Sestrin 2 Protein Regulates Platelet-derived Growth Factor Receptor β (Pdgfrβ) Expression by Modulating Proteasomal and Nrf2 Transcription Factor Functions*

Background: Sestrin 2 is a redox-dependent repressor of Pdgfrβ signaling and thereby interferes with lung injury repair. Results: Sestrin 2 is a positive regulator of proteasomal function and activates transcription of Nrf2-regulated antioxidant genes. Conclusion: Sestrin 2 is a component of a novel Sestrin 2/Pdgfrβ repressor pathway. Significance: The Sestrin2/Pdgfrβ repressor pathway is likely to play role in the pathogenesis of chronic obstructive pulmonary disease (COPD). We recently identified the antioxidant protein Sestrin 2 (Sesn2) as a suppressor of platelet-derived growth factor receptor β (Pdgfrβ) signaling and Pdgfrβ signaling as an inducer of lung regeneration and injury repair. Here, we identified Sesn2 and the antioxidant gene inducer nuclear factor erythroid 2-related factor 2 (Nrf2) as positive regulators of proteasomal function. Inactivation of Sesn2 or Nrf2 induced reactive oxygen species-mediated proteasomal inhibition and Pdgfrβ accumulation. Using bacterial artificial chromosome (BAC) transgenic HeLa and mouse embryonic stem cells stably expressing enhanced green fluorescent protein-tagged Sesn2 at nearly endogenous levels, we also showed that Sesn2 physically interacts with 2-Cys peroxiredoxins and Nrf2 albeit under different reductive conditions. Overall, we characterized a novel, redox-sensitive Sesn2/Pdgfrβ suppressor pathway that negatively interferes with lung regeneration and is up-regulated in the emphysematous lungs of patients with chronic obstructive pulmonary disease (COPD).

We recently identified the antioxidant protein Sestrin 2 (Sesn2) as a suppressor of platelet-derived growth factor receptor ␤ (Pdgfr␤) signaling and Pdgfr␤ signaling as an inducer of lung regeneration and injury repair. Here, we identified Sesn2 and the antioxidant gene inducer nuclear factor erythroid 2-related factor 2 (Nrf2) as positive regulators of proteasomal function. Inactivation of Sesn2 or Nrf2 induced reactive oxygen species-mediated proteasomal inhibition and Pdgfr␤ accumulation. Using bacterial artificial chromosome (BAC) transgenic HeLa and mouse embryonic stem cells stably expressing enhanced green fluorescent protein-tagged Sesn2 at nearly endogenous levels, we also showed that Sesn2 physically interacts with 2-Cys peroxiredoxins and Nrf2 albeit under different reductive conditions. Overall, we characterized a novel, redox-sensitive Sesn2/ Pdgfr␤ suppressor pathway that negatively interferes with lung regeneration and is up-regulated in the emphysematous lungs of patients with chronic obstructive pulmonary disease (COPD).
Sestrin 2 belongs to a family of highly conserved, stress-inducible proteins implicated in the regulation of metabolic homeostasis. Mammalian cells express three sestrin isoforms: sestrin 1 (Sesn1 2 ; also known as PA26), sestrin 2 (Sesn2; also known as Hi95), and sestrin 3 (Sesn3) (1,2); Sesn1 and Sesn2 are involved in the defense against reactive oxygen species (ROS). The antioxidant function of the sestrins was originally attributed to their sulfinic acid reductase activity required for regenerating over-oxidized 2-Cys-peroxiredoxins (Prxs) (3). Prxs are highly conserved and ubiquitously expressed antioxidant proteins that are inactivated by hyperoxidation of their peroxidatic cysteine, thus yielding sulfinic acid in the presence of high peroxide concentrations (4,5). However, it was later shown that the primary Prx reactivating enzyme is sulfiredoxin, which unlike the typical cellular reductants such as glutathione or thioredoxin can reduce sulfinic acid to thiol (6,7). Therefore, the ability of Sesn1 and -2 to catalyze sulfinic to sulfenic acid is still a matter of debate.
Antioxidant activities of Sesn1 and -2 were recently attributed to their activation of the Nrf2 (nuclear factor erythroid 2-related factor 2)-Keap1 (Kelch-like ECH-associated protein 1) pathway (7). Nrf2 is a global antioxidant gene inducer whose activity is tightly controlled by cytoplasmic association with its inhibitor, Keap1. Keap1 targets Nrf2 for polyubiquitination and degradation by the 26 S proteasome, thereby keeping its basal levels low. Oxidative stress disrupts the Nrf2-Keap1 protein complex, and Nrf2 translocates into the nucleus where it heterodimerizes with its small Maf binding partner to transactivate antioxidant gene expression by binding the antioxidant response elements typically present in promoter regions of * This work was supported by the SFB815 "Redox Signaling" of the Deutsche antioxidant genes (8). Sesn2 interfered with this pathway by stimulating the autophagic degradation of Keap1 (7).
Recently, we and others identified Sesn2 as a repressor of Pdgfr␤ signaling (9,10) and Pdgfr␤ signaling as an inducer of lung tissue regeneration and injury repair (9). In mice, mutational inactivation of Sesn2 prevents development of cigarette smoke-induced pulmonary emphysema by up-regulating Pdgfr␤-controlled alveolar maintenance programs (9). Although we showed that this up-regulation is mediated by ROS (9), the molecular mechanism(s) of this regulation was not determined.
The proteasome is the major site for removal of shortlived, unfolded, and damaged proteins. With a molecular weight of 150 MDa, the 26 S proteasome is one of the largest known catalytic complexes inside all eukaryotic cells. Composed of the catalytically active 20 S core and 19 S regulatory particles, the 26 S proteasome governs ubiquitin-and ATPdependent protein degradation (11,12). Three subunits of the 19 S proteasome (Rpt5, Rpn10, and Rpn13) bind polyubiquitinated substrates that earmark them for degradation (13)(14)(15)(16). The catalytic centers are located inside the 20 S proteasome where three of its seven ␤-subunits show proteolytic activity, namely ␤1, ␤2, and ␤5, exhibiting caspase-, trypsin-, and chymotrypsin-like activities, respectively (17,18). Oxidative stress leads to proteasome disassembly into its 19 S and 20 S subunits and to the accumulation of polyubiquitinated proteins (19).
Here we identify Sesn2 as a positive regulator of proteasomal function and the proteasome as a major component of a redoxsensitive signal transduction pathway controlling Pdgfr␤ signaling. Within this pathway, we show that Nrf2 provides a crucial redox switch that, when activated by Sesn2, down-regulates Pdgfr␤ by enhancing its proteasomal degradation. Finally, we show that this Sesn2/Pdgfr␤ suppressor pathway is up-regulated in the emphysematous lungs of individuals with advanced chronic obstructive pulmonary disease (COPD), where it likely interferes with lung regeneration.
Cell Cultures and Cell Transductions-MLFs and HeLa, MRC5, and HEK293T cells were grown in Dulbecco's modified Eagle's medium (Life Technologies) supplemented with 10% (v/v) fetal calf serum (Life Technologies), 100 units/ml penicillin, and 100 g/ml streptomycin (Life Technologies). HTBH-tagged hRpn11 expressing HEK293T cells (provided by L. Huang, University of California, Irvine) were used for affinity purification of human proteasomes. nLAP-Sesn2, Prx1, and Prx2 HeLa cell lines obtained from the BAC Interactomics resource (Hyman Lab, Max Planck Institute of Molecular Cell Biology and Genetics, Dresden, Germany) were grown in the presence of 400 g/ml G418 (Life Technologies). nLAP-Sesn2 fibroblasts (nLAP-Sesn2 MF) were derived from murine nLAP-Sesn2 embryonic stem cells (21) by in vitro differentiation as described by Hewit et al. (22). Primary nLAP-Sesn2 MFs were conditionally immortalized by infecting the cells with the doxycycline-inducible pRRL-ppT-SV40 large T antigen lentiviral vector. Conditionally immortalized nLAP-Sesn2 MFs were maintained in standard tissue cultures supplemented with 1 g/ml doxycycline (Sigma). Both wild type (WT) and Nrf2 KO MLFs were isolated from the lungs of 4-month-old WT and Nrf2 KO mice as described previously (23) and immortalized with SV40 large T antigen as described by Zalvide et al. (20). For transient transfections, HEK293T cells were plated on a 6-well plate, grown to 90% confluence, and transfected with 0.5 g of ␤1-pEGFP-N1 plasmid using Lipofectamine 2000 according to the manufacturer's instructions (Life Technologies). Sesn2 KO MLFs stably reexpressing Sesn2 were obtained by infecting the cells with Sesn2-pBabe-Puro retrovirus and selecting in 2 g/ml puromycin (Life Technologies). shRNA knockdown of Sesn2 in MRC5 cells or Nrf2 in HeLa cells was performed by using the Mission Lentiviral shRNA system (Sigma).
Cell Exposure to Growth Factors, Translational Inhibitors, Oxidants, and Antioxidants-For the activation of Pdgfr␤ signaling, MLFs were serum-starved for 24 h before adding 25 ng/ml PDGF-BB to the cultures, as previously described (9). For Pdgfr␤ half-life measurements, cells were exposed to 10 g/ml cycloheximide (CHX with and without 5 mM Tempol or 10 M MG132 or 100 M chloroquine). Finally, for ROS level manipulations, cells were incubated with 125 M H 2 O 2 for 6 h unless otherwise stated or with 5 mM Tempol or NAC for 7 and 8 h, respectively.
Glutathione S-Transferase (GST) Protein Expression and Pulldown Assays-GST fusion proteins expressed in BL21 Escherichia coli strain were purified and coupled to glutathione-Sepharose beads (GE Healthcare) as previously described (25). For GST pulldown assays, MLFs and HEK293T cells were lysed on ice for 30 min in 50 mM Tris, pH 7.4, 0.15 M NaCl, 2 mM EDTA, 1% (v/v) Nonidet P-40 lysis buffer supplemented with Protease Inhibitor Mixture (Roche Applied Science). Cell lysates were incubated with either GST or GST-tagged proteins immobilized on glutathione-Sepharose beads overnight at 4°C. The beads were washed 3 times with 1 ml of lysis buffer. The samples were resuspended in loading buffer containing 10% (v/v) ␤-mercaptoethanol, boiled for 5 min at 95°C and separated by SDS-PAGE.
Co-immunoprecipitation-Cells were lysed for 30 min on ice in IP lysis buffer (10 mM Tris-HCl, pH 8, 150 mM NaCl, 5 mM EDTA, pH 8, 0.5% (v/v) Triton X-100, 60 mM N-octyglucopyranoside) supplemented with Roche Applied Science protease inhibitor mixture. For HTBH-hRpn11 IPs, lysates from HTBH-hRpn11 HEK293 cells were incubated with streptavidin agarose beads (Thermo Fisher Scientific, Waltham, MA) at 4°C overnight, and unbound proteins were removed by washing once with IP lysis buffer supplemented with 1 mM ATP and 1 mM sodium orthovanadate and twice in Wash buffer (50 mM Tris-HCl, pH 7.5, 10% (v/v) glycerol) supplemented with 1 mM ATP and 1 mM sodium orthovanadate. For nLAP-Sesn2 IPs, cell lysates were incubated at 4°C overnight with GFP Trap-A beads (Chromotek, Martinsried, Germany) and subsequently washed in IP lysis buffer. IPs were analyzed by SDS-PAGE and Western blotting.
Proteasome and Nrf2 Activity Assays-For measuring proteasome activity, cells were lysed on ice in proteasome activity lysis buffer (100 mM NaCl, 50 mM Na 3 PO 4 , 10% (v/v) glycerol, 5 mM MgCl 2 , 0.5% (v/v) Nonidet P-40) supplemented with Protease Inhibitor Mixture (Roche Applied Science), 1 mM sodium orthovanadate, and 5 mM ATP (Sigma). The lysates were homogenized on ice by using a 23-gauge needle and cleared by centrifugation at 13,000 rpm for 15 min. Chymotrypsin-like activity was measured in duplicate using 50 g of proteins and 100 M succinyl-LLVY-amidomethylcoumarin (Enzo Life Sciences) fluorogenic substrate per reaction as previously described (14). Assays were carried out at 37°C for 30 min, and activities were measured using a Viktor2 (PerkinElmer Life Sciences) spectrofluorometer with 380-nm excitation and 460-nm emission filters. For measuring proteasome activity in human lung samples, proteasome was purified from the tissue lysate by using the GST-tagged ubiquitin-like (UBL) domain of hHR23B. GST pulldown assays were performed overnight at 4°C before proteasome activity measurements.
ROS Measurements-Superoxide release from MLFs was measured by EPR as described (26). Briefly, EPR measurements were performed at Ϫ170°C using an EMXmicro Electron Spin Resonance (ESR) spectrometer (Bruker, Karlsruhe, Germany) and 1-hydroxy-3-methoxycarbonyl-2,2,5,5-tetramethylpyrrolidine (CMH) (Noxygen, Elzach, Germany) as the spin probe for detecting intra-and extracellular superoxide production. Because CMH reacts with superoxide and peroxynitrite, parallel samples containing either CMH alone or CMH and superoxide dismutase (SOD) conjugated to polyethylene glycol (PEG-SOD) were measured. This enabled the assessment of the superoxide signal as part of the total CMH signal. Thus, duplicate samples of 2 ϫ 10 5 cells were incubated with 15 units/ml PEG-SOD (Sigma) for 2 h at 37°C followed by the addition of CMH (500 M) Ϯ PEG-SOD. After incubating for another 20 min, the samples were shock-frozen and stored in liquid nitrogen. Spectrometry was performed on frozen samples using a g-factor of 2.0063, a center field of 3349.95G, a microwave power of 200 milliwatt, a sweep time of 20 s, and a sweep number of 5.
Nucleic Acids and Protein Analyses-Total RNA was isolated by using the TriReagent kit (Sigma) according to the manufacturer's instructions. Lysates were centrifuged at 25,000 ϫ g for 10 min at 4°C. One microgram of RNA was reverse-transcribed in 20 l of reverse transcription buffer containing 2 M random primers (New England Biolabs, Frankfurt am Main, Germany) and 200 units of RevertAid H Minus Reverse Transcriptase (Thermo Fisher Scientific). Real time PCRs (Opticon 2 quantitative PCR machine, MJ Research, Ramsey, MN) were performed in duplicate in 25 l of quantitative PCR SYBR Green Mix (Thermo Fisher Scientific) containing 5 l of 5-fold diluted reverse transcription product using an annealing temperature of 60°C. RNA polymerase II was used for normalization. Primer sequences are available on request.
Cell lysates and Western blotting for protein quantification were performed as previously described (9,27). For protein analysis in PDGF-BB stimulation experiments, lysis buffers were supplemented with 1 mM each of sodium fluoride and sodium orthovanadate.
Patient Characteristics-Human lung tissues were obtained from COPD transplant patients (GOLD stage IV) and healthy donor controls ( Table 1). The studies were approved by the Ethics Committee of the Justus-Liebig-University School of Medicine (AZ 31/93), Giessen, Germany.
Statistics-Comparisons between more than two groups were performed using analysis of variance (ANOVA) with the Bonferroni post-hoc test. For comparisons of two groups, Student's two-tailed t test was used. p values below 0.05 were considered significant.

Sesn2-depleted Mouse Lung Fibroblasts Accumulate Proteasomal Target Proteins-We and others showed that ROS and
Pdgfr␤ accumulate in MLFs from Sesn2 knock out mice (Sesn2 KO MLFs) (3,9,24). To test whether Pdgfr␤ accumulation results from its decreased turnover, we estimated Pdgfr␤ levels in MLFs at various time intervals after exposing the MLFs to the protein synthesis inhibitor CHX. Fig. 1A shows that Pdgfr␤  APRIL 10, 2015 • VOLUME 290 • NUMBER 15 levels declined about five times slower in Sesn2 KO MLFs than in WT MLFs. However, the extended Pdgfr␤ half-life in Sesn2 KO MLFs could be readily reversed to normal by the ROS scavenger Tempol (Fig. 1B), indicating that the Pdgfr␤ half-life prolongation is ROS-dependent.

Sesn2 and Pdgfr␤ Expression
Earlier studies showed that Pdgfr␤ can undergo both lysosomal and proteasomal degradation (28,29). To distinguish between these possibilities in our system, we treated MLFs exposed to CHX with chloroquine or MG132 to selectively inhibit lysosomal and proteasomal degradation, respectively. Fig. 1, C and D, show that MG132, but not chloroquine, prolonged the Pdgfr␤ half-life in WT MLFs, indicating that Pdgfr␤ degradation is primarily proteasomal.
Because excessive ROS negatively interfere with proteasomal function (19), we assumed that Sesn2 KO MLFs exhibit defects in the proteasomal protein degradation system. To test this, we first estimated c-myc and p53 protein levels in WT and Sesn2 KO MLFs by Western blotting. Because c-myc and p53 are exclusively degraded by the proteasome (30, 31), their steady state levels indirectly reflect proteasome activity. As shown in Fig. 2A, both c-myc and p53 accumulated in Sesn2 KO MLFs, suggesting reduced proteasomal activity. This effect was independent of Pdgfr␤ signaling, as c-myc and p53 levels remained unaffected by PDGF-BB ligand stimulation ( Fig. 2A). Consistent with the reduced proteasomal activity, Sesn2 KO MLFs accumulated polyubiquitinated proteins including Pdgfr␤ (Fig. 2,  B and C).
Next, we tested whether decreased receptor endocytosis might contribute to the elevated Pdgfr␤ levels in Sesn2 KO MLFs by monitoring subcellular Pdgfr␤ localization at various time intervals after PDGF-BB stimulation. As shown in Fig. 2D, the kinetics of Pdgfr␤ internalization after ligand stimulation were similar in WT and Sesn2 KO MLFs, suggesting that its endocytosis is not affected by the Sesn2 mutation. Taken together, the results suggest that Pdgfr␤ accumulates in Sesn2 KO MLFs as a result of a ROSinduced proteasome dysfunction.
Sesn2 Binds the Proteasome and Modulates Its Function-To confirm that proteasomal activity is reduced in the Sesn2-de- pleted cells, we measured chymotrypsin-like activity (CLA) in WT and Sesn2 KO MLFs using the fluorogenic chymotrypsin substrate succinyl-LLVY-amidomethylcoumarin. As shown in Fig. 3A, Sesn2 KO MLFs exhibited significantly decreased CLA when compared with WT MLFs. This was similar in human fetal lung fibroblasts (MRC5 cells) in which SESN2 was inactivated by shRNA (Fig. 3B), suggesting that proteasomal inhibition is not restricted to murine cells.
Because expression levels of the ␣, ␤1, and ␤2 subunits of the 20 S proteasome as well as the levels of PSMD1 subunit of the 19 S proteasome were similar in Sesn2 KO and WT MLFs (Fig.  3C), we concluded that the activity of the proteasome rather than the expression of the proteasomal subunits is affected by the Sesn2 mutation. This activity could be restored in Sesn2 KO MLFs by re-expressing Sesn2 (Fig. 3D) or by exposing the MLFs to the ROS scavengers NAC or Tempol (Fig. 3, E and F), suggesting that ROS contribute to proteasomal inhibition.
To test whether Sesn2 binds to the proteasome, we used HEK293T cells stably expressing a His-tagged, biotinylatable Rpn11 (HTBH-hRpn11) subunit of the 19 S proteasome (32) and pulled down Rpn11 by using streptavidin-coupled agarose beads. Western blot analysis identified Sesn2 in the pulldown in addition to the 19 S PSMD1 and 20 S ␣ and ␤1 subunits, indicating that Sesn2 physically associates with the native proteasome (Fig. 4A). This association was supported further by IF stainings, which are suggestive of partial cytoplasmic and nuclear Sesn2/proteasome co-localization (Fig. 4B).
To investigate which proteasomal subunits bind to Sesn2, we performed GST pulldown assays from WT MLFs using a purified Sesn2-GST fusion protein bound to glutathione-Sepharose beads. As shown in Fig. 4C, Sesn2 bound to the ␣ and catalytically active ␤1 and ␤5 subunits but not to the ␤2 and PSMD1 subunits of the 20-and 19 S proteasome, respectively. However, when testing ␤1 subunit binding, we observed a ␤1 band shift in the Sesn2-GST pulldown when compared with input ( Fig. 4C). To verify that ␤1 indeed interacted with Sesn2, we performed additional GST pulldown assays from HEK293T cells transiently transfected with a GFP-tagged human ␤1 subunit expression plasmid. Fig. 4D shows that Sesn2-GST readily pulled down ␤1-GFP, but not GFP, confirming that Sesn2 physically interacts with the ␤1 subunit of the 20 S proteasome. Overall, these results suggest that Sesn2 may directly interfere with the proteolytic functions of the proteasome.
Sesn2 Interacts with 2-Cys Peroxiredoxins in a Redox-dependent Manner-It was suggested that Sesn2 reduces oxidative stress by binding to and reducing over-oxidized 2-Cys peroxiredoxins, in particular Prx1 and Prx2 (3). Because the Sesn2/ Prx interactions were thus far detected only in exogenous protein overexpression experiments (3), we asked whether the interactions could also be detected under physiological conditions. For this, we used a HeLa cell line from the Bacterial Artificial Chromosome (BAC) Interactomics Resource stably expressing a BAC transgene in which the N terminus of SESN2 was tagged with an enhanced green fluorescent protein (EGFP) containing a localization and affinity purification (LAP) tag (nLAP-SESN2) (33). As nLAP-SESN2 is expressed and regulated like endogenous SESN2, these cells were amenable to generic tagbased proteomics under nearly physiological conditions. We immunoprecipitated nLAP-SESN2 from the nLAP-Sesn2 HeLa cells using GFP-TrapA beads and probed the IPs with anti-Prx1, Prx2, and over-oxidized Prx (Prx-SO3) antibodies. Fig. 5A shows that PRX1 and PRX-SO3, but not PRX2, co-immunoprecipitated with nLAP-SESN2. Consistent with this finding, IF stainings of nLAP-Prx1 and nLAP-Prx2 HeLa cells obtained from the same BAC Interactomics resource showed cytosolic SESN2-PRX1 but not SESN2-PRX2 co-localization (Fig. 5B). Similarly, in nLAP-Sesn2 HeLa cells, SESN2 co-localized with endogenous PRX1 and PRX1-SO3 (Fig. 5C). However, this Sesn2/Prx interaction was inhibited by hydrogen peroxide (H 2 O 2 ) (Fig. 5, A and C), which also induced translocation of SESN2 from the cytoplasm into the nucleus (Fig. 5C), implying that Sesn2 is not a Prx reductase.
We repeated these experiments using conditionally immortalized murine fibroblasts derived from an embryonic stem cell line expressing nLAP-Sesn2 from its endogenous locus (nLAP- Sesn2-ESC) (21). Unlike the HeLa cells from the BAC Interactomics resource, the nLAP-Sesn2-ESCs were obtained by inserting the nLAP tag into a multipurpose Sesn2 gene trap allele by recombinase-mediated cassette exchange (21). Similar to the HeLa cells, the nLAP-Sesn2 of the conditionally immortalized fibroblasts (nLAP-Sesn2-MF) interacted with Prx1 and Prx-S03 in a redox-dependent manner (Fig. 5D). However, although nLAP-Sesn2-MFs did not express higher levels of Prx2 than the corresponding HeLa cells (Fig. 5E), Sesn2 interacted with Prx2 (Fig. 5D), suggesting that the basal Sesn2/Prx2 interaction is likely cell type-specific.
Nrf2 Activity Is Reduced by Sesn2 Inactivation-A recent study showed that Sesn2 activates the antioxidant transcription factor Nrf2 by binding and inactivating its inhibitor Keap1 (7). To test whether Nrf2 activity is compromised in Sesn2-depleted cells, we first estimated Nrf2 and Keap1 expression in WT and Sesn2 KO MLFs by quantitative real-time-PCR and Western blotting. As shown in Fig. 6A, neither Nrf2 nor Keap1 expression was affected by the loss of Sesn2. However, when the MLFs were subjected to antioxidant response elements (ARE)-luciferase reporter assay, Sesn2 KO MLFs developed significantly less luciferase activity than the WT MLFs under both basal and oxidative stress conditions (Fig. 6B) and showed reduced expression of hemoxigenase 1 (Ho1), an endogenous Nrf2 target gene (Fig. 6C). These results suggest that ROS accumulate in Sesn2-deficient MLFs as a result of reduced Nrf2 activity and, consequently, inhibit proteasomal function.
Sesn2 Directly Interacts with Nrf2 in a Redox-dependent Manner-As noted above, Sesn2 activates Nrf2 by directly interacting with its inhibitor Keap1. However, this interaction was described in HEK293T cells overexpressing Sesn2 and Keap1 cDNAs (7). To test whether the Sesn2/Keap1 interaction can be replicated in cells expressing nearly normal levels of Sesn2, we immunoprecipitated nLAP-SESN2 from HeLa cells either treated or untreated with H 2 O 2 and probed the IPs with anti-Keap1 and Nrf2 antibodies. Fig. 6D shows that NRF2, but not KEAP1, co-immunoprecipitated with nLAP-SESN2, especially under oxidative stress conditions. Likewise, as revealed by IF staining, H 2 O 2 -induced SESN2-NRF2 co-localization and cytoplasmic nuclear translocation (Fig. 6E), suggesting that SESN2 may act as a transcriptional co-activator of Nrf2 target genes independently of Keap1.
As there is some concern in the literature whether the size of the mRNA predicted Nrf2 translation product (Ϸ65 kDa) represents bona fide Nrf2 (34), we subsequently validated the anti-Nrf2 antibody employed in these experiments using immortalized MLFs from Nrf2 knock out mice (23) and HeLa cells transduced with anti-Nrf2 shRNA. As shown Fig. 6F, the 65-kDa band detected by the R&D monoclonal antibody was missing in Nrf2 KO MLFs, and its intensity was reduced in NRF2 KD HeLa cells, confirming that this band represents monomeric Nrf2.
Nrf2 Inactivation Up-regulates Pdgfr␤-To test whether Nrf2 inactivation replicates the Pdgfr␤ induction phenotype of Sesn2 KO MLFs, we estimated ROS levels and Pdgfr␤ expression in MLFs derived from Nrf2 KO mice. As shown in Fig. 6G, Nrf2 KO MLFs accumulated ROS and expressed significantly higher Pdgfr␤ levels than the corresponding WT MLFs (Fig. 6, H and I), resulting in Pdgfr␤ signal amplification in response to ligand stimulation (Fig. 6, J and K). This suggests that Sesn2, Nrf2, and the proteasome are all part of a redox-sensitive Pdgfr␤-suppressing pathway whose inactivation up-regulates Pdgfr␤. However, unlike previously reported Pdgfr␤ activation by ROS-mediated inhibition of dedicated protein-tyrosine phosphatases (35,36), the elevated phosphorylated Pdgfr␤ (p-Pdgfr␤) levels in Nrf2 KO MLFs were largely attributable to Pdgfr␤ accumulation rather than enhanced phosphorylation. Thus, the Pdgfr␤/Gapdh ratio but not the p-Pdgfr␤/Pdgfr␤ ratio was elevated in Nrf2 KO MLFs (Fig. 6K).
Sesn2 Inactivation Up-regulates Nf-ya-We showed previously that the up-regulation of Pdgfr␤ in the lungs of Sesn2 KO mice occurred at both the mRNA and protein level (9). To test the possibility that mRNA induction is mediated by a Pdgfr␤ transcription factor whose proteasomal degradation is delayed in Sesn2-depleted cells, we focused on the Nf-ya subunit of the Nf-y transcription factor complex, which controls basal Pdgfr␤ gene transcription (37) and is degraded exclusively by the proteasome (38) . Fig. 7, A and B, shows that Nf-ya and Pdgfr␤ levels were significantly increased in Sesn2 KO MLFs. Most importantly, Nf-ya was also up-regulated in Nrf2 KO MLFs (Fig. 7, C  and D), suggesting that Nf-ya and Nrf2 operate within the same Sesn2/Pdgfr␤ suppressor pathway (Fig. 7E).
Nrf2 Is Up-regulated in the Lungs of Individuals with COPD-Recently, we showed that SESN2 is up-regulated and PDGFR␤ expression repressed in the lungs of individuals with advanced COPD (9), suggesting that the entire Sesn2/Pdgfr␤ suppressor

Sesn2 and Pdgfr␤ Expression
9746 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 290 • NUMBER 15 • APRIL 10, 2015 pathway (Fig. 7E) might be up-regulated in COPD lungs. Indeed, in contrast to previous reporting (39), we found NRF2 significantly up-regulated and KEAP1 repressed in COPD lungs when compared with healthy donor lungs (Fig. 8, A and B; see Table 1 for patient characteristics). Moreover, consistent with enhanced antioxidant defense, Nrf2 target genes such as GCLC, HO1, and PRX2 were also up-regulated in the COPD lungs, whereas the expression of the positive Nrf2 regulator DJ1 was not significantly different from normal lungs (Fig. 8, A and  B). Finally, consistent with recent reporting (40) and in contrast to a previous study (53), proteasomal activity and subunit expression were not decreased in COPD lungs (Fig. 8, C-F).

DISCUSSION
We recently reported that Sesn2 interferes with the development of pulmonary emphysema by suppressing Pdgfr␤-regulated alveolar maintenance programs (9). Here, we identified the transcription factor Nrf2 and the 26 S proteasome as key components of a Sesn2/Pdgfr␤ suppressor pathway and found that this pathway is highly up-regulated in lungs of individuals with advanced COPD.
In lung fibroblasts of mouse and human origin, inactivation of Sesn2 decreased proteasomal activity by inducing ROS formation via inhibition of Nrf2-controlled antioxidant gene expression. Decreased proteasomal degradation of Pdgfr␤ and of its transcriptional activator Nf-ya resulted in Pdgfr␤ accumulation and signal amplification in response to ligand stimulation (9).
In vitro and in vivo protein binding studies revealed that Sesn2 physically interacts with several 20 S proteasomal subunits including the catalytically active ␤1 and ␤5 subunits. Other oxidoreductases were also reported to associate with the 26 S proteasome. For example, thioredoxin-like 1 (Txn1) or thioredoxin-related protein of 32 kDa (TRP32) interacts with the 19 S regulatory subunit near the substrate binding site and was suggested to play a role in proteasome assembly (41)(42)(43). Likewise, the human flavin-dependent NADPH:quinone reductase (NQO1) interacts with the 20 S proteasome and protects transcription factors p53 and p73 from proteasomal degradation (31). Accordingly, Sesn2 could directly stimulate proteasomal activity by serving as a maturation factor that recruits proteasomal subunits for assembly. To our knowledge, the only other known interaction partner of the catalytically active proteasomal ␤ subunits is proteassemblin (Ump1), which is involved in 20 S proteasome assembly (44,45). In yeast, inactivation of the Ump1 homolog causes disordered proteasome assembly with abnormal subunit stoichiometry, incomplete propeptide processing, and reduced proteolytic activity (44). Similarly, the inactivation of Sesn2 might cause proteasome dysfunction by preventing its orderly assembly. Alternatively, Sesn2 interaction with 20 S proteasome might affect the structure of 20 S active sites and thus have an impact on proteasomal activity. Although this cannot be excluded by the present experiments, the recovery of proteasomal function in Sesn2-depleted cells by treatment with antioxidants such as NAC and Tempol suggests that Sesn2 also interferes indirectly with proteasomal function via ROS.
To scrutinize possible mechanisms of ROS accumulation in Sesn2-depleted cells, we used HeLa cells and mouse ESC-derived fibroblasts expressing EGFP-tagged Sesn2 from its endogenous chromosomal location, which enables generic tag-based proteomics under nearly physiological conditions. Although we found that Sesn2 physically interacts with peroxiredoxins, including over-oxidized peroxiredoxins, these interactions were effectively blocked by ROS. Thus, although the biological significance of the Sesn2/Prx interactions remains to be established, their inhibition by oxidative stress does not support the sulfinic acid reductase activity previously attributed to sestrins (3).
In agreement with a recent report (7), we found that Sesn2 regulates ROS metabolism by modulating the activity of Nrf2. Both Sesn2 KO-and Nrf2 KO MLFs accumulated ROS and up-regulated Pdgfr␤, confirming that Nrf2 is a downstream target of Sesn2. Indeed, experiments with nLAP-SESN2 expressing HeLa cells revealed that SESN2 co-localizes and physically interacts with NRF2, especially under oxidative stress conditions that simultaneously promoted a nuclear SESN2/NRF2 translocation. Because we could not detect a SESN2/KEAP1 interaction deemed essential for Nrf2 activation (7), this ROSinduced nuclear SESN2/NRF2 translocation suggests that Sesn2 may act as a transcriptional co-activator of Nrf2 target genes independently of Keap1.
Overall, we identified Nrf2, ROS, and the proteasome as central components of a novel signal transduction pathway employed by Sesn2 to suppress Pdgfr␤ expression. Activation of NRF2 in the lungs of individuals with advanced COPD, which were shown to overexpress SESN2 (9), suggests that the Sesn2/Pdgfr␤ suppressor pathway (Fig. 7E) is induced in COPD lungs and is responsible for repressing Pdgfr␤-controlled alveolar maintenance programs (9). In this context, it should be mentioned that activation of antioxidant defense mechanisms   in COPD lungs is still a matter of debate. Although induction of antioxidant genes, including Nrf2, by chronic exposure to cigarette smoke was clearly documented in both airway epithelial cells and lung tissue (46 -49), the picture is less clear in lungs of patients with COPD. Several genome-wide expression studies showed either up-regulation or no change in the expression of some (but not all) Nrf2-related antioxidant genes in COPD lungs (46,50,51), whereas other studies reported a decline in Nrf2 activity correlating with the severity of the disease (39). Our data showing that NRF2 is up-regulated in the lungs of individuals undergoing transplantation for late-stage COPD suggest that pharmacological NRF2 activators recently proposed for COPD treatment (52) should be used with caution.