Dynamic oscillation of translation and stress granule formation mark the cellular response to virus infection

Abstract

Virus infection-induced global protein synthesis suppression is linked to assembly of stress granules (SGs), cytosolic aggregates of stalled translation preinitiation complexes. To study long-term stress responses, we developed an imaging approach for extended observation and analysis of SG dynamics during persistent hepatitis C virus (HCV) infection. In combination with type 1 interferon, HCV infection induces highly dynamic assembly/disassembly of cytoplasmic SGs, concomitant with phases of active and stalled translation, delayed cell division, and prolonged cell survival. Double-stranded RNA (dsRNA), independent of viral replication, is sufficient to trigger these oscillations. Translation initiation factor eIF2α phosphorylation by protein kinase R mediates SG formation and translation arrest. This is antagonized by the upregulation of GADD34, the regulatory subunit of protein phosphatase 1 dephosphorylating eIF2α. Stress response oscillation is a general mechanism to prevent long-lasting translation repression and a conserved host cell reaction to multiple RNA viruses, which HCV may exploit to establish persistence.

Figure 1. HCV Infection Induces SG Formation in IFN-α-Treated Human Hepatoma Cells

(A) Huh7 YFP-TIA1 cells (green) were infected with HCV for 24 hr and subsequently cultured with or without IFN-α. Characterization of HCV-induced SGs 18 hr after addition of 100 IU/ml IFN-α. Cells were fixed and stained for NS5A (blue) and TIA-1 (red). Shown are representative confocal micrographs. Scale bars, 25μm. See also Figure S1. (B) Quantification of the kinetics of HCV-induced SGs. Cells were treated with 100 IU/ml IFN-α and fixed at given time points. NS5A was detected by fluorescence microscopy. Shown are mean values ± SD from three independent experiments with at least 100 cells analyzed each. Statistical significance is given on the top. n.s., no statistical significance. (C) Quantification of HCV-induced SGs upon treatment with increasing concentrations of IFN-α. Cells were analyzed as described in (B). (D) Quantification of PB docking to HCV-induced SGs in IFN-α treated cells and comparison to SGs induced by the indicated stresses. The PB marker Hedls (red) was detected by confocal microscopy. Shown are maximum-intensity Z-projections of representative cells. Deconvolved optical sections were analyzed using 3D surface rendering (column “3D”). Red and yellow dots indicate PB docking to the front or the back side of SGs, respectively. A quantification of PB docking to SGs is shown s in the right of each panel. Number of cells and SGs analyzed are given in the top. See also Figure S2.

Figure 2. SG Components Promote HCV Replication but Are Dispensable for IFN-α-Mediated Suppression of HCV

(A) Schematic of the RNAi-based screen set-up to measure rescue of HCV replication in IFN-α treated cells upon knockdown of a given SG- or PB-related gene. (B and C) Results of RNAi screen. A STAT2-specific siRNA (green box) and a nontargeting siRNA (NT; red box) were included as positive and negative control, respectively. Red and green dashed lines correspond to Z-score = 0 as determined by NT siRNA and to Z-score ± 2, respectively. (B) Boxplots of quantile raw values of HCV replication in absence of IFN-α normalized to cells transfected with NT siRNA. siRNAs leading to an efficient knockdown (Z-score > 2) and a significant change of HCV replication (p value < 0.05) are highlighted with blue boxes. These genes were excluded from the calculation shown in panel (C). (C) Boxplots of quantile ratios between HCV replication in IFN-α treated and untreated cells (mock). Only knockdowns giving rise to a p value < 0.05 and a Z-score > 2 in both calculations (raw values and ratios) were considered as hit candidates. In case of CNOT1 and Rck1, two dependency factors (see B), knockdown and IFN-α treatment reduced replication to nonmeasurable levels and therefore, values could not be determined (n.a., not applicable). See also Figure S3.

Figure 3. SGs Induced in HCV-Infected Cells upon IFN-α Treatment Are Highly Dynamic

(A) Representation of the imaging procedure to visualize the dynamics of HCV-induced SGs. Movies were created from 3D time series by using maximum intensity Z-projections (MIP) and acquired for 72 hr at 1 hr intervals. See also Figure S4. (B) Representative MIP of the first acquisition time point (T0). White square represents a cropped section shown as example in (C). Scale bar, 50μm. (C) Cropped section of a MIP of the YFP-TIA1 channel representing the first 23 hr of time-lapse acquisition with 1 hr intervals. Two representative cells (1* and 2*) are labeled. The white arrow indicates phases of detectable SGs in cell 1*. (D) Dynamics of SGs in HCV-infected cells treated with 100 IU/ml IFN-α. SG oscillation plots indicate presence or absence of SGs in individual infected cells during the 72 hr observation period. Cell 1* and 2* correspond to the ones highlighted in (C). See also Movie S1. (E and F) Scatter plots of SG oscillation frequency and interval from at least 100 randomly selected HCV-infected cells in presence or absence of IFN-α. Shown in red are mean values ± SD. Statistical significance and the number of analyzed cells (n) are given in the top. (E) SG oscillation frequency is given as the number of stress peaks per 24 hr observation time. (F) SG oscillation interval corresponds to the mean stress peak duration in infected cells showing SGs. Percentage of the total population showing SGs is given in the top (see also Figure S5). (G) SG oscillation plots of two representative HCV-infected cells in absence of IFN-α. See also Movie S2.

Figure 4. SG Induction by HCV Coincides with Delayed Cell Division and Translational Arrest

(A and B) Cell division properties of naive or HCV-infected Huh7 YFP-TIA1 cells during the 72 hr observation period. “Mother cell” corresponds to the cell chosen for analysis before its division. Sister cells result from a mother cell division and correspond to one generation. A randomly selected uninfected (A) and HCV-infected (B) cell treated with IFN-α is shown. Black dots represent stress phases. (C) Quantification of the generation number of Huh7 YFP-TIA1 cells. Shown is a comparison between uninfected (upper left panel) and HCV-infected (upper right panel) cells treated with IFN-α and untreated HCV-infected cells (lower left, cells without SGs, and lower right, cells with SGs). (D) SG oscillation plot of a representative IFN-α-treated HCV-infected mother cell that divides after a 16 hr observation period. (E) Reduction of RNA translation in HCV-infected cells during SG periods. Left panel: Huh7 cells infected for 24 hr with HCV and treated 18 hr with IFN-α were incubated 1 hr in methionine-free medium in presence or absence of 200 μM cycloheximide (CHX). De novo synthesized proteins were labeled by incubating cells with Click-iT metabolic labeling reagent coupled to Alexa Fluor 594 (AHA-594, red). After fixation of cells, SGs were immunostained for eIF3B (green). Three exemplary view fields are shown. “1”: cell with SGs and a de novo protein synthesis level reduced to that of cells treated with CHX. “2”: cells with SGs and residual de novo protein synthesis. Right panel: scatter plot of de novo protein synthesis measured by mean fluorescence of AHA-594 intensity and shown as arbitrary unit (a.u.). “1” and “2” refer to cell populations specified above. Shown in red are mean fluorescence intensities ± SD. The number of analyzed cells (n) is given in the top.

Figure 5. The eIF2α Kinase PKR Is Necessary and Sufficient to Induce SG Formation in HCV-Infected Huh7 Cells

(A) Western blot analysis of cells infected with HCV for 24 hr (lane 2) or cells infected for 24 hr and cultured for 18 hr in the presence (lane 4, 6) or absence of IFN-α (lane 3, 5). Naive cells cultured for 24 hr in parallel were used as reference (lane 1). (B and C) Transient silencing of PKR in Huh7 YFP-TIA1 cells. Cells were transfected without siRNA (Mock), with nontargeting siRNA (siRNA NT) or with PKR-specific siRNA (siRNA PKR). Twenty four hours later cells were infected with HCV for 24 hr and treated or not with IFN-α for 18 hr. (B) Western blot analysis. (C) Cells were fixed, stained for NS5A, and analyzed. Shown are the mean values ± SD from three independent experiments with at least 100 cells analyzed each. (D and E) Overexpression of PKR in Huh7 YFP-TIA1 cells. (D) Upper panel: Western blot analysis of Huh7 YFP-TIA1 control cells (Ctrl) treated or not with 100 IU/ml IFN-α for 18 hr or Huh7 YFP-TIA1-PKR cells. Lower panel: characterization of HCV permissiveness of Huh7 YFP-TIA1-PKR cells. Cells were infected with a HCV Renilla luciferase reporter virus, lysed at different time points post-infection and luciferase activity was measured. HCV replication was normalized to the 4 hr values to correct for transfection efficiency (RLU, relative light units). (E) Quantification of HCV-induced SGs in Huh7 YFP-TIA1-PKR cells. Cells were infected with HCV for 24 hr and subsequently treated or not with IFN-α for 18 hr. Cells were fixed, stained for NS5A and analyzed. Shown are the mean values ± SD from three independent experiments with at least 100 cells analyzed each. (F–H) Scatter plots of SG oscillation frequency, interval, and total stress duration from randomly selected HCV-infected Huh7 YFP-TIA1-PKR cells. Shown in red are mean ± SD. Statistical significance and the number of analyzed cells (n) are given in the top. HCV results (gray symbols) are shown as reference. (F) SG oscillation frequency in HCV infected cells. (G) SG oscillation interval in HCV-infected cells. Number of analyzed cells (n) and corresponding percentage of the total population showing oscillating SGs are given in the top. (H) Total stress duration (i.e., the ratio between total stress time and the observation period for an individual cell) in HCV-infected cells is given in percent. See also Movie S3.

Figure 6. GADD34 Controls SG Formation and Cell Viability

(A) mRNA quantification of GADD34 in Huh7 cells infected with HCV for 24 hr and cultured for 18 hr with or without IFN-α. All values were normalized to GAPDH mRNA levels. Results represent fold induction relative to uninfected cells. Shown are means of triplicate measurements ± SD of a representative experiment. Statistical significance is given in the top. n.s., no statistical significance. (B) mRNA quantification of PKR and GADD34 in Huh7 cells infected with HCV as given in panel A. Cells were harvested at given time points and mRNA amounts were quantified relative to GAPDH mRNA levels. ISG56 mRNA served as control for IFN-α response of Huh7 YFP-TIA1 cells. Results represent fold induction relative to uninfected cells. Shown are mean values ± SD from three independent experiments. See also Figure S6. (C–E) Characterization of GADD34 overexpressing cells. (C) Quantification of GADD34 mRNA in parental Huh7 (Ctrl) and Huh7-GADD34 cells, relative to GAPDH mRNA levels. Results represented fold induction relative to Ctrl cells. Shown are means of triplicate measurements ± SD of a representative experiment. (D) Ectopic expression of GADD34 does not affect HCV replication. Cells were transfected with in vitro transcripts of a HCV Renilla reporter virus, lysed 4, 24, 48, 72, and 96 hr later and luciferase activities were determined (Relative Light Units, RLU). (E) Quantification of HCV-induced SGs in Huh7-GADD34 cells (see also Movie S4). Cells were infected with HCV for 24 hr and subsequently treated or not with IFN-α for 18 hr. Cells were fixed, stained for NS5A, and analyzed. Shown are the mean values ± SD from three independent experiments with at least 100 cells analyzed each. (F) Time-lapse series of Huh7 YFP-TIA1cells that were left untreated or treated with 50 or 75 μM Guanabenz, a GADD34-specific inhibitor. Incubation times with the drug are given in the top. (G) Cytotoxicity of Guanabenz treatment. Huh7 YFP-TIA1 cells were treated with increasing amounts of Guanabenz for 2, 6, or 10 hr. Cell viability was determined by measuring release of lactate dehydrogenase. Values were normalized to untreated cells (0μM) and are represented as percentage of cell survival. (H) mRNA quantification of GADD34 in Huh7 YFP-TIA1 cells treated with 50 μM Guanabenz. Cells were collected at given time points and GADD34 mRNA amounts were determined relative to GAPDH mRNA levels. Results represent fold induction relative to untreated cells. Shown are means of triplicate values ± SD. of a representative experiment. (I) Western blot analysis of Huh7 YFP-TIA1 cells treated for 4 hr with 75 μM Guanabenz (lane 2). Untreated cells were used as reference (lane 1).

Figure 7. SG Oscillations Are Triggered by dsRNA

(A) Oscillation plots of three representative Huh7 YFP-TIA1 cells infected with SFV, DENV, SeV, or NDV. Red cross, cell death. (B) Scatter plots of SG oscillation frequencies; see also Figure S7 and Movies S5 for SFV; Figure S6 for NDV; and Figure S7 for poly(I:C). Analyses are based on randomly selected Huh7 YFP-TIA1 cells infected with SFV, DENV, SeV, NDV, or transfected with 4 μg poly(I:C). Shown in red are means ± SD. Number of analyzed cells (n) is given in the top. SG oscillation frequencies of HCV-infected cells with SGs are shown as reference. (C) Quantification of death of Huh7 YFP-TIA1 cells containing SGs. Histograms represent the percentage of death of cells infected with HCV (±IFN-α, respectively), or transfected with poly(I:C). Uninfected cells treated with IFN-α (mock) were used as control. (D) Model of cell homeostasis and its perturbation by HCV and IFN-α treatment. Left panel: basal levels of phospho-eIF2α are continuously regulated by the PP1 complex and its regulatory subunit GADD34. Under these conditions, sufficient amounts of eIF2α are available for initiation of protein synthesis (translation ON) and SGs are not formed (no SG). This cytoprotective state reflects homeostasis and thus cell survival. 1: Upon HCV infection, dsRNA is formed (red waved lines). Addition of IFN-α limits HCV replication and induces expression of PKR that binds viral dsRNA. Activated PKR phosphorylates eIF2α, thus arresting RNA translation (translation OFF). 2: GADD34 transcription and translation are increased in response to eIF2α hyper-phosphorylation. GADD34 causes rapid dephosphorylation of eIF2α by PP1, restoring RNA translation and promoting SG disassembly and thus stress recovery. Under this condition GADD34 is rapidly downregulated resulting in accumulation of phosphorylated eIF2α. This regulatory circuit sustains SG oscillation that correlate with active and stalled phases of RNA translation.