A miR-494 dependent feedback loop regulates ER stress

Defects in stress responses are important contributors in many chronic conditions including cancer, cardiovascular disease, diabetes, and obesity-driven pathologies like non-alcoholic steatohepatitis (NASH). Specifically, endoplasmic reticulum (ER) stress is linked with these pathologies and control of ER stress can ameliorate tissue damage. MicroRNAs have a critical role in regulating diverse stress responses including ER stress. Here we show that miR-494 plays a functional role during ER stress. ER stress inducers (tunicamycin & thapsigargin) robustly increase the expression of miR-494 in vitro in an ATF6 dependent manner. Surprisingly, miR-494 pretreatment dampens the induction and magnitude of ER stress in response to tunicamycin in endothelial cells. Conversely, inhibition of miR-494 increases ER stress de novo and amplifies the effects of ER stress inducers. Using Mass Spectrometry (TMT-MS) we identified 23 proteins that are downregulated by both tunicamycin and miR-494. Among these, we found 6 transcripts which harbor a putative miR-494 binding site. We validated the anti-apoptotic gene BIRC5 (survivin) as one of the targets of miR-494 during ER stress. Finally, induction of ER stress in vivo increases miR-494 expression in the liver. Pretreatment of mice with a miR-494 plasmid via hydrodynamic injection decreased ER stress in response to tunicamycin in part by decreasing inflammatory chemokines and cytokines. In summary, our data indicates that ER stress driven miR-494 may act in a feedback inhibitory loop to dampen downstream ER stress signaling. We propose that RNA-based approaches targeting miR-494 or its targets may be attractive candidates for inhibiting ER stress dependent pathologies in human disease.


Introduction
The endoplasmic reticulum (ER) is the site of mRNA translation alongside proper folding and post-translational modifications of proteins destined for secretion or localization to various cellular membrane systems. In addition, the ER is the center of lipid biosynthesis, detoxification, homeostasis of intracellular Ca 2+ and redox balance. Several pathologies including neurodegenerative diseases, diabetes, atherosclerosis and cancers have attributed ER stress as a critical driver of disease. When the folding capacity of the ER is challenged, the accumulation of un-or mis-folded proteins in the lumen of the ER triggers the unfolded protein response (UPR) (1,2). This response activates a trio of transducers: the PKR-like ER kinase (PERK), inositol requiring enzyme 1 α (IRE1α) and activating transcription factor 6 (ATF6), which work synergistically to control transcriptional and translational programs to either alleviate the burden of unfolded proteins and return to protein homeostasis or initiate apoptosis. It is thought that acute ER stress can trigger feedback mechanisms that protect cells from death by suppressing global translation and increasing ER chaperone levels, whereas persistent UPR activation or chronic unmitigated ER stress leads to increased oxidative stress, inflammation and eventual apoptosis (3,4).
Endothelial cells (ECs) encounter a variety of stressors and stimuli during development and disease (5). Recent studies have implicated ER stress and the UPR as drivers of endothelial dysfunction in cardiovascular disease (6,7). For instance, ER stress is thought to promote both EC inflammation and apoptosis in atherosclerosis (8)(9)(10). Furthermore, ER stress has been shown to contribute to vascular dysfunction and cardiac damage in preclinical models of hypertension (11). Similarly, oxidative stress and ER stress pathways have been shown to be interlinked in ECs in metabolic syndromes such as diabetes and non-alcoholic fatty liver disease (NAFLD) (12)(13)(14)(15)(16). Therefore, ECs have adopted several mechanisms to regulate cell fate decisions in response to both acute and chronic ER stress (17).
Our previous work identified a cohort of miRs that are induced by DNA damage in ECs (24,31,32). We further characterized one of these miRs, miR-494, as a regulator of endothelial senescence in response to genotoxic stressors (24). In this study, we show that ER stress is a potent inducer of miR-494 likely via ATF6. Surprisingly, we find that miR-494 operates in a feedback loop to dampen the ER stress response potentially by targeting the anti-apoptotic protein survivin. Our in vivo experiments suggest miR-494 pretreatment diminishes tunicamycininduced ER stress in the liver. Overall, our studies illuminate a novel function for miR-494 and open new avenues for further investigations into mechanisms by which miRs modulate stress responses.

TMT-Mass Spectrometry
TMT labeling and mass spectrometry were performed by the OHSU proteomics core facility as described in detail elsewhere (35). Briefly, HUVECs were treated with microRNAs or tunicamycin as described above. After 24h-48h, samples were lysed in 50 mM triethyl ammonium bicarbonate (TEAB) buffer (50 µg of protein/ sample) followed by a Protease Max digestion, a microspin solid phase extraction and TMT labeling. Multiplexed TMT-labeled samples were separated by two-dimensional reverse-phase liquid chromatography. Tandem mass spectrometry data was collected using an Orbitrap Fusion Tribrid instrument (Thermo Scientific). RAW instrument files were processed using Proteome Discoverer (PD) (Thermo Scientific). Searches used a reversed sequence decoy strategy to control peptide false discovery and identifications were validated by Percolator software. Search results and TMT reporter ion intensities were processed with in-house scripts. Differential protein abundance was determined using the Bioconductor package edgeR.

Statistical Analysis
All statistical analysis was performed using Prism software (GraphPad Software, San Diego, CA  Results:

ER stress induces miR-494 in vitro
We previously showed that miR-494 is responsive to radiation and chemical inducers of genotoxic stress and functions to increase endothelial senescence during DNA damage responses (24). Given the intricate relationship between radiation, oxidative stress and ER stress (14,17), we asked if ER stress affected miR-494 expression and function. First, we confirmed a robust ER stress response to known inducers, tunicamycin (TCN) and thapsigargin (TG), in human umbilical vein endothelial cells (HUVECs) by measuring the level of the transcription factors spliced XBP1 (mRNA: sXBP1, protein: XBP1s) and DDIT3 (CHOP), which are well characterized markers of the ER stress response (Fig. 1A-B). We observed that TCN significantly increased the levels of mature miR-494 (Fig. 1C) and to a lesser extent, the primary unprocessed miR-494 transcript (Fig. 1D). Similarly, TG also induced sXBP1 and DDIT3 in parallel with the primary and mature forms of miR-494 ( Supplementary Fig. 1). We also saw a comparable increase in both ER stress and the primary miR-494 transcript in another EC line (Human Microvascular Endothelial cells -HMVECs) ( Supplementary Fig. 2).
A triumvirate of transducers (ATF6, IRE1α & PERK) co-ordinate ER stress response programs to promote cell recovery, or in the case of excessive or chronic stress, initiate apoptosis. We asked which of these transducers was responsible for miR-494 induction in response to ER stress using siRNAs. We found that knockdown of ATF6 but not IRE1α (ERN1) or PERK (EIF2AK3) significantly decreased the induction of miR-494 6h after TCN treatment ( Supplementary Fig. 3). These data indicate that ER stress induces the expression of miR-494 in an ATF6 dependent manner.

miR-494 diminishes ER stress
To understand the functional relevance of miR-494 during ER stress, we performed gain-and loss-of function experiments utilizing miR-494 mimics or inhibitors, respectively. Pretreatment of HUVECs with miR-494 mimic suppressed the TCN based induction of ER stress responsive genes DDIT3 (Fig. 2A). Conversely, inhibition of miR-494 robustly increased levels of DDIT3 ( Fig. 2B). Similarly, miR-494 mimic decreased levels of sXBP1 (Fig. 2C) whereas, inhibition of miR-494 increased sXBP1 mRNA both de novo and in combination with TCN (Fig. 2D). We validated that the miR-494 mediated decrease in these transcription factors also persisted at the protein level (Fig. 2E). Taken together, our data suggest that miR-494 diminishes the induction of ER stress-associated transcription factors and plays a protective role in HUVECs during ER stress.

Mass spectrometry identifies putative targets of miR-494 relevant for ER stress
miRs typically are thought to regulate numerous targets in a context dependent manner. We have previously shown that miR-494 targets the Mre11a, Rad50, Nbn (MRN) complex in the DNA damage repair pathway in response to genotoxic stressors (24). To identify the targets relevant for miR-494 in the context of ER stress, we undertook a proteomics-based approach.
We used Tandem mass tag-based mass spectrometry (TMT-MS) to compare changes in the proteome with either miR-494 mimic treatment or TCN treatment. We found 23 proteins were commonly downregulated between the TCN and miR-494 treatment conditions (Fig. 3A).
Among these, we identified six proteins whose mRNAs harbored putative miR-494 binding sites in their 3'UTRs as predicted by the TargetScan algorithm. We validated the expression of these six genes using qRT-PCR ( Fig. 3B and Supplementary Fig. 4A-D). All six targets were validated and demonstrated a substantial decrease in expression in response to miR-494 gain-of-function activity. Conversely, inhibition of miR-494 restored BIRC5, GINS4, MINA and to a lesser extent, UHRF1 mRNA levels in HUVECs treated with TCN ( Supplementary Fig 4E). Since BIRC5 and GINS4 were consistently regulated in a miR-494 dependent fashion, we chose to further validate these two targets at the level of protein expression. We found survivin (gene: BIRC5), and GINS4 protein levels were significantly decreased upon treatment with miR-494 mimic alone, TCN treatment alone, and the sequential combination of miR pre-treatment (24h) followed by TCN (24h) (Fig. 3C). Finally, we used immunofluorescence staining and confocal microscopy to evaluate survivin expression in HUVECs. Consistent with our western blot data, we observed a significant decrease in the amount of survivin in cells transfected with either miR-494 mimic or TCN treatment alone or in the sequential combination (Fig. 3D). Multiple studies have demonstrated that miR-494 binds to the 3'UTR and leads to subsequent degradation of the survivin (BIRC5) transcript (36)(37)(38). Our observations support these data and further demonstrates this relationship in ECs during ER stress.

miR-494 pretreatment ameliorates ER stress in mouse liver
ER stress is thought to be a significant driver of liver diseases including hepatic fibrosis, nonalcoholic fatty liver disease (NAFLD) that can progress to steatohepatitis (NASH) and hepatocellular carcinoma (3,39,40). Endothelial dysfunction plays a critical role in the pathology of NAFLD and NASH (41). Indeed, liver sinusoidal endothelial cells (LSECs) can function to provoke inflammation and fibrosis in NAFLD (16). To delve further into the function of miR-494 during ER stress in vivo, we utilized a well characterized murine model of NASH (42). It has been shown that injection of TCN (2µg/g) increases acute ER stress in the liver with upregulation of CHOP and spliced XBP1 (42). We observed that TCN treatment induced endogenous miR-494 expression in mouse livers (Fig. 4A). To address the relationship between miR-494 and ER stress, we used hydrodynamic injection of plasmid DNA to transiently increase miR-494 levels in the liver. Compared to the GFP control plasmid, miR-494 plasmid injection significantly increased the levels of miR-494 (Fig. 4B). Importantly, when followed by TCN treatment 24h later, this increase in miR-494 levels was sufficient to decrease ER stress in the livers as assessed by DDIT3 levels (Fig. 4C).
ER stress induction by TCN treatment has been shown to induce inflammation in the liver, which is a hallmark of NASH (3,7). To explore the mechanisms by which miR-494 may impact the activity of TCN in vivo, we utilized a Nanostring ® IO360 mouse profiling array to evaluate the immune response and inflammation in these animals. Compared to the plasmid control, miR-494 treated mice exhibited a significant decrease in several immune response and inflammatory genes in the liver (Supplementary Fig. 5A). Gene Ontology analysis identified cytokine response, immune cell migration and immune cell adhesion to vascular endothelium as the most significantly downregulated pathways in the miR-494 treatment group (Supplementary Fig.   5B).

Discussion:
Here  Further, TCN consistently generated a rapid and robust upregulation of miR-494 in multiple EC cell lines ( Fig. 1 and Supplementary Fig. 2). TG, a non-competitive inhibitor of the Sarco/Endoplasmic Reticulum Calcium ATPase (SERCA) disrupts calcium homeostasis in the ER. In our experiments, TG treated ECs experienced similar signaling patterns, with an increase in both spliced XBP1, CHOP transcription and induction of miR-494 ( Supplementary Fig 1).
Interestingly, TCN but not TG induced a biphasic expression of mature miR-494 with a slight increase at 1h and a more significant (~30 fold) increase at 6h. Moreover, primary miR-494 transcript expression mirrored that of the mature transcript in TG treatment but not TCN (Fig. 1 & Supplementary Fig.1). It is unclear if these differences reflect differential regulation of miR processing pathways by these two inducers. Indeed, the canonical miR processing protein Dicer has been shown to localize and interact with proteins in the ER (45). Therefore, it is possible that ER stress pathways can differentially impact miR transcription and processing and it will be pertinent to investigate the impact of ER stress in the setting of EC-associated induction of miR-494.

Our data from gain-and loss-of-function experiments show that miR-494 decreases ER stress in
ECs and in mouse liver (Fig. 2, 4). These results bolster other reports which also demonstrate miR-mediated responses to ER stress in different tissues (26,(46)(47)(48)(49). While miRs have a recognized role in liver disease (50), our studies are the first to show the ER stress reducing activity of miR-494 in the context of liver disease (NASH). miR-494 is known to be oncogenic and drives tumor progression and drug resistance in specific cancer types including colorectal cancer (51) and hepatocellular carcinoma (52,53). Furthermore, miR-494 has been shown to attenuate ischemia reperfusion injury in the liver by regulating the PI3K/Akt signaling pathway (54). We previously reported a unique role for miR-494 as a mediator of endothelial senescence by decreasing the MRN DNA repair protein complex. Other studies have shown that miR-494 is involved in vascular inflammation in atherosclerosis (55). Our data indicates a novel function of miR-494 in a complex pathological process.
Using a proteomics approach, we identified six putative miR-494 targets that are relevant during the ER stress response in ECs. Of these, survivin (BIRC5) has been previously shown to be a bona fide target of miR-494 in different disease models (37,38). Interestingly, the cross-talk between ER stress and survivin has been shown to downregulate inflammatory genes in a mouse model of chronic ER stress in the colon (56). Consistent with our observations of a miRmediated reduction in ER stress combined with a decrease in survivin (BIRC5) and markers of inflammation, Gundamaraju et al., demonstrated that pharmacological inhibition of survivin is comparable to ER stress inhibition and attenuates inflammatory gene expression (56).
We used a Nanostring Mouse PanCancer™ profiling array to query the in vivo response to miR-494 pre-treatment prior to induction of ER stress. Our results reveal that a number of inflammatory genes are downregulated in the liver with miR-494 plasmid pre-treatment. Very few of the array-identified genes harbor miR-494 binding sites and therefore, the mechanism(s) by which miR-494 diminishes ER stress and subsequent inflammation is unclear. Further mechanistic complexity arises from recent reports which indicate miR-494 can be found in exosomes from patients with cirrhosis (57,58). Pertinent to this body of work, the relative contributions of miR-494 from hepatocytes in comparison to liver ECs requires further investigation. In this setting, it is conceivable that exosomal miR-494 from sinusoidal or other ECs in the liver can orchestrate inflammation and mediate paracrine effects during ER stress.
In summary, our work shows that miR-494 is induced during acute ER stress and functions to attenuate ER stress in vitro and in vivo. Our observations elucidate a new potential mechanistic role for miR-494 in the ER stress response pathway in ECs and likely in other cell types. These studies offer opportunities to inspire new hypotheses to understand the link between miR activity and the response to stressors which influence cell fate decisions in many human diseases.