

Where “High Control” represents wells containing 5HB only; while “Low Control” represents wells containing DMSO and 5HB and HR2 and finally the “Data Wells” contain the same as the low control but also with test compounds. The Z’ and S:B were calculated using the High Control and Low Control wells. In each case, a Z’ value greater than 0.5 was required for a plate to be considered acceptable [21]. The high throughput assay design and principle can be found in Fig. 4.

Fig. 4. Assay conditions.

(A) The biochemical assay was performed by adding 5HB and a small molecule to fluorescently labeled HR2 peptide. After incubation the FP signal is measured. (B) No inhibition yields a high FP signal and an inhibitor will decrease the FP signal similar to the lack of 5HB.

3. Results

3.1. Maybridge pilot screen

The first step of the HTS campaign was screening the assay against a subset of the UF SDDL library. In this assay, 14,306 compounds from the Maybridge library were tested at a single concentration in triplicate at a final nominal concentration 8.3 μM. Raw assay data was imported into the UF Scripps database and activity of each compound was calculated on a per-plate basis using the equation shown in the methods.

Assay performance was excellent with an average Z’ of 0.80±0.02 and an average signal-to-background ratio (S:B) of 6.78±0.42 (n=36 plates). A summary of the results of the Maybridge pilot assay are shown in Table 2. A mathematical algorithm was used to determine active compounds. Here we applied what we call an “Interval Cutoff” which is a routinely used practice in HTS [22]. In this case the average activity of all wells tested that fall within −3SD of the low control and +3SD of the high control was used as a cutoff parameter, i.e., any compound that exhibited greater percent activation than the cutoff parameter was declared active. Using this “Interval Cutoff” criteria of 10.52 % the pilot assay yielded 9 active compounds (“hits”). The assay showed a low hit rate, but demonstrated that the assay is stable and ready for full HTS.

Table 2.

5HB HTS campaign.

5HB (Pentamer) HTS Campaign	# compounds Tested	Replicates	Screening Concentration	Hit Cutoff	# Hits	Hit Rate
Maybridge pilot	14,306	3X	8.3uM	10.52 %$	9	0.06 %
Primary HTS	635,262	1X	8.3uM	13.69 %$	1,530	0.24 %
Confirmation Screen	1,528	3X	8.3uM	13.79 %&	121	7.92 %
Late Pin Counterscreen	1,528	3X	8.3uM	13.79 %&	104	6.81 %
Titration	130	3 × 10 point 1:3 dilution	20.75uM start	50 % activity cutoff	61	46.90 %
Late pin Titration Counterscreen	130	3 × 10 point 1:3 dilution	20.75uM start	50 % activity cutoff	44	33.80 %
Machupo FLUC Titration	130	3 × 10 point 1:3 dilution	20.75uM start	50 % activity cutoff	26	20.00 %
SARS NLUC Titration	130	3 × 10 point 1:3 dilution	20.75uM start	50 % activity cutoff	22	16.90 %
SARS Entry CTG Cytotoxicity Titration	130	3 × 10 point 1:3 dilution	20.75uM start	50 % activity cutoff	15	11.50 %
5HB (Monomer)
Maybridge pilot	14,306	3X	8.3uM	10.47 %$	16	0.11 %
5HB (Monomer) Titration	52	4 × 10 point 1:3 dilution	20.75uM start	50 % activity cutoff	5	1.92 %
5HB (Pentamer) Titration	52	4 × 10 point 1:3 dilution	20.75uM start	50 % activity cutoff	5	1.92 %

^$Interval Cutoff.

^&DMSO Cutoff

3.2. Primary HTS

The full HTS campaign ran similarly to the pilot but now included 635,262 tested in singlicate at a nominal concentration of 8.3 μM. Assay performance was excellent with an average Z’ of 0.76±0.02 and an average signal-to-background ratio (S:B) of 8.09±1.86 (n=510 plates). A summary of the results of the primary HTS are shown in Table 2. The “Interval Cutoff” criteria of 13.69 % yielded 1,530 active compounds (“hits”). The assay performance was stable throughout the full HTS. (Fig. 5A-B) as can be seen by the Z score analysis from the primary HTS along with the potency of P083 reference.

Fig. 5. 5HB (Pentamer) HTS campaign.

(A) Z scores analysis of the controls and active compounds found during the HTS campaign. Low control (green), high control (yellow) and hits (red) are shown in the z score plot. (B) The P083 reference inhibitor response from batch to batch during the HTS. PRUN= primary run number; B= batch number during that run. The IC50 over the campaign was 1.59±0.43ηM. (C) The Venn overlap of active compounds from the confirmation and counterscreen. 23 compounds were found selective for the primary assay. (D) The Venn overlap of 4 of the assays (Late pin not included). 41 compounds inhibit only 5HB assay but 3 inhibit the Machupo assay and 4 inhibited both the Machupo assay and the SARs assays.

3.3. Secondary assays

3.3.1. Confirmation screen

The secondary assays employed the same reagents, protocols, and detection systems as the primary screen, but tested the 1,528 (2 unavailable) selected compounds in triplicate. The 5HB FP assay performance was consistent with an average Z’ of 0.76 ± 0.02 and a S:B of 5.04 ± 0.45 (N = 6 plates). In this case we used a DMSO cutoff, where we take the average and 3SDs of plates with just DMSO to establish the hit cutoff. This cutoff of 13.79 % gave 121 active compounds. The results can be found in Table 2.

3.3.2. Late pin counterscreen

The late pin counterscreen assay employed the same reagents, protocols, and detection systems as the primary screen, but pinned the 1,528 selected compounds in triplicate just after the reaction end point and directly before reading the FP signal on the Pherastar. In this way we should be able to identify compounds that exhibit fluorescence properties that affect the signal only and not the biology of the assay. Hits in this assay would be removed from further pursuit. The SCV-2 5HB FP INH late-pin assay performance was consistent with an average Z’ of 0.78 ± 0.01 and a S:B of 5.25 ± 0.31 (N = 6 plates). The same DMSO cutoff was also used here which identified 104 active compounds. The results can be found in Table 2. A comparison of the actives found in the confirmation and counterscreen (Fig. 5C) determined that 23 compounds selectively inhibit the 5HB assay without responding in the late pin counterscreen.

3.4. Concentration response assays

The concentration response assay employed the same reagents, protocols, and detection systems as the primary screen, but tested the selected compounds as 10-point dose-response titrations (3-fold dilutions) in triplicate. In this assay 130 compounds were tested. These are analogous to the 121 confirmed active from the assay described above plus the 9 hits found in the Maybridge pilot HTS. We chose not to eliminate any of the compounds moving forward based on the late pin results at this stage. The SCV-2 5HB FP INH titration assay performance was consistent with an average Z’ of 0.79 ± 0.01 and a S:B of 6.15 ± 0.32 (N = 6 plates). 61 compounds achieved a threshold of 50 % inhibition or greater and thus were determined active.

3.4.1. Late pin counterscreen

Similarly, we ran the SCV-2 5HB FP INH late pin concentration response assay and its performance was consistent with an average Z’ of 0.78 ± 0.03 and a S:B of 6.00 ± 0.31 (N = 6 plates). 44 compounds reached 50 % inhibition or greater and were determined active.

3.4.2. Additional titration assays

At this stage we added three additional counterscreens. This was facilitated in part using the assays from a parallel HTS project being conducted in our lab for the Midwest AViDD center that tested Machupo (FLUC) and SARS (NLUC), separately in this case, in cell based luciferase assays looking for inhibitors [11]. These assays were previously optimized using hACE2-H1299 cells infected using pseudoviruses of MACV-FLUC and SARS2-N-LUC.

The Machupo FLUC titration assay performance was robust with an average Z’ of 0.58 ± 0.05 and a S:B of 20.96 ± 2.01 (N = 6 plates). 26 compounds achieved >50 % inhibition and were determined active. The SARS NLUC titration assay performance was also robust with an average Z’ of 0.64 ± 0.04 and a S:B of 36.29 ± 5.45 (N = 6 plates). 22 compounds achieved >50 % inhibition and were determined active.

Using the same cells described in the pseudovirus assays above, we were also able to test for cytotoxicity of the compounds using CellTiter-Glo (CTG) using formerly published methods [23]. The CTG concentration response assay performance was robust with an average Z’ of 0.79 ± 0.03 and a S:B of 19.22 ± 4.39 (N = 6 plates). 15 compounds reached >50 % inhibition and were determined active. All the results from all the concentration response assays can be found in Table 2. The Venn overlap from 4 of the assays is shown in Fig. 5D. The assay showed some selectivity with 41 compounds found active in just the 5HB assay. There were also 7 compounds that were found active in either the Machupo assay or the SARS assay or both, while not being cytotoxic.

3.5. Additional screening

3.5.1. 5HB monomer assay

Following the HTS campaign, a monomer version of the 5HB was made to test its effectiveness under the same assay conditions. The assay was run similarly to the pentamer assay except 2.5nM 5HB and 2.5nM HR2 was used. The monomer assay was run as a point of validation against the Maybridge Hitfinder library to help potentially reveal hits that the pentamer assay was insensitive too. The same14,306 Maybridge compounds were tested at a single concentration in triplicate at a final nominal concentration 8.3 μM. Assay performance was excellent with an average Z’ of 0.81±0.03 and an average signal-to-background ratio (S:B) of 10.00±1.42 (n=36 plates). A summary of the results of the Maybridge pilot assay are shown in Table 2. 16 compounds were found with activity greater than the interval hit cutoff of 10.47 %. A Venn overlap of the active inhibitors was done for the pentamer vs monomer assay (Fig. 6A). Notably 4 of the hits overlapped between these two assays which was encouraging. Upon in depth review of the results from both Maybridge pilot HTS campaigns by Thomas Bannister, an industry trained medicinal chemist and leader of the Midwest AViDD Chemistry Core, commercially available confirmation samples of the most tractable hits and close analogs, totaling 52 compounds, were purchased and then tested in titration assays in quadruplicate at a starting nominal concentration of 20.7 uM.

Fig. 6. 5HB Pentamer vs 5HB monomer assays.

(A) Venn overlap of both assays run in triplicate vs the 14.4K Maybridge hitfinder collection with actives based on the hit cutoff from the assays (Table 2). (B) The 5 compounds that showed overlapping activity from the 52 compound titration assays. The top 3 compounds showed significant inhibition in both assays.

3.5.2. Monomer assay

Assay performance was excellent with an average Z’ of 0.78±0.01 and an average signal-to-background ratio (S:B) of 8.70±0.17 (n=4 plates). Results can be found in Table 2.

3.5.3. Pentamer assay

Assay performance was excellent with an average Z’ of 0.64±0.02 and an average signal-to-background ratio (S:B) of 5.78±0.20 (n=4 plates). Results can be found in Table 2.

Five compounds showed inhibition that repeated in both assays. These 5 compounds were tested in the cell-based pseudovirus assays, but unfortunately, none of them showed any inhibition. The compounds with both monomer and pentamer assay activity are found in Fig. 6B.

4. Discussion

Originally, we used the pentamer assay for a full HTS campaign. During that campaign, 121 confirmed actives from single point assays were confirmed in triplicate and selected for concentration response analysis. We also added 9 compounds from the Maybridge pilot to make 130 compounds. All 130 compounds were tested in dose response in the pentamer assay as well as the late pin counterscreen assay. At this stage we added three additional assays (Machupo FLUC, SARS NLUC, Cell titer Glo Cytotoxicity assay). Fig. 5D displays the Venn overlap of actives from all these titration assays. We anticipated that the 5HB pentamer assay and the pseudovirus SARs entry assay would show overlap and reveal compounds that would be selective for SARs entry. However, none did, and all others showed overlap in the assays that ruled out specificity for SARs entry; ie the cytotoxicity assay and the Machupo entry assay. Thus, we concluded with the full large scale 630K+ compound HTS campaign.

At this point, in order to potentially identify more compounds of interest using an assay that still monitored 5HB interaction, albeit possibly in a more sensitive way, we developed the monomer assay to help identify molecules that the pentamer was insensitive to. The monomer assay was only run against the Maybridge Hitfinder collection of 14.4K compounds. This allowed the comparison of the data from the 5HB pentamer Maybridge pilot to the monomer Maybidge pilot (Fig. 6A Venn). 5 compounds hit only the pentamer and 4 hit both, making these 4 hits of particular interest. However, to expand on the number of hits from these pilots, our medicinal chemist reviewed the hits from both assays and selected a total of 52 compounds. All 52 compounds were tested in both monomer and pentamer assays for IC50 determination. In conclusion, 5 of these compounds showed inhibition in both the monomer and pentamer assays, and those were selected to be tested in the cell based pseudovirus assay, which would provide confirmation that these compounds would be potential SARs entry inhibitors. Unfortunately, none of them showed any inhibition in that assay. At this stage the AViDD program was sunsetting, and we concluded all HTS efforts directed at the 5HB SARS entry target.

While, we were successful at rapidly screening large libraries of drugs to find potential inhibitors of virus entry, it is unlikely we found any that truly block infection. While this may be disappointing it also allows for readers to learn from the lessons described herein. There may be several reasons that this occurred. As noted, the reaction was extremely fast so the affinity for the 5HB to the HR2 may be too difficult to disrupt at the current concentrations. Although we titered the protein concentration to address this we were unable to find a low enough amount of 5HB that demonstrated a kinetic response over time. Secondly, many of the compounds that showed a decrease in FP, also seemed to have similar effects in the late pin counterscreen indicating that they are likely fluorescence artifacts. This assertion is also supported by the rapid affiliation of 5HB and HR2, wherein the activity of a compound in the late pin assay would indicate an innate effect of its own fluorescence and not the binding. However, knowing this, we decided not to disclude these from concentration response profiling which preserved compounds that may have had subtle effects on the 5HB interaction but may also have artifactual effects due to fluorescence. Although not available at the time of this publication, an alternative counterscreen that does not rely on the same fluorescence detection would likely have been better, and we may have ruled out compounds with the innate fluorescence issues. Finally, the conformational nature of the spike protein makes the inhibition of the helix bundle difficult to fully inhibit. The spike protein contains several spikes, not just one [24]. The inhibitors we found might be blocking some of the HR binding, but may not be capable of blocking all of the spike proteins on the virus. A remedy to this may be to treat the pseudovirus in the cell-based assays with the compound for a defined and extended period before adding to the cells. This may allow for the inactivation of the spike proteins rendering them unable to bind to the cell to obtain entry. We did test this idea, altering the pseudovirus assay by pre-incubating just the pseudovirus with the compounds for 1 h prior to adding it to the cells. Unfortunately, it also did not yield any inhibition of infection with any of the compounds. At this stage of the HTS, we concluded all further testing since the results indicated that none of the hits were truly leads worthy of pursuit. While not the outcome we had hope for, our strategy allowed for rapid determination to halt the project quickly resulting in cost effectively decision making.