Prolonged Pseudohypoxia Targets Ambra1 mRNA to P-Bodies for Translational Repression

Hypoxia has been associated with several pathological conditions ranging from stroke to cancer. This condition results in the activation of autophagy, a cyto-protective response involving the formation of double-membraned structures, the autophagosomes, in the cytoplasm. In this study, we investigated the cellular mechanisms regulating the autophagy gene Ambra1, after exposure to a hypoxia mimetic, cobalt chloride (CoCl2). We observed that, upon CoCl2 administration, activation of the apoptotic machinery was concomitant with down-regulation of the pro-autophagic factor Ambra1, without affecting transcription. Additionally, co-treating the cells with the caspase inhibitor z-VAD-FMK did not restore Ambra1 protein levels, this implying the involvement of other regulatory mechanisms. Partial re-localization of Ambra1 mRNA to non-translating fractions and cytoplasmic P-bodies was further detected. Thus, in this pseudohypoxic context, Ambra1 mRNA translocation to P-bodies and translational suppression correlated with increased cell death.


Introduction
Multicellular organisms have developed oxygen-sensing systems to maintain oxygen homeostasis, which is essential for survival of the organism [1]. In fact, hypoxia has been detected in many pathological conditions such as coronary heart disease, stroke, and organ transplant injection [2]. Hif-1α (Hypoxia-inducible factor alpha) is a transcription factor that is rapidly activated in response to decreased levels of O 2 ; this switches on, in turn, a subset of genes ensuring cell survival in hypoxic conditions [3]. Among the genes controlled by Hif-1α are VEGF (Vascular Endothelial Growth Factor) and EPO (Erythropoietin), both involved in erythropoiesis and angiogenesis to increase oxygen delivery to the hypoxic site [4,5]. Hypoxia also generally activates the cyto-protective response of autophagy in a Hif-1α dependent manner [3].
Autophagy is a highly conserved process of self-digestion through the lysosomal pathway. It involves the delivery of cytoplasmic components and organelles to the lysosomes through specific vesicles, termed autophagosomes [6]. Autophagy and the molecular components involved in this process have been widely studied. Ambra1 (Activating Molecule in Beclin 1-Regulated Autophagy) is a key molecule in autophagy upstream regulation and plays a role in neurodevelopment [7][8][9]. It encodes a 1,300 amino acid long protein that is highly conserved among vertebrates. Functional inactivation of the gene results in embryonic lethality due to severe neural tube defects [7].
Considering the involvement of defective autophagy in various pathological conditions, such as cancer and neurodegenerative disorders, several mechanisms regulating this process have been intensively investigated [10]. The most studied mode of regulation occurs through protein-protein interactions and post-translational modifications; However, new regulatory mechanisms are emerging, such as mRNA localization and processing through the miRNA machinery [11,12].
Once a stressor affects the cell, this should be able to respond rapidly by changing its proteome/transcriptome [13,14]. One mechanism suggested to this aim is mRNA localization. When one mRNA product is not required by the cell, it exits polysomes and assembles in an mRNP complex lacking translation initiation factor. This multi-protein RNA complex accumulates in processing bodies (P-bodies) to then proceed with decapping followed by degradation [15]. P-bodies are cytoplasmic structures composed of several core proteins (Edc3, decapping enzymes Dcp1/2, exonuclease Xrn1, etc.), some additional factors (such as GW182 and Ago1-4, etc.) and various mRNAs [16]. P-bodies also act as a storage site for those mRNAs that are not needed for a certain period of time. These silent intact mRNAs can exit P-bodies upon various stimuli to enter the translation machinery [15].
Cobalt Chloride (CoCl 2 ) is a hypoxia mimetic which increases Reactive oxygen species (ROS) production and stabilizes HIF-1α protein through inhibition of prolyl hydroxylases [17], a process also known as pseudohypoxia. In this study we observed that prolonged exposure to CoCl 2 resulted in a block of autophgy flux and induction of apoptosis, accompanied by gradual decrease of Ambra1 protein levels and re-localization of its transcripts to non-translating fractions and cytoplasmic P-bodies.

RNA extraction and reverse transcription
RNA was extracted from HeLa cells using QIAGEN mini RNeasy kit following the manufacturer's instruction. Two μg of extracted RNA was used for reverse transcription and 500 ng of random primers for each μg of RNA was added to each tube. Samples were heated to 70°C for 5 min followed by immediate incubation on ice. M-MLV reaction buffer (1X), dNTPs, RNase inhibitor (1U/μl), and M-MLV reverse transcriptase (8U/reaction) (Promega) were added to the mix and were incubated for one hour at 37°C. Real-time PCR was performed using ABI lightcycler. 0.5 μg cDNA were used with 10 μl SYBR Green Master Mix (ABI), and 200 nM of each primer were

Cell cultures and Chemical Treatments
HeLa and HEK-293 cells were grown in Dulbecco's modified Eagle's medium with 10% fetal calf serum (FCS), 100 units/ml penicillin, 100 mg/ml streptomycin, and 2 mM glutamine, in a 5% CO 2 atmosphere at 37°C. For CoCl 2 treatment, cells were incubated with 500 μM CoCl 2 from 0 to 24 hours. For z-VAD-FMK (10 μM) and MG132 (5 μM) treatments cells were incubated for the indicated time-points. For autophagy flux evaluation, chloroquine was added for the last hour of the CoCl 2 treatment in the medium at a concentration of 20 μM. For Actinomycin D treatment, the drug was added for the last 4 hours of the CoCl 2 treatment at a final concentration of 10 μM. For NAC treatment, this was added for the last 4 hours of CoCl 2 treatment at a final concentration of 1 and 5 mM, respectively.

ROS evaluation
To evaluate ROS relative concentration upon CoCl 2 treatment, 30 min before the end of the experimental procedure, cells were incubated with 5μM DHE (Life Technologies) at 37°C. Cells were then washed twice in ice-cold PBS and collected. The fluorescence intensity of DHE, enhanced by the reaction with ROS, was analyzed by recording FL-2 fluorescence on a FACS-Verse (BD Bioscences) flow cytometer.

RNA-FISH
QuantiGene View RNA ISH cell assay kit was obtained from Affymetrix. The experiment was conducted as instructed. In brief, HeLa cells were grown on poly-lysine coated glass slides, 96-well plates, or 384-well plates to 70% confluence. Cells were washed with PBS and fixed in 4% formaldehyde for 30 min. After washing with PBS 1X, they were permeablized with Detergent Solution for 5 min followed by protease digestion (1:8000) for 10 min. Working Probe Sets were added and the slides were incubated at 40°C for 3 hours. Consequently, they were incubated with pre-Amplifier, Amplifier, and Label Probe Mix, each step for 30 min at 40°C. Finally, the slides were counterstained with DAPI and the images were acquired by using a confocal microscope (see below). RNA probes were designed by Affymetrix against human Ambra1 and human Beclin 1.

Immunocytochemistry
Cells were washed with PBS 1X and fixed with 4% paraformaldehyde for 15 min at room temperature. After several washes with PBS they were permeabilized with 0.5% Triton X-100 in PBS for 5 min before immunostaining. After incubation with 5% BSA for one hour at room temperature, the cells were incubated over night with anti-Ge-1 (1:400) primary antibodies. Detection of the primary antibodies was performed using 1:400 Alexa Fluor 488 goat anti-rabbit IgG and Alexa Fluor 555 goat anti-mouse IgG secondary antibodies for 1 hour at room temperature. For nuclei detection, cells were counterstained with DAPI. Confocal images were acquired with Confocal laser scanning microscope Olympus FV 1000 using laser 405 nm (diode) for blue channel (DAPI), 488nm (argon) for green channel (FITC) and 543nm (HeNe) for red channel (TRITC). Oil objective 60x (NA 1,35) with optical zoom 3x were used. Fluorochrome unmixing was performed by automated-sequential collection of multi-channel images to reduce spectral cross-talk between channels. Ambra dots colocalization with GE-1 has been done using ImageJ software; For control and pseudohypoxia conditions 26 (5 fields) and 20 cells (4 fields) have been counted, respectively. For apoptotic nuclei analysis, DAPI was added in the medium of living cells; after 10 minutes of incubation, 10 random images per condition were taken and apoptotic and normal nuclei were counted.

Linear sucrose gradient fractionation
Polysome profile analysis was performed using linear sucrose gradient fractionation (LGS). First, cells were incubated with 100 μg/ml cycloheximide (Sigma) for 15 minutes at 37°C. Then, they were washed with PBS containing 100 μg/ml cycloheximide and lysed with 500 μl LGS buffer (100 mM KCl, 20 mM Tris, pH 7.5, 5 mM MgCl 2 , 0.4% NP-40, 100 μg/ml cycloheximide, 0.1 U RNase inhibitor (Promega) and complete EDTA-free protease inhibitor). Cell lysate were centrifuged at 14,000 rpm for 15 min at 4°C and supernatant was recovered. The supernatant was loaded onto 15-50% sucrose gradient. Ultra-centrifugation was performed for 110 min at 37,000 rpm at 4°C in a Beckman SW41 rotor. Each gradient was collected as 10 fractions (1 ml each) in a tube containing SDS with the final concentration of 1% and RNA-spike (20 pg/fraction). Fractions were monitored by continuous OD 254 measurement; 100 μg/ml Proteinase K was added to the samples followed by 1 hour incubation at 37°C. Total RNA was extracted from each fraction using phenol-chloroform and isopropanol precipitation.

Results
To study the regulatory mechanisms of hypoxia-induced autophagy, HeLa cells were treated with 500μM CoCl 2 for 24 hours, a condition termed 'pseudohypoxia'. Induction of pseudohypoxia was confirmed by detecting in a time-course Hif1-α protein levels (Fig 1A), which were increased as early as 2-6h after treatment with CoCl 2 . To further verify hypoxia induction we performed a Real-Time PCR on p62 and Nqo1 mRNAs, two targets of the transcription factor Nfr2, known to be activated during hypoxia; Indeed, both after 8 and 18 hours of CoCl 2 treatment, the two mRNAs are significantly upregulated (Fig 1B). Autophagy was then monitored by western blot analysis of two common markers, p62 and LC3. As shown in Fig 2A, p62 protein levels decreased with time, accompanied by more conversion of LC3I to LC3II. In order to understand whether the accumulation of LC3II was due to an activation of autophagy or rather to a block of this process, which would impair its degradation, we analyzed the autophagy onrate/off-rate (autophagy flux) by using chloroquine, an inhibitor of the autophagosome/lysosome fusion. We found that 8 hours after treatment with CoCl 2 there is no production of new autophagosomes; instead, the autophagy flux is blocked, leading to LC3II accumulation ( Fig 2B). To further corroborate this data, we next analysed LC3 dots by immunofluoresce in cells treated with CoCl 2 , both in basal and autophagy-blocked conditions; as shown in Fig 2C, CoCl 2 inhibits LC3 dots formation after chloroquine, this confirming a block of the autophagy flux (Fig 2C and S1 Fig). Of note, an expected ROS increase due to pseudohypoxic conditions, is induced only 16 hours after CoCl 2 administration, this indicating that ROS are not responsible of the observed effects on autophagy, whose block is detectable as soon as 8 hours after treatment (Fig 3A). In agreement with this conclusion, treatement with the ROS scavenger NAC has modest if any effects on LC3II accumulation (Fig 3B).
Since several studies show the activation of apoptosis following CoCl 2 treatment [18], this process was also analyzed in treated cells. As shown in Fig 4A, at later time points, cleavage of caspase 3 (Casp3) indicates apoptosis induction. As a consequence of caspase activation, after 24 hours of CoCl 2 treatment, we observe cleavage of PARP [Poly (ADP-ribose) polymerase], a  hallmark of apoptosis (Fig 4B). In line with this finding, a count of apoptotic nuclei revealed an increased cell death following the treatement, confirming that, indeed, CoCl 2 treatment leads to apoptosis (Fig 4C).
To investigate the regulation of autophagy genes upon hypoxia, the protein levels of Ambra1 and Beclin 1, two upstream proautophagic proteins, were analyzed by western blotting (WB; Fig 5A and 5B). A gradual decrease of Ambra1 and Beclin 1 protein levels was observed upon CoCl 2 treatment. This finding is inversely related with what shown in Fig 2A, where we show an increase of LC3II protein levels starting at the 2h time-point and a continued increase until the final measurement at 24h. Next, we analyzed the Ambra1 mRNA levels to verify whether these changes were due to transcriptional or translational regulation. Quantitative real-time PCR was performed using primers against Ambra1 and Tubulin α-6 (Tuba6) (as a control). Interestingly, no significant changes in Ambra1 mRNA levels were observed at 18h,   Ambra1 Repression in Pseudohypoxia the time-point at which protein levels decreased (Fig 6A). In order to understand if Ambra1 mRNA was somehow stabilized or instead transcribed and then degraded in a continuous loop, we performed an experiment using Actinomycin D (ActD), a drug inhibiting mRNA transcription. In the control sample, a treatment of 4 hours of ActD is sufficient to observe a significant decrease of Ambra1 mRNA; However, on cells treated for 8 hours with CoCl 2 , ActD does not lead to a significant Ambra1 mRNA decrease, this indicating that upon CoCl 2 treatment there is a block of Ambra1 mRNA production and degradation (Fig 6B). A recent study has shown that long-term stress, such as starvation or staurosporin treatment, leads to Ambra1 protein degradation through caspase cleavage, this inhibiting the prosurvival role of autophagy [19]. To exclude the involvement of this event on the decrease of Ambra1 protein during prolonged pseudohypoxia, cells were treated with the pan-caspase inhibitor z-VAD-FMK along with CoCl 2 for 18h. Interestingly, as shown in Fig 7A, inhibition of caspases had little if any effects on Ambra1 protein levels, implicating that other mechanism(s) must be responsible for Ambra1 protein decrease. Thus, we checked if the proteasome could account for Ambra1 decrease, by adding to the cells the proteasome inhibitor MG132 together with CoCl 2 ; after 18 hours we observed that Ambra1 protein levels are still low when the proteasome is inhibited, this suggesting that the proteasome system is not responsible for AMBRA1 decrease (Fig 7B).
In principle, an apparent discrepancy between protein and mRNA levels implies the involvement of a translational control in this phenomenon. Therefore, we investigated the localization of Ambra1 mRNA in normal and stress conditions. To check whether the transcription/ translation rate of Ambra1 mRNA was impacted by its translational control, Ambra1 polysome profile was analyzed in both control conditions and 18h after induction of pseudohypoxia in HeLa cells. Ten fractions were obtained from each sample with continuous OD 254nm measurement. As shown in Fig 8A, pseudohypoxia resulted in a general decrease in translation. RNA was extracted from each fraction and the RNAs obtained from 'translating' fractions (#1-5) were pooled together. The same was done for 'non-translating' fractions (#6-10), followed by reverse transcription. Q-PCR showed a significant decrease in the amount of translating Ambra1 mRNA after CoCl 2 treatment, whilst control mRNA (tuba6) was unchanged in control and pseudohypoxic conditions (Fig 8B).
Non-translating mRNAs are often stored in cytoplasmic P-bodies. P-bodies are cytoplasmic structures that are involved in storage and processing of mRNAs [20]. They also play a role in translational regulation in different conditions [3]. We thus analyzed the localization of Ambra1 mRNA in the cytoplasm before and after exposure to pseudohypoxic conditions (Fig 9). P-bodies were detected using an antibody against Ge-1 (a marker for this compartment) and appeared as fewer but much larger spots after CoCl 2 treatment (Fig 9C and 9G). To analyze the localization of Ambra1 mRNA relative to P-bodies, RNA-FISH was followed by immunostaining of Ge-1 (Fig 9B and 9F). Interestingly, while there is no co-localization at all between Ambra1 mRNA and P-bodies in control conditions, pseudohypoxia induced by CoCl 2 resulted in a partial re-localization of Ambra1 mRNA to these cytoplasmic structures (see arrows in Fig 9D and 9H, Figure

Discussion
Here we propose the existence of a translational regulation of Ambra1, a pro-autophagy molecule, upon pseudohypoxia induction. We observed that Ambra1 protein levels decreased upon prolonged exposure to the hypoxia mimetic CoCl 2 , without any significant changes in mRNA levels. Along with data previously reported, we observed induction of pseudohypoxia in our cell system upon treatment with CoCl 2 , as confirmed by stabilization of Hif1-α protein.
Although it is widely accepted that Hif1-α activation leads to autophagy induction, after 8 hours of treatment with CoCl 2 we found that LC3II accumulation is due to a block rather than to an induction of the autophagy flux, as revealed using chloroquine in western blot and in immunofluorescence experiments. This can be explained by the fact that, after a stress, autophagy is increased in order to preserve the cell, even though-if the stimulus persists for a long   Ambra1 Repression in Pseudohypoxia period, autophagy is blocked. Therefore, we can speculate that after 8 hours of pseudohypoxia, autophagy inhibition is due to the long exposure to CoCl 2 . On the contrary, p62 is slightly decreased after treatment, even if an increase in its transcription is observed, this being due, most likely, to hypoxia-mediated Nrf2 activation. This suggests a block in its translation or its protein degradation by caspases or the proteasome; alternatively, when autophagy is inhibited, p62 can forms aggregates that are insoluble in lysis buffer containing Triton. In addition, we show here that ROS are activated only after 16 hours of CoCl 2 treatment, indicating that probably they are not early induced by CoCl 2 . Most likely, since an impaired degradation of damaged mitocondria leads to ROS production, we can speculate that the block of autophagy due to prolonged pseudohypoxia causes indirect ROS induction that, in turn, leads to apoptosis at later time points.
Looking at autophagy genes, we detected a gradual decrease in Ambra1 and Beclin 1 protein levels. We then proceeded with the analysis of Ambra1, a more upstream regulator of autophagy. Since Ambra1 mRNA levels remain unchanged, the first hypothesis to test was degradation of the Ambra1 protein. A recent study by Piacentini and colleagues demonstrated that treating different cell lines such as HeLa and 2FGTH with Staurosporine, a DNA-damage inducer, resulted in a gradual decrease of Ambra1 protein over time, without affecting the mRNA levels [19]. Interestingly, this decrease happened concomitantly with an increase in the PARP cleaved form, a marker of apoptosis induction. They further observed that Ambra1 was a target of caspases, since treating the cells with the caspase inhibitor z-VAD-FMK together with staurosporine could partially restore the level of Ambra1. Ambra1 degradation occurs when the stress is too intense or prolonged and autophagy cannot rescue the cell; therefore, by degradation of Ambra1, the pro-survival machinery of autophagy switches to apoptosis [19]. Although the direct interaction of Ambra1 with caspases has been reported, and inhibition of caspases prevents Ambra1 degradation in overexpressing conditions, it does not completely restore the endogenous protein levels [19]. This suggests the involvement of other mechanisms in down-regulation of Ambra1.
In this study, we have observed that long-term exposure to pseudohypoxic conditions results in the activation of apoptosis along with decreased Ambra1 protein levels, whilst mRNA level remains unchanged. Also, applying caspase inhibitor to the cells does not restore Ambra1 protein levels. We also showed that Ambra1 degradation is not due to the ubiqutine-proteasome system; in addition, we showed that MG132 together with CoCl 2 leads to an almost complete depletion of Ambra1, suggesting that probably a protein involved in Ambra1 decrease is stabilized by blocking the proteasome-mediated degradation. Moreover, our experiments with ActD indicate that Ambra1 mRNA transcription and turnover are blocked after the treatment with CoCl 2 .
Our data implies the involvement of a translational control in this phenomenon. To test this hypothesis, polysome profile analysis was performed and we observed translocation of Ambra1 mRNA from polysomes (translating fraction) to non-translating fraction upon pseudohypoxia, resulting in the decrease of Ambra1 protein levels. Since un-translating mRNAs are often stored in cytoplasmic P-bodies, we analyzed the localization of Ambra1 mRNA in the cytoplasm before and after exposure to hypoxic conditions. Even though Ambra1 mRNA is stable in stress conditions, our data suggest that, at least in part, it is sequestered and stabilized in Pbodies, thus becoming inaccessible to the protein translation machinery. This sequestration may then result in a decrease in protein levels, in time with activation of cell death and suppression of the pro-survival pathway of autophagy. By contrast, Beclin 1 mRNA does not relocate to P-bodies at all in similar conditions, indicating that the mechanism here described does not concern all autophagy regulators. Although in general the decay of mRNAs is a consequence of prolonged stress conditions that lead to apoptosis, it is plausible that the maintenance of a pool of Ambra1 mRNA may represent an important reserve for the cell in case normal conditions are restored. Indeed, this would allow Ambra1 prompt translation.
As for the mechanism by which Ambra1 mRNA moves to P-bodies, it certainly needs further investigation. There is evidence for interaction of P-bodies with actin filaments and/or microtubule structures [21,22]. Also, association of certain RNAs with the dynein motor complex for dendritic localization has been postulated [23]. Therefore, the next step would be studying the effect of microtubule disruption on localization of Ambra1 mRNA with or without CoCl 2 treatment and its effect on cell death and survival in hypoxic conditions. Furthermore, since the co-localization of Ambra1 mRNA and P-bodies after CoCl 2 treatment is only partial, it would be interesting to know if this messenger can also localize in other specific foci, such as stress granules.