SCFFBXO17 E3 ligase modulates inflammation by regulating proteasomal degradation of glycogen synthase kinase-3β in lung epithelia

Glycogen synthase kinase-3β (GSK3β) has diverse biological roles including effects on cellular differentiation, migration, and inflammation. GSK3β phosphorylates proteins to generate phosphodegrons necessary for recognition by Skp1/Cullin-1/F-box (SCF) E3 ubiquitin ligases leading to subsequent proteasomal degradation of these substrates. However, little is known regarding how GSK3β protein stability itself is regulated and how its stability may influence inflammation. Here we show that GSK3β is degraded by the ubiquitin-proteasome pathway in murine lung epithelial cells through lysine 183 as an acceptor site for K48 polyubiquitination. We have identified FBXO17 as an F-box protein subunit that recognizes and mediates GSK3β polyubiquitination. Both endogenous and ectopically expressed FBXO17 associate with GSK3β, and its overexpression leads to decreased protein levels of GSK3β. Silencing FBXO17 gene expression increased the half-life of GSK3β in cells. Furthermore, overexpression of FBXO17 inhibits agonist-induced release of keratinocyte-derived cytokine (KC) and interleukin-6 (IL-6) production by cells. Thus, the SCFFBXO17 E3 ubiquitin ligase complex negatively regulates inflammation by targeting GSK3β in lung epithelia.

The ubiquitin-proteasome system (UPS) 3 is the primary mechanism for protein degradation in eukaryotic cells (1).
Ubiquitination of protein substrates involves the stepwise transfer of ubiquitin from an E1 ubiquitin-activating enzyme to an E2 ubiquitin-conjugating enzyme, and finally to an E3 ubiquitin ligase complex. In the final step of the reaction, the E3 ubiquitin ligase transfers ubiquitin chains to the substrate to facilitate degradation by the proteasome or sorting to the endosome-lysosome pathway (2). Two E1 enzymes, almost 40 E2 enzymes, and more than 1000 E3 enzymes have been identified in mammalian cells (3). The SCF superfamily represents the largest group of E3 ligases that mediate critical roles in cell biology, including tumorigenesis and inflammation. The SCF E3 ligase is a modular complex comprised of core proteins Rbx1, Skp1, Cul1, and an F-box protein (1,2). The F-box protein subunit confers specificity for the target substrate. Approximately 70 human F-box proteins have been identified in the genome, but most have not been characterized (4 -6). Proteins targeted by the ubiquitin-proteasome pathway contain primary sequences, post-translational modifications, or structural changes named "degrons" that allow recognition by ubiquitin ligases in the appropriate context (7,8). Regulation of protein homeostasis is a growing area of interest in cancer biology and inflammatory diseases. F-box proteins in particular are attractive therapeutic targets because recognition of degrons serves as a key regulatory point for protein stability (1,4,9). A few F-box proteins have been implicated in the pathogenesis of acute respiratory distress syndrome (ARDS) (9 -11). This devastating illness is characterized by an initial robust inflammatory response in the host that is linked to the activation of key regulatory kinases in effector cell populations in the setting of sepsis or severe pneumonia.
An important regulatory step for commitment of a protein substrate to proteasomal degradation is phosphorylation (8,12). Glycogen synthase kinase 3␤ (GSK3␤) is a serine-threonine kinase with over 50 known protein substrates (13,14). Although basal constitutive kinase activity is high in cells, substrates of GSK3␤ usually require priming with phosphorylation by upstream kinases, leading to optimal kinetics of GSK3␤mediated phosphorylation (14,15). Many substrates of SCF E3 ligases are phosphorylated by GSK3␤, generating the phosphodegron required for E3 ligase recognition, ubiquitination, and proteasomal degradation (12). For example, GSK3␤ phosphorylates the ST2L receptor for IL-33 leading to degradation by SCF FBXL19 (11). TRAF proteins are phosphorylated by GSK3␤ prior to degradation by SCF FBXL2 (16), and the F-box protein FBXL2 itself is phosphorylated by the same kinase prior to ubiquitination and degradation by another SCF complex, SCF FBXO3 (12,16,17).
In recent years there has been a resurgence in interest in GSK3␤ as its list of substrates grows in the literature (12,18). GSK3␤ was initially discovered in the context of glucose metabolism, but it has now been established as a point of convergence for adaptive and innate immunity pathways (19,20). GSK3␤ has been shown to phosphorylate serine 468 on the p65 subunit of NF-B, inhibiting binding to promoters (21). TNF␣-induced expression of IL-6 and CXCL1 and p65 recruitment to the promoter of select inflammatory genes also requires GSK3␤ activity in mouse embryonic fibroblast cells (22). GSK3␤ is required for TNF␣-mediated NF-B signaling, and suppression or deletion of GSK3␤ enhances sensitivity to TNF␣-induced apoptosis (23,24). GSK3␤ also negatively regulates NF-B activity and mediates cross-tolerance to LPS in macrophages (25). Overall, the role of GSK3␤ in downstream inflammatory pathways varies depending on the cell type and stimuli used (13,19).
A growing number of studies demonstrate a pathogenic role for GSK3␤ in end-organ damage in murine models of sepsis. Liver injury has been shown to be mediated by GSK3␤ activity and hepatocellular injury decreases with the use of GSK3␤ inhibitors (26,27). More recently, murine models of ARDS demonstrate an active role of GSK3␤ in mediating acute lung inflammation and alveolar damage (28,29). Despite critical roles in phosphorylating proteins to direct their fate to the UPS and studies supporting its role in inflammation and sepsis, very little is known about the regulation of GSK3␤ stability itself. Studies show that GSK3␤ is degraded by the proteasome in some cell types but the mechanism has not been well characterized (30). Regulation of GSK3␤ protein stability in lung biology has not been studied, and to date, no E3 ubiquitin ligase complex has been identified that targets GSK3␤ to the proteasome.
In this study we elucidate the mechanism of GSK3␤ polyubiquitination and degradation by the proteasome. We have identified a previously uncharacterized F-box protein subunit, FBXO17, which targets GSK3␤ for polyubiquitination by the SCF complex. We also show that FBXO17 attenuates inflammatory responses in lung epithelial cells by facilitating degradation of GSK3␤ protein. These studies describe a mechanism for regulation of GSK3␤ protein stability and potentially has broad implications in the regulation of immune responses in acute lung injury.

GSK3␤ degradation occurs through the ubiquitin-proteasome pathway
We examined the half-life of GSK3␤ in lung epithelial cells and determined whether degradation was primarily dependent on the UPS or endosome-lysosome pathway. Using cycloheximide (CHX) to inhibit protein synthesis, we treated cells with the proteasome inhibitor, MG132, or a lysosome inhibitor, leupeptin. The half-life of GSK3␤ exceeded ϳ8 h. Protein levels were stabilized in the presence of MG132, but not leupeptin, supporting the supposition that GSK3␤ is processed primarily through the UPS (Fig. 1, A and B). To determine whether GSK3␤ is polyubiquitinated, control or HA-ubiquitin (HA-Ub) plasmids were transfected into mouse lung epithelial (MLE) cells. Endogenous GSK3␤ levels decreased in a dose-dependent manner in cells with increasing amounts of ectopically expressed HA-Ub plasmid (Fig. 1C). It has been demonstrated that lysine 48-linked ubiquitin chains are associated with targeting proteins to the UPS, whereas lysine 63 chains facilitate protein sorting to the endosome-lysosome pathway (31,32). After treating MLE cells with MG132 to allow for accumulation of polyubiquitinated proteins, endogenous GSK3␤ was immunoprecipitated. GSK3␤ was shown to be polyubiquitinated by lysine 48-linked ubiquitin, but not lysine 63-linked ubiquitin chains (Fig. 1D). The data support the UPS as the dominant mechanism for GSK3␤ degradation in lung epithelia.

Lysine 183 is an acceptor site for K48 polyubiquitination in GSK3␤
Ubiquitination most often occurs on lysine residues in target proteins. Human and mouse GSK3␤ share 99% homology and contain 23 lysine residues. Based on prediction algorithms (7, 33) 16 sites were screened and selected for site-directed mutagenesis. Plasmids expressing GSK3␤ with single lysine to arginine mutations were generated and transfected into MLE cells. CHX chase assays were performed with all mutations. Expression of a variant GSK3␤ plasmid harboring a K183R mutation in cells conferred stability of protein levels in CHX chase assays compared with expression of a wild-type or a K205R plasmid (Fig. 2, A and B). Cells transfected with plasmids expressing HA-tagged wild-type, K183R, or K205R variants were also subjected to co-immunoprecipitation and immunoblotting. First, after immunoprecipitation of ubiquitinated products and probing with HA antibody, polyubiquitination of mutant K183R-GSK3␤ was reduced compared with wild-type (Fig. 2C). In other experiments the antibodies were reversed using co-immunoprecipitation demonstrating similar findings (Fig. 2D). Last, in separate studies, immunoprecipitation with HA antibody followed by probing with the K48 antibody demonstrated that GSK3␤ was modified by K48 polyubiquitin chains. The intensity of this signal on immunoblots was reduced after analysis of expressed K183R-GSK3␤ plasmid (Fig.  2E). Of note, Lys-205 had been previously predicted as a ubiquitin acceptor site by the Phosphosite database (34).

FBXO17 mediates GSK3␤ degradation
GSK3␤ is a phosphoenzyme and SCF E3 ligases target phosphoproteins for ubiquitination. Thus, with data supporting GSK3␤ degradation through the UPS, we sought to identify a potential F-box subunit of the SCF E3 ligase complex that could recognize and target GSK3␤ for polyubiquitination. Using a library of over 30 F-box protein expression plasmids, HEK293 cells were transfected with these plasmids and immunoblotted

FBXO17 mediates degradation of GSK3␤
to analyze GSK3␤ protein levels. FBXO17 was identified in this screen as being associated with lower expression levels of GSK3␤ (data not shown). We then transfected a FBXO17-V5 plasmid in MLE cells and observed a trend for a decrease of GSK3␤ protein levels with no effect on mRNA transcript expression (Fig. 3, A and B). To confirm that this F-box protein was a subunit of the canonical SCF E3 ubiquitin ligase apparatus, we overexpressed FBXO17-V5 and performed co-immunoprecipitation. Here, Skp1 was shown to be associated with FBXO17 (Fig. 3C).
To further assess behavior of SCF FBXO17 to target GSK3␤ for degradation, plasmids expressing HA-tagged wild-type, K183R, or K205R variants were co-transfected with or without FBXO17-V5 plasmid into MLE cells. Expression of wild-type FBXO17-V5 decreased wild-type and K205R protein levels but not K183R protein (Fig. 3D). Additional studies were conducted expressing an empty vector or ⌬FboxFBXO17-V5 in MLE cells in CHX chase studies. The latter construct lacks the ability to engage other components of the SCF apparatus. Here, unlike effects of FBXO17-V5, expression of ⌬FboxFBXO17-V5 did not alter GSK3␤ lifespan nor did it induce kinase polyubiq-uitination ( Fig. 4, A and C). When FBXO17 gene knockdown is performed, the GSK3␤ half-life increases and polyubiquitination of the kinase decreases ( Fig. 4, B and D). These experiments suggest that FBXO17, as a component of the SCF complex, mediates K48 polyubiquitination of GSK3␤ at a distinct molecular site (Lys-183) for its proteasomal degradation.

FBXO17 associates with GSK3␤ through a docking motif
Although our data supports the hypothesis that GSK3␤ is a substrate for FBXO17, GSK3␤ phosphorylates many substrates of the SCF E3 ligase complexes. Thus, we sought to clarify if FBXO17 specifically associates with GSK3␤ as a substrate for polyubiquitination. We examined this issue by expressing FBXO17-V5. Immunoprecipitation of GSK3␤ demonstrated an association between endogenous GSK3␤ and ectopically expressed FBXO17 (Fig. 5A). Colocalization of both proteins in the cytoplasm was demonstrated by immunofluorescence ( Fig.  5B). To further characterize molecular interactions between FBXO17 and V5-tagged GSK3␤, GSK3␤ truncation mutants were cloned and used in in vitro transcription and translation assays. When FBXO17 was immunoprecipitated, all mutants were found to associate with FBXO17 by immunoblotting with the exception of loss of binding of the 201-420 amino acid NH 2 -terminal mutant (Fig. 5C). These experiments demonstrate that amino acids 151-200 contain a putative binding domain required for association of GSK3␤-V5 with FBXO17.

Degradation of GSK3␤ by FBXO17 abrogates inflammatory cytokine production
To determine whether FBXO17 has a role in inflammatory responses in lung epithelial cells, we first examined if endogenous levels of FBXO17 in MLE cells changed after stimulation with lipopolysaccharide (LPS). Kinetics studies of expression showed no significant changes in FBXO17 or GSK3␤ protein levels (Fig. 6A). Next, we overexpressed various plasmids in MLE cells and evaluated responses to LPS and tumor necrosis factor ␣ (TNF␣). We found that IL-6 and KC production was significantly reduced in response to LPS and TNF␣ when FBXO17 is ectopically expressed in cells (Fig. 6, B and C). Partial rescue of KC production in response to LPS was observed with overexpression of HA-GSK3␤ (Fig. 6D). FBXO17 expression did not modulate IL-10 levels with any stimuli used, and the F-box protein did not significantly alter IL-6 and KC production by MLE cells in response to lipoteichoic acid (LTA) or PamCys3 (data not shown). These results differ from a genome-wide study by Alper et al. (35) in which FBXO17 was identified as a possible proinflammatory gene in innate immune responses to LTA. In that study, knockdown of FBXO17 expression in RAW macrophages resulted in a blunted response to LTA stimulation, and its overexpression led to enhanced NF-B promoter activity. However, our data suggest that FBXO17 has anti-inflammatory effects through negative regulation of GSK3␤ stability in lung epithelial cells.

Discussion
This study elucidates a unique mechanism for attenuation of immune responses by regulation of GSK3␤ protein stability through a poorly characterized SCF E3 component. FBXO17, also known as FBG4, is an F-box protein and one of five members of the F-box-associated family thought to be involved in glycoprotein ER-associated degradation (36,37). Glenn et al. (38) identified sulfated and galactose-terminated glycoproteins as likely binding substrates of FBXO17 by a glycan array, with no specific proteins identified as substrates for ubiquitination. This latter study also demonstrated FBXO17 expression in mouse brain and weaker expression in the lung, but not muscle, liver, pancreas, adipose tissue, or kidney, contrary to broader human tissue expression identified by others (37). Mouse and human FBXO17 share 85% homology. We confirmed that this

FBXO17 mediates degradation of GSK3␤
F-box protein subunit is an authentic subunit associated with the core SCF E3 ubiquitin ligase machinery. Our identification of a specific protein substrate, GSK3␤, targeted by the SCF FBXO17 E3 ligase complex has potentially important biologic implications for a wide array of fundamental processes. Given the many biological roles of GSK3␤, this study raises additional questions regarding how altered stability of this kinase by FBXO17 influences downstream cellular pathways in different cell types. Moreover, as with other F box proteins (9,16), small molecule FBXO17 targeting might be a strategy to modulate various processes by maintaining cellular levels of GSK3␤.
There is growing interest in understanding how dysregulation of protein stability contributes to the pathogenesis of lung diseases (39 -41). Findings from our group have identified several F-box protein subunits with anti-inflammatory effects. IL-33 stimulates inflammation in asthma and acute lung injury through the receptor ST2L. GSK3␤ phosphorylates ST2L and subsequently SCF FBXL19 targets the receptor for polyubiquitination, internalization, and degradation (11). FBXL19 overexpression decreases ST2L levels and reduces inflammation while improving animal survival in these studies. Another example is FBXL2, an F-box subunit that targets TRAF proteins in NF-B signaling (17). FBXO3 was identified as an F-box subunit that targets FBXL2 for degradation after phosphorylation by GSK3␤, thus promoting inflammation. Pharmacologic inhibition or depletion of FBXO3 leads to normal levels of FBXL2, decreased cytokine levels, and improved survival in murine models of sepsis (9,16). These studies represent a subset of F-box protein subunits involved in modulating inflammatory processes. They also support prior studies that show GSK3␤mediated phosphorylation as a critical step for substrate recognition by SCF E3 ubiquitin ligases and commitment to the UPS pathway (12).
Lysine 183 appears to be the critical acceptor site for most, but not all K48-linked polyubiquitin chains covalently attached to GSK3␤. However, other lysine residues may provide finetuning of regulation of kinase degradation by the UPS. In our studies, expression of the K183R mutant protein reduces responses to LPS compared with wild-type GSK3␤, despite increased stability of mutant protein levels (data not shown). This may be attributed to reduced catalytic activity of the kinase possibly impacted by the K183R mutation given its close proximity to the activation loop (15,42). Interestingly, a recent study demonstrated that GSK3␤ is ubiquitinated by K63, but not K48-linked ubiquitin chains on lysine 183 in response to poly(I: C)-mediated TLR3 stimulation of RAW macrophages (43). The K183R mutant was shown to function as a dominant-negative protein in this study. K63 polyubiquitination was required for TLR3 signaling, and the physiologic role of protein degradation by the proteasome was not suggested by this study. Another study showed that SCF FBXO7 mediates K63 polyubiquitination of GSK3␤ and negatively regulates its activity in HEK293T cells (44). In our experiments, we did not observe significant levels of K63-linked polyubiquitin chains when GSK3␤ was immunoprecipitated from MLE cells after MG132 treatment, indicating mechanistic variability of GSK3␤ processing by the UPS  (IB). B, MLE12 cells were transfected with 2 g of FBXO17-V5 plasmid for 48 h. RNA was isolated and analyzed by RT-PCR using primers against GSK3␤ and GAPDH as an internal control. Inset, FBXO17-V5 protein expression was confirmed by immunoblotting using V5 antibody. C and D, FBXO17-V5 expression plasmid (2 g) was transfected into MLE-12 cells and cells were cultured for 48 h. Immunoprecipitation (IP) of lysates was performed using Skp1 antibody. Samples were immunoblotted with Skp1, V5, and ␤-actin (loading control) antibodies. The data in each panel are representative of at least n ϭ 3 independent experiments. D, MLE-12 cells were co-transfected with HA-tagged wild-type, K183R, or K205R mutant GSK3␤ plasmids with or without FBXO17-V5 plasmid. Samples were immunoblotted with HA, V5, and ␤-actin (loading control) antibodies.
depending on the experimental context. Proteomic data suggest ubiquitination at lysine 205, but the K205R mutation was less stable than wild-type GSK3␤ (34). This finding may reflect the importance of lysine 205 for overall stability because it is located within the catalytic loop of GSK3␤ (14,15).
GSK3␤ kinase activity is integral to Toll-like receptor and NFB signaling in diverse cell types. Of note, ectopic expression of FBXO17 in MLE cells did not substantially abrogate cytokine release in response to TLR2 agonists suggesting a more restricted effect for the F-box protein in modulating TLR4 stimulation. Martin et al. (45) was the first group to show that GSK3␤ is required for TLR4-induced responses in RAW macrophages and in murine models of sepsis. In murine models of ARDS, Akt-mediated phosphorylation of GSK3␤ inhibits inflammation, and pharmacologic inhibition of GSK3␤ reduces inflammation and improves survival (29,46). GSK3␤ also pro-motes neuroinflammation in glial cells from mice treated with LPS through STAT3 activation (47). A recent study by Peng and colleagues (48) demonstrated that FBXO17 reduces type I IFN signaling through recruitment of protein phosphatase 2A (PP2A) for dephosphorylation of interferon recruitment factor 3 (IRF3) in A549 and HEK293T cells. Intriguingly, these effects were not related to SCF E3 ligase function and were independent of the F-box domain and ubiquitin-proteasome pathway. These data support FBXO17 as a negative regulator of inflammation. Dexamethasone treatment of pancreatic islet cells has been associated with proteasomal degradation of GSK3␤, indicating a relationship between suppression of inflammation and decreased stability of GSK3␤ (30). Our study suggests that GSK3␤ promotes inflammatory responses in lung epithelial cells, and FBXO17 is a negative regulator of these pathways.

FBXO17 mediates degradation of GSK3␤
In summary, this study is the first to identify an SCF E3 ligase that targets GSK3␤ for its proteasomal degradation. Physiologically, our data suggest that SCF FBXO17 is a negative regulator of GSK3␤-mediated inflammatory responses in lung epithelia. It is likely that FBXO17 recognizes other substrates, and additional studies may implicate other targets of this F-box protein subunit. Drugs targeting the ubiquitin proteasome pathway are already being used in the clinical setting to treat cancer and rejection in organ transplantation (1,3). Further studies of this pathway may reveal key therapeutic targets to exploit GSK3␤ degradation as a means for therapy in neoplasia and inflammatory disorders.

Antibodies and reagents
FBXO17 antibody was obtained from Protein Tech Group, Inc. (Rosemont, IL). Anti-mouse IgG and anti-rabbit IgG were obtained from Santa Cruz Biotechnology (Santa Cruz, CA). Antibodies against ubiquitin, K48 ubiquitin, K63 ubiquitin, Skp1, and Cullin 1 were purchased from Cell Signaling (Danvers, MA). ␤-Actin antibodies, lipoteichoic acid (Staphylococcus aureus), and CHX were obtained from Sigma. QuikChange II site-directed mutagenesis kits were purchased from Agilent Technologies (Santa Clara, CA). Pierce protein A/G-agarose beads and protease inhibitors were purchased from Thermo Scientific (Rockford, IL). MG132 and leupeptin was from Calbiochem (Darmstadt, Germany). Mouse V5 antibodies, miniprep, midiprep, and pcDNA3.1 directional cloning kits were purchased from Invitrogen and New England Biolabs (Ipswich, MA). In vitro transcription and translation (TNT) kits were from Promega (Madison, WI).

FBXO17 mediates degradation of GSK3␤
cultured in HITES medium supplemented with 10% FBS. Prior to CHX treatment, cells were starved for 1 h with 0% FBS, HITES medium. CHX treatment was carried out at a concentration of 40 g/ml at varying time points in 0% FBS medium, with or without MG132 (20 g/ml) or leupeptin (20 g/ml).

Immunoprecipitation and immunoblotting
Cell lysates in 150 l of lysis buffer (20 mM Tris-HCl (pH 7.4), 150 mM NaCl, 2 mM EGTA, 5 mM ␤-glycerophosphate, 1 mM MgCl 2 , 1% Triton X-100, 1 mM sodium orthovanadate, 10 g/ml of protease inhibitors, 1 g/ml of aprotinin, 1 g/ml of leupeptin, and 1 g/ml of pepstatin) were sonicated on ice for 12 s and centrifuged at 10,000 ϫ g for 10 min at 4°C in a microcentrifuge. For immunoprecipitation, equal amounts of cell lysates (1 mg) were incubated with 5 g/ml of specific primary antibodies overnight at 4°C followed by the addition of 40 l of protein A/G-agarose for 4 h at 4°C. For immunoblotting, equal amounts of supernatant (20 g) were subjected to 10% SDS-PAGE gels, transferred to nitrocellulose membranes, blocked with 5% (w/v) nonfat milk in TBST (25 mM Tris-HCl, pH 7.4, 137 mM NaCl, and 0.1% Tween 20) for 1 h, and incubated with primary antibodies in 5% (w/v) BSA in TBST for 1-2 h. The membranes were washed at least three times with TBST at 10-min intervals followed by a 1-h incubation with mouse, rabbit, or goat horseradish peroxidase-conjugated secondary antibody (1:2,000). The membranes were developed with an enhanced chemiluminescence detection system according to manufacturer's instructions (49,50).

RT-PCR
RNA was isolated from cells using RNeasy Mini Kits (Qiagen) per the protocol provided. Isolated RNAs were immediately converted to cDNA using High Capacity RNA-to-cDNA Kits (Life Technologies) after their concentrations were measured. Real-time PCR assays were performed using SYBR Select Master Mix for CFX (2X) (Life Technologies) per the protocol provided and assays were performed with the C1000 Thermal Cycler (Bio-Rad).

Cloning and mutagenesis
Human GSK3␤ plasmid was purchased from Addgene (Cambridge, MA) for subsequent cloning into pcDNA3.1D/ V5-His vector (Invitrogen). Site-directed mutagenesis of lysine to arginine residues was carried out via a PCR-based approach using the QuikChange II XL kit from Agilent Technologies (Santa Clara, CA) and appropriate primers. Primers used for lysine to arginine mutagenesis were designed with the following sequences: Lys-183 forward, 5Ј-AGGTTCTGCGGTCTAAT-ATCCCGATGGCAGATTCCA-3Ј; reverse, 5Ј-TGGAATC-TGCCATCGGGATATTAGACCGCAGAACCT-3Ј; K205R

Gene silencing
Small interfering RNAs (siRNA) for human FBXO17 were purchased from Santa Cruz (sc-97555). FBXO17 siRNA was transfected into BEAS-2B cells by electroporation. Concentrations used ranged from 10 to 100 nM. Cells were cultured for 72 h prior to lysate collection and immunoblotting.

Transfection
2.5 ϫ 10 6 MLE cells were suspended in 100 l of 20 mM HEPES in PBS and mixed with 2 g of DNA in a cuvette. Cells were nucleofected on the T-013 protocol using an Amaxa Nucleofector II device (Basel, Switzerland). Immediately following nucleofection, 1 ml of 10% HITES medium was added to the cuvette and the cell solution was plated in 2 ml of 10% HITES culture medium in a 6-well plate. Cells were allowed to grow until 80% confluent prior to half-life experiments (48 -72 h). Cells were then treated with 40 g/ml of CHX at different time points for half-life studies and were harvested in lysis buffer using a rubber policeman for subsequent immunoblot analysis.

Immunofluorescence staining
MLE cells grown on glass bottom dishes were fixed with 3.7% formaldehyde for 20 min, and immunostained with anti-HA or anti-FBXO17 antibody. Cells were washed three times, and incubated with fluorescent-conjugated secondary antibodies. Images were captured by a Nikon ECLIPSE TE 300 inverted microscope.

In vitro transcription/translation and binding assays
The details of this assay were described previously and adapted from the manufacturer's protocol (9). Briefly, 1 g of plasmid expressing full-length V5-GSK3␤, truncated V5-GSK3␤ mutants, or FBXO17 was used in 25-l reactions using the TNT Coupled Reticulocyte Lysate System (Promega). FBXO17 antibody was incubated with protein A/G-agarose resin for 1 h at room temperature and washed with PBS prior to adding reactions containing in vitro synthesized FBXO17. Reactions were incubated with beads for 1 h at room temperature and after washing with PBS, in vitro synthesized V5-GSK3␤ reactions (full-length and truncation mutants) were added. After incubating for 1 h at room temperature and washing, the proteins were eluted and processed for V5-GSK3␤ immunoblotting.

Enzyme-linked immunosorbent assay (ELISA)
IL-6 and IL-10 ELISA kits were purchased from Affymetrix/ Ebioscience (San Diego, CA). KC ELISA Duoset kits were purchased from R & D Biosystems (Minneapolis, MN). After treating MLE cells with TNF␣ (10 ng/ml), LPS (10 g/ml), or LTA (10 g/ml) for 6 or 16 h in serum-free medium, supernatants were collected. The quantitative assays were performed following protocols provided by the manufacturers.

Statistical analysis
Descriptive statistics were reported with mean Ϯ S.D. or S.E. as indicated. As our sample sizes of each experimental group were all less than 10, the sample sizes limited the ability to study a normal sample distribution. Thus, we employed appropriate non-parametric methods such as a Mann-Whitney U test and a Kruskal-Wallis equality-of-populations rank test to compare multiple groups within and between experiments. Also, a Wilcoxon-type test for trend was used where appropriate to check trends in data significance. Using these methods provided conservative analysis to determine statistical significance. All analyses were performed using one-way analysis of variance with Dunnett's post test using GraphPad Prism version 7.0 for Windows and Mac OS (GraphPad Software, La Jolla, CA).
Author contributions-T. S. performed most of the experiments, data analysis, and wrote the manuscript. J. W. and A. J. assisted with in vitro protein-binding assays and ELISA experiments. I. N. assisted with experiments. Y. Z. assisted with experiment design and data analysis. J. Z. and R. M. supervised the project and provided constructive review and edited the manuscript.