Dengue Virus Perturbs Mitochondrial Morphodynamics to Dampen Innate Immune Responses

Abstract

With no antiviral drugs or widely available vaccines, Dengue virus (DENV) constitutes a public health concern. DENV replicates at ER-derived cytoplasmic structures that include substructures called convoluted membranes (CMs); however, the purpose of these membrane alterations remains unclear. We determine that DENV nonstructural protein (NS)4B, a promising drug target with unknown function, associates with mitochondrial proteins and alters mitochondria morphology to promote infection. During infection, NS4B induces elongation of mitochondria, which physically contact CMs. This restructuring compromises the integrity of mitochondria-associated membranes, sites of ER-mitochondria interface critical for innate immune signaling. The spatio-temporal parameters of CM biogenesis and mitochondria elongation are linked to loss of activation of the fission factor Dynamin-Related Protein-1. Mitochondria elongation promotes DENV replication and alleviates RIG-I-dependent activation of interferon responses. As Zika virus infection induces similar mitochondria elongation, this perturbation may protect DENV and related viruses from innate immunity and create a favorable replicative environment.

Graphical abstract

Figure 1

DENV Infection Induces Mitochondria Elongation (A) Huh7 cells were infected with the DENV2 strain 16681 (DVs) at MOI = 1 or left uninfected. Three days later, cells were fixed, permeabilized, and indicated proteins were visualized by immunofluorescence using confocal microscopy. Incubation with MitoTracker was performed right before fixation. (B) Quantification of immunofluorescence shown in (A). Upper panel: the mitochondrial network of ∼100 cells per condition and experiment (n = 4) was examined and classified into three morphological categories (normal, fragmented, elongated). Lower panel: the morphology of the mitochondrial network was analyzed in uninfected, DVs- or WNV-infected cells at 1, 2, and 3 days post infection. Error bars indicate SEM calculated using four independent experiments. (C) Upon optical sectioning and deconvolution, Z-stacks were used for 3D reconstruction of control and DVs-infected cells stained for NS3, NS4B, and mitochondria. (I and II) Magnifications of regions of interest (indicated with white cuboids) show the proximity between NS3/NS4B-containing punctae and elongated mitochondria. The viewing angles are indicated with the white arrows in the cuboids. (D) Live cell imaging frame captures of DVs-infected Huh7 cells (MOI = 10) expressing a mitochondria-targeted mTurquoise2 fluorophore. Captures were made 20, 28, 36, 44, and 52 hr post infection and show progressive elongation of mitochondria over time. To facilitate tracking, two cells of interest are labeled with α and β. Scale bar, 10 μm, except (CI) and (CII).

Figure 2

DENV2 Induces Mitochondria Elongation in the Vicinity of Viral Convoluted Membranes and Alters Mitochondria-ER Contacts (A) Huh7 cells were infected with DENV2 NGC (MOI = 5). 24 hr later, cells were processed for FIB/SEM. Tridimensional reconstruction for mitochondria (red) and CM (green) was performed. Scale bar, 2 μm. (B) Sections of control and DVs-infected cells were analyzed by TEM. Magnification of regions of interest, red squares) are given below each top panel to highlight that ER-mitochondria contacts sites are mostly disrupted. In addition, CMs are connected to mitochondria (M) at distinct sites via ER membranes (red arrows). (C) The length of mitochondria was measured with the ImageJ software package using multiple images acquired by TEM; lengths were classified into three categories. (D and E) Elongation of CM-proximal mitochondria. The distance between mitochondria and CMs (d) was measured. When this distance was below or above 400 nm, a given mitochondria was considered proximal (PROX.) or distal (DIST.), respectively. Mitochondria lengths (y axis) of these two groups are shown and were classified as in (C). Further examples of EM images are given in Figure S4B. (F) Huh7/mito-mTurquoise2/YFP-Sec61β cells were left uninfected or infected with DVs. Three days later, cells were fixed and stained for NS3 (not shown). Optical sections were acquired with a spinning disc confocal microscope and after deconvolution, Z-stacks were used for 3D reconstruction and 3D colocalization analysis between mitochondria and YFP-Sec61β-containing ER networks. Note that NS3-positive cells were considered infected. White areas on reconstructed mitochondria indicate sites of colocalization with YFP-Sec61β (Coloc.). Scale bar, 1 μm. (G) A 3D colocalization analysis quantifying the extent of mitochondria/Sec61β contacts in uninfected (n = 30) and DVs-infected cells (n = 33).

Figure 3

CMs Contain Both NS4B and NS3, and Their Biogenesis Is Linked in Time and Space with Mitochondria Elongation in DENV-Infected Cells (A–C) Huh7 cells were infected with DVs(NS4B-HA^∗). Three days post infection, cells were processed for immunolabeling of thawed cryosections using NS3- or HA-specific primary antibodies to detect NS3 (A) or NS4B-HA^∗ (B and C), respectively. Red arrows: immune-gold-labeled NS3 (A) or NS4B-HA^∗ (B) within DENV-induced CMs. Blue arrows: immune-gold-labeled NS4B-HA^∗ (C) associated with VPs. (D–F) Huh7/mito-mTurquoise2/YFP-Sec61β cells were infected with DVs and processed for CLEM three days post infection. The mitochondria signal and NS3/NS4B-containing YFP-Sec61β puncta are shown in red and green, respectively. Mitochondria morphologies were used to allocate the red fluorescent signal of a Z-stack to the proper ultrastructure of the EM image and hence, for overall correlation. Regions of interest, yellow dashed squares, were selected for higher magnification analyses that are shown in (F). (G and H) Huh7/mito-mTurquoise2/YFP-Sec61β cells were infected with DVs (MOI = 10) and analyzed by live-cell imaging. For 140 cells from 13 movies, the time of first appearance of CMs was determined (G) and correlated with the time when a given mitochondrial network started to elongate. Cells in which YFP-Sec61β accumulated within puncta were considered infected with DENV. (H) Infected cells were classified into four defined categories: tElong. ≥ tCM: Mitochondria elongation occurred after CM appearance; tElong. ≤ tCM: CM appeared after mitochondria elongation; tElong. = tCM. The two processes occurred simultaneously. CM w/o elong.: CM formation without detectable mitochondria elongation.

Figure 4

DENV2 NS4B Induces Mitochondria Elongation (A) Schematic representation of used constructs. The subgenomic replicon is depicted in the top. The constructs used for expression with the T7-based system are indicated below. Black lines represent the DENV and HCV untranslated regions (NTRs). The HA-tag is shown as red oval. (B) Huh7 cells were electroporated with capped in vitro transcripts of a sub-genomic RNA. Three days later, cells were processed for detection of ATP5B and DENV NS4B by immunofluorescence using confocal microscopy. (C and D) Huh7 cells stably expressing T7 RNA polymerase only (Huh7-T7) (C and D) or T7 RNA polymerase and DENV NS2B-3 (Huh7-T7/NS2B-3) (not shown) were transfected with the expression constructs specified in the left of each panel. 24 hr later, cells were incubated with MitoTracker, fixed, immunostained for the indicated proteins and analyzed by confocal microscopy. Scale bar, 10 μm. (E) Based on the analysis of mitochondria morphology of at least 150 cells per condition and at least three independent experiments, cells were classified into three groups as follows: elongated, normal and fragmented mitochondrial network. Error bars indicate SEM calculated using three or four independent experiments.

Figure 5

Mitochondria Elongation Favors DENV Replication (A) Huh7 cells were transduced with shRNA-expressing lentiviruses. Four days post transduction, cells were collected and the expression of both DRP1 and MFN2 was analyzed by western blotting. (B) The morphology of mitochondria in shRNA-transduced cells was analyzed by confocal microscopy after staining with MitoTracker or by TEM (lower panels). White scale bar, 10 μm; black scale bar, 1 μm. (C) Cell viability of transduced cells was evaluated 2, 4, or 5 days post-transduction using the CellTiter-Glo assay, which is a measure for ATP levels in the cells. (D–F) Two days post transduction, cells were infected with (D) the DVs-R2A reporter virus (MOI = 0.1), (E) DVs (MOI = 0.01), (F) NGC (MOI = 0.01). (G) ZIKV H/PF/2013 (MOI = 0.05) or (H) WNV NY99 (MOI = 0.05). At the indicated time points, Renilla luciferase activity (RLU) reflecting DENV RNA replication (D) or virus titers (E–H) were determined. Mean values and SD are indicated in (C) displaying a representative of at least three independent replicates. Mean values and SEM are indicated in (C) and (F)–(H) based on at least three independent experiments. ^∗∗p ≤ 10⁻³; ^∗p ≤ 10⁻²; N.S: not significant (p ≥ 0.05). (I) Huh7-T7 cells were transduced with shRNA-expressing lentiviruses and transfected three days post transduction with a plasmid encoding a DENV2 NS1-5 polyprotein. 16 hr later, cells were analyzed for knockdown efficiency and NS4B expression using western blotting. (J) Cells from (I) were fixed, permeabilized and NS4B was visualized by immunofluorescence using confocal microscopy. (K) Optical sections were acquired for 50 cells in two independent experiments and, after 3D reconstruction, number and mean volume of NS4B punctae per cell were determined.

Figure 6

DENV2 Inhibits the Activation and Mitochondrial Translocation of the Fission Factor DRP1 (A) Huh7 cells were infected with DVs (MOI = 1). Three days post infection, cells were labeled with MitoTracker, DAPI and antibodies specified on the top. Dotted boxes indicate areas that are enlarged in the respective adjacent panel. Scale bar, 10 μm. (B) Extracts of naive or DVs-infected cells were prepared 72 hr after infection and analyzed by western blotting using antibodies recognizing differentially phosphorylated (Ser616 or Ser637) or total DRP1. ^∗Non-specific band. (C) Extracts of Huh7 and HA-DRP1-overexpressing Huh7 cells were subjected to immunoprecipitation using an HA-specific antibody. Captured complexes were analyzed by western blotting for phosphorylated and total DRP1. (D) Cell lysates were prepared as in (B) and analyzed by western blotting. (E) Huh7 cells were transduced with lentiviruses expressing the indicated DENV proteins (MOI = 5). Four days later, cell extracts were prepared and analyzed for the expression of the indicated proteins using western blotting.

Figure 7

Mitochondria Elongation Alleviates DENV2-Induced Activation of the Interferon Response Huh7 and A549 cells were transduced with DRP1- or MFN2-specific shRNA-encoding lentiviruses. Two (Huh7) or three (A549) days post-transduction, cells were infected with DENV2 DVs or NGC (MOI = 5). (A and B) 48 (Huh7) or 24 (A549) hr later, cells were collected and mRNA amounts of IFN-λ1, IFN-β, and ISG56 were quantified using qRT-PCR. Quantifications are representative of three independent replicates. (C) Knockdown efficiency in A549 cells was monitored 4 days post transduction by western blotting using MFN2- and DRP1-specific antibodies. Actin served as loading control. (D) Huh7 RIG-I knockout cell pools were generated by CRISPR/Cas9 technology using two different guide RNAs and lysates of cells were analyzed by western blotting. (E) RIG-I knockout cells were infected with DVs or left uninfected. Two days later, abundance of IFN- λ1 and β-actin mRNA was measured using qRT-PCR. IFN- λ1 induction levels upon infection were calculated after normalization to β-actin. For qRT-PCR experiments (A, B, E), error bars indicate SD calculated using three technical replicates. (F) Huh7 cells were transduced and infected with DVs as in (A). Cell extracts were subjected to subcellular fractionation to isolate MAMs. Equal amounts of proteins from each fraction were analyzed by western blotting using antibodies recognizing the indicated proteins. FACL4 and calnexin served as markers enriched in the MAM fraction and α-tubulin as cytosolic marker.