Histone deacetylase inhibitors prevent H2O2 from inducing stress granule formation
A B S T R A C T
Reactive Oxygen Species (ROS) are generated as by‐products of aerobic metabolism. The production of ROS increases during xenobiotic stress and under multiple pathological conditions. Although ROS are considered harmful historically, mounting evidence recently indicates a signaling function of ROS, preceding to and reg- ulating transcriptional or post‐transcriptional events, contributing to cell death or cell survival and adaptation. Among the cellular defense mechanisms activated by ROS is formation of stress granules (SGs). The stalled translational apparatus, together with mRNA, aggregates into microscopically detectable and molecularly dynamic granules. We found that with H2O2, the dose most potent for inducing SGs in HeLa cells is 400–600 μM. With 200 μM H2O2, 2 h treatment induced the highest percentage of cells containing SGs. Whether ROS signaling pathways regulate the formation of SGs was tested using pharmacological inhibitors. We probed the potential role of PI3K, MAPKs, PKC or histone deacetylation in SG formation. Using deferoxam- ine as a positive control, we found a lack of inhibitory effect of wortmannin, LY‐294002, JNK‐I, SB‐202190, PD‐ 98059, or H89 when the percentage of cells containing SGs was counted. About 35% inhibition was observed with HDAC6 inhibitor Tubastatin A, whereas general HDAC inhibitor Trichostatin A provided a complete inhi- bition of SG formation. Our data point to the need of investigating the role of HDACs in SG formation during oxidative stress.
1.Introduction
Multiple cellular defense mechanisms have evolved to deal with environmental and pathogenic stresses. Stopping protein synthesis in general serves to conserve energy yet avoid generating incorrect sequence or conformation of proteins. Upon cessation of protein trans- lation, ribonucleoprotein complexes containing 40S ribosomes aggre- gate rapidly to form stress granules (SGs) for sheltering translation apparatus and mRNA species (Anderson and Kedersha, 2008; Kedersha et al., 2002; Buchan and Parker, 2009; Souquere et al., 2009; Protter and Parker, 2016; Panas et al., 2016). Over 150 proteins have been identified in SGs, including Ras GTPase Activating Protein Binding Proteins (G3BPs), Heterogeneous Nuclear Ribonucleoproteins (HNRNPs), ELAVL1, FUS and FUBP1 (Markmiller et al., 2018). Aggregation prone proteins such as T‐Cell Intracytoplasmic Antigen‐ 1 (TIA‐1) and TIA‐1 related protein (TIAR) are essential for SG forma- tion (Kedersha et al., 1999; Gilks et al., 2004). Experimentally, SGs have been detected in tissues from diseases with increased oxidative stress, such as Amyotrophic Lateral Sclerosis, Alzheimer’s, stroke and viral infection (Mahboubi and Stochaj, 2017; McCormick andKhaperskyy, 2017). Formation of SGs serves as a self‐defense mecha- nism and a step for cell survival under oxidative stress.Reactive oxygen species (ROS) are by‐products of aerobic metabo- lism due to incomplete mitochondrial electron transport.
An increase of ROS production has been observed from five sources: 1) environ- mental stress; 2) diseases involving inflammatory response; 3) ischemia and reperfusion; 4) metabolism of xenobiotics; and 5) alkylating agents or certain cancer chemotherapeutic agents, which deplete endogenous glutathione. At the molecular level, several endocrine factors and cytokines have been reported to induce ROS generation via mecha- nisms not completely understood, including Epidermal Growth Factor,Insulin, Angiotensin II, Tumor Necrosis Factor‐alpha (TNF‐α) and Interleukin‐1β (Droge, 2002). With cells in culture, H2O2 serves wellas a convenient tool for studying the consequence of oxidative stress. ROS induce a plethora of signaling events to regulate various cellu-lar endpoints. Many of these signaling events were first discovered as an immediate response to receptor binding of growth factors or endocrine factors (Sugden and Clerk, 2006). The common signaling events of ROS include activation of Phosphoinositide 3‐Kinase (PI3K), Mitogen‐Activated Protein Kinases (MAPKs), and ProteinKinase C (PKC). The PI3K phosphorylates Akt, which in turn controls for cell survival response by inactivating pro‐apoptotic molecule Bad and inducing pro‐survival factor Bcl‐2 (Barthel and Klotz, 2005). MAPKs have three branches: Extracellular Signal‐Regulated Kinases (ERKs), c‐Jun N‐terminal Kinases (JNKs), and p38 MAPKs. Activation of ERKs is associated with cell survival response, while JNKs or p38 activation has been linked to apoptosis regulation in general (Matsuzawa and Ichijo, 2008).
Family members of PKC are involved in signaling of cell growth, cell death and stress response leading to adaptation (Gopalakrishna and Jaken, 2000; Mochly‐Rosen et al., 2012). Pharmacological inhibition of these kinases serves as the first step for understanding their participation in a specific cellular event.were counted from 20–100 cells under 100x objective lens. The diam- eters of 60 SGs were measured under 100× objective lens with “arbi- trary line” tool of the Olympus CellSens Standard software. Two‐tail Student’s t test or one way ANOVA was used to measure the signifi- cance of data sets. A value of p < 0.05 was used to determine the sta- tistical significance.Cells were seeded at a density of 3x104 per well in a 24‐well platecontaining 1 mL culture medium. When reaching 80% confluency, cells were treated with H O for 2 h and placed in fresh medium forHistone deacetylases (HDACs) can be an important regulator of 2 2ROS signaling and cellular response. Classic function of HDACs is enhancing binding of histones to DNA by removing acetyl groups from ε‐N‐acetyl lysine in histones. Nuclear HDACs regulate the expression of a long list of genes, including those encoding transcription factors, sig- naling molecules, and DNA repair enzymes. HDAC6, a cytosolic form of HDACs, can deacetylate α‐tubulin, regulating microtubule assemblyor stability. An increase activity of HDACs has been observed withinduction of oxidative stress (Miura et al., 2008). Inhibitors of HDACs alter oxidative stress induced genes (Zelko and Folz, 2015). Previous work from our laboratory showed an association of HDAC6 with 40/43S ribosomes and HDAC6 inhibitor Tubastatin prevented stress induced protein translation (Kappeler et al., 2012). Because 40/43S ribosomes are an important component of SGs, the association of HDACs with this fraction supports a need to test the involvement of HDACs in SG formation. In this study, we characterized SG formation following induction of oxidative stress by H2O2, and address which sig- naling pathway is upstream of SG formation. 2.Materials and methods HeLa cells were maintained under tissue culture condition with weekly subculture in high‐glucose Dulbecco’s Modified Eagle’s med- ium (DMEM) supplemented with 10% fetal bovine serum (FBS), 100 U/ml penicillin and 100 U/ml streptomycin. Cells were seeded into 24‐well plates at 6.5 × 104 cells per well and reached 80% con- fluency before treatment with inhibitors and H O . Inhibitors were additional 16 or 24 h of culture. MTT was added to the culture media to a final concentration of 0.16 mg/mL. Following 30 min incubation in a tissue culture incubator at 37 °C with 5% CO2, purple crystals of reduced formazan were observed inside cells under a phase contrast microscope. The culture medium was removed and the formazan crys- tals were dissolved in isopropanol containing 4 mM HCL and 0.1% NP‐ 40 for measuring cellular metabolic state by comparing the absorbance at 570 nm. 3.Results SGs are commonly viewed as stalled translation complexes contain- ing mRNAs, 40S ribosomes, eukaryotic Initiation Factors (eIFs), poly A binding proteins (PABPs), and additional RNA binding proteins. It is thought that either eIF2 α phosphorylation or non‐translating mRNA can prepare the core containing pre‐initiation complex, i.e. 40S ribo-somes plus mRNA, eIFs and the polyA binding protein, into the origin of a membrane‐free SG structure. This core recruits aggregation‐prone proteins, such as TIA‐1/TIAR, to form SG seed. Fusion of the seeds along microtubules produces microscopically visible SGs (Panas et al., 2016). TIA‐1 is an RNA‐binding protein containing a prion‐ like domain necessary for formation of SGs (Gilks et al., 2004). Immunocytochemistry staining for TIA‐1 represents a standard method for detection of SGs in the cytoplasm. Since SGs contain the 40S small subunit of ribosome due to the fact that SGs are derived from stalled translation initiation complex, we validate the specificityadded to culture medium 1 h before H2O2 treatment.Cells grown on glass coverslips were fixed in 4% formalin and washed two times with phosphate buffered saline (PBS) followed by additional wash with PBS containing 0.25% Triton X‐100. Following 40 min incubation with blocking buffer [PBS containing 1% Bovine Serum Albumin (BSA) and 0.1% Tween 20], cells were incubated with the primary antibodies against TIA‐1, S6 or L36 (Santa Cruz Biotech- nology, CA, 1:100 dilution) for 1 h with shaking. After 5 washes with blocking buffer, the cells were incubated 1 h in dark with Alexa Fluor 488 or Alexa Fluor 568 conjugated secondary antibody (Invitrogen Molecular Probes, CA, 1:600 dilution). After removal of unbound anti- body by 5 washes in PBS, the coverslips were mounted onto a micro- scopic slide with Vectashield Antifade Mounting Medium (Vectashield Laboratories, CA). Images were acquired with Olympus fluorescence microscope under 60x or 100x oil objective lenses.Cells and SGs were randomly selected for counting. To measure the percentage of cells containing SGs, about 300 cells were counted under a 60× objective lens. To determine the number of SGs per cell, SGsSG formation (Fig. 1A) and colocalization of TIA‐1 with small riboso- mal subunit protein S6 (Fig. 1B) but not large ribosomal subunit pro- tein L36 (Fig. 1C), supporting the specificity of TIA‐1 staining for SGs. Dose response studies were performed to define the dose range of H2O2 for SG formation. An early report had shown that H2O2, when added to cells in culture, disappears in culture medium in a doseand time dependent manner. With one million cells treated with 100 or 200 μM H2O2, about 90% H2O2 were eliminated from the culture medium within 60 mins (Yakes and Van Houten, 1997). This means that by 2 h, the majority if not all H2O2 has been depleted by cells. For cells treated 2 h with non‐lethal doses of H2O2, i.e. 100, 200, 300, 400, 600 or 800 μM (Fig. 2A), we found that 400–600 μMH2O2 caused SG formation in 80% cells, whereas a higher dose, i.e.,800 μM H2O2 is not as effective for inducing SG formation, caused the percentage of cells containing SGs to a level similar to that of 300 μM (Fig. 2B). Among the cells containing SGs, the number of SGs per cell does not appear to change significantly from dose to dose(Fig. 2C). Metabolic activities of the cells were measured by MTT assay at 2 h and 24 h after H2O2 treatment to show that SGs form at the doses, i.e. 200–400 μM, which do not suppress the metabolic activityof the cells significantly, when the SGs are detectable within 2 h ofH2O2 treatment, or even after 24 h. Although morphologically cells appeared normal at high doses of H2O2, i.e. 600–800 μM, the meta- bolic activity as measured by MTT absorbance declined (Fig. 2D), cor-relating with less than optimal dose for SG formation. When the dimension of SGs were measured with various doses of H2O2 treat- ment, we observed a dose dependent increase in the size of SGs from100 to 300 μM H2O2, with the overall size of SGs being the biggest with 300 μM H2O2 treatment (Fig. 3B).To determine how soon or time dependence of SG formation, we treated cells with 200 μM H2O2 for 0.5, 1, 2, 4, 8, 12 and 24 h, then counted the percentage of cells with SGs versus the number of SGs per cell. A time dependent increase of the percentage of cells contain-ing SGs was observed within the first 2 h, with SG first detectable at0.5 h and fewer cells containing SGs after 2 h (Fig. 2B). The number of SGs per cell showed significant increase at 0.5 h and maintained for 2 h (Fig. 2C). The trend showed a decrease in the number of SGs per cells in those cells containing SGs after 2 h. By 24 h, most SGs had disappeared based on the percentage of cells containing SGs or the number of SGs per cell (Fig. 4A‐C).Oxidants are known to activate a plethora of signaling events, and inhibitors of these signaling pathways are available for testing the involvement of PI3K/Akt, MAPKs, PKCs and HDACs. Deferoxamine (50 μM) was included as a positive control, which chelates iron and blocks formation of reactive hydroxyl radical from H2O2. Previouswork from our laboratory showed that PI3K/Akt was inhibited by Wortmannin (40 μM) and LY294002 (40 μM) (Tu et al., 2002),whereas MAPKs were inhibited by JNK inhibitor I (JNK‐1, 12 μM, for JNKs), SB202190 (30 μM for p38) and PD98059 (80 μM for ERKs) (Barr et al., 2004; Davies et al., 2000; Tu and Chen, 2003; Coronella‐ Wood et al., 2004). H89 (30 μM) is a widely used general inhibitor of protein kinase C. Tubastatin A (10 μM) is a potent inhibitor with high specificity for HDAC6 (Kappeler et al., 2012; Butler et al., 2010). Tri- chostatin A (10 μM) is a general inhibitor of HDACs and has a broadspectrum of activity reducing function of HDACs (Drummond et al., 2005).Without H2O2 treatment, some inhibitors caused an increase in the number of cells containing SGs, such as LY294002, SB202190, and H89 (Fig. 5A). The effect of H89 in inducing the fraction of cells con- taining SGs or the number of SGs per positive cell is most dramatic, over 7 fold compared to the control without any drug treatment (Fig. 5A&B). H89 caused significant inhibition of cellular metabolism (Fig. 5C). In contrast, Trichostatin A did not induce SG formation despite of inhibition of cell metabolism as measured by MTT (Fig. 5C).With H2O2 treatment, Deferoxamine showed a complete blockage of SG formation (Fig. 5A). While Tubastatin A showed about 35% inhi- bition when the percentage of cells containing SGs was counted, Tri- chostatin A completely blocked H2O2 from inducing SG formation based on the percentage of cells containing SGs (Fig. 5A). When the number of SGs per cell was counted in those cells containing SGs, the inhibitors did not significantly affect this measure, even with Deferoxamine or Trichostatin A. A downward trend in the number of SG per cell was observed with LY294002 and SB202190 for H2O2treated cells compared to the cells treated with H2O2 without inhibi- tors (Fig. 5B).The inhibitory effect of Trichostatin A on SGs was detected when cells were exposed to H2O2 for 2 h. If an inhibitor blocks cytotoxic effort of H2O2, such as with Deferoxamine, an inhibition of SG forma- tion is expected. To excluding the possibility that Trichostatin A blocks SG formation due to inhibition of H2O2 cytotoxicity, MTT assay was performed to confirm Trichostatin A did not affect metabolic inhibi- tion or cytotoxicity of high dose H2O2 (Fig. 5C). At 16 h after800 μM H2O2 treatment, although Deferoxamine prevented H2O2 frommetabolic inhibition, Trichostatin A failed to do so. These data suggest that the inhibitory effect of Trichostatin A on H2O2 induced SG forma- tion is independent of cytotoxicity inhibition. 4.Conclusion Our data indicate that formation of SGs as a result of H2O2 exposure is a dynamic process, depending on the dose as well as the time frame of stress. The optimal dose for SG induction is 400–600 μM. The time point for the highest level of SGs is 2 h of H2O2 exposure. Among the signaling pathways of oxidative stress, inhibition of HDACs appears to prevent SG formation. 5.Discussion The time frame of SG formation observed here is consistent with a live cell imaging study (Martin and Tazi, 2014). The process of SG assembly or de‐assembly appears to be ATP driven since ATP‐ dependent disaggregases, DEAD‐box helicases and autophagy proteins have been found in SGs (Jain et al., 2016). Autophagy of SGs, a process termed granulophagy, consists a mechanism of SG removal, providing one explanation for the disappearance of SGs over time (Buchan et al., 2013). In addition, cells may resume translation through converting SGs to processing bodies, i.e. P bodies, for disassembly of the granular structure and decaping RNA for translation initiation (Protter and Parker, 2016). Therefore, SGs serve to control the timing of protein translation in addition to protection of RNA and translational machinery.Not all forms of stress can induce SGs. Some of the pharmacologicalinhibitors such as LY‐294002 and Trichostatin A, while exhibited cyto- toxicity and inhibit cellular metabolism, did not induce SG formation (Fig. 5). In contrast, H89, an inhibitor of PKC, was an effective inducer of SGs for a mechanism yet to be determined (Fig. 5). While inhibition of cellular metabolism was observed with H89, H89 did not affect SG induction by H2O2, arguing against a role of PKC in H2O2 induced SGformation. Aulas et al. (2017) tested a variety of stressors and found that oxidative stress and hyperosmotic stress being most effective for inducing SGs compared to heat shock, ER stress, proteasome inhibi- tion, UV radiation or inhibition of eIF4A. While H89 can be added to the list of SG inducers, oxidants are most effective for SG induction, therefore remaining the top choice for studying the mechanism of SG formation.SG formation serves as an antagonistic mechanism against apopto- sis (Takahashi et al., 2013). Although PI3K/Akt pathway has been reported as an important pathway for cell survival; PI3K inhibitors did not affect the percentage of cells containing SGs following H2O2 treatment. However, a trend of decrease in the number of SGs per cell was observed with LY‐294002 (Fig. 5B). Since mTOR complex down- stream of PI3K/Akt inhibits the assembly of autophagosomes, inhibi-tion of PI3K/Akt results in increased autophagosome formation, pro- viding a possible explanation for the observed decrease in the number of SGs per cell with LY‐294002. A second possibility is related to the report showing mTOR‐S6 kinase promotes SG assembly due to eIF2a phosphorylation by S6 kinase 1 (Sfakianos et al., 2018). Inhibition of PI3K/Akt leads to S6 kinase 1 inhibition and therefore reduction of SG formation.The role of MAPKs in SG assembly is controversial. Consistent with our finding, JNK inhibition did not affect oxidative stress induced SGs as measured by RNA binding protein HuR as a biomarker of SGs (Meyerowitz et al., 2011;6:57.). Although p38 MAPK inhibitor did not affect the percentage cells containing SGs following H2O2 treat- ment, a downward trend was observed in the number of SGs per cell in those cells containing SGs, to the level similar to that with LY‐ 294002 (Fig. 5B). A well‐established signaling pathway for SG forma- tion is 50 ‐AMP‐activated protein kinase (AMPK) (Mahboubi et al., 2016, 2015). Since p38 activation can be a downstream event of AMPK, there is a possibility that p38 MAPK mediates the interplay between AMPK and SGs.Trichostatin A is a broad HDAC inhibitor, can inhibit HDAC6 as well as nuclear HDACs. HDAC6 has been shown to play a role in SG biology during arsenite stress or T‐cell leukemia virus infection (Kwon et al., 2007; Legros et al., 2011). Association with G3BP and deacetylation of α ‐tubulin appear to mediate SG formation(Mahboubi et al., 2016; Kwon et al., 2007). Our data from TubastatinA is consistent with a role of HDAC6 in oxidants induced SGs (Fig. 5), related to the CC-90001 fact that HDAC6 regulates microtubule assembly and fusion of microscale SGs along the microtubules to produce microscop- ically visible SGs. Compared to Tubastatin A, Trichostatin A is 300 fold more selective for class IIa nuclear HDACs than cytoplasmid HDACs (Lobera et al., 2013). Little is known about the role of class IIa HDACs in SG formation. Functionally, an inhibitory effect of Trichostatin A in H2O2 induced SGs does not translate to an increase in cell death (data not shown). Since the process of SG formation requires ATP, there is a possibility that inhibition of metabolic activity by Trichostatin A may compound its effect on HDAC inhibition for effective blockage of SG formation.