Identification of entry inhibitors with 4-aminopiperidine scaffold targeting group 1 influenza A virus

Abstract

Influenza A viruses (IAVs) cause seasonal flu and occasionally pandemics. The current therapeutics against IAVs target two viral proteins - neuraminidase (NA) and M2 ion-channel protein. However, M2 ion channel inhibitors (amantadine and rimantadine) are no longer recommended by CDC for use due to the emergence of high level of antiviral resistance among the circulating influenza viruses, and resistant strains to NA inhibitors (oseltamivir and zanamivir) have also been reported. Therefore, development of novel anti-influenza therapies is urgently needed. As one of the viral surface glycoproteins, hemagglutinin (HA) mediates critical virus entry steps including virus binding to host cells and virus-host membrane fusion, which makes it a potential target for anti-influenza drug development. In this study, we report the identification of compound CBS1116 with a 4-aminopiperidine scaffold from a chemical library screen as an entry inhibitor specifically targeting two group 1 influenza A viruses, A/Puerto Rico/8/34 (H1N1) and recombinant low pathogenic avian H5N1 virus (A/Vietnam/1203/04, VN04_Low). Mechanism of action study shows that CBS1116 interferes with the HA-mediated fusion process. Further structure activity relationship study generated a more potent compound CBS1117 which has a 50% inhibitory concentration of 70 nM and a selectivity index of ~4000 against A/Puerto Rico/8/34 (H1N1) infection in human lung epithelial cell line (A549).

1. Introduction

Influenza A viruses (IAVs) are negative-sense, single-stranded, segmented RNA viruses from the Orthomyxoviridae family. IAV has caused four influenza pandemics in recent history and the latest H1N1 pandemic (2009–2010) spread to 199 countries with at least 151,700 respiratory and cardiovascular deaths globally (Dawood et al., 2012). Seasonal influenza also results in 3–5 million severe cases with 290,000 to 650,000 deaths worldwide annually (WHO, 2018). In addition, a few avian IAVs such as H5N1, H7N9, and H9N2, have crossed the species barrier and caused human infections (Short et al., 2015). These emerging viruses can cause high mortality rates in humans due to the lack of pre-existing immunity and limited therapeutic options. The concern about potential pandemic risk of the interspecies transmission of avian IAVs has been heightened.

Vaccination is currently the major strategy to prevent IAV spread. However, vaccines are of little use when a rapid pandemic emerges, because 1) a vaccine can only be developed after the characterization of the pandemic strain and 2) manufacturing vaccines takes 5–6 months (Wong and Webby, 2013). Hence, antivirals represent a complementary strategy to fight against influenza pandemics. Two classes of antiviral drugs have been approved by U.S. Food and Drug Administration (FDA), targeting viral proteins - M2 ion-channel and neuraminidase (NA)(Julianna et al., 2018). However, IAV strains resistant to these antiviral drugs (particularly M2 ion-channel inhibitors e.g. amantadine and rimantadine) have emerged throughout the world. Almost all seasonal viruses show resistance to adamantanes, and these M2 ion-channel inhibitors are no longer recommended by U.S. Centers for Disease Control and Prevention for treatment of IAV infection. Regarding NA-targeting drugs (oseltamivir, zanamivir, laninamivir and peramivir), though most recently circulating IAVs in US have been susceptible to these NA inhibitors, high rates of oseltamivir resistance (>90%) were observed in the United States during the 2008 to 2009 influenza season (Dharan et al., 2009). In addition, baloxavir marboxil, which inhibits the cap-dependent endonuclease activity of the PA protein of influenza A and B viruses, was approved for the treatment of uncomplicated influenza in Japan and the US in 2018 (Mifsud et al., 2019). However, resistant IAV strains quickly emerged during the first influenza season in Japan after baloxavir had been licensed (Takashita et al., 2019). Another polymerase inhibitor favipiravir was approved in Japan in 2014, but the use has been strictly regulated due to its risk for teratogenicity and embryotoxicity (Furuta et al., 2017). These facts underscore the urgent need of developing novel anti-influenza therapies targeting other viral factors or host factors.

Hemagglutinin (HA), the viral surface glycoprotein of IAV, mediates virus entry, and plays an important role in host immune responses by harboring the major antigenic sites. Based on the antigenic properties of HA, IAVs can be classified into 18 different HA subtypes (H1-H18), which can be further divided into group 1 (H1, H2, H5, H6, H8, H9, H11, H12, H13, H16, H17, H18) and group 2 (H3, H4, H7, H10, H14, H15) phylogenetically (Wu and Wilson, 2018). The mature HA is a spike-like homotrimer, composed of a globular head region and a stem region. The receptor binding site (RBS) located in the head region of HA binds to sialic acids on the cell surface and initializes virus entry via endocytosis. Once inside endosome, the acid environment induces conformational change of HA stem region, resulting in the fusion of virus membrane with host endosomal membrane and the release of viral RNA genomes into the cytoplasm. Because of their essential functions in virus entry, evolution of RBS and the stem region is severely constrained, making them desirable targets for antiviral development (Wu and Wilson, 2017).

To identify potential inhibitors targeting HA-mediated IAV entry, we performed a comparative high-throughput screening (HTS) assay to screen a small molecule library of 19,200 compounds against the infections of pseudotyped influenza A/H5N1, Marburg or Lassa virus in human lung epithelial cell line (A549). In this study, we report the discovery and characterization of a novel small-molecule IAV entry inhibitor. We demonstrate that 2,4-dichloro-N-(1-isopropyl-4-piperidinyl) benzamide can specifically inhibit the infection of two group 1 IAVs (H1N1 and H5N1) in A549 and Madin-Darby canine kidney (MDCK) cell lines, by interfering with the HA-mediated membrane fusion process.

2. Materials and methods

2.1. Cell Culture and Plasmids

Human embryonic kidney cell line (293T), human lung epithelial cell line (A549) and Madin-Darby canine kidney cell line (MDCK) were cultured in Dulbecco’s modified Eagle’s medium (DMEM; Cellgro, Manassas, VA) supplemented with 10% fetal bovine serum (GIBCO, Carlsbad, CA), 100 μg/mL streptomycin, and 100 U penicillin (Invitrogen, Carlsbad, CA). The following plasmids were used for the production of pseudovirions for HTS: hemagglutinin (HA), isolated from a highly pathogenic avian influenza virus, A/Goose/Qinghai/59/05 (H5N1) strain, Marburg virus (MARV) glycoprotein (GP), Lassa virus (LASV) envelope GP and the HIV-1 proviral vector pNL4–3.Luc.R^–E^–.

2.2. Generation of pseudoviruses

All types of pseudovirions- HIV/MARV, HIV/H5N1 and HIV/LASV were produced by transient cotransfection of human 293T cells using a polyethylenimine–based transfection protocol. Plasmids encoding MARV GP, HA/NA, LASV GP and replication-defective HIV-1 vector (pNL4–3.Luc.R^–E^–) were used for transient co-transfection into 293T cells. Six hours after transfection, cells were washed with phosphate-buffered saline (PBS), and 40 mL of fresh medium was added to each plate (150 mm). Forty-eight hours after transfection, the supernatants were collected and filtered through a 0.45-μm pore size filter (Nalgene, Rochester, NY). The pseudovirion stocks were stored at 4 °C prior to use.

2.3. Infectious viruses

A/Puerto Rico/8/34(H1N1), recombinant low pathogenic avian H5N1 virus (A/Vietnam/1203/04, VN04_Low), A/HongKong/2/68 (H3N2), A/rhea/North Carolina/39482/93 (H7N1) and H1N1-GFP (A/Puerto Rico/8/34, NS1-GFP) were used in this study. The H5N1 virus based on a human isolate (A/Vietnam/1203/2004) was generated using the reverse genetics system (Steel et al., 2009). This is a recombinant virus without polybasic cleavage site in HA and is considered low pathogenic. The remaining 7 segments are wild-type sequences. It was tested and confirmed by USDA as a low pathogenic virus. All viruses were grown in 10-day old eggs at 37 °C for 2–3 days before collection.

2.4. High-Throughput Screening (HTS)

The Chembridge small molecule library (19,200 compounds) was screened following a previously published protocol with modifications. Briefly, the screen was performed using 384-well plates with 320 compounds arrayed in each plate. The following controls were used in the screen: azidothymidine (AZT; Sigma, St. Louis, MO) was used as a positive control and DMSO was used as a negative control.

A549 cells were seeded into white, flat-bottom, 384-well plates (CulturPlate-384; PerkinElmer, Waltham, MA) at a density of ~1000 cells/well in 30-μL assay medium using JANUS liquid handler MDT (Modular Dispense Technology; PerkinElmer) and incubated at 37 °C, 5% CO₂. Twenty-four hours later, 0.2 μL of each compound was transferred into wells containing 79.8 μL DMEM in an intermediate 384-well plate through a pin tool (V&P Scientific, San Diego, CA) and mixed thoroughly using a JANUS liquid handler MDT. Twenty μL compound plus 20 μL pseudovirus was transferred to the target cell plates. Plates were incubated at 37 °C, 5% CO₂ for 48 h. After incubation, 20 μL of neolite luciferase substrate (PerkinElmer) was added to each well, and plates were incubated at room temperature for 5 to 10 min. Luciferase activity was measured by an EnVision plate reader (PerkinElmer). The data were analyzed using the statistical programming language R as previously described.

2.5. Antiviral evaluation of lead compounds

Four different protocols were used to evaluate the potency of the lead compounds. Please see the supplementary methods for details.

2.6. Cell viability assay

A549 cells were seeded in 96-well plates at density 5×10³ cells/well and incubated at 37 °C. After 24 hours, serial dilutions of each compound (from 100 μM to 45 nM) in 100μl DMEM were added to cells. DMSO was used as a negative control (0.2%). After 48 hours incubation, 50 μL of CellTiter-Glo (Promega, Madison, WI) was added to each well and plates were incubated at room temperature for 5 to 10 min. Luciferase activity was measured by an EnVision plate reader (PerkinElmer). Sample signals were normalized to DMSO control wells. The 50% toxicity concentration (CC₅₀) value, which is the concentration of the compound toxic to 50% of cells, was calculated using Graphpad prism.

2.7. Time-of-addition experiment

Time of addition experiment was performed as previously described with modifications (Cheng et al., 2015). Briefly, A549 cells (5×10³/well) were seeded into 96-well plates 24h before the assay. At −1h, the cells were incubated with H5N1 pseudovirus or H1N1-GFP (multiplicity of infection (MOI)=10, i.e., 10 plaque-forming unit per cell) for 1 hour at 4°C. At infection time point 0 h, the virus was removed, cells were washed twice with PBS and cultured in fresh media. Temperature was shifted to 37°C to trigger virus entry (0h). CBS1116 (10 μM) were introduced at different time points of virus infection. At −1h, the cells were incubated with 10 μM CBS1116 for 1 hour and then replaced by fresh media. At 0h, +1h, +2h and +3h time points, 10 μM CBS1116 was introduced and incubated with the cells till the end of the assay. The plates were incubated for 24 (H1N1-GFP) or 48 hours (H5N1). For H1N1-GFP infection, the cells were fixed with 4% paraformaldehyde (Alfa Aesar, Ward Hill, MA) and stained with DAPI (Sigma, St. Louis, MO). Fluorescence was detected using a Keyence BZ-X710 microscope (Keyence, Itasca, IL) with a 10x objective lens. GFP or DAPI positive cells were counted with BZ-X analyzer (Keyence, Itasca, IL). For H5N1 infection plates, 50 μL/well of neolite luciferase substrate (PerkinElmer) were added, and infectivity was measured by EnVision plate reader (PerkinElmer) after 5–10 minutes incubation at room temperature. DMSO drug vehicle was used as a control at the same time points as the compound. Signals were normalized and percentage of infectivity (% infectivity) was calculated as 100 × (normalized signal). H1N1-GFP was treated with 5 μg/mL tosylsulfonyl phenylalanyl chloromethyl ketone (TPCK)-treated trypsin (Sigma, St. Louis, MO) at 37°C for 30 min and cooled down on ice for 30 min before the assay.

2.8. Cell surface binding inhibition assay

The cell surface binding inhibition assay experiment was designed as follows. The purified HA proteiwith a polyhistidine tag at the C-terminus (1μg) was produced as described previously(Antanasijevic et al., 2016). HA was incubated with different concentrations of the compound (1.6 μM-1000μM) in 50 μl PBS/2%BSA on ice for 30 minutes. After incubation this mix was added to 5×10⁵ A549 cells on ice for 1.5 h. Cells were washed twice with PBS/2%BSA and re-suspended in 30μL anti His-tag antibody conjugated with FITC (Thermo Fisher Scientific 6xHis,FITC conjugate) on ice for 45 minutes. Cells were washed three times with PBS/2%BSA, and then fixed in PBS/2% paraformaldehyde, and subjected to flow cytometry analysis. The signals were normalized by background fluorescence (cells alone) and fluorescence in positive control (1μg his-tagged HA protein with cells). 2,3 and 2,6 sialyllactose receptor analogues (Carbosynth, UK) were also used as controls.

2.9. Hemolysis inhibition assay

The experiment was done following a protocol by Vanderlinden et al (Vanderlinden et al., 2010) with minor modifications. Briefly, 100 μl of HIV/H5N1 pseudovirus was incubated with 100 μL of compounds (serially diluted in PBS) at 37°C for 1 hour, 200 μL of chicken red blood cells (RBCs, 2% in PBS, Innovative Research, USA) was then added. The samples were incubated at 37°C for 10 minutes and centrifuged at 2000 rpm, 4°C for 10 min. The pellets were re-suspended in 500 μL of PBS buffer (PH 5.0) containing the same concentrations of the compounds. Samples were incubated at 37°C for 25 min, and centrifuged at 2000 rpm, 4°C for 10 min. Three hundred microliters of the supernatant were measured for their optical density at wavelength 535nm with an EnVision plate reader (PerkinElmer). 2,6 and 2,3 sialyllactose receptor analogues (Carbosynth,UK) were used as controls. The data was normalized to samples containing DMSO, virus and RBCs only. The IC₅₀ values were calculated from dose response curves.

2.10. Evaluation of the lead compound analogues

Fifty-eight analogues of the lead compound (2,4-dichloro-N-(1-isopropyl-4-piperidinyl benzamide) were purchased (ChemBridge,San Diego, CA) and evaluated for their inhibitory activity using both HIV/H5H1 pseudiovirus and H1N1-GFP virus as described above.

3. Results

3.1. High-Throughput Screening to identify entry inhibitors of IAV

A chemically diverse small molecule library (19,200 compounds) was screened to identify entry inhibitors for three enveloped viruses including Marburg virus (HIV/MARV), Lassa (HIV/LASV) and IAV H5N1 (HIV/H5N1) using an HIV pseudovirus-based entry assay protocol (Wang et al., 2014). Since the screen was performed using three pseudo-viruses in a comparative way, we were able to identify putative influenza virus-specific entry inhibitors which show inhibition against the entry of IAV but not Marburg virus or Lassa virus.

As summarized in Table 1, 55 putative IAV-specific hits (0.28% hit rate) were identified which showed > 80% luciferase activity reduction in HIV/H5N1 plates, but < 30% reduction in HIV/MARV and HIV/LASV plates at compound concentration of 12.5 μM. These compounds were cherry-picked into new 384-well plates and tested for confirmation. The antiviral activities of 48 compounds (0.25% hit rate) were confirmed and they did not show cell toxicity at 12.5 μM. Among them, 9 hits with >90% inhibition against HIV/H5N1 infection were picked for dose-response titration evaluation. Seven hits were further selected based on their high potency against HIV/H5N1 entry with IC₅₀<5μM and high antiviral selectivity index value (SI>10) in A549 cells. The chemical structures, IC₅₀s, CC₅₀s and SIs of the seven hit compounds are summarized in Table 2 and Figure 1A.

Table 1.

HTS Summary

	Number (%)
Compounds screened	19,200
Primary HIV/H5N1a Hits >80% inhibition at 12.5μM	55(0.28%)
Nontoxic hits	48(0.25%)
Specific HIV/H5N1 hits > 90% inhibition at 12.5μM	9(0.04%)
Specific hits with SIb >10	7(0.034%)

^aHIV/H5N1 was generated by transfecting 293T cells with HIV-l expression vector, H5 and N1 plasmids.

^bSelectivity index (SI) values were determined from CC₅₀/IC₅₀.

Table 2.

Hits specific for influenza virus inhibition

Compound No.	Compound structure	HIV/H5N1a
		IC50 (μM)	CC50 (μM)	SI
1		1.2	60.6	50.5
2		1.7	105	61.7
3		3.9	160	44.4
4		0.5	60.6	121.2
5		0.6	99.4	168.4
6		0.5	27.3	69.2
7		0.5	59.2	118.4

^aAntiviral profiles of the top seven hit compounds were obtained from pseudotyped H5N1 infection assay in A549 cells.

Figure 1. The specificity and inhibitory effect of the seven compounds on pseudo and infectious virus.

A. The compounds were tested against HIV/AIV(H5N1), HIV/MARV and HIV/LASV pseudotyped viruses (pv) at 12.5 μM. The compounds were mixed with the pseudoviruses and then incubated with A549 cells for 48 hours. Three independent experiments were performed, and data are means±SD (n=3).

B. Validation of the 7 hits against infectious VN04_Low (H5N1). VN04_Low (MOI=0.001) was incubated with 50 μM of each compound for one hour at room temperature and the mixtures were used to infect A549 cells at 37 °C. After 1-hour infection, cells were washed with PBS twice and fresh media containing the same concentration of the compounds were added. The supernatant was collected after 48 hours, and virus titers (PFU/ml) was measured by plaque assay. Data are means±SD (n=3).

To confirm the potency of these seven hit compounds, they were tested with a recombinant low pathogenic avian H5N1 strain (A/Vietnam/1203/04, VN04_Low) in A549 cells (Tundup et al., 2017). As shown in Figure 1B, 6 out of 7 compounds showed significant inhibition against VN04_Low at 50 μM, among which compound 3 and compound 6 stood out with the strongest inhibition. Compound 3 and 6 showed almost 1,000-fold and 10,000-fold reduction in virus replication respectively. Dose-response titrations (ranging from 0.2μM to 50μM) were then performed for both compounds. Strong antiviral effect of compound 3 against H5N1 infection was observed at 5μM (Figure 2A) while that of compound 6 was observed at 25μM (data not shown) which was very close to its CC₅₀ value (27.3μM). This suggests that the observed antiviral effect of compound 6 (Figure 1B) is likely due to its cytotoxicity. Thus compound 3 (which is also referred to as CBS1116) was chosen for further studies.

Figure 2. CBS1116 is group 1 IAV specific inhibitor.

Dose-response titration of CBS1116 in A549 cells against the infection of A) VN04_Low H5N1 strain (group 1) at MOI of 0.001, B) A/HongKong/2/68(H3N2)(group 2) at MOI of 0.01, C) A/rhea/North Carolina/39482/93(H7N1) (group 2) at MOI of 0.1, D) and E) H1N1-GFP (A/Puerto Rico/8/34, NS1-GFP) at MOI of 0.5. Significant inhibition was observed for H5N1 and H1N1 strains at 5μM. DMSO and Oseltamivir phosphate were used as controls in (D). GFP signals were quantified in Envision plate reader at excitation wavelength of 485nm and emission wavelength of 535nm and normalized with DMSO control in (E). Dose-response titrations were evaluated in MDCK cells for CBS1116 against the infection of (F) A/Puerto Rico/8/34 (H1N1) and (G) VN04_Low (H5N1). Each virus (MOI=0.0001) was incubated with different concentrations of the compound for 1h before infecting MDCK cells for 1h. The viruses were then removed, and the cells were incubated for another 40–72h. The percentage of inhibition was calculated as described in plaque reduction assay in materials and methods. Both inhibition (red) and viability (green) curves are shown and the IC₅₀ and CC₅₀ values were generated by fitting the dose-response curves with four-parameter logistic regression in GraphPad Prism software. Data are means±SD (n=3).

3.2. CBS1116 is a group 1 IAV-specific inhibitor

To evaluate the spectrum of anti-influenza activity of CBS1116, several influenza A viruses, including VN04_Low (H5N1) (group 1), A/HongKong/2/68 (H3N2) (group 2), and A/rhea/North Carolina/39482/93 (H7N1) (group 2), were tested in A549 cells. As shown in Figure 2, CBS1116 displayed no inhibition against H3N2 (Figure 2B) and H7N1 (Figure 2C) at the maximum concentration tested (50μM), while it displayed inhibitory activity against VN04_Low strain in a concentration-dependent manner (Figure 2A).

CBS1116 was further evaluated against the infection of a recombinant H1N1-GFP strain (A/Puerto Rico/8/34, NS1-GFP) to confirm its group-1 IAV specificity. As shown in Figure 2D and 2E, H1N1-GFP infection was completely blocked at 5 μM in A549 cells. These results strongly suggest that CBS1116 is a specific inhibitor of group 1 IAVs.

To compare the antiviral activities of CBS1116 against the group 1 IAV subtypes, we performed dose-response titrations for CBS1116 with A/Puerto Rico/8/34 (H1N1) and VN04_Low (H5N1) strains using plaque reduction assay in MDCK cells. CBS1116 exhibited more potency against H1N1 infection with an IC₅₀ of 0.4 μM (Figure 2F) than H5N1 infection with an IC₅₀ of 13.8 μM (Figure 2G). The compound showed no cell toxicity to MDCK cells at the highest concentrations tested at 100 μM.

3.3. CBS1116 interferes with HA-mediated membrane fusion process but not virus binding to cell surface

Influenza entry starts with (1) virus receptor binding step followed by endocytosis and then (2) pH-dependent HA-mediated membrane fusion step. A time-of-addition experiment was performed with HIV/H5N1 pseudovirus to determine the target stage blocked by CBS1116. As described in Materials and methods, CBS1116 (10 μM) was introduced at various time points of the entry process (Figure 3A). DMSO treated wells at the same time points were used as the controls to normalize data. Interestingly, CBS1116 displayed no inhibition of HIV/H5N1 infection when it was co-incubated with virus at the virus attaching to the cell surface step (−1h). However, CBS1116 showed maximum inhibition (~99%) against virus infection when endocytosis was triggered (0h) and this antiviral effect diminished to 75% and 41% inhibition when it was added at +1h or +2h respectively (Figure 3B). To confirm this finding, we further performed the time-of-addition assay with infectious H1N1-GFP. The ratios of GFP-positive cell number to DAPI-positive cell number were used to evaluate virus infection. As shown in Figure 3C, 47% of DMSO treated cells were infected by H1N1-GFP after 24h. When CBS1116 was incubated with the virus at the attachment step (−1h), 38% cells were infected. CBS1116 exhibited the strongest antiviral effect at the time point 0h, with 0% cell infected by H1N1-GFP. The compound’s protective activity faded away when it was added later at +1h, +2h and +3h time points, with 17%, 26% and 33% cells infected by H1N1-GFP respectively. Figure 3D presents representative images from H1N1-GFP infection assay. These results suggest that CBS1116 does not act at the virus-cell surface binding step, but at an early viral entry step after endocytosis starts.

Figure 3. Time of addition experiment.

(A) Schematic diagram of time-of-addition experiment. A549 cells were incubated with pseudotyped HIV/H5N1 (B) or H1N1-GFP (MOI=10) (C) at 4°C at the −1h time point for 1h. At time point 0h, the virus was removed, and temperature was shifted to 37°C to trigger virus internalization. CBS1116 (10 μM) was introduced at different time points of virus infection as depicted in (A), and the compound’s effects on viral infection are shown. Two independent experiments were done, and data are shown as means±SD (n=3 for HIV/H5N1 infection and n=6 for H1N1-GFP infection). (D) Representative images from H1N1-GFP infection assay (C) are shown.

Cell surface binding inhibition assay was then carried out to probe the effect of CBS1116 on virus-cell binding. Briefly CBS1116 was incubated with 1μg of H5 (HA) protein with polyhistidine-tag at various compound concentrations on ice for 30 minutes. The mixture was then incubated with A549 cells for 1 hour on ice to allow HA to bind to the cell surface. Unbound HA proteins were removed by PBS/2%BSA wash twice. Bound HA were probed by addition of FITC conjugated anti his-tag antibody. After 45 minutes incubation on ice, cells were washed three times with PBS/2%BSA, fixed in PBS/2% paraformaldehyde and subjected to flow cytometry. The results were normalized as percentages of binding of 1μg HA (H5) his-tag protein with cells alone. If CBS1116 inhibits HA protein binding to the cells, the FITC fluorescence will decrease as CBS1116 concentration increases in the mixture. 2,3-sialyllactose and 2,6-sialyllactose receptor analogues were used as controls. Both sialyllactose controls inhibited his-tagged HA (H5) protein binding to A549 cells in a dose-dependent manner (Figure 4A). Interestingly 2,3-sialyllactose inhibited protein binding completely at 9 mM (IC₅₀=1.8 mM) while 2,6-sialyllactose only inhibited protein binding by 47% at 27 mM, the highest test concentration (IC₅₀=29.5mM). This phenomenon could be explained as HA (H5) protein of avian origin has a strong preference for α2,3-receptor analogue. However, 100% of the HA protein still bound to the cells in the presence of CBS1116 up to 1mM (Figure 4B), which is much higher than its anti-H5N1 IC₅₀ of 3.9 μM in the pseudovirus assay, indicating that CBS1116 does not act by blocking the virus-cell binding step.

Figure 4. Effects of CBS1116 on HA binding to cell surface and HA-mediated fusion process.

Binding of polyhistidine-tag HA (H5) protein to the surface of A549 cells was probed by his-tag specific FITC-conjugated antibody and quantified with flow cytometry. The effects of A) 2,3-sialyllactose or 2,6-sialyllactose and B) CBS1116 on the binding are shown. Chicken red blood cells (2% in PBS) were incubated with different concentrations of (C) 2,3-sialyllactose or 2,6-sialyllactose and (D) CBS1116. The pH was lowered to 5.0 to trigger HA-mediated hemolysis, which was quantified by optical density reading at 535nm using Envision plate reader. The IC₅₀s were calculated from the dose response curve fitting in GraphPad Prism software. Three independent experiments were performed, and data are means±SD (n=3).

To further investigate the role of CBS1116 in HA-mediated membrane fusion process, we performed a hemolysis inhibition assay. HA can bind to sialic acid of chicken blood red cells, and HA conformation change triggered by low pH makes ‘holes’ on the cell surface which releases hemoglobin and results in hemolysis. H5N1 pseudovirons were used in the hemolysis assay. As shown in Figure 4D, CBS1116 inhibited hemolysis in a dose-dependent manner with an IC₅₀ value of 6 μM which is very close to its IC₅₀ value of 3.9 μM in the HIV/H5N1 pseudovirus infection assay, strongly indicating that CBS1116 prevents virus entry by inhibiting the fusion process. In Figure 4C, 2,6-sialyllactose and 2,3-sialyllactose receptor analogues also inhibit the hemolysis by 43% and 26% at 4 mM, the highest test concentration, with estimated IC₅₀ values of 5.8 mM and 14.2 mM respectively. These IC₅₀s are comparable to the IC₅₀s of the analogues in the HA-binding assay, suggesting both analogues inhibit hemolysis through preventing virus binding to the cells, but not through inhibiting the fusion step.

3.4. Structure activity relationship study of CBS1116 generates CBS1117 with better antiviral profile

We then performed a structure-activity relationship (SAR) analysis for CBS1116. A CBS1116 analogue library of 58 compounds was made in which the backbone was preserved but variations were made in the phenyl ring and the piperidine ring to investigate the active part in the chemical structure. The library was screened with HIV/H5N1 pseudovirus and H1N1-GFP reporter virus in A549 cells at the final compound concentration of 12.5 μM. The results for the designed substitutions in the piperidine and the phenyl ring are shown in supplementary Table 1 and 2 respectively. As summarized in Figure 5, at the piperidine ring, several alkyl substitutes are acceptable for activity in order of preference: iso-Pro>ET>Pr>Bn>Me; at the phenyl ring, variety of substituents are tolerated: 1) halogens are better than alkyl or alkyloxy; 2) distributions are generally preferable to mono or tri substitutions; 3) premier patterns are dihalogen at 2,3–2,4–2,5–2,6 positions.

Figure 5. Structure activity relationship (SAR) developed for CBS1116.

From the initial screen, 11 analogs with >90% inhibition against H1N1 infection were further characterized with dose-response titrations. In comparison with the original compound, one compound CBS1117 (2,6-dichloro-N-(1-isopropyl-4-piperidinyl benzamide) exhibited increased potency with an IC₅₀ value of 0.07μM, less toxicity with a CC₅₀ value of 274.3 μM and improved SI value of 3914 (Table 3). Thus, CBS1117 represents a promising lead compound for further drug development.

Table3.

Anti-H1N1 activities of CBS1116 and CBS 111

Compound	Compound structure	IC50 (μM)a	CC50 (μM)	SI
CBS1116		0.4	160	400
CBS1117		0.07	274	3914

^aA/H1N1/Puerto Rico/8/34 GFP strain was used at MOI of 0.5 in A549 cells.

4. Discussion

As the first stage of viral replication cycle, cell entry is a promising target for antiviral development. So far, seven entry inhibitors have been approved by US FDA to treat infectious diseases caused by HIV, herpes simplex virus and respiratory syncytial virus(De Clercq and Li, 2016). However, for IAV, the entry inhibitors are still limited. To find novel entry inhibitors, we performed a screen of 19,200 compounds with pseudotyped H5N1, Marburg and Lassa viruses and identified a few potent inhibitors of H5N1 entry. The best compound, CBS1116 / 2,4-dichloro-N-(1-isopropyl-4-piperidinyl) benzamide, specifically inhibits two group 1 IAVs, likely by interfering with the HA-mediated viral-host membrane fusion.

Multiple evidences indicate that CBS1116 targets the HA-mediated membrane fusion step. Firstly, in the time-of-addition experiment, CBS1116 did not protect cells from IAV infection when it was introduced at the virus binding step (Fig. 3). In addition, CBS1116 did not interfere with HA (H5) protein binding to cell surface in the flow cytometry assay (Fig. 4B). Secondly, it has been reported that IAVs undergo uncoating through fusion with late endosomes at 30 to 90 minutes post-infection (Qin et al., 2019). CBS1116’s antiviral activity peaked within this time window, suggesting that the inhibitor probably acts at virus trafficking or virus-endosome fusion step. Thirdly, CBS1116 blocked HA-mediated hemolysis at low-pH which mimics the fusion step in the endosome. The IC₅₀ value of 6 μM in the hemolysis assay is very close to the IC₅₀ value of 3.9 μM in the pseudotyped H5N1 infection assay (Fig.4D and Table 2), strongly indicating that CBS1116 acts by inhibiting HA-mediated membrane fusion step. Thus, CBS1116 likely acts via binding to the stem region of HA and interfering with the HA conformation change triggered by low-pH in the endosome.

The HA stem region is highly conserved among HA subtypes, making it a suitable target for antiviral therapy development. Recently identified stem-binding broad neutralizing antibodies (bnAbs) FI6v3 and CR9114 can bind and/or neutralize every HA subtypes from both group 1 and 2(Corti et al., 2011; Dreyfus et al., 2012). However, due to large size of bnAbs, their conformational freedom to approach the stem regions may be restricted by the proximity of the stem region to the viral membrane. Furthermore, viruses may evolve to mask the stem region by glycosylation, making it inaccessible by the bnAbs. Small molecules can easily overcome these two disadvantages of bnAbs and represents a complementary strategy. Recently, JNJ4796 has been identified from a compound screen. The compound binds to the stem region of HA in a similar way as the broadly neutralizing antibody CR6261 and it protects mice against A/Puerto Rico/8/34 (H1N1) challenge after oral administration(van Dongen et al., 2019). In addition, a few fusion inhibitors (BMY 27709, LY-180299, RO5464466) also show group-specific inhibition against either group 1 or group 2 IAVs, though the exact binding sites for these inhibitors remain elusive(Hoffman et al., 1997; Luo et al., 1996; Zhu et al., 2011). Also, the crystal structures of HA complexed with tert-butylhydroquinone or Arbidol have been determined, revealing that these two chemicals bind to a hydrophobic pocket further up the stem region recognized by the bnAbs(Kadam and Wilson, 2017; Russell et al., 2008). We have obtained a crystal structure of H5 HA complexed with CBS1117. The structure clearly shows that CBS1117 binds to the stem region of HA near the fusion peptide, which overlaps with the binding site of the fusion inhibitor JNJ4796 (Antanasijevic et al., 2020). Further structural characterization of the interaction between the compound and HA may provide more insights about mechanism of action which can be utilized to modify the chemical structure to enhance the compound’s antiviral activity.

Another way to improve CBS1116’s potency is through SAR analysis. Variations in the phenyl ring and the piperidine ring of CBS 1116 were made to evaluate their effects on antiviral activity. It appears that the N-isopropyl group may be filling a specific hydrophobic pocket in HA (supplemental table 1). Alkyl groups that are smaller, e.g., the ethyl group in compound 2 and the methyl group in compound 5, lead to significantly less potent compounds. The benzyl group in compound 4, which is significantly larger than the isopropyl group of CBS1116, also leads to reduced potency. The n-propyl group in compound 3 is isomeric with the isopropyl group in CBS1116, but its shape is different. This small difference leads to a compound that is less potent but which is closest in potency to CBS1116. For the variations on the aromatic ring, two substituents appear to be superior than one, three or no substituents; halogen substituents are better than alkyl or alkoxy substituents. The patterns of 2,4-, 2,6-, 2,3- and 2,5-dihalo substitution produce the most potent inhibitors (supplemental table 2). It is clear that the presence of a finely tuned aromatic ring is important for potent H1N1 inhibition. From this SAR study, we successfully identified a new compound CBS1117 with better potency than CBS1116. CBS1117 has a lower IC₅₀ value of 70 nM and higher CC₅₀ value of 274 μM, leading to an excellent SI value of 3914, almost 10-fold improvement from that of CBS1116 (Table 3).

In summary, we have identified CBS1116 with 4-aminopiperidine scaffold as a novel specific inhibitor targeting group 1 IAVs. CBS1116 acts by interfering with the HA-mediated membrane fusion process. Preliminary SAR analysis generated CBS1117 with better antiviral activity which can serve as a starting point for further medicinal chemistry optimization and may be developed as an effective anti-influenza therapy.

Highlights.

• A 19,200 compound library was screened to identify compounds targeting influenza A virus entry.

• Hit compound CBS1116 specifically inhibits the cell entry of two group 1 influenza A viruses: H1N1 and H5N1.

• CBS1116 acts by interfering with the HA-medicated fusion process.

• Structure activity relationship studies dramatically improved the antiviral activities of CBS1116.