key: cord-0693860-xhznxdg4
authors: Ianevski, Aleksandr; Yao, Rouan; Zusinaite, Eva; Lello, Laura Sandra; Wang, Sainan; Jo, Eunji; Yang, Jaewon; Ravlo, Erlend; Wang, Wei; Lysvand, Hilde; Løseth, Kirsti; Oksenych, Valentyn; Tenson, Tanel; Windisch, Marc P.; Poranen, Minna; Nieminen, Anni I.; Nordbø, Svein Arne; Fenstad, Mona Høysæter; Grødeland, Gunnveig; Aukrust, Pål; Trøseid, Marius; Kantele, Anu; Vitkauskiene, Astra; Legrand, Nicolas; Merits, Andres; Bjørås, Magnar; Kainov, Denis E.
title: Interferon alpha-based combinations suppress SARS-CoV-2 infection in vitro and in vivo
date: 2021-04-28
journal: bioRxiv
DOI: 10.1101/2021.01.05.425331
sha: 32ca54741eb7ec589d57896bc4e35c86b738b978
doc_id: 693860
cord_uid: xhznxdg4

There is an urgent need for new antivirals with powerful therapeutic potential and tolerable side effects. In the present study, we found that recombinant human interferon-alpha (IFNa) triggers intrinsic and extrinsic cellular antiviral responses, as well as reduces replication of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) in vitro. Although IFNa alone was insufficient to completely abolish SARS-CoV-2 replication, combinations of IFNa with remdesivir or other antiviral agents (EIDD-2801, camostat, cycloheximide, or convalescent serum) showed strong synergy and effectively inhibited SARS-CoV-2 infection in human lung epithelial Calu-3 cells. Furthermore, we showed that the IFNa-remdesivir combination suppressed virus replication in human lung organoids, and that its single prophylactic dose attenuated SARS-CoV-2 infection in lungs of Syrian hamsters. Transcriptome and metabolomic analyses showed that the combination of IFNa-remdesivir suppressed virus-mediated changes in infected cells, although it affected the homeostasis of uninfected cells. We also demonstrated synergistic antiviral activity of IFNa2a-based combinations against other virus infections in vitro. Altogether, our results indicate that IFNa2a-based combination therapies can achieve higher efficacy while requiring lower dosage compared to monotherapies, making them attractive targets for further pre-clinical and clinical development.

Viral diseases continue to pose a serious threat to public health due to a paucity of effective, rapidly 45 deployable, and widely available control measures [1, 2] . Viruses are submicroscopic agents that 46 replicate inside living organisms. When viruses enter and replicate in the cells, viral pathogen- 

IFNs are a large class of host proteins that are activated during innate antiviral immune response 55 [4, 5] . They are classified into three types, according to the cellular receptor they bind [6] (Fig. S1 ). Type 56 I IFNs consist of IFN-alpha (IFNa), IFN-beta (IFNb), IFN-epsilon, IFN-kappa and IFN-omega (IFNw) 57 and bind to the IFN-alpha/beta receptor (IFNAR1/2). Type II IFNs consist of IFN-gamma (IFNg) and 58 interact with the IFN-gamma receptor (IFNGR1/2). Finally, type III IFNs, consisting of IFN-lambda- 

Notably, mutations in IFN-signaling pathway genes have resulted in increased susceptibility to viral 67 infections and reduced patient survival [9] [10] [11] [12] . However, the exact role of each IFN pathway and their 68 crosstalk remain unclear.

The use of recombinant human IFNs has been approved for treatment of hepatitis C virus (HCV) 70 and hepatitis B virus (HBV) infections [13] . Additionally, IFNs have been shown to be effective against 71 a variety of other viruses in clinical trials and in laboratory settings (Fig. S2 ) [14] [15] [16] . Unfortunately,

IFNs possess limited efficacy when administered as antiviral treatments [17, 18] and can cause adverse 73 effects when used at established doses [19] .

IFN-related toxicity can be reduced by combining IFNs with other antiviral drugs that act 75 synergistically, thus allowing for the use of smaller doses of each component (Fig. S3) . Moreover, 76 4 synergistic combinations can often have higher efficacy against viral infections than individual 77 components administered as monotherapies, even at lower doses. Indeed, combination treatment of 78 IFNa and ribavirin was the "gold standard" for treatment of chronic HCV infection for more than 79 decade. Similarly, several rhIFN-based drug combinations have been tested against COVID-19. Of note, 80 combinations of IFNb1b/lopinavir-ritonavir/ribavirin, IFNa2b/IFNg, and IFNa/umifenovir were all 81 shown to be effective for treatment of patients with . However, despite these 82 promising data, the mode in which IFNs can be optimally combined with other drugs to maximize 83 antiviral and minimize side effects remains unclear.

Here, we have identified several novel synergistic IFNa2a-based drug combinations against SARS-

CoV-2, HCV, HEV, FluAV and HIV-1 infections. These treatment combinations are effective at lower 86 concentrations compared to monotherapies. These combinations have powerful treatment potential, 87 which can be leveraged for use in response to imminent viral threats including the emergence and re-88 emergence of viruses, immune-evading or drug-resistant variants, and viral co-infections.

Type I IFNs reduce SARS-CoV-2 replication more efficiently than type II and III IFNs

Although dexamethasone has been shown to improve survival of patients with severe or critical 92 , there are currently no curative therapies against SARS-CoV-2. However, previous 93 studies have uncovered several potent antiviral agents, including IFNs, against SARS-CoV-2 in vitro 94 and in vivo [14, 15, 25 ]. Here, we tested type I, II, and III IFNs against wild type SARS-CoV-2 95 (multiplicity of infection (moi) 0.01) in Calu-3 and Vero-E6 cells using cell viability and virus plaque 96 reduction assays as readouts. We observed that type I IFNs rescued both cell types from virus-mediated 97 death and reduced SARS-CoV-2 replication more efficiently than type II and III IFNs. However, the 98 rescue was only partial, and virus replication was reduced only by 2-3 common logarithms (Fig. 1 ).

To identify the type I IFN with most activity against SARS-CoV-2 infection, we infected IFN-100 treated and untreated Calu-3 cells with SARS-CoV-2-mCherry (moi 0.01) and collected media from the 101 5 cells (p1) after 48 h. The media were diluted 25-fold and applied to noninfected cells for another 48 h 102 (p2). Mock-infected cells were used as controls (Fig. 2a) . Fluorescence microscopy, fluorescence 103 intensity analysis, and cell viability assay of p1 and p2 cells showed that IFNa1b, IFNa2a and IFNw1 104 were more effective inhibitors of SARS-CoV-2 infection than IFNb1a. However, none of the IFNs tested 105 were able to inhibit virus infection completely ( Fig. 2b-d) .

Type I IFNs are encoded by multiple genes and vary slightly from one another in their protein 107 structure. In basic research, IFNa2a is widely used to elucidate the biological activities, structure, and 108 mechanism of action of such type I IFNs. Thus, we next tested IFNa2a against various doses of SARS- this, cells were treated with 1 μg/mL of IFNa2a, other type I IFNs, or vehicle; then infected with virus 120 or mock. After 24 h, we analyzed polyadenylated RNA using RNA-sequencing. We found that IFNa2a 121 and other type I IFNs attenuated production of viral RNA (Fig. 4a) , while increasing expression of many 122 ISGs in cells, regardless of virus-or mock-infection (Fig. 4b) . These ISGs include IFIT1, IFIT2 and IFIT3, 123 which play a role in recognition of viral RNA; OASL and OAS2, which are involved in RNase L-124 mediated RNA degradation; and IDO1, which is essential for kynurenine biosynthesis [26] [27] [28] [29] .

Interestingly, IFNa2a and other type I IFNs boosted virus-activated expression of type III IFNs (IFNl1, 

Next, we studied the effect of IFNa2a on the metabolism of mock-and SARS-CoV-2-infected Calu-131 3 cells. A total of 93 mainly polar metabolites were quantified at 24 hpi (Fig. S4 ). We found that tyrosine 132 and 4-hydroxyproline levels were substantially lowered during viral infection (log2FC<-2). 

We first confirmed antiviral activities of these known viral inhibitors on Calu-3 cells using SARS-

CoV-2-mCherry (Fig. 5a, Fig. S5a ). Then, we tested the antiviral efficacy and toxicity of these agents in 

We plotted synergy distribution maps, showing synergy (higher than expected effect) at each pairwise 155 dose. For each drug pair, we calculated average ZIP synergy scores for the whole 6×6 dose-response 156 matrices and for most synergistic 3×3 dose-regions, summarizing combination synergies into single 157 metrics (Fig. 5e ). We observed that all combinations showed a strong synergy (synergy scores >10) at 158 various combination doses. Thus, the observed synergy allows us to substantially decrease the 159 concentration of both components to achieve antiviral efficacy that was comparable to those of 160 individual drugs at high concentrations.

Both remdesivir and rhIFNa2a (Pegasys) were approved for the treatment of COVID-19 infection 162 in several countries. Therefore, we evaluated the antiviral effect of IFNa2a-remdesivir combination on 163 iPSC-derived lung organoids (LOs). Thirty-day-old LOs were treated with 5 ng/mL IFNa2a, 0.5 μM 164 remdesivir, or a combination thereof, then infected with SARS-CoV-2-mCherry. At 72 hpi, the 165 organoids were analyzed for viral reporter protein expression (mCherry) and cell death 166 (CellToxGreen). We found that IFNa2a-remdesivir substantially attenuated virus-mediated mCherry 167 expression without affecting cell viability (Fig. 6a ).

We also evaluated the effect of the combination treatment on viral and cellular RNA expression in 169 LOs. RNA-sequencing revealed that at 48 hpi IFNa2a-remdesivir substantially reduced production of 

Next, we examined whether IFNa and remdesivir can affect the replication of SARS-CoV-2 in vivo.

Four groups of 8 six-week-old female Syrian hamsters were injected IP with recombinant mouse IFNa, 182 remdesivir, IFNa-remdesivir combination or vehicle thereof. After 2 h of drug treatment, animals 183 received SARS-CoV-2 intranasally. After 3 days, animals were anesthetized and euthanized, and the 184 lungs were collected (Fig. 7a ). Virus titers from hamster lung homogenates in each treatment group 185 were determined using plaque reduction assays (Fig. 7b ). In addition, viral RNA was extracted and 186 sequenced. Sequencing results were validated using RT-qPCR (Fig. 7c,d) . The IFNa-remdesivir 187 combination attenuated the SARS-CoV-2 production and the synthesis of some viral RNAs more 188 efficiently than individual agents. 

Next, we studied IFNa2a in combination with known HEV inhibitors, NITD008 and ribavirin, 201 against HEV infection in Huh-7.5 cells (Fig. S9, Fig. 8 ). Both NITD008 and ribavirin are nucleoside 202 9 analogs which inhibit viral RNA synthesis. We observed that IFNa2a-NITD008 and IFNa2a-ribavirin 203 were synergistic against HEV infection (ZIP synergy scores: 11 and 8; the most synergistic area scores: 204 14 and 19, respectively) while remaining nontoxic at synergistic doses for either drug.

We also tested IFNa2a in combination with known influenza inhibitor pimodivir against FluAV 

Here, we have reported several novel IFNa2a-based combination therapies that have better 223 efficacy and lower toxicity than single drugs. In particular, we report novel in vitro activities of IFNa2a 224 combinations with remdesivir, EIDD-2801, camostat, cycloheximide, and convalescent serum against 225 SARS-CoV-2, with sofosbuvir or telaprevir against HCV infection, with NITD008 or ribavirin against 226 HEV infection, with pimodivir against FluAV, as well as with lamivudine against HIV-1 infection. Our

10 results indicate that other IFNa could be as efficient as IFNa2a when combined with these antivirals.

Moreover, they expand the spectrum of antiviral activities of these combinations and emphasize the 229 potential of IFNa-based combinatorial approach (Fig. 8b) . Interestingly, pimodivir, lamivudine, 230 remdesivir, EIDD-2801, NITD008, ribavirin and sofosbuvir interfere with synthesis of viral nucleic 231 acids, whereas camostat, cycloheximide, telaprevir and convalescent serum inhibit other steps of viral 232 replication cycle [33, 36, 41, 45] , indicating that IFNa could be combined with virus-and host-directed 233 agents targeting different stages of virus replication.

Based on our experiments, we propose the following mechanism of action of the IFNa-based 

263 Table S1 lists IFNs and other antiviral agents, their suppliers and catalogue numbers. Lyophilized

IFNs were dissolved in sterile deionized water to obtain 200 μg/mL concentrations. Compounds were 265 dissolved in dimethyl sulfoxide (DMSO; Sigma-Aldrich, Hamburg, Germany) or milli-Q water to 266 obtain 10 mM stock solutions. The convalescent serum (G614) from a recovered COVID-19 patient has 267 been described in a previous study [25] .

The propagation of wild-type SARS-CoV-2 (hCoV-19/Norway/Trondheim-S15/2020), recombinant 269 mCherry-expressing SARS-CoV-2 strains (SARS-CoV-2-mCherry), wild type human influenza 

The lung organoids (LOs) were generated as described previously (10.3390/v13040651). Briefly, 293 induced pluripotent stem cells (IPSCs) were subjected to embryoid body induction using embryoid 294 bodies (EB)/primitive streak media (10 μM Y-27632 and 3 ng/mL BMP4 in serum-free differentiation 295 (SFD) media consisting of 375 mL Iscove's Modified Dulbecco's Medium (IMDM), 100 mL Ham's F-12, 296 2.5 mL N2, 5 mL B27, 3,75 mL 7.5% BSA, 5 mL 1% penicillin-streptomycin, 5 mL GlutaMax, 50 μg/mL 297 ascorbic acid, and 0.4 μM monothioglycerol) in ultra-low attachment plates. After 24 h the media was 298 replaced with endoderm induction media (10 μM Y-27632, 0.5 ng/mL BMP4, 2.5 ng/mL FGF2, and 100 299 ng/mL Activin A in SFD media). Extra media was added every day for 3 days. The embryoid bodies 300 were collected and dissociated using 0.05% Trypsin/EDTA and plated on fibronectin-coated plates with 301 a cell density of 85,000 cells/cm 2 . Cells were then incubated in anteriorization media-1 (100 ng/mL 302 Noggin, and 10 μM SB431542 in SFD media), followed by an incubation with anteriorization media-2 303 (10 μM SB431542, and 1 μM IWP2 in SFD media). The anteriorization media-2 was replaced with 304 13 ventralization media (3 μM CHIR99021, 10 ng/mL FGF10, 10 ng/mL FGF7, 10 ng/mL BMP4, and 50 nM 305 all-trans Retinoic acid in SFD media) and incubated for two days. The cell monolayer was then lifted 306 by gentle pipetting, and the suspended cells were transferred to an ultra-low attachment plate where 307 they would form the lung organoids.

Thirty-two 6-week-old healthy female Syrian hamsters were obtained from Janvier Labs. The 

At 48 hpi, the media was removed, and a CTG assay was performed to measure cell viability. 

Madison, WI, USA). In a parallel experiment, a CTG assay was performed to measure cell viability.

We also examined cytotoxicity and antiviral activity of drug combinations using GFP-expressing 346 HCV in Huh-7.5 cells by following previously described procedures [56] . For testing compound toxicity 347 and efficacy against HEV, electroporated Huh-7.5 cells were seeded in the 384-well plate (3x10 3 

KGaA, Darmstadt, Germany) was used as a chromatographic separation column.

Gradient elution was carried out with a flow rate of 0.1 mL/min with 20 mM ammonium carbonate, 440 adjusted to pH 9.4 with ammonium solution (25%) as mobile phase A and acetonitrile as mobile phase 441 B. The gradient elution was initiated from 20% mobile phase A and 80% mobile phase B and maintained 442 for 2 min. Then, mobile phase A was gradually increased up to 80% for 17 min, followed by a decrease 443 to 20% over the course of 17.1 min. and sustained for up to 24 min.

The column oven and auto-sampler temperatures were set to 40 ± 3 °C and 5 ± 3 °C, respectively.

The mass spectrometer was equipped with a heated electrospray ionization (H-ESI) source using 

Metabolite peaks were confirmed using the mass spectrometry metabolite library kit MSMLS-1EA

(Sigma Aldrich supplied by IROA Technologies).

For data processing, final peak integration was done with the TraceFinder 4.1 software (Thermo

Fisher Scientific, Waltham, MA, USA) and for further data analysis, the peak area data was exported as 454 19 an Excel file. Data quality was monitored throughout the run using pooled healthy human serum as 455 Quality Control (QC), which was processed and extracted in the same manner as unknown samples.

After integration of QC data with TraceFinder 4.1, each detected metabolite was checked and %RSD 457 were calculated, while the acceptance limit was set to ≤ 20%.

Blank samples were injected after every five runs to monitor any metabolite carryover. A carryover 459 limit of ≤ 20% was set for each metabolite. Percentage background noise was calculated by injecting a 460 blank sample at the beginning of the run. The acceptance limit for background noise was set at ≤ 20% 461 for each metabolite. 

Inborn errors of anti-viral interferon immunity in humans

Peginterferon alfa-2a plus ribavirin for chronic hepatitis C virus 517 infection

Type 1 interferons as a potential treatment against COVID-19

Interplay between SARS-CoV-2 and the type I interferon response

Safety and efficacy of inhaled nebulised interferon beta-1a

Pathogenesis of 1918 pandemic and H5N1 influenza virus 602 infections in a guinea pig model: antiviral potential of exogenous alpha interferon to 603 reduce virus shedding

Discovery and development of safe-in-man broad-spectrum 605 antiviral agents

Novel antiviral activities of obatoclax, emetine, niclosamide, 607 brequinar, and homoharringtonine. Viruses

Identification and Tracking of Antiviral Drug Combinations

A plasmid DNA-launched SARS-CoV-2 reverse genetics system and 613 coronavirus toolkit for COVID-19 research

A Novel Inhibitor IDPP Interferes with Entry and Egress of HCV by 615

SynergyFinder: a web application for analyzing drug combination 617 dose-response matrix data

SynergyFinder 2.0: visual analytics of multi-619 drug combination synergies

Figure 1. Type I IFNs rescue Calu-3 and Vero-E6 cells from SARS-CoV-2-mediated death and attenuate virus 623 replication. (a) The effect of different doses of IFNs on viability of SARS-CoV-2-infected (moi = 0.01) Calu3 and 624

Cell viability was determined using the CTG assay at 72 hpi. Mean ± SD; n = 3. The anti-SARS-CoV-625 2 activity of the IFNs was quantified using drug sensitivity scores (DSS). (b) The effects of IFNs on viral replication, 626 measured by plaque reduction assay

Figure 2. IFNa1b, IFNa2a and IFNw1 are more effective than IFNb1a against SARS-CoV-2-mCherry infection 630 in Calu-3 cells. (a) Schematic representation of the experimental setup. (b) Fluorescent images of non-treated 631 (Ctrl) and IFN-treated (1 μg/mL) SARS-CoV-2-mCherry-infected Calu-3 cells (p1) and cells (p2) treated with 632 25-fold diluted media from P1 cells taken at 48 hpi. (c, d) Fluorescence intensity and viability analysis of p1 633 and p2 cells at 48 hpi

Anti-SARS-CoV-2 activity of IFNa2a depends on moi and time of administration. (a) Calu-3 cells were 637 treated with 1 μg/mL IFNa2a and infected with indicated moi of SARS-CoV-2-mCherry

cell viability were measured after 48 h (Mean ± SD; n = 3). (b) Calu-3 cells were treated with 1 μg/mL IFNa2a prior, 639 simultaneously or post infection with SARS-CoV-2-mCherry (moi 0.01). Fluorescence intensity and cell viability 640 were measured after 48 h (Mean ± SD

Transcriptomic and metabolomic analysis of mock-and SARS-CoV-2-infected Calu-3 cells non-644 treated or treated with type I IFNs. (a) Calu-3 cells were stimulated with IFNs (1 μg/mL) or non-stimulated 645 and infected with SARS-CoV-2 (moi = 0,01). A heatmap of viral RNAs affected by treatment is shown. Each 646 cell is colored according to the log2-transformed expression values of the samples

A 649 heatmap of the most variable cellular genes affected by treatment and virus infection is shown. Each cell is 650 colored according to the log2-transformed expression values of the samples, expressed as fold-change relative 651 to the nontreated mock-infected control. (c) Cells were treated as for panel b. After 24 h, the cell culture 652 supernatants were collected

Synergistic IFNa2a-based combinations against SARS-CoV-2-mCherry infection in Calu-3 cells

Mean ± SD; n = 3. (c) The 6 × 6 dose-response matrices and interaction landscapes of IFNa2a and 662 remdesivir obtained using fluorescence analysis of SARS-CoV-2-mCherry-infected Calu-3 cells. ZIP synergy score 663 was calculated for the drug combinations

IFNa2a and remdesivir obtained using a cell viability assay (CTG) on mock-, and SARS-CoV-2-mCherry-infected 665

Calu-3 cells. The selectivity for the indicated drug concentrations was calculated (selectivity = efficacy

ZIP synergy scores were calculated for indicated drug combinations. (e) ZIP synergy scores

Evaluation of antiviral effect of IFNa2a-remdesivir combination in human lung organoids (LOs). (a)

LOs were treated with 0,5 µM remdesivir, 5 ng/mL IFNa2a, their combination or vehicle

SARS-CoV-2-mCherry (moi = 0,1) or mock. Fluorescence of drug-or carrier-treated SARS-CoV-2-mCherry-675

Virus infection, cell nuclei, and cytotoxicity are shown in red, blue, and 676 green, respectively. Scale bars, 200 μm. (b) LOs were treated with 0,5 µM remdesivir, 5 ng/mL IFNa2a, their 677 combination or vehicle, and infected with SARS-CoV-2-mCherry (moi = 0,1)

After 48 h, total RNA was extracted 681 and sequenced. A heatmap of the most variable cellular genes affected by treatment and virus infection is 682 shown. Each cell is colored according to the log2-transformed expression values of the samples, expressed as 683 fold-change relative to the nontreated mock-infected control. Cut-off -3.75. (d) Cells were treated as for panel 684 a

A heatmap of the most affected metabolites is shown. Each cell is colored according to the log2-686 transformed profiling values of samples, expressed as fold-change relative to the mock control

Evaluation of antiviral activity of recombinant mouse IFNa-remdesivir combination in vivo

Schematic representation of the experimental setup. (b) The effects of IFNa-remdesivir combination on viral 690 replication in hamster lungs, measured by plaque reduction assay. Mean ± SD; n = 8. (c) A heatmap of viral 691

Each cell is colored according to the log2-transformed expression values of the 692 samples, expressed as log2 fold-change relative to the nontreated control. Mean, n = 8. (d) RT-qPCR analysis 693 of selected viral RNA. Expression of viral RNA was normalized to b-actin control

Statistically significant differences in viral gene expression between non-treated and treated animals are 695

Synergy and the most synergistic 698 area scores of IFNa2a-based combinations against HCV, HEV, FluAV and HIV-1. (b) Structure-activity 699 relationship of antivirals from known and novel (in bold) IFNa-based combinations

JavaScript library. The broad-spectrum antiviral activities of the compounds are shown as bubbles

The authors declare no conflicts of interest.