Oxidative stress specifically inhibits replication of dengue virus

Abstract

Reactive oxygen species (ROS) are chemically active species which are involved in maintaining cellular and signalling processes at physiological concentrations. Therefore, cellular components that regulate redox balance are likely to play a crucial role in viral life-cycle either as promoters of viral replication or with antiviral functions. Zinc is an essential micronutrient associated with anti-oxidative systems and helps in maintaining a balanced cellular redox state. Here, we show that zinc chelation leads to induction of reactive oxygen species (ROS) in epithelial cells and addition of zinc restores ROS levels to basal state. Addition of ROS (H₂O₂) inhibited dengue virus (DENV) infection in a dose-dependent manner indicating that oxidative stress has adverse effects on DENV infection. ROS affects early stages of DENV replication as observed by quantitation of positive and negative strand viral RNA. We observed that addition of ROS specifically affected viral titres of positive strand RNA viruses. We further demonstrate that ROS specifically altered SEC31A expression at the ER suggesting a role for SEC31A-mediated pathways in the life-cycle of positive strand RNA viruses and provides an opportunity to identify drug targets regulating oxidative stress responses for antiviral development.

Introduction

Reactive oxygen species (ROS) are oxygen derived molecules involved in the regulation of various cellular processes. Depending upon the concentration, ROS display bimodal effects; at lower concentrations they mediate signalling processes whereas their elevated levels lead to inactivation of critical cellular processes. Innate immune cells such as phagocytes and neutrophils utilize ROS as a defence mechanism against invading pathogens by generating H₂O₂ via different signalling pathways [1, 2]. ROS deficiency is associated with hyperinflammation and increased immune activation leading to autoimmune diseases [3]. Excessive ROS levels adversely affect cellular components such as lipids, proteins and DNA [4]. Oxidative stress plays distinct roles in different viral infections and influences viral infectivity. In the case of human immunodeficiency virus (HIV) infection, oxidative stress is induced due to the increase of reactive oxygen intermediates (ROIs) which leads to activation of NF-κB-dependent virus transcription. This oxidative environment has been implicated in the transition of HIV from quiescent to activated state [5]. Oxidative stress and ROS have been implicated in hyperin-flammation and lung pathogenesis observed infections with respiratory viruses such as SARS-CoV, respiratory syncytial virus (RSV) and influenza A virus [6–9]. Human hepatitis C virus (HCV) NS5A alters Ca²⁺ homeostasis leading to oxidative stress which in turn activates NF-κB and STAT-3 [10]. ROS has been demonstrated to interfere with the formation of replication complexes leading to suppression of HCV replication [11]. Thus, redox balance and oxidative stress may play a divergent role in viral infections depending on the threshold levels of ROS which either facilitate or inhibit RNA virus infection and viruses from different families with positive, negative or double-stranded RNA genomes may show varied susceptibilities to ROS levels depending on the host factors and pathways hijacked by these viruses for replication.

Dengue virus (DENV) is a positive-strand RNA virus which belongs to Flaviviridae family. Flavivirus infections in general are known to modulate cellular ROS level via induction of endoplasmic reticulum (ER) stress [12]. In the case of DENV infection, oxidative stress induction leads to activation of multiple signalling factors such as interferon regulatory factor 3 (IRF-3), IRF-7, signal transducer and activator of transcription 1 (STAT-1) and NF-κB-driven antiviral responses. It has also been demonstrated that decrease in ROS levels due to inhibition of NOX-complex renders the innate immune responses inefficient promoting DENV replication [13]. Despite these evidences, the effects exerted by oxidative stress environment on DENV life-cycle has not been explored. We had earlier demonstrated that zinc chelation inhibited dengue infection in Caco-2 cells. In this study, we further investigated the mechanism of action of zinc chelation and discovered that the perturbation of antioxidant functions of zinc may be responsible for inhibition of dengue infection. We found that induction of ROS by zinc chelator N,N,N′,N′-tetrakis(2-pyridinylmethyl)-1,2-ethanediamine (TPEN) inhibits dengue infection which could be reversed by zinc supplementation. Addition of exogenous ROS (H₂O₂) had the same effect as zinc chelation and inhibitory effect of oxidative stress appears to be specific to positive-strand RNA viruses as negative-strand and double-strand RNA viruses were not affected. Thus, our study assigns an antiviral role to oxidative stress that is specific to positive strand RNA viruses and may be useful in development of therapeutic strategies against these viruses.

Methods

Screening and enrolment of dengue patients

The study was approved by the Institutional Ethics committees of all the three participating institutes (IEC/NP-352/08.10.2014, THS 1.8.1/(32)dt. 5 April 2015 and 134/A/11/16/A/Academic/NC/2016/95, 18 June 2018). All patients in this study were enrolled between September 2016 and January 2019 at the Department of Paediatrics, All India Institute of Medical Sciences (AIIMS), New Delhi and at the Employees State Insurance Corporation Hospital and Medical College, Faridabad. Screening and enrolment of patients were exactly as described previously [14, 15]. Written informed consent for the study was taken from parents/guardians to collect blood samples at the time of admission. Children aged between 4–14 years with symptoms suggestive of dengue were screened using a ‘Dengue Day 1 Test’ which is a rapid solid phase immuno-chromatographic test for the qualitative detection of dengue NS1 Antigen (J. Mitra and Co. Pvt. Ltd, New Delhi). Patients testing positive for either NS1 or IgM were informed and requested for consent for the study. Consent was also requested from patients who were negative in NS1/IgM test for other febrile illness (OFI) controls. We classified all patients based on the latest WHO grading into three groups — Severe Dengue (SD), Dengue with Warning signs (DW) and Dengue Illness (DI) [16]. Complete blood counts were performed on automated COULTER COUNTER analyser (Beckman Coulter).

Cells and viruses

Caco-2, A549, BHK-21, C6/36, HEp-2 and MA104 cells were cultured as described previously. Calu-3 cells (obtained from ATCC) were cultured in MEM medium whereas Vero-E6 (obtained from National Centre for Cell Science, Pune, India) were cultured in in DMEM medium. African CHIKV isolate used in this study was obtained from National Institute of Virology, Pune. Other virus strains used in the study and infection procedures have been described before [17, 18]. A plaque-purified clinical isolate of SARS-CoV-2 was used in this study [19]. Caco-2 cells were infected with DENV-2 and rotavirus at 10 MOI and 0.5 MOI respectively. Rotavirus was activated using 10 μg ml^-1 trypsin (Worthington Biochemical Corporation) for 1 h at 37 °C. After adsorption cells were washed twice with PBS and cells were grown in serum-free medium containing trypsin at 0.5 μg ml^-1. A549 cells were infected with DENV, JEV and CHIKV, RSV at 1 MOI. Plaque assay for DENV, JEV and CHIKV was setup in BHK-21 cells. RSV in HEp-2 cells and RV was set up in MA104 cells as described previously [18]. Calu-3 cells were infected with SARS-CoV-2 at 0.3 MOI. SARS-CoV-2 plaque assays were carried out in Vero-E6 cells using the same protocol as DENV except that the plaque assays were fixed at 60 h pi.

Treatment of cells

Caco-2 cells were infected with DENV at 10 MOI. After viral adsorption, cells were treated with DMSO or TPEN alone or in combination with ebselen (25 μM) (TOCRIS), VAS2870 (10 μM) (Sigma-Aldrich), S3QEL 2 (10 μM) (TOCRIS) and l-NAME (100 μM) (Calbiochem) in serum-free medium. At 24 h pi, cells were fixed and stained using DENV-envelope antibody using immunofluorescence assay described in the following sections. Cells were treated with hydrogen peroxide (H₂O₂) in 10 % serum containing media for inducing oxidative stress. To study the effect of H₂O₂ on virus infection, different concentrations of H₂O₂ were added in media after infection. At indicated time points, supernatant was collected for estimating virus titre by plaque assay and total RNA was isolated from cells and used for quantitative real-time PCR (qRT-PCR) as described in the following sections. Cells were also treated with peroxynitrite in increasing concentrations to check its effect on DENV titres. For short duration experiments, H₂O₂ was added for 1 and 4 h after virus adsorption, cells were washed and complete media was added. Supernatants were collected at 24 h pi for measuring viral titres. For time kinetics experiments, cells were treated with 150 μM of H₂O₂ for 1, 2, 4, 8 and 16 h pi. Cells were used for RNA isolation and RT-PCR as described in the later sections. To check the specificity of H₂O₂, cells were infected with JEV, CHIKV, SARS-CoV-2, RSV and RV and infection titres were assessed by plaque assay.

Cell viability assay

Cell viability assay was performed using CellTiter-Glo Assay kit (Promega) according to user manual. This method determines the number of viable cells in culture based on quantification of the ATP present, in metabolically active cells. The homogeneous CellTiter-Glo reagent results in cell lysis and generates a luminescent signal proportional to the amount of ATP present. Briefly, A549 and Caco-2 cells were seeded in 96 well-plates and grown for 24 and 48 h respectively. Cells were treated with H₂O₂ and peroxynitrite at indicated concentrations. After 24 h post-treatment, CellTiter-Glo reagent was added to the well and luminescence was measured by microplate reader (Synergy HT - BioTek).

Flow cytometry

Gating strategy for whole blood

Gating strategy was essentially as described before with minor modifications to include neutrophils [14, 15]. Peripheral blood samples were stained as described in the methods section. Leukocytes were gated by their scatter properties in a SSC-A vs FSC-A gate which were further gated based on granulocyte marker CD66b. CD66b negative population was gated for live cells. T cells and B cells were gated from live cells using CD3 vs CD19 marker. CD3 and CD19 negative population was gated for monocytes using CD14 vs CD16 marker (to gate for classical, non-classical and intermediate monocytes). CD66b positive population was gated for neutrophils using CD16 marker.

ROS staining in dengue patients

Whole blood was collected in Na-Citrate tubes (around 4 ml) and centrifuged at 50 g for 15 min with no breaks to separate platelet-rich plasma (PRP). PRP was removed and remaining blood was subjected to RBC lysis using RBC lysis buffer (155 mM NH₄Cl, 12 mM NaHCO₃, 0.1 mM EDTA) added at a ratio of 1 : 10 (1 ml blood and 10 ml RBC lysis buffer). Cells were incubated for 15 min at RT in dark followed by centrifugation at 200 g for 10 min. Cell pellet was washed twice with PBS by centrifugation at 200 g for 10 min. Cell viability and counts were estimated by trypan blue staining. One million cells were stained using H₂DCFDA in DMEM without phenol red supplemented with 2 mM l-glutamine in a final volume of 100 μl. Cells were incubated at 37 °C in a CO₂ incubator for 30 min and washed with DMEM without phenol red at 500 g for 5 min. Cell pellet was incubated with human TruStain FcXTM (BioLegend) to block Fc receptors for 15 min at room temperature (RT) followed by incubation with fluorescent tagged antibodies to identify T cells [CD3+(PE)], B cells [CD19+(APC)], classical monocytes [CD14+(BV421) CD16-(PE/Cy7)], non-classical monocytes [CD14dim (BV421) CD16+(PE/Cy7)], intermediate monocytes [CD14+(BV421) CD16+(PE/Cy7)], neutrophils [CD66b+(PerCP/Cy5.5) CD16+(PE/Cy7)] and eosinophils [CD66b+(PerCP/Cy5.5) CD16-(PE/Cy7)] for 30 min on ice. Fixable viability stain eFluor780 was added in the antibody cocktail at 1 : 250 dilution. All the antibodies were procured from BD Biosciences or BioLegend. Cells were washed with FACS buffer and fixed using IC fixation buffer (eBiosciences). Cells were acquired in FACS Canto II. ROS levels in dengue patients and other febrile illness was expressed as fold change relative to healthy controls. Intracellular ROS levels in cell lines were measured by flow cytometry as described previously [20] by using DMSO as control for normalization and relative fold-change expression in treatment conditions.

Organelle markers staining

For free zinc localization studies, A549 cells were seeded on coverslips at a cell density of 80 000 cells per well in a 24-well plate. Cells were transduced using fusion constructs specific for Golgi apparatus, mitochondria, lysosomes and endoplasmic reticulum (ER), packaged in insect virus baculovirus (CellLight RFP BacMam 2.0 reagents; Molecular probes). Cells were washed once with PBS and transduced using 50 particles per cell (PPC) for mitochondria, lysosomes and ER and 100 PPC for Golgi. Cells were incubated for 1 h on a rocker in a 37 °C CO2 incubator. After 1 h, complete media was added directly to the wells to make up the volume to 500 μl and incubated overnight. At 24 h, cells were washed with DMEM (without phenol red) once and stained using ZinPyr-1 (ZP-1), a zinc fluorophore that specifically binds to vesicular free zinc pools, followed by staining with 4′,6-diamidino-2-phenylindole (DAPI) at 1 : 10000 dilution for 10 min to stain the nuclei. Cells were fixed using IC fixation buffer (eBiosciences). Coverslips were washed with PBS twice and mounted in ProLong Gold Antifade reagent (Molecular Probes). Slides were imaged using confocal microscope FV3000 (Olympus) and single slice images converted to TIFF using CellSens software (Olympus).

Immunofluorescence staining of cells for confocal microscopy was performed essentially as described before [21]. Caco-2 cells were seeded on coverslips at 100 000 cells per well in 24-well plate. Cells were treated with H₂O₂ at a concentration of 150 μM for 1 h in complete media. After 1 h treatment, cells were fixed using ice cold methanol at −20°C for 20 min or 3 % paraformaldehyde (PFA) at RT for 10 min. Cells were washed twice using PBS, permeabilized with 0.1 % Triton X-100 for 15 min followed by blocking with 1 % BSA in PBS for 1 h at RT. Cells were incubated with antibodies against LC3B (1 : 100; CST, 3868S), EEA1 (1 : 100; BD Biosciences, 610456), Sec31a (1 : 400, BD Biosciences, 612350), GM130 (1 : 100, BD Biosciences, 610822), MPR-300 (1 : 100, a kind gift from Dr. Peter Schu) in 0.1 % BSA for 1 h at RT. This was followed by incubation with secondary antibody conjugated with Alexa flour 488 for 30 min at RT in dark. Cells were washed with PBS three times and stained with DAPI. Cells were washed with PBS, mounted using Prolong antifade solution and images were captured using FV3000 fluorescence microscope. For mitotracker and lysotracker staining, cells were washed twice with PBS followed by incubation with 100 nM MitoTracker Red CMXRos (Invitrogen, M7512) and LysoTracker Red DND-99 (Invitrogen, M7528) prepared in serum-free growth media for 30 min. After incubation, cells were washed three times with PBS and stained with DAPI at a dilution of 1 : 10000 for 5 min. Cells were washed, and coverslips were mounted on a glass slide and left to dry in the dark overnight. Images were acquired using 100× oil immersion objective. Images were processed by background correction using cellSens software and z projection images are shown.

Quantitative real-time PCR assay

Caco-2 cells were infected with DENV-2 at a MOI of 10 for 1 h. After 1 h of virus adsorption cells were washed twice with PBS and H₂O₂ was added. At indicated time points, supernatant was collected for estimating viral titres by plaque assay and cells were harvested. For DENV genome level detection, RNA was isolated from cells collected at indicated time points. RT-PCR was set up using TaqMan RNA-to-Ct one step kit and GAPDH was used to normalize DENV genome levels. For positive and negative strand detection, RNA was isolated using nucleospin kit (MACHEREY-NAGEL GmbH) and reverse-transcribed using forward or reverse primer. RT product was further amplified using primer and probe mix using TaqMan RNA-to-Ct one step kit (Applied Biosystems) as described previously. GAPDH primer probe mix (Applied Biosystems) was used as housekeeping control. Data was analysed using ΔΔCt method.

Statistics

Data was analysed and charts were prepared using GraphPad Prism (Prism 7e) software. All experiments were performed with two or more replicates and graphs have been prepared representing data from at least two independent experiments with n ≥6. Error bars represent mean±SD. Statistical significance was estimated by t-test (unpaired, non-parametric) using Mann-Whitney test.

Results

Zinc-deficiency induces oxidative stress in epithelial cells

We have shown previously that zinc chelation by TPEN led to specific inhibition of dengue virus but not other RNA viruses such as respiratory syncytial virus or rotavirus [17, 20]. We were further interested in understanding the effect of ROS on DENV infection in epithelial cells. We first verified our previous results by treating Caco-2 cells with 0.5 μM TPEN and ROS levels were estimated by FACS using H₂DCFDA at indicated time points. We found ~two-fold increase in ROS levels at 4 and 8 h which increased up to ~four-fold at 24 h post-treatment (Fig. 1a). To assess whether Zn supplementation could reverse ROS induction, cells were treated with 0.5 μM TPEN for 4 h followed by addition of 10 μM ZnSO₄ for 2 h in the same medium and ROS levels were measured. We observed that ZnSO₄ inhibited TPEN induced ROS response (Fig. 1b). These data suggest that zinc-chelation using TPEN leads to increase in ROS levels in Caco-2 cells. This effect was reversed after zinc addition thus indicating the antioxidant properties of zinc.

Fig. 1.

Zinc deficiency induces oxidative stress. ROS levels were measured in Caco-2 cells using H₂DCFDA by flow cytometry. (a) Graph indicates fold change in ROS levels in Caco-2 cells treated with TPEN at 0.5 μM concentration at indicated time points (b) Cells were treated with TPEN for 4 h followed by addition of 10 μM ZnSO₄ for 2 h and ROS levels were measured. All the data are from at least two independent experiments. Data are presented as mean+SD. **P <0.01.

Antioxidants reverse the effect of Zinc chelation

To further demonstrate the specificity of the role of Zn as an anti-oxidant, we assessed the effect of different divalent cationic salts; zinc (Zn), magnesium (Mg), manganese (Mn), copper (Cu) and iron (Fe), on TPEN-induced ROS levels and DENV inhibition. Caco-2 cells were infected with DENV and after viral adsorption, 0.5 μM TPEN was added alone and with combination of indicated salts (1 μM). At 24 h pi, cells were fixed and stained using DENV-envelope antibody. We observed that ZnSO₄ was able to block the inhibitory effect of TPEN whereas no effect was observed with other salts (Fig. 2a). This correlated with viral titres at 24 h in the supernatants (Fig. 2b) suggesting that the antioxidant functions of Zn plays a specific role in reversing inhibition of dengue infection with TPEN treatment. To further confirm that ROS generation is responsible for the inhibitory effects of TPEN, cells were treated with TPEN with or without ebselen, an antioxidant, to assess for reversion of TPEN-induced ROS levels and DENV inhibition. Similar to Zn treatment, ebselen was capable of blocking ROS production in TPEN treated cells and reversing virus inhibition (Fig. 2c, d). Interestingly, salu-brinal, an ER stress inhibitor, which was shown to mitigate the effects of TPEN [17] was capable of reversing the effect of TPEN also blocked ROS induction and reversed virus inhibition suggesting that the ER stress may contribute to ROS production under zinc chelation conditions (Fig. 2c, d). We further demonstrated the specificity of this effect by showing that the inhibitors of NADPH oxidase (NOX) family namely and VAS2870 had no effect on TPEN mediated ROS induction or DENV inhibition (Fig. 2e). Similarly, S3QEL 2 and N _ω-Nitro-l-arginine methyl ester hydrochloride (l-NAME), suppressors of superoxide and reactive nitrogen species (RNS) respectively, did not play any role in preventing TPEN induced DENV inhibition (Fig. 2e). Treatment with inhibitors alone had no effect in DENV infection (Fig. S1, available in the online version of this article). These results suggest that ROS induction via ER stress pathways may contribute to DENV inhibition during zinc chelation conditions.

Fig. 2. Antioxidants treatment block TPEN-induced ROS and facilitates DENV infection.

(a) Caco-2 cells were infected with DENV at 10 MOI and treated with 0.5 μM TPEN or TPEN with indicated salts at 1 μM concentration. At 24 h pi, cells were fixed and stained using DENV-envelope by immunofluorescence assay. (b) Viral titres in the supernatants were measured at 24 h pi. (c) Caco-2 cells were treated with TPEN alone and in combination with ebselen and salubrinal at 25 and 50 μM respectively. ROS levels were measured at indicated time points using H₂DCFDA by flow cytometry. Graph indicates fold change values in mean fluorescence intensity (MFI) normalized to DMSO. (d) Caco-2 cells were infected with DENV at 10 MOI and treated with TPEN alone and with ebselen or salubrinal as above. Viral titres were measured at 24 h pi. (e) Caco-2 cells were infected at 10 MOI of DENV. After viral adsorption, cells were treated with TPEN alone and in combination with ebselen (EBS) (25 μM), S3QEL 2 (10 μM), l-NAME (100 μM). At 24 h pi, cells were fixed and stained using DENV-envelope antibody using immunofluorescence assay. Data are from at least two independent experiments and error bar represents standard deviation. Scale Bar: 100 μm.

Oxidative stress leads to reduction in DENV infection

We next set out to determine the direct effect of inducing oxidative stress by addition of ROS on DENV infection in Caco-2 cells. H₂O₂ treatment was used to create an oxidative environment in vitro. Initially, we examined cell viability in cells treated with increasing concentrations of H₂O₂. We observed cytotoxicity in cells which were treated with >600 μM concentration of H₂O₂ (Fig. S2a). Next, Caco-2 cells were infected with DENV at a MOI of 10 and after virus adsorption, H₂O₂ was added at a concentration of 150 and 300 μM. Viral titres in the supernatant were determined by plaque assay at 24 h pi and DENV genome levels were determined using quantitative real time PCR. H₂O₂ treatment led to significant reduction in DENV titres and DENV genome levels in a dose-dependent manner (Fig. 3a, b). The effect of H₂O₂ was reversed by co-incubating with ebselen (Fig. S3) further suggesting that the effect is due to addition of ROS. To further verify whether inhibitory effect of oxidative stress is specific to H₂O_2, we tested the effect of peroxynitrite, an oxidizing agent on DENV infection in a similar manner. Peroxynitrite treatment above 200 μM showed cytotoxicity (Fig. S2b). At 24 h pi, DENV infection was determined by immunofluorescence staining using DENV-envelope antibody. We could observe reduction in the percentage of cells infected with DENV infection in peroxynitrite treatment in a dose-dependent manner (Fig. 3c). This correlated with viral titres (Fig. 3d). The results suggest that oxidative stress created due to exogenous H₂O₂ and peroxynitrite leads to reduction in DENV infection.

Fig. 3.

Oxidative stress inhibits DENV infection: (a) Caco-2 cells were infected with DENV at 10 MOI and treated with 150 and 300 μM of H₂O₂. Supernatants were collected at 24 h pi and viral titres were measured by plaque assay. (b) Cells were collected for RNA isolation and checked for DENV genome levels by real time PCR. GAPDH was used for normalization. (c) Caco-2 cells infected with DENV and peroxynitrite was added at indicated concentrations after viral adsorption. At 24 h pi, cells were fixed and stained with anti-DENV-envelope antibody. (d) Supernatants were collected and viral titres were determined by plaque assay. Data are from at least two independent experiments. Data are presented as mean±SD. **P <0.01, ***P <0.001, ns: not significant. Scale bar 100 μm.

ROS blocks DENV replication

As we observed reduced viral RNA in DENV infected cells, we were next interested in determining whether a short duration treatment with ROS at early stages of infection was sufficient to inhibit DENV replication. Caco-2 cells were infected with DENV and after viral adsorption for 1 h, and culture medium was incubated with H₂O₂ for 1 or 4 h post-infection and further incubated in normal growth medium up to 24 h pi. We found that H₂O₂ treatment for only 1 h post-infection led to marked reduction in DENV titres and viral titres reduced by two orders of magnitude in samples that were treated with H₂O₂ for 4 h (Fig. 4a). Further, we examined the dose-dependence of H₂O₂ on DENV replication by adding indicated concentrations of H₂O₂ after virus adsorption. After 1 h treatment, medium was removed and cells were cultured in normal growth media for 24 h. Viral titres and viral RNA levels were estimated at 24 h pi. We observed a reduction in DENV titres and DENV RNA levels in 75 μM and 150 μM treatment conditions (Fig. 4b, c). DENV genome is a positive-sense RNA strand which replicates via production of negative strand intermediates. Therefore, estimation of negative strand intermediates by qRT-PCR is a measure of virus replication. Cells were infected with DENV and H₂O₂ was added after 1 h of virus adsorption. We collected cells at indicated time points and RNA was isolated followed by detection of positive and negative strand by qRT-PCR. We found that positive strand RNA levels started declining from 4 to 16 h in a time-dependent manner. There was moderate increase in negative-strand RNA levels initially but no further production of fresh negative strand intermediates could be observed at 16 h pi (Fig. 4d, e). These results suggest H₂O₂ affects early stages of DENV replication post-viral-entry.

Fig. 4. H₂O₂ treatment inhibits DENV replication.

(a) Caco-2 cells were infected with DENV at 10 MOI. After viral adsorption, cells were treated with H₂O₂ for 1 and 4 h in the culture medium. After treatment, medium was replaced and cells were cultured in normal growth media. Supernatants were collected at 24 h pi and viral titres were measured. (b) Caco-2 cells were infected with DENV at 10 MOI. After virus adsorption, H₂O₂ was added as indicated for 1 h followed by replacement with normal growth media. Supernatants were collected at 24 h pi to estimate viral titres and (c) DENV RNA levels were estimated by qRT-PCR. (d) Caco-2 cells were infected with DENV at 10 MOI and cultured in the presence of 150 μM of H₂O₂ and cells were collected for RNA isolation at different time points for estimation of positive strand and negative strand (e) RNA by qRT-PCR. Data are from at least two independent experiments. Data are presented as mean+SD. *P <0.05, **P <0.01, ns: not significant.

Oxidative stress specifically inhibits positive-strand RNA virus infections

ROS triggers a number of signalling cascades and affects metabolic functions that has either a positive or a negative impact on RNA viruses [20, 22]. Therefore, we were next interested in understanding the specificity of the effect of ROS treatment on RNA viruses from different families. We included JEV, CHIKV and SARS-CoV-2 as other representatives of positive-strand viruses along with DENV. RSV, a negative-strand virus and rotavirus (RV), a double-stranded RNA virus were used. A549 cells were infected with DENV, JEV, CHIKV at MOI of 1 and with RSV at an MOI of 0.3. Calu-3 cells were infected with SARS-CoV-2 at 0.3 MOI. Caco-2 cells were infected with RV at a MOI of 0.5. After viral adsorption, H₂O₂ was added at a concentration of 150 μM. Viral titres were determined by plaque assay in supernatants collected at 24 h pi for DENV, JEV, CHIKV, SARS-CoV-2, RSV and 16 h pi for RV. We found a ten-fold reduction in titres of positive-strand RNA viruses in H₂O₂ treatment whereas RSV and RV titres were unaffected (Fig. 5). These results suggest that H₂O₂ treatment has inhibitory effect only on positive-strand RNA viruses suggesting that this class of viruses are more susceptible to oxidative stress.

Fig. 5.

Hydrogen peroxide specifically inhibits positive-strand RNA viruses. A549 cells were infected with DENV, JEV, CHIKV, RSV at 1 MOI. Caco-2 cells were infected with RV at 0.5 MOI. Calu-3 cells were infected with SARS-CoV-2 at 0.3 MOI. Then 150 μM H₂O₂ was added after 1 h of viral adsorption. Cell supernatants were collected for DENV, JEV, RSV, CHIKV and SARS-CoV-2 at 24 h pi. For RV, supernatants were collected at 16 h pi. Data are from at least two independent experiments. Data are presented as mean±SD. *P <0.05, **P <0.01, ns: not significant.

H₂O₂ affects SEC31A distribution at the ER

Flavivirus replication complex drives viral RNA replication within the virus-induced membranous compartments derived from the host ER membranes [23]. We observed that the inhibitor of ER stress, salubrinal, was capable of reversing ROS induction and inhibition of infection caused by zinc chelation suggesting that the components of the ER may be involved in zinc homeostasis and ROS-induced inhibition of DENV replication. The localization and the size of free/labile zinc pools seems to differ based on the cell types used [24, 25]. We sought to verify the localization of free zinc in A549 cells by transducing cells with baculovirus constructs expressing red fluorescent protein signal peptide fusions and labelling cells with ZP-1 after 24 h. We used constructs expressing RFP-tagged proteins with N-acetylgalactosaminyl transferase 2 (Golgi-resident enzyme), leader sequence of E1 alpha pyruvate dehydrogenase (mitochondrial enzyme), lysosomal associated membrane protein 1 (LAMP-1) and ER signal sequence of calreticulin and KDEL (ER retention signal), respectively to mark the respective compartment. ZP-1 stain colocalized mostly with the Golgi and the ER fraction and was absent from mitochondria and lysosomes (Fig. 6). Therefore, we hypothesized that TPEN treatment may disrupt zinc homeostasis at the ER and Golgi leading to ROS induction. Therefore, ROS induced due to zinc chelation or due to direct H₂O₂ treatment may have similar effect at the ER and in-turn disrupt the biogenesis of ER-derived membranes required for virus replication. We tested this by examining the localization of organelle markers in cells treated with H₂O₂ by immunofluorescence assay using antibodies and fluorescent dyes specific for various cellular compartments. We stained cells for markers of endoplasmic reticulum (ER) (SEC31A), Golgi complex (GM130), early endosome (EEA1), autophagosome (LC3B), late endosome/lysosome (MPR-300 and LysoTracker), and mitochondria (MitoTracker). H₂O₂ treatment led to reduction in signal intensity of SEC31A, a component of coat protein complex (COPII), while no effect was observed on any other organelle markers (Fig. 7a, b). These results suggest that induction of ROS has profound effects specifically on ER homeostasis which affects DENV replication.

Fig. 6.

Localization of labile zinc pools stained by ZP-1. A549 cells were infected with 50 PPC of CellLight RFP lysosomes/mitochondria/endoplasmic reticulum and 100 PPC of CellLight RFP Golgi overnight and stained for free zinc using ZP-1. Images captured using confocal microscope and single slice images used for analysis. From top to bottom: Golgi, mitochondria, lysosomes and endoplasmic reticulum. From left to right: DAPI, ZP-1, RFP-tagged organelle marker and merged images. (DAPI: blue; ZP-1: green; organelles: red). Arrows indicate regions of colocalization. Scale bar 10 μm.

Fig. 7.

H₂O₂ treatment affects SEC31A localization at the ER. Caco-2 cells were treated with H₂O₂ for 1 h and fixed and stained to visualize various organelle markers. (a) Cells were stained with primary antibodies (EEA1, MPR-300, SEC31A, GM130 and LC3b; left to right) followed by incubation with secondary antibodies conjugated with Alexa fluor 488 (green). Nuclei were stained with DAPI (blue). (b) Graph showing quantification of confocal data. (c) MitoTracker Red CMXRos and LysoTracker Red DND-99 dyes were used to stain mitochondria and lysosomes respectively. (d) Graph showing quantification of data. Data are from at least two independent experiments. Data are presented as mean+SD. ***P<0.001. Scale bar is 10 μm.

DENV infection leads to reduction in ROS formation in blood cell subsets

Oxidative stress may influence viral infections in more than one way depending on the host pathways and compartments that are usurped by the viruses for genome replication, translation, assembly and egress [26, 27]. Our observations in vitro suggest that DENV replication is sensitive to ROS induction. Therefore, we speculated that DENV may adopt strategies to circumvent oxidative stress. We measured intracellular ROS levels in blood cell subsets from whole blood samples collected from patients with dengue or other febrile illnesses (OFI). Clinical features of patients enrolled into the study are described in Table 1. Out of 23 patients enrolled, eight patients had mild dengue (DI), nine had dengue with warning signs (DW) and six patients were with severe dengue (SD), classified as per WHO guidelines as described before [14]. We also included 12 samples from patients with other febrile illness (OFI) and from healthy volunteers for relative comparison of intracellular ROS levels by flow cytometry. We observed a significant reduction in ROS levels in dengue samples relative to samples from OFI or healthy controls. As the sample numbers were low in each of the three disease conditions, we could not detect any association of ROS levels with disease severity, however, we observed a negative trend between disease severity and ROS levels with most severe cases showing lowest levels of ROS (Figs 8a and S5a). We measured DENV RNA levels in total RNA isolated from whole blood by quantitative RT-PCR as described earlier from these patients [14]. Viral RNA levels were not significantly different between the DI, DW and SD conditions (Fig. S5b). Although ROS levels were lower in samples with high viremia (Fig. S6a) and viral RNA levels negatively correlated with ROS levels, this correlation was not statistically significant (Fig. S6b). We were interested to further probe which of the peripheral blood mononuclear cell (PBMC) subsets from dengue patients show altered ROS induction in DENV patients. Therefore, total leukocytes in whole blood were stained with cell surface markers to identify neutrophils (CD66b⁺ CD16⁺), eosinophils (CD66b⁺ CD16^-), T cells (CD3⁺), monocyte subsets (classical (CD14⁺), intermediate (CD14⁺ CD16⁺) and non-classical (CD14^dim CD16⁺) and B cells (CD19⁺). Intracellular ROS levels in these subsets were measured by flow cytometry. Interestingly, relative to healthy controls, almost all the leucocyte subsets showed increased ROS production in OFI cases but samples from DENV patients did not show any elevated levels of ROS and, contrary to OFI samples, showed downregulation of ROS in most of the cellular subsets (Fig. 8b). Of all the leucocyte subsets, T cells and non-classical monocytes (CD14^dim CD16⁺) showed a significant reduction in ROS levels between dengue and OFI samples (Fig. 8b). These data suggest that redox balance during DENV infection may be altered and modulation of pathways dealing with oxidative stress may have an important role in the clinical outcome of dengue disease.

Table 1. Clinical features of dengue patients enrolled into the study.

	DI (8)	DW (9)	SD (6)	P value	Statistical test
Sex (M:F)	3 : 5	8 : 1	3 : 3	0.052	Fisher’s exact
Age, yrs: Mean (SD)	10 (2.5)	11.4 (1.4)	10.4 (1.2)	0.257	ANOVA
Median Hours of fever (Min-Max)	75 (61–120)	96 (48–120)	96 (95–168)	0.150	Kruskal-Wallis
Median Platelet counts (×1000) μl−1 (Min-Max)	137 (90–201)	60 (36–79)	17 (11–68)	<0.001	Kruskal-Wallis
Haemoglobin (g dl−1); Mean (SD)	12.3 (0.7))	12.4 (2)	12.4 (2.6)	0.994	ANOVA
Hematocrit (%); Mean (SD)	37.2 (2.5)	37.9 (6.8)	36.9 (6.8)	0.945	ANOVA
Median TLC μl-1 (Min-Max)	3550 (2000–6900)	3900 (2000–5900)	9150 (7100–14900)	0.002	Kruskal-Wallis
Dengue serotype	DEN2 (2), DEN3 (3), DEN4 (2) DEN2 and 4 (1)	DEN2 (2), DEN3 (5) DEN2 and 4 (1) Undetermined (1)	DEN1 (2), DEN2 (1), DEN3 (1), DENV4 (1) DEN2 and 4 (1)	–	–

Fig. 8. Detection of ROS in blood cell subsets from dengue patients.

(a) Peripheral blood samples from healthy volunteers, dengue patients or patients with other febrile illness (OFI) were processed as described in the experimental procedures and stained using H₂DCFDA by flow cytometry. Disease severity in patients was classified as per the WHO guidelines as DI-Dengue infection, DW-Dengue with warning signs and SD-Severe dengue. (b) Indicated blood cell subsets were identified using cell surface markers and ROS levels in these subsets were estimated using H₂DCFDA by flow-cytometry. Fold change in MFI of peripheral blood leucocyte subsets relative to healthy controls is shown. **<I>P<0.01.+indicates the mean value.

Discussion

Reactive oxygen species (ROS) are chemical intermediates involved in various cellular and signalling processes to maintain redox homeostasis. However, overproduction of ROS leads to oxidative stress which can interfere with critical physiological processes. We had shown previously that zinc chelation by TPEN leads to inhibition of dengue replication by induction of interferon response [17]. Zinc is one of the most relevant elements associated with the regulation of redox environment and zinc-deficiency leads to excessive ROS formation [28]. Therefore, in this study, we further investigated whether zinc chelation leads to ROS production and if ROS is involved in anti-dengue activity. We show that TPEN treatment leads to induction of ROS which could be rescued by zinc but not by other divalent cations. Furthermore, we show that addition of zinc or antioxidants counteract the effect of TPEN on DENV infection. We found that ROS has an antiviral role and specifically inhibits some of the positive-strand RNA viruses that were tested in this study. This effect of ROS appears to be at the level of ER homeostasis or trafficking pathways as most of the free zinc was found to be associated with ER and Golgi. Many of the positive-strand RNA viruses are known to utilize ER or Golgi-derived membranes for replication [29]. Our results show that H₂O₂ treatment specifically affected SEC31A protein, a component of COPII complex involved in the cellular trafficking from ER to Golgi apparatus. Most of the previous studies have implicated COPII complex in transporting viral proteins through the secretory pathway thereby affecting virus assembly, maturation or release [30–33]. In the case of CHIKV, silencing of COPI and COPII components led to pronounced reduction in viral RNA replication suggesting that disruption of COP complexes may affect virus replication most probably by perturbing the biogenesis of membranous compartments required for segregation of replication complexes from the virus assembly and packaging processes [34]. This is specifically relevant for positive strand RNA viruses as most of these viruses are dependent on the ER homeostasis for replication [29]. Interferon-induced genes are known to affect secretory pathways at the ER and disrupt production of infectious particles in flavivirus infection [32]. As our previous report had shown induction of interferon-dependent pathway upon zinc chelation, we speculate that zinc chelation may trigger the cellular antiviral response by employing both the interferon and the oxidative stress responses which affect the ER. Alternatively, it has been shown that transient induction of oxidative stress leads to partitioning of mRNA transcripts to stress granules leading to selective translation of genes for stress adaptation [35]. Therefore, induction of oxidative stress at early stages of viral infection may disrupt the formation of ER-derived compartments and/or segregation of viral replication complexes from cellular dsRNA sensing machinery thus leading to inhibition of viral replication. This provides a strategy to transiently target the redox balance of the cells to intervene with dengue virus and other positive-strand RNA virus replication during early stages of infection or disease.

Several studies have highlighted the effect of oxidative stress and ROS in the modulation of cellular pathways and, depending upon the replication strategy the virus adapts, ROS has divergent effects on viral life cycle. Coronaviruses have been shown to induce oxidative stress which is suggested to be a major contributor in the clinical manifestations of severe disease [8, 9]. However, we show that in our cell culture models of infection, addition of H₂O₂ was inhibitory to SARS-CoV-2 infection indicating that ROS may play an antiviral role if induced early on in infection. In the case of HCV infection, it has been demonstrated that increased ROS led to viral inhibition [11]. Similar to this observation, we show that H₂O₂ treatment inhibits dengue infection and our data suggests that short bursts of ROS at early stages of infection can block virus replication without any cellular damage. Therefore, it is plausible that dengue virus and other positive-strand RNA viruses would adopt strategies to counter oxidative stress. We observed that ROS levels in total leukocytes from dengue patients were lower as compared to controls. Although T cells from OFI samples showed induction of ROS levels, the same was not observed in dengue samples suggesting that dengue infection may not trigger an oxidative stress response. ROS acts as a signalling intermediate in T cells and has been shown to regulate the differentiation of a naive T cell towards Th1 and Th17 or Th2 pathway [36, 37]. Therefore, lack of oxidative stress may influence the activation and differentiation of T cells in dengue infection. We had earlier shown that dengue virus primarily replicates in monocyte subsets [14, 15]. ROS has been shown to be required for monocyte proliferation and maturation [38, 39]. ROS may also mediate cross-talk between neutrophils and patrolling monocytes during inflammation [40]. Interestingly, all the three monocyte subsets, the classical, intermediate and non-classical subsets, showed a trend towards downregulation of ROS however the non-classical or the patrolling monocytes showed the most significant downregulation relative to non-dengue samples. It has been shown that patrolling monocytes monitor blood vessels and inflammatory triggers leads to interaction of patrolling monocytes with neutrophils which leads to activation of neutrophils and production of ROS leading to tissue damage. Therefore, it would be interesting to investigate the effect of viremia on the patrolling behaviour of non-classical monocytes and endothelial leakage observed in severe dengue infection. Macrophages are one of the primary producers of ROS. Therefore, we hypothesize that lower ROS levels measured in DENV-infected samples may be due to one or more of the following reasons: (i) decreased activation of macrophages, (ii) suppression of oxidative stress pathways or, (iii) upregulation of antioxidant pathways by DENV.

Our data suggests an opposite effect with respect to ROS as compared to previous studies with dengue patients or YF17D vaccinees [41–43]. We propose that induction of ROS by an attenuated YF17D suggests a positive role for ROS in shaping the host response to this virus which culminates in long-term immunity. Since there is no data on the effect of wild-type YFV on ROS, it would be difficult to directly extrapolate these findings in the context of a natural dengue infection. However, there are major differences between some of the previous studies and our study, in terms of patient population, timelines of sample collection and assays performed to assess oxidative stress. Other studies have reported data in adult population [42, 43], with hospitalized dengue patients who exhibited classical symptoms of DHF and DSS at days 5 or 7 post-enrolment by measuring serum markers of oxidative stress such as malondialdehyde and protein carbonyls and show an association between these serum markers and severe dengue. This likely indicates a role for oxidative stress in disease progression and suggests that oxidative stress may be induced at later stages of infection contributing to pathogenesis in severe dengue. Our samples are from paediatric patients who were enrolled early (less than 3 days of symptoms) in infection to measure viremia and detect any correlation between viral replication kinetics and oxidative stress. We measured ROS directly in PBMC subsets and show that although other febrile illness samples showed higher levels of ROS in PBMCs, dengue samples did not show any increase. Furthermore, samples with higher viral load showed lesser ROS induction suggesting a link between dengue viremia and ROS at early stages of the disease. Molecular mechanisms behind the role of ROS in dengue and other flavivirus pathogenesis warrants further investigation. Our future studies will focus on the tripartite interactions between zinc homeostasis, reactive oxygen species and ER homeostasis which together may regulate DENV replication and immune responses.

Abbreviations

CHIKV

chikungunya virus

DENV

dengue virus

ER

endoplasmic reticulum

H₂DCFDA

2′,7′-dichlorodihydrofluorescein diacetate

JEV

Japanese encephalitis virus

OFI

other febrile illness

ROS

reactive oxygen species

TPEN

N,N,N′,N′-tetrakis(2-pyridinylmethyl)-1,2-ethanediamine