HTML
-
Lung epithelial cells, neutrophils, and monocytes/macrophages all play important roles in the cytokine storm induced by influenza virus infection and the subsequent effects of ALI and ARDS (Short et al. 2014). In order to identify a human cell line that can support influenza virus infection and the production of important proinflammatory cytokines, while also being suitable for screening anti-influenza drugs suppressing cytokine release, we examined seven human cell lines, including the epithelial lung A549 cells, promyelocytic HL-60 cells, neutrophil-like iHL-60 cells, monocytic TPH-1 cells, macrophage-like iTPH-1 cells, monocytic U937 cells, and macrophage-like iU937 cells, for their abilities in supporting influenza virus infection and the production of proinflammatory cytokines based on signal-to-background ratios (SBRs) during virus replication and CCL2/CXCL10/ IL-8/IL-6 expression levels (Table 1). Among the cell lines tested, A549 has lower SBRs in viral replication and proinflammatory cytokine secretion when infected with H1N1 PR8 influenza virus at MOI of 0.1 and cultured for 48 h, while HL-60 and iHL-60 can support influenza virus replication under the same conditions, but they only induce specific cytokine expression (CCL2 for HL-60, IL-6 for iHL-60). Therefore, these cell lines are not suitable for developing cell models for screening of anti-influenza drugs suppressing cytokine release. The human monocytic U937 cell line was found to support efficient replication of influenza virus H1N1 PR8 strain and the production of the four pro-inflammatory cytokines tested; additionally, it also had the highest SBRs for both viral replication and proinflammatory cytokine expression. Interestingly, THP-1 cell line, which is also monocytic cell, was shown to support influenza virus replication; however, it only expressed two of the four pro-inflammatory cytokines with very low expression levels under the same conditions. It is worth noting that the U937 and THP-1 cells were less supportive of influenza virus infection after being induced into macrophage-like cells and the SBRs of viral replication decreased from 7.3 and 5.8 to 2.4 and 1.5, respectively. Although the total expression levels of inflammatory cytokines in iU937 and iTHP-1 were significantly increased (data not shown), the amount of cytokine expression in the uninfected group was also greatly increased, which resulted in the lower SBRs of the three pro-inflammatory cytokines. These results indicate that the un-induced monocyte cell line U937 has the greatest potential to be a cell model for screening anti-inflammatory agents among the seven cell types tested.
Name Cell-type Signal to background ratio (SBR)a NAb CCL2c CXCL10c IL-8c IL-6c A549 Lung epithelial cell line 2.3±0.5 1.6±0.1 1.5±0.3 1.9±0.1 3.6±0.1 THP-1 Monocyte cell line 5.8±2.1 3.2±0.7 1.7±0.2 2.0±0.2 1.2±0.1 iTHP-1d Macrophage-like cell 1.5±0.6 1.4±0.2 3.4±0.2 1.2±0.1 1.1±0.1 U937 Monocyte cell line 7.3±1.1 16.2±2.5 23.6±3.3 17.9±2.3 13±3.7 iU937d Macrophage-like cell 2.4±0.7 1.1±0.1 2.5±0.2 1.2±0.2 1.6±0.2 HL-60 Promyeloblast 6.1±0.6 13.0±2.7 1.2±0.2 4.4±0.1 1.3±0.3 iHL-60e Neutrophil-like cell 4.1±0.4 1.4±0.3 1.0±0.1 2.2±0.3 13.2±2.6 aCells were incubated with 0.1 MOI of A/PuertoRico/8/1934 (H1N1) influenza virus and tested for the signal to background ratios (SRBs) at 48 h post infection. The SBR was calculated from the fluorescence value of the infected wells (signal) divided by the uninfected control wells (background) for each assay. b NA (neuraminidase), measured by neuraminidase activity assay, represents the replication level of the influenza virus. cCytokine expression was measured by AlphaLISA. diU937 and iTHP-1 cells were prepared from U937 and THP-1 cells induced by 100 ng/mL TPA for 24 h. eiHL-60 was prepared from HL-60 induced by 1.25% DMSO for 5 days. Table 1. Identification of U937 cell line as a cell-based model for drug discovery against cytokines induced by influenza virus infection.
-
To confirm whether the U937 cell line is capable of supporting influenza virus replication and the secretion of proinflammatory cytokines, they were infected with the influenza virus PR8 strain at different MOIs. Viral replication levels were determined according to neuraminidase (NA) levels and viral titers in a time course assay. As shown in Fig. 1A, NA activities in the supernatants increased with increasing MOIs of the viruses and reached a peak level at 48 hpi (hours post infection). Interestingly, both the culture time and the MOI can affect the viral titers of the supernatants of infected U937 cells (Fig. 1B). We found that the viral growth curves in U937 cells determined by the TCID50 endpoint dilution assay were consistent with what was determined using the NA activities when the MOIs were equal to or less than 0.1, suggesting that under such infection MOI conditions, the NA activities correlate well with the influenza virus titers. The TCID50 growth curves at high MOIs are partially inconsistent with the NA growth curves, in that the TCID50s reached peak levels at 24 hpi, possibly due to the cell death caused by virusinduced cytopathic effects with longer culture times (Fig. 1A–1C). Next, we tested the mRNA levels of the seven major pro-inflammatory cytokines induced postinfection in the influenza-infected U937 cells. As shown in Fig. 1D, at 12 hpi, the mRNA levels of CCL2, CCL5, CXCL10, IL-6, IL-8, IFN-β and TNF-α in U937 cells increased by 26.3, 1.7, 318.3, 27.5, 7.5, 344.2 and 2.15 times, respectively, indicating that PR8 infection can stimulate the transcription of pro-inflammatory cytokines in U937 cells. To study the correlation between the production of cytokines and virus MOIs, we then measured the protein levels of CCL2 and CXCL10 in the supernatants of the PR8 virus infected U937 cells. As shown in Fig. 1E and 1F, there was no apparent production of cytokines observed at 12 hpi, and the protein levels of CCL2 and CXCL10 were found to be at low levels at 24 hpi, reaching their highest levels at 48 hpi. It is worth noting that the levels of both CCL2 and CXCL10 increased with decreasing MOIs and reached their highest levels when the MOIs were around 0.025–0.05. However, as the MOI decreased below that range, i.e. 0.0125, the productions of the pro-inflammatory cytokines CCL2 and CXCL10 were observed to greatly decrease. These results indicate that U937 cells can support the replication of the influenza virus PR8 as well as the production of multiple cytokines. More importantly, infection of U937 cells with a relatively low MOI is critical for the production of the influenza-induced cytokine storm; infection with a high MOI can cause severe CPE and a low MOI can result in a low level of viral replication, both conditions can attenuate influenza-induced cytokine expression.
Figure 1. U937 cells support H1N1 virus replication and express multiple cytokines. U937 cells were infected with the PR8 (H1N1) virus at different MOIs and incubated at 37 ℃ for 1 h. The uninfected virus was washed away through centrifugation and cells were cultured at 37 ℃ for 72 h with complete medium (RPMI1640 + 10% FBS). Supernatants were collected at the indicated time points for the detection of NA activities (A) and virus titers (B). The cell viabilities were determined using the MTS method (C). The mRNA levels of 7 important pro-inflammatory cytokines in the U973 cells were determined by quantitative real-time PCR analysis at 12 h post infection (D). The protein levels of CCL2 (E) and CXCL10 (F) in the supernatants of U937 cells infected with the PR8 virus at different MOIs were determined using AlphaLISA. Each data point in the figure was repeated three times and the values were presented as mean ± SEM. NC represents cell control.
-
Influenza viruses are classified into types A, B, C and D and make up four of the seven genera of orthomyxoviridae. The type A viruses are the most virulent pathogens and can also be subdivided into different serotypes based on the antibody response to the viral components HA and NA. Influenza B viruses almost exclusively infect humans and can cause severe disease; they have only one serotype. Each of these influenza viruses may replicate and induce a cytokine storm in host cells differently (Hay et al. 2001). To investigate whether U937 cells can support the replication of different subtypes of influenza viruses and as well as the expression of the major pro-inflammatory cytokines in addition to the IAV H1N1 strain we initially used, three other subtypes of influenza virus strains from type A and one strain from type B, including H3N2, H7N8, and IBV, were selected and used to infect U937 cells. All four virus strains were able to replicate in U937 cells, although IBV was shown to replicate much less efficiently, as shown in Fig. 2A. Next, we examined the expression of CCL2, CXCL10, and IL-8 in the supernatants of the infected U937 cells. H1N1, H3N2, and H7N8 virus strains were shown to induce the expression of CCL2, CXCL10, and IL-8 (Fig. 2B–2D). Interestingly, IBV, which was shown to replicate inefficiently in U937 cells, activates the expression of the three cytokines as efficiently as the other three IAV strains (Fig. 2). These results indicate that U937 cells can support the replication of different types and subtypes of influenza viruses and induce the expression of multiple pro-inflammatory cytokines in response to these virus infections. In addition, the infection of a relatively low MOI of influenza virus in U937 cells is critical and sufficient for the production of multiple cytokines.
Figure 2. U937 cells support the replication of multiple subtypes of influenza viruses and the expression of pro-inflammatory cytokines. Four subtypes of influenza viruses, including A/PuertoRico/8/1934 (H1N1), A/Human/Hubei/3/2005 (H3N2), A/Duck/Hubei/216/1983 (H7N8), and B/Human/Hubei/1/2007 (IBV), were used to infect U937 cells at different MOIs at 37 ℃ for 1 h, and the uninfected viruses were washed away using centrifugation. The cells were cultured at 37 ℃ for 48 h with complete medium (RPMI1640+ 10% FBS). The supernatants were collected and used to detect the levels of NA activity (A), and the protein levels of CCL2 (B), CXCL10 (C) and IL- 8 (D) using a neuraminidase assay and AlphaLISA, respectively. Each data point in the figure was repeated three times and the values were presented as mean ± SEM. NC represents cell control.
-
To validate the U937 cell model, we tested a panel of immunomodulatory and antiviral agents with known in vivo activities (Table 2). These immunomodulatory agents, including the TPK inhibitor ibrutinib, PAR1 antagonist vorapaxar, S1PR1 agonist ozanimod, PPARγ agonist rosiglitazone, and macrolides antibiotics have been reported to improve the survival rate and pathology in influenza-infected mice. As expected, most of these immunomodulatory agents, with the exception of rosiglitazone, showed anti-inflammatory activities in the U937 cell model, as determined by their inhibition of CCL2 and CXCL10 expression. The antiviral agents, including the macrolide antibiotic ascomycin and the antiviral drugs ribavirin and Oseltamivir, were shown to have antiviral effects in the U937 cell model, as determined by the neuraminidase activity assay (Table 2). In contrast, some of the anti-inflammatory agents, including the COX non-selective inhibitor ibuprofen and the COX2-specific inhibitor etoricoxib, which have been reported to have no improvement on the survival rate of influenza-infected mice when used alone, were shown to have no inhibitory effect on the production of CCL2 and CXCL10 in U937 cell model (Table 2). Though the diversity and numbers of agents tested here were limited, these results suggest that, when using NA as an antiviral readout and CCL2/CXCL10 expression as anti-inflammatory readouts, the U937 cell model can be used to screen antiviral and immunomodulatory agents against influenza infection.
Chemical
nameCC50a IC50b
NASId
NAIC50c
CCL2IC50c
CXCL10SId
CCL2SId
CXCL10Target and possible mechanisme In vivo f Referencesg Ibrutinib 26.5 –h – 0.62 0.96 42.7 27.6 TPK inhibitor +i Florence et al.(2018) Vorapaxar > 80 – – 5.2 13.3 > 15.4 > 6 PAR1 antagonist + Khoufache et al.(2013) Ozanimod 26 – – 7.7 10.7 3.4 2.4 S1PR1 agonist + Teijaro et al. (2011) Ascomycin 62 21.3 2.9 8.1 9.9 7.7 6.3 Macrolides antibiotics + Sato et al. (1998) Ribavirin > 160 26.2 > 6.1 – – – – Anti-influnenza + Smee et al. (2008) Oseltamivir > 100 0.58 > 172.4 – – – – Anti-influnenza + Tsai et al. (2015) Rosiglitazone > 100 – – – – – – PPARγ agonist + Moseley et al.(2010) Etoricoxib > 50 – – – – – – COX-2 inhibitor ±j Zheng et al. (2008) Ibuprofen > 50 – – – – – – COX non-selective inhibitor ± Lauder et al. (2011) aCC50: 50% cytotoxic concentration (μmol/L) determined by MTS assay. bNA IC50: 50% inhibition concentration (μmol/L) determined by NA activity assay. cCCL2/CXCL10 IC50: 50% inhibition concentration (μmol/L) determined by AlphaLISA. dSI (Selective Index): a ratio of CC50/ IC50. eTarget and possible mechanism attributed to the compound as stated on either its "label" or literature. fIn vivo: the protective effect of mice from lethal influenza infection. gRef.: the reference of the reported protection effect of antiviral or immunomodulation agents in vivo. h"–": lower than 50% inhibition or not effective. i"+": Agent used alone can protect mice against lethal influenza virus infection according to the literature. j"±": Agent is effective in combination therapy, but when used alone cannot protect mice against lethal influenza virus infection. Table 2. Validation of U937 cell model with a panel of antiviral and immunomodulatory agents.
-
After the optimization of several parameters, including the number of cells per well, the MOI, and culture time, a U937 cell-based HTS model was established. Briefly, U937 cells were cultured in 384-well plates at a density of 1 × 106/mL, followed by infection with 0.05 MOI of influenza virus. The plates were cultured for 48 h at 37 ℃. Viral replication and the expression levels of cytokines were determined using a neuraminidase assay and AlphaLISA, respectively. The performance of our U937 cellbased HTS platform was evaluated by screening the FDAdrug library to identify agents that can inhibit the induction of the expression of CCL2 and CXCL10 by influenza infection (Fig. 3A). At a concentration of 20 μmol/L, 1280 compounds were screened on four 384-well plates. The SBRs of NA in the four plates were between 7.5 and 15, as shown in Fig. 3B. The SBRs of CCL2 and CXCL10 were between 17.6 and 21.5, and 20.8 and 28.3, respectively. The average Z' factor for NA, CCL2, and CXCL10 were 0.74, 0.75 and 0.84, respectively; all of the Z' factors were greater than 0.65, suggesting that the U937 cell model is robust and desirable for the HTS of immunomodulatory agents against the production of pro-inflammatory cytokines induced by influenza virus infection.
Figure 3. Establishment and implementation of a quantitative HTS using the U937 cell model. A Flow chart for the HTS of the 1280 drug library for immunomodulatory agents against influenza infection. The flow chart includes the steps and conditions for the entire screening process and also lists the selection criteria of the hit compounds and the number of hits obtained after each screening. The Venn diagram in the flow represents the cross-check of anti-CCL2 hits and antiCXCL10 hits and the results. B Fluorescence values representing NA activity, CCL2 and CXCL10 levels from infected and non-infected cells from all four 384-well plates were used to calculate the signalto-background ratio and Z'-factor scores. The calculation method used for the Z' factor is described in the statistical analysis section in materials and methods.
In the screen of the FDA-approved drug library, a hit was defined as a compound with cell viability > 75%, and an inhibition rate for CCL2 or CXCL10 > 70%, and an NA inhibition rate < 75%. We identified 60 hits with an anti-CCL2 effect with a hit rate of 4.7%, and 33 hits with an anti-CXCL10 effect with a hit rate of 2.6%. Interestingly, when analyzing the composition of the hit compounds with anti-CCL2 and anti-CXCL10 effects, listed in Supplementary Table S1, we found that most of the traditional anti-inflammatory drugs, such as steroidal anti-inflammatory drugs, cox inhibitors, prostaglandin synthetase inhibitor, histamine antagonist, 5-HT2 receptor antagonist, b-adrenergic blocker, and cholinergic receptor inhibitor, were identified as hits with anti-CCL2 effects, but not as hits with anti-CXCL10 effects. In addition to traditional antiinflammatory drugs, many agents with new targets were identified as well, which include the majority of the 16 hits found in both the anti-CCL2 and anti-CXCL10 hits, including HMG-CoA reductase inhibitors, DNA gyrase/topoisomerase IV inhibitors, ion channel inhibitors, and tyrosine kinase inhibitors. It appears that the target distributions of CCL2 and CXCL10 are different but related to some extent, suggesting that they are co-regulated by certain signaling pathways, and specifically regulated by their own signaling pathways.
-
In the primary single dose (20 μmol/L) screen using the influenza virus infected U937 cell model on 384-well plates, we identified 16 hit compounds with an anti-inflammatory effect (inhibiting both CCL2 and CXCL10) without an antiviral effect. To further confirm and evaluate their efficacies in repressing cytokine expression, we tested all 16 hit compounds for their kinetic inhibition of CCL2 and CXCL10 and cell viabilities in U937 cells. Their IC50s and CC50s were determined. Compounds with an IC50 > 10 μmol/L or shown to have no effect were considered to be inactive, 10 out of 16 hit compounds met this criterion (data not shown). Compounds with an IC50 < or = 10 μmol/L were considered to be active; 6 out of the 16 primary hit compounds met this criterion, they are dasatinib, pitavastatin, simvastatin, fluvastatin, protriptyline, and levofloxacin (Fig. 4A–4F). The cytotoxicity of the six hit compounds on uninfected U937 cells was determined, as shown in Fig. 4 (right panel). The CC50s were all greater than 90 μmol/L, except for simvastatin which had a CC50 of 66.8 μmol/L. To exclude compounds with antiviral activities that may lead to a reduction in CCL2 and CXCL10 levels, we examined the antiviral effects of the six hit compounds in the same confirmation experiments simultaneously (Fig. 4 right panels). Surprisingly, all 6 of the hits were shown to promote viral amplification to varying degrees, and viral replication was shown to increase with increasing drug concentrations, suggesting that these compounds may also inhibit the expression or functions of certain host factors associated with the innate antiviral response that correlated to the influenza-induced cytokine storm. Based on the kinetic inhibition of the two important cytokines CCL2 and CXCL10, and the IC50s and SIs (Fig. 4 and Table 3), we identified 6 compounds with potent inhibitory effects on influenza-induced expression of cytokines in U937 cells. Taken together, our results suggest that the U937 cell model is suitable for high throughput screening of antiinflammatory drugs against influenza-induced cytokine release.
Figure 4. Six hit compounds were confirmed to inhibit the production of CCL2 and CXCL10 in the U937 cell model. The kinetics of the inhibition of CCL2 and CXCL10 expression of the primary hit compounds was tested in the U937 cell model for Dasatinib (A), Pitavastatin (B), Simvastatin (C), Fluvastatin (D), Protriptyline (E), Levofloxacin (F). The structures of the six compounds are shown in the left panel. The inhibitory effects on CCL2 and CXCL10 are shown in the middle panels. The compound toxicity (green) and antiviral activity (red) expressed as a histogram are shown in the right panel. Each data point represents the mean ± SEM. The IC50 values for inflammatory factors are listed in the figure and were calculated using Graph Pad Prism 5.0 software.
Chemical name IC50
CCL2IC50
CXCL10CC50 SI
CCL2SI
CXCL10Approved of intended use Possible mechanism and target Dasatinib 0.63 0.71 > 90 > 142.9 > 126.8 Antineoplastic Tyrosine-protein kinases inhibitor Pitavastatin 0.85 0.87 > 90 > 105.9 > 103.4 Antilipemic HMG-CoA reductase inhibitor Simvastatin 1.16 1.75 66.8 57.6 38.2 Antilipemic HMG-CoA reductase inhibitor Fluvastatin 2.43 2.95 > 90 > 37.0 > 30.5 Antilipemic HMG-CoA reductase inhibitor Protriptyline 2.23 3.15 > 90 > 40.4 > 28.6 Antidepressant Serotonin reuptake inhibitor Levofloxacin 8.94 3.12 > 90 > 10.1 > 28.8 Antibiotic DNA gyrase and topoisomerase IV inhibitor Table 3. Six compounds were confirmed to inhibit influenza-induced CCL2/CXCL10 expression in U937 cell model.