Dengue Virus Induces and Requires Glycolysis for Optimal Replication

Cells and viruses.

Primary HFFs (ATCC PCS-201-010) were propagated and maintained in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 2 mM l-glutamine, 100 U/ml penicillin, and 100 μg/ml streptomycin. Human telomerase reverse transcriptase-immortalized microvascular endothelial (TIME) cells were maintained in EGM-2 medium (Lonza). For the glucose and glutamine depletion studies, DMEM lacking d-glucose, l-glutamine, sodium pyruvate, and phenol red (catalog number A14430-01; Invitrogen) was used. This medium was supplemented with 2% dialyzed fetal bovine serum (catalog number SH30079.03; HyClone), and for replete medium, 1 g/liter d-glucose and 2 mM l-glutamine were added. Dialyzed serum was thoroughly depleted of small molecules, including glucose and glutamine. The glucose concentration reported for the lot of dialyzed HyClone serum used was <20 mg/dl; therefore, when it is used at 2%, the glucose level in the medium at less than 0.004 g/liter.

Stocks of DENV type 2/NG-C (provided by Michael Gale, Jr., University of Washington) were propagated, and titers were determined by plaque assays on Vero cells. Experimental infection rates were determined by immunofluorescence assay. DENV-infected cells were fixed at 24 h postinfection (hpi) with 10% (wt/vol) formalin, permeabilized with 70% (vol/vol) ethanol, and blocked with 3% bovine serum albumin in Dulbecco's phosphate-buffered saline (DPBS). Cells were then immunostained with 1 μg/ml human monoclonal antibody (MAb) 1.6D (a generous gift from Sharon Isern and Scott Michael, Florida Gulf Coast University) and 2 μg/ml Alexa Fluor 488-conjugated goat anti-human antibody (Invitrogen) and counterstained with Vectashield mounting medium (Vector Laboratories).

Reagents.

Sodium oxamate (catalog number O2751), 2-deoxy-d-glucose (2DG; catalog number D8375), and deferoxamine mesylate salt (DFO; catalog number D9533) were obtained from Sigma-Aldrich. Sodium oxamate and 2DG were directly solubilized in cell culture medium and used at the specified concentrations. DFO was solubilized in deionized water and used at the final concentration indicated.

Virus infections for metabolic analysis.

HFFs were washed with DPBS and serum starved in DMEM for ∼16 h prior to infection. Cells were then mock or DENV infected (multiplicity of infection [MOI] of 9) in serum-free DMEM for 3 h, after which the inoculum was replaced with DMEM supplemented with 2% fetal bovine serum, 2 mM l-glutamine, 100 U/ml penicillin, and 100 μg/ml streptomycin. Cells were harvested with a cell scraper at the times indicated and washed once in cold DPBS, and pellets were snap-frozen in a dry ice-ethanol bath. Samples were stored at −80°C until shipment to Metabolon (Durham, NC).

A multitude of curation procedures were conducted by Metabolon to guarantee the quality of the data presented. Metabolon data analysts used proprietary visualization and interpretation software to confirm peak identification consistency and to limit system artifacts, misassignments, and background noise among the samples. Bradford assays were performed to normalize all samples by protein concentration, and each biochemical measured in the mock samples was rescaled to yield a median of 1. Following log transformation and imputation with minimum observed values for each compound, two-way analysis of variance (ANOVA) with contrast tests was used to determine which metabolites were significantly altered by DENV infection of HFFs at each time point across the four biological replicates.

Nutrient starvation and glycolytic inhibitor treatment studies.

HFFs were washed with DPBS and infected with DENV at an MOI of 3 in serum-free DMEM for 2 h. For nutrient starvation experiments, cells were washed with DPBS and fed replete medium, glucose-free medium, or glutamine-free medium. For 2DG treatment studies, DENV-infected HFFs were treated with 0, 10, or 50 mM 2DG. For oxamate treatment experiments, DENV-infected cells were treated with 0, 50, or 100 mM oxamate. TIME cells were infected with DENV at an MOI of 1 for 2 h and treated with 0 or 100 mM oxamate. Supernatant was collected from infected cells at 24 hpi and centrifuged to remove cellular debris, and titers were determined by focus-forming-unit reduction assays on Vero cells. Error bars reflect standard errors of the means from three separate experiments.

Focus-forming unit reduction assays.

Monolayers of Vero cells in 6- or 12-well plates were infected with supernatants collected from DENV-infected cells. Supernatants were diluted in serum-free DMEM and allowed to infect the cells for 1 h, after which the monolayers were overlaid with plugs consisting of a 1:1 ratio of 1.8% Agar, Noble (BD Biosciences) and 2× MEM (Lonza) supplemented with 20% FBS, 4 mM l-glutamine, 200 U/ml penicillin, and 200 μg/ml streptomycin. Four days postinfection, cells were fixed with 10% (wt/vol) formalin, permeabilized with 70% (vol/vol) ethanol, and subjected to immunostaining. Virus foci were detected with mouse anti-DENV MAb 3H5.1 (catalog number MAB8702; Millipore) or 4G2 (generous gift from Sharon Isern and Scott Michael, Florida Gulf Coast University), followed by horseradish peroxidase-conjugated goat anti-mouse immunoglobulin (catalog number 31410; Pierce), and developed with AEC chromogen substrate (Dako) or 3,3′-diaminobenzidine tetrahydrochloride (Sigma-Aldrich).

Glucose uptake assays.

HFFs in 12-well plates were serum starved for ∼16 h prior to infection. Cells were then mock or DENV infected at an MOI of 9 in serum-free DMEM for 2 h, after which the inoculum was replaced with DMEM supplemented with 2% fetal bovine serum, 2 mM l-glutamine, 100 U/ml penicillin, and 100 μg/ml streptomycin. Complete medium was replaced at 22 hpi with serum-free DMEM. At 24 hpi, cells were washed twice with DPBS (containing magnesium and calcium) and fed 0.5 μCi of deoxy-d-glucose, 2-[1,2-³H(N)] (catalog number NET549A250UC; PerkinElmer) in 500 μl of DPBS (containing magnesium and calcium) for 5 min at 37°C. Cells were then washed twice in ice-cold DPBS (without magnesium and calcium) and lysed in 300 μl of 1% SDS for 20 min at room temperature. Next, 150 μl of lysate was added to 4 ml of Biofluor Plus (PerkinElmer) and activity (disintegrations per minute) was measured by liquid scintillation counting. Cell-associated radioactivity was normalized to the protein concentration, which was determined with the BCA Protein Assay Reagent kit (Pierce). Error bars reflect standard errors of the means from three separate experiments.

Real-time RT-qPCR.

Total RNA was isolated from HFFs and TIME cells with the NucleoSpin RNA II kit (Macherey-Nagel). Two-step quantitative real-time reverse transcription-PCR (RT-qPCR; Bio-Rad) was used to measure transcript or DENV RNA levels. iScript Reverse Transcription Supermix for RT-qPCR (Bio-Rad) was used to synthesize 1 μg of cDNA from 1 μg of total RNA according to the manufacturer's instructions. One hundred nanograms of cDNA was used in SsoAdvanced SYBR green Supermix (Bio-Rad) according to the manufacturer's protocols. The primers used for hexokinase 2 (HK2) (22), glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (23), DENV RNA (24), and hypoxanthine-guanine phosphoribosyltransferase (HPRT) (10) have been described previously. Relative abundances of HK2 mRNA or DENV RNA were normalized by the delta threshold cycle method to the abundance of GAPDH or HPRT mRNA, with mock-infected cells set to 1 or DENV-infected cells fed replete medium set to 100. Error bars reflect standard errors of the means from three separate experiments for the HK2 expression studies. Error bars reflect standard errors of the means from at least two independent experiments for the DENV RNA quantitation studies.

Western blot analysis.

HFFs were mock or DENV infected at an MOI of 9 in serum-free DMEM for 2 h. Following infection, cells were treated with 0 or 150 μM DFO and harvested with a cell scraper at 24 hpi. Cytoplasmic and nuclear fractions were obtained with the NE-PER nuclear and cytoplasmic extraction kit (Thermo Scientific). Proteins were separated by SDS-PAGE and transferred to Immobilon-FL polyvinylidene difluoride membranes (Millipore). Blots were incubated with the indicated primary antibody (anti-HK2 [catalog number sc-6521; Santa Cruz Biotechnology], anti-GLUT1 [catalog number G3900-01J; US Biological], and anti-β-actin [catalog number A5441; Sigma-Aldrich]) and subsequently incubated with IRDye secondary antibody (LI-COR). Proteins were visualized and differences in band intensity were quantified with the Odyssey CLx Infrared Imaging System (LI-COR).

Statistical analysis.

Standard errors of the means are shown, and statistically significant differences between groups were analyzed with Student's t test (two-tailed) or one-way ANOVA. A P value of ≤0.05 was considered significant and is indicated by an asterisk in the figures. A P value of ≤0.01 is indicated by a double asterisk and a P value of <0.0001 is indicated by a triple asterisk in the figures.

DENV infection alters glucose metabolism.

To identify DENV-induced alterations in host cellular metabolism, we utilized an intracellular metabolic profiling approach. Primary cells were used for these experiments, as transformed cells exhibit extensively altered metabolism and are therefore not ideal for investigation of virus-induced metabolic alterations (reviewed in references 13, 25, and 26). HFFs were the most permissive primary cell type tested for DENV infection and thus were chosen for these studies. HFFs were serum starved for approximately 16 h prior to infection to synchronize the cells in the G₀/G₁ phase. Cells were then mock or DENV infected at an MOI of 9 and harvested at 10, 24, and 48 hpi for metabolic analysis. When infected at an MOI of 9 (determined as the number of PFU/Vero cell), ∼50% of the HFFs expressed detectable levels of DENV envelope (E) glycoprotein by 24 hpi, as quantified by immunofluorescence. Each sample consisted of approximately 4 × 10⁶ cells, and samples from four independent experiments were analyzed on both gas chromatography-mass spectrometry (MS) and liquid chromatography-tandem MS platforms. DENV induced dynamic perturbations in the host cellular metabolome following infection of HFFs (see Data Set S1 in the supplemental material). Table 1 summarizes the changes observed in the superpathways of amino acid, carbohydrate, lipid, and nucleotide metabolism during DENV infection. Of particular interest, we noted striking alterations in glucose and glutamine utilization, with glycolysis representing one of the most dramatically perturbed metabolic pathways in DENV-infected cells. As shown in Fig. 1A, the levels of all of the glycolytic metabolites detected were significantly different in DENV-infected cells from those in mock-infected cells during the first 48 h of infection. Levels of several intermediates of glycolysis were significantly increased early during DENV infection, whereas glycolytic metabolites were measured at significantly decreased levels only at the latest time point examined postinfection. Interestingly, we observed two main trends in metabolite levels throughout the course of DENV infection. Early glycolytic intermediates, such as glucose 6-phosphate and fructose 6-phosphate, increased over time, whereas late glycolytic intermediates, including 3-phosphoglycerate and phosphoenolpyruvate, were initially higher at 10 hpi but decreased over time (Fig. 1B). It is important to note that elevated levels of a given intracellular metabolite may reflect either an increase in its production or a decrease in its consumption. Although an upregulation in glycolytic flux cannot be inferred from our global metabolomic screen, these data indicate that DENV infection modulates glucose utilization in HFFs.

In addition, we observed a considerable perturbation of glutamine metabolism during the course of DENV infection. Glutamine is catabolized via glutaminolysis, a pathway consisting of two deamination steps. Glutamine is first deaminated to glutamate and subsequently deaminated to the TCA cycle intermediate α-ketoglutarate. Figure 1C shows that both glutamine and glutamate levels were higher in DENV-infected cells than in their mock-infected counterparts. Glutamine levels were significantly increased at 10 hpi, whereas glutamate levels were significantly elevated at all three time points examined. Taken together, these data reveal that DENV implements a distinct carbon utilization program during infection.

Exogenous glucose is required during DENV infection.

Since our metabolomic screen identified DENV-induced perturbations in central carbon metabolism, we investigated the impact of glucose and glutamine deprivation on DENV production. HFFs were infected with DENV at an MOI of 3 and subsequently fed replete medium containing both glucose and glutamine or medium lacking either glucose or glutamine. At 24 hpi, the release of infectious extracellular virus was quantified by focus-forming unit reduction assays. As shown in Fig. 2A, depriving DENV-infected cells of exogenous glucose reduced the production of infectious virus by greater than 2 logs. We have previously shown that 24 h of glucose or glutamine starvation has only a modest impact on HFF proliferation and no significant impact on cell viability (10). Furthermore, depriving HFFs of exogenous glucose does not inhibit VACV replication, indicating that the block in DENV production was not due to the effects of glucose deprivation on cellular integrity. Interestingly, depriving DENV-infected cells of exogenous glutamine, the other primary carbon source for mammalian cells, reduced infectious virus production by only approximately 60% (Fig. 2A). To determine if exogenous glucose starvation has an impact on DENV genome replication, we measured viral RNA levels in DENV-infected cells fed replete medium or medium lacking glucose. Figure 2B shows that DENV RNA accumulation is significantly reduced under conditions of glucose deprivation. These results indicate that DENV demonstrates a selective dependence on exogenous glucose during infection.

Glycolysis is activated during DENV infection.

The metabolic profiling data show that the glycolytic pathway of glucose metabolism is perturbed in DENV-infected cells, and we found that DENV requires exogenous glucose for optimal replication. To determine if DENV activates glycolysis, we first examined glucose consumption following DENV infection. HFFs were mock or DENV infected at an MOI of 9, and uptake of the radiolabeled glucose analog 2-deoxy-d-[³H]glucose was measured at 24 hpi. As shown in Fig. 3A, DENV-infected cells exhibited approximately 16% greater glucose uptake than their mock-infected counterparts. Importantly, an average infection rate of 50% was achieved in these experiments. Therefore, the population of uninfected cells in the culture masked the full impact DENV infection has on glucose consumption.

Exogenous glucose uptake into mammalian cells is mediated by a family of at least 14 isoforms of facilitative transporter membrane proteins known as the glucose transporters (GLUTs). GLUT1 is ubiquitously expressed in many different cell types of various tissues and is responsible for basal glucose transport (reviewed in reference 27). To determine if the increase in glucose consumption observed in DENV-infected cells is due to upregulation of GLUT expression, we measured GLUT1 protein levels via Western blot analysis. Because GLUT1 expression is enhanced by hypoxia, we treated mock- and DENV-infected cells with DFO, a hypoxia-mimetic agent, to better visualize the relative levels of GLUT1. HFFs were mock or DENV infected at an MOI of 9, and cytoplasmic extracts were harvested from normoxic and DFO-treated cells at 24 hpi. As shown in Fig. 3B, GLUT1 protein levels were elevated in DENV-infected cells, with the greatest increase in GLUT1 expression detected under conditions of DFO treatment. These results suggest that GLUT1 induction is likely responsible, at least in part, for enhanced glucose consumption during DENV infection.

To further examine if DENV activates glycolysis, we also measured hexokinase expression in mock- and DENV-infected cells. Hexokinase is the first rate-limiting enzyme of the glycolytic pathway and the HK2 isoform, in particular, has been shown to be a key mediator of aerobic glycolysis (22, 28). We utilized real-time RT-qPCR to measure HK2 mRNA levels during DENV infection. HFFs were mock and DENV infected at an MOI of 5, and total RNA was isolated from cells harvested at 24 hpi. Figure 3C shows that HK2 expression was significantly higher in DENV-infected cells than in mock-infected cells. Levels of the three isoforms of phosphofructokinase 1, the second rate-limiting enzyme of glycolysis, were not significantly increased during DENV infection (data not shown). We next confirmed HK2 upregulation via Western blot analysis. As HK2 levels are low in normal cells, we again utilized DFO to enhance HK2 protein expression for a comparison of mock- and DENV-infected cells at 24 hpi. Figure 3D shows that HK2 protein levels were increased in DENV-infected cells, which, similar to GLUT1 levels, were enhanced following DFO treatment. Again, the GLUT1 and HK2 protein expression increases detected are diminished by the lower infection rates obtained with DENV infection of HFFs. Combined with the metabolic profiling data, these results reveal that glycolysis is induced during DENV infection.

Glycolysis is required for optimal infectious DENV production.

Because glycolysis is activated during DENV infection and exogenous glucose is essential for DENV production, we hypothesized that glycolysis is necessary for efficient viral replication. Thus, we next examined DENV replication following treatment with oxamate, a glycolytic inhibitor. Oxamate is a pyruvate analog that specifically inhibits lactate dehydrogenase, the enzyme that catalyzes the conversion of pyruvate to lactate (29–31). Figure 4A shows that DENV-infected HFFs treated with increasing concentrations of oxamate exhibited a dose-dependent decrease in the release of extracellular infectious virus. Moreover, viral RNA levels were similarly reduced following oxamate treatment (Fig. 4B), indicating that glycolysis is required at an early stage in the DENV life cycle. We confirmed these results with 2DG, a glucose analog that inhibits hexokinase, the first enzyme in the glycolytic pathway. As shown in Fig. 4C, treatment with 2DG significantly impaired DENV RNA synthesis. To determine if glycolysis is necessary for DENV replication in cell types other than HFFs, we measured the impact of glycolytic inhibition on infectious virus production and DENV genome synthesis in TIME cells. Although it is immortal, this cell line has been shown to exhibit metabolism similar to primary endothelial cells (9, 32). DENV-infected endothelial cells have been found in dengue patients and murine models of DENV disease and may contribute to DENV pathogenesis (33–40). Infected TIME cells treated with oxamate showed a block in both production of infectious DENV progeny and viral RNA levels (Fig. 4D and E), revealing that the requirement for glycolysis is not HFF specific. Together, these data indicate that glycolysis is a critical metabolic pathway for optimal DENV replication.