ATF3 represses PINK1 gene transcription in lung epithelial cells to control mitochondrial homeostasis

Summary PINK1 (PTEN‐induced putative kinase 1) is a key regulator of mitochondrial homeostasis that is relatively depleted in aging lungs and in lung epithelial cells from patients with idiopathic pulmonary fibrosis (IPF), a disease linked with aging. Impaired PINK1 expression and accumulation of damaged mitochondria in lung epithelial cells from fibrotic lungs were associated with the presence of ER stress. Here, we show that ATF3 (activating transcription factor 3), a member of the integrated stress response (ISR), negatively regulates transcription of the PINK1 gene. An ATF3 binding site within the human PINK1 promoter is located in the first 150 bp upstream of the transcription start site. Induction of ER stress or overexpression of ATF3 inhibited the activity of the PINK1 promoter. Importantly, overexpression of ATF3 causes accumulation of depolarized mitochondria, increased production of mitochondrial ROS, and loss of cell viability. Furthermore, conditional deletion of ATF3 in type II lung epithelial cells protects mice from bleomycin‐induced lung fibrosis. Finally, we observed that ATF3 expression increases in the lung with age and, specially, in lung epithelial cells from IPF lungs. These data provide a unique link between ATF3 and PINK1 expression suggesting that persistent stress, driven by ATF3, can dysregulate mitochondrial homeostasis by repression of PINK1 mRNA synthesis.

delineated (Mora, Bueno & Rojas, 2017;Selman & Pardo, 2014). One of the most well-accepted theories in IPF pathogenesis is the vulnerability to injury of type II alveolar epithelial cells (AECIIs) which is associated with the presence of endoplasmic reticulum (ER) stress markers (Noble, Barkauskas & Jiang, 2012) and markers of accelerated epithelial cell senescence (Minagawa et al., 2011).
Recently, we have discovered that aging lungs exposed to ER stress are highly susceptible to developing mitochondrial dysfunction, similar to IPF lungs (Bueno et al., 2015;Torres-Gonzalez et al., 2012). Hyperplasic AECIIs in IPF lungs show accumulation of swollen and dysfunctional mitochondria linked to significantly reduced levels of PINK1 (PTEN-induced putative kinase 1). In mouse lung samples, PINK1 mRNA levels were significantly diminished with aging and exposure to the ER stressor, tunicamycin (TM) (Bueno et al., 2015). PINK1-deficient mice showed higher frequency of enlarged swollen mitochondria associated with remodeling in the lung and high levels of the pro-fibrotic factor TGF-b, with a concomitant higher susceptibility to injury and fibrosis (Bueno et al., 2015). However, the mechanistic basis for the ability of ER stress to downregulate PINK1 expression is unknown.
Protein folding abnormalities induce ER stress and trigger the adaptive unfolded protein response (UPR) to halt protein synthesis to facilitate proper protein folding by increasing chaperone production. Mammalian cells present three distinct UPR signaling pathways mediated by the inositol-requiring protein 1 (IRE1), the pancreatic ER kinase (PERK), and the activating transcription factor 6 (ATF6) acting as ER proteostasis sensors. When misfolded proteins accumulate, activated PERK phosphorylates eiF2a, attenuating global translation.
Phosphorylation of eIF2a also induces preferential translation of the activating transcription factor 4 (ATF4) and subsequently activating transcription factor 3 (ATF3), both integral members of the integrated stress response (ISR) (Jiang et al., 2004). Activating transcription factor 3 is a member of the ATF/CREB (cAMP-responsive element-binding protein) family of transcription factors which share a basic leucine zipper DNA binding motif and a binding consensus sequence TGACGTCA (Hai, Wolford & Chang, 2010). Transient transfection and in vitro transcription assays indicate that ATF3 represses transcription as a homodimer. Conversely, ATF3 can activate transcription when co-expressed with its heterodimeric partners from the AP-1 family such as c-Jun and Chop10 (Chen, Wolfgang & Hai, 1996). ATF3 has a critical role in acute stress responses; however, its sustained activation can be detrimental (Hai, Wolfgang, Marsee, Allen & Sivaprasad, 1999). This study shows a new ER-mitochondria functional relationship, linking ATF3 to PINK1 expression. Our data suggest that ER stress, via ATF3, regulates mitochondrial homeostasis by repression of PINK1 gene transcription.

| ER stress decreases PINK1 expression
Previously, we have shown that in vivo treatment with the ER stressor tunicamycin (TM) reduces transcript levels of PINK1 in lungs of wild-type mice (Bueno et al., 2015). To analyze the role of ER stress in PINK1 transcription, we treated A549 cells with tunicamycin. TM treatment induced upregulation of genes involved in the unfolded protein response (UPR) such as the ER chaperone immunoglobulin-binding protein (BiP/Grp78, 15-fold to 20-fold induction), transcription factors XBP1 (fourfold to sixfold induction), CCAAT-enhancer-binding protein homologous protein (CHOP, 40-fold to 80-fold induction) (Figure S1A), and ATF3 (50-to 100-fold induction) (Figure 1a). In sharp contrast, transcript levels of PINK1 measured by qRT-PCR were significantly reduced in A549 cells exposed to increased concentrations of tunicamycin ( Figure 1b). Differences in PINK1 mRNA levels between control and TM-treated cells were eliminated in the presence of actinomycin D (2 lg/ml), an inhibitor of transcription (Figure 1c), suggesting that ER stress mediates PINK1 transcriptional repression. These changes in relative abundance of ATF3 and PINK1 can be found at the protein level ( Figure 1d, Figure S1B) and not only in A549 but also in primary human pulmonary alveolar epithelial cells (AECs). AECs exposed to a low dose of TM upregulate ER stress markers ( Figure S1C). They also recapitulate the upregulation of transcript levels of ATF3 ( Figure 1e) and reduction in PINK1 ( Figure 1f). Finally, cell stress can induce premature senescence (Pascal et al., 2005;Toussaint et al., 2002), accordingly, TM-treated AECs show increased mRNA levels of senescence markers p16, p19, and p21 ( Figure 1g). Taken together, these data indicate that tunicamycin triggers UPRs in A549 and AECs and that ER stress mediates transcriptional repression of PINK1 in epithelial cells.

| ATF3 represses the regulator of mitochondrial homeostasis PINK1
Activating transcription factor 3 is a member of the ATF/CREB family of transcription factors induced by the ER stress PERK-ATF4 pathway and, as a homodimer, can act as a transcriptional repressor (Hai & Hartman, 2001). We examined levels of PINK1 expression in A549 cells after ATF3 overexpression. Enhanced expression of ATF3 was confirmed by immunoblotting, alongside reduction of PINK1 protein levels ( Figure 2a). ATF3-driven PINK1 reduction in vitro also drives upregulation of ER stress and fibrotic markers ( Figure S2A-D) as previously shown for the PINK1-deficient AECIIs (Bueno et al., 2015). Also, it is complemented with an increase in the senescence marker p21 ( Figure S2E). To analyze whether ATF3 was required for ER stress-mediated repression of PINK1 transcription, A549 cells were ATF3-depleted and transcript levels of PINK1 were measured by qRT-PCR. Cells transfected with siATF3 showed reduced ATF3 mRNA expression before and after tunicamycin exposure (Figure 2b). Cells exposed to TM have significantly reduced PINK1 expression. Enhanced PINK1 transcript levels were observed in cells treated with siATF3 despite TM treatment ( Figure 2c). Finally, siATF3 was able to reduce ATF3 protein upregulation after 24 hr TM treatment (Figure 2d, Figure S2F). These results suggest that ATF3 is required for transcriptional repression of PINK1 after ER stress induction.
2.3 | ATF3 binds to the PINK1 promoter: characterization of the ATF3 binding sites within the PINK1 promoter Next, we evaluated interaction of endogenous ATF3 to the PINK1 promoter in A549 cells. Chromatin immunoprecipitation (ChIP) assay performed with an anti-ATF3 antibody showed positive binding of ATF3 to the PINK1 promoter. ChIP assays performed with a preimmune IgG detected no such enrichment of ATF3. Signals obtained in the input sample were used to normalize the data. ATF3 binding on the PINK1 promoter was positive in cells treated with vehicle control and significantly increased following TM (1 lg/ml) stimulation ( Figure 3a, Figure S3A). These data show that ATF3 binds to the PINK1 promoter under ER stress conditions. The human PINK1 promoter activity was monitored in A549 cells using a Gaussia luciferase reporter driven by a sequence of 1,251 bases located upstream of the transcription starting site (TSS) of PINK1. Reduced luciferase activity was observed after 24 hr exposure to TM when compared with its basal activity ( Figure 3b). The internal control reporter (SEAP, secreted alkaline phosphatase under a generic CMV promoter) activity was not significantly altered after TM treatment ( Figure S3B). PINK1 promoter activity was fully F I G U R E 1 ER stress-mediated transcriptional repression of PINK1. A549 cells show upregulation of ATF3 mRNA levels (a) after tunicamycin (TM) treatment. (b) PINK1 mRNA transcript levels are lower after TM treatment. (c) qRT-PCR assay for PINK1 transcript stability after inhibition of transcription activity by actinomycin D does not display any differences. (d) Immunoblot analysis (see Figure S1B) of ATF3 and PINK1 protein levels at different time points after TM treatment confirmed upregulation of ATF3 and decreased PINK1. Primary human AECs exposed to low concentrations of TM show upregulation of ATF3 mRNA levels (e) and reduction in PINK1 transcript (f), concomitantly with upregulation of senescence markers (g). Data represent mean AE SEM of four (a-c) and three (d-g) independent experiments. *p < .01, two-way ANOVA with multiple comparison test restored when the cells were depleted of ATF3 before TM exposure No inhibition of activity by ATF3 was observed in cells expressing the À78 to +28 PINK1 promoter sequence (pPINK-C) in line with previous studies (Duan et al., 2014) that identified this region as the minimal PINK1 promoter (Figure 3d). The slight TM-mediated inhibition of this minimal PINK1 promoter sequence is probably due to the activation of other ER stress pathways leading to global cell transcription arrest (that are not activated when only overexpressing ATF3). Altogether, these data support that a key ATF3 binding site is located in the first 150 bp upstream PINK1 transcription start site.

| Overexpression of ATF3 compromises mitochondria homeostasis
We examined whether the modulation of ATF3 expression could rescue (by silencing) or mimic (by overexpression) the accumulation of depolarized mitochondria observed with ER stress or in PINK1deficient cells (Bueno et al., 2015). To analyze mitochondria homeostasis under those conditions, ATF3-silenced A549 were treated with TM for 24 hr (Figure 4a-d), and also, A549 were independently transfected with increasing amounts of ATF3 expression plasmid (Figure 4e-h) or transfection plasmid control ( Figure S4). As we previously reported, TM treatment caused decreased in cell viability and mitochondrial membrane potential, and increase in mitochondrial mass and production of mitochondrial ROS (Bueno et al., 2015). This Transfected cells also showed dose-dependent mitochondrial accumulation ( Figure 4f) with significant depolarization (Figure 4g). In F I G U R E 2 Inactivation of ATF3 potentiates PINK1 transcription. (a) Representative immunoblot analysis of ATF3 and PINK1 in total cell lysates of A549 cells, transfected with GFP (transfection control) or ATF3. Cells overexpressing ATF3 for 48 hr show lower levels of PINK1 in whole cell lysates. A549 cells transfected with siRNA scramble control or ATF3 siRNA for a total of 48 hr and exposed to tunicamycin the last 24 hr (b-d). Less ATF3 mRNA after 24 hr TM treatment (b) and a recovery of the basal PINK1 transcript levels (c) were measured in knockdown ATF3 cells. (d) At 48 hr, protein levels of ATF3 also reflect these changes after TM treatment in the presence or absence of ATF3 silencing (see Figure S2F). Data represent mean AE SEM of four (a-c) and three (d) independent experiments. *p < .01, two-way ANOVA with multiple comparison test parallel, ATF3 overexpression in cells exhibited higher levels of mitochondrial ROS (Figure 4h). In summary, these data corroborate that upregulation of ATF3, present in persistent ER stress, can affect mitochondrial homeostasis through PINK1-reduced transcription.
Furthermore, the in vitro phenotype previously reported (Bueno et al., 2015) can be rescued by silencing ATF3.

| ATF3 is upregulated in fibrotic and aged mice lungs
To examine the role of ATF3 in the fibrotic lung, young mice F I G U R E 3 ATF3 binds to the PINK1 promoter to repress gene transcription. (a) A549 cells were treated with a low concentrations of DMSO (vehicle) or TM (1 lg/ml) for 5 hr. Chromatin immunoprecipitation (ChIP) assays on the PINK1 promoter were performed using antibodies against ATF3 and an IgG isotype control. Data represent mean AE SEM. *p < .01, two-way ANOVA with multiple comparison test.
(Negative locus, see Figure S3). (b) A549 cells transfected with a human PINK1 luciferase promoter reporter construct were treated with 10 lg/ml TM in the absence or presence of siATF3, and cotransfected with an ATF3 overexpressing plasmid. Luciferase and SEAP (secreted alkaline phosphatase) activities were measured after 24-hr stimulation. (c) Schematic diagram of the 1.2 kb PINK1 promoter region in the 5 0 flanking region upstream of the transcriptional starting site (TSS). (d) Deletion constructs of the 1.2 kb cloned PINK1 promoter luciferase reporter plasmid (pPINK-A). Arrows represent the direction of transcription and the numbers detail the endpoint of each construct. The deletion plasmids were transfected in A549 cells, and after 24 hr of TM (10 lg/ml) treatment or overexpression of ATF3 plasmid (or GFP as control), promoter activity was measured. Luciferase activities were normalized to SEAP activities. Values obtained for the untreated (or GFP cotransfected) sample of each construct represent 100%. Data are reported as mean AE SEM of four independent experiments. One-way ANOVA with multiple comparison test; *p < .05 vs. untreated, † p < .01 vs. TM 10 lg/ml (b). *p < .05 and **p < .01, unpaired, two-tailed Student's t test vs. each corresponding untreated (or GFP cotransfected) sample (d)

| Conditional and selective ATF3 deletions from mice AECIIs protect from bleomycin-induced lung fibrosis
To explore the potential role of ATF3 on epithelial injury and senescence in vivo, we generated an inducible conditional type II lung epithelial cells ATF3 knockout mice (ATF3 spc-KO) and recombination was confirmed after 2 to 4 weeks of doxycycline administration (625 mg/kg) in chow ( Figure S6). Recombination was positive in the lung AECIIs ( Figure S6A and C) and was not detected in other tissues, such as liver ( Figure S6B).   Table S1). Levels of ATF3 transcript and immunoblot analyses (Figure S10A) in total lung lysates showed that ATF3 is highly expressed in IPF lungs in comparison with young and old age-matched control donor lungs (Figure 7a-b). Using immunohistochemistry studies, we confirmed the presence of nuclear ATF3 in lung epithelial cells in aging normal lungs (Figure 7c). In addition, strong nuclear staining was found in AECII lining honeycomb areas in IPF lungs (Figure 7c).
We confirmed that in fact, in the honeycombing areas of the IPF lung, the ATF3 overexpressing lung cells are alveolar type II epithelial cells by colocalization of ATF3-and ABCA3-positive staining (Figure 7d, Figure S10B). These data suggest that high expression of ATF3 might be relevant in aging cell perturbations in lung epithelial cells and in the pathogenesis of lung fibrosis.

| DISCUSSION
Recently, we unveiled a key role for PINK1 in the maintenance of mitochondrial quality control in AECII and in activation of pulmonary pro-fibrotic responses (Bueno et al., 2015). Levels of this homeostatic protein diminish with age and are greatly reduced in IPF lungs.
Here, we demonstrated that ER stress transcriptionally inhibits of the ISR is the activation of the a-subunit of eukaryotic translation initiation factor 2 (eIF2a) that inhibits protein synthesis and simultaneously facilitates the expression of specific stress response genes such as ATF4 and ATF3. Expression of ATF3 triggers rapid induction of a large number of target genes to restore proper cellular function (Hai et al., 1999); however, the ISR has a hormetic feature and depending on the severity of stress this can lead to pro-survival or pro-apoptotic outcomes, thus compromising mitochondrial homeostasis. Our data suggest that ATF3 is a key component of the ISR in the lung epithelial cells that modulates mitochondrial stress through PINK1.
Activating transcription factor 3 expression increases with a wide spectrum of cellular stresses including ER stress, DNA damage, nutrition deprivation, hypoxia, ROS, TLR activation, TGF-b, IFN, HDL, and wounding. Thus, ATF3 is a critical adaptive response gene in different cells and organs. For instance, ATF3 is required to adjust glucose metabolism in mice exposed to high-fat diet (Zmuda et al., 2010). ATF3 also promotes cell stress-induced senescence by regulation of the Id1 transcription factor that control p16 expression (Chambers, Leoni, Kaminski, Laurent & Heller, 2003;Hara et al., 1994 (Duan et al., 2014;Mei et al., 2009;Murata et al., 2015). Interestingly, others and we have shown that PINK1 expression decreases with age (Bueno et al., 2015;Sosulski et al., 2015). In parallel, we have found ATF3 expression is higher in aging human lungs. Although further studies are necessary, it will be of interest to determine whether aging-associated mitochondrial dysfunction might be linked to age-related increase in ATF3 expression and repression of PINK1.
PTEN-induced putative kinase 1 expression has been found also to be induced by the canonical PTEN and its isoform PTENa (Liang et al., 2014). Fibroblasts and epithelial cells from IPF lungs have been found that express low levels of PTEN with persistent Akt activation (Miyoshi et al., 2013;Xia et al., 2008Xia et al., , 2010. PINK1 expression is also induced by TGFb in lung epithelial cells, and immunofluorescence studies in the IPF lung have shown positive staining for PINK1 protein potentially associated with the accumulation of damaged mitochondria (Patel et al., 2015). Nevertheless, in vivo deficiency of PINK1 results in increased susceptibility to fibrosis and TGF-b1-induced cell death (Patel et al., 2015).
In conclusion, in addition to ER stress, ATF3 is a hub of adaptive stress responses but also in age-related cell stress (Kim, Park, Rhee & Pyo, 2015). It is a versatile transcription factor in which activity and outcome are deeply dependent on the biological context of the stress, severity, and persistence. Acute upregulation of ATF3 is in most cases beneficial; however, sustained expression of ATF3 due to unresolved stress is usually linked with pathological states. The data presented here link ATF3 to PINK1 expression which in turn suggests that ER stress, via ATF3, regulates mitochondrial homeostasis by repression of PINK1 gene transcription. This connection may be critical in devising strategies designed to augment PINK1 expression or inhibit ATF3 levels to lessen severity of pulmonary fibrosis, especially in the context of the aging lung.   (Hartman et al., 2004;Wolford et al., 2013). All mice used in this study were geno-

| Quantitative PCR (qPCR)
RNA was extracted using RNeasy kit (Qiagen) according to manufacturer's protocol. Quantitative real-time reverse transcription-PCR (qRT-PCR) was then performed as previously described (Bueno et al., 2015). Data analysis was based on the 2 ÀDCT method using RNA18S as normalization control. The primers used for the real-time PCR are listed in Table S2. PCR was conducted in a QuantStudio5 real-time A549 cells were transiently transfected with each of the luciferase reporter plasmid and cotransfected with pRK-ATF3 (ATF3 overexpression) or pGFP-V-RS (GFP overexpression) or treated with tunicamycin (1 or 10 lg/ml). After 48 hours of transfection and 24 hr post-treatment, the cell culture medium was collected.
The SEAP signal was used as a transfection efficiency internal control. Luminescence intensities were detected using a Synergy Microplate Reader (BioTek). The normalized GLuc activity (GLuc/SEAP ratio) of all samples was compared.

| Hydroxyproline assay
Dried-pulverized frozen lung tissue samples (~5 mg per sample) were resuspended in 10 volume of deionized water and 10 volume of 12N HCl, and then hydrolyzed at 110°C overnight. Assay reactions were performed following the manufacturer's protocol (Sigma). A standard curve was generated using trans-4-hydroxy-L-proline (Sigma). Results were expressed as micrograms of collagen per milligram of lung tissue.

| Immunohistochemistry and immunofluorescence staining
After sacrifice, mice lungs were perfused with 2% paraformaldehyde followed by paraffin embedding. Paraffin sections were stained with Masson trichrome to determine collagen deposition. Collected human lungs samples were perfused with 2% paraformaldehyde, followed by saturation in 30% sucrose for frozen sections or paraffin embedding. Antibodies against ATF3 (HPA001562, Sigma) were used to perform the immunohistochemistry studies in paraformaldehyde fixed paraffin embedded lung tissue. Immunofluorescence analyses were performed in paraformaldehyde-fixed sucrose-saturated frozen sections of lung tissue using antibodies against ATF3 (HPA001562, Sigma) and ABCA3 (WMAB-ABCA-17, Seven Hills).

| Statistical analysis
Statistical analysis was performed using Prism 7 (GraphPad). Differences between groups were calculated by two-tailed Student's t test or by one-or two-way ANOVA followed by post hoc tests. Results are presented as mean AE SEM. p-Values less than .05 are considered significant.
4.11 | Study approval for human samples