Introduction

Obesity has been recognized as one of the most serious public health problems [1]. An increased adipocyte number (hyperplasia) and/or size (hypertrophy) contributes to development of obesity and its related insulin resistance [2]. Both adipocyte number and size determine the lipid storage capacity of the body in humans. When the demand for lipid storage exceeds its capacity, the number of adipocytes increases through proliferation and/or adipocyte differentiation from pre-adipocytes [3, 4]. In vitro adipocyte differentiation from 3T3-L1 cells is mediated by a well-programed cascade of transcriptional events, which are induced by the activation of major transcription factors, including the CCAAT/enhancer binding protein family and peroxisome proliferator-activated receptor γ (PPARγ) [5, 6]. PPARγ is an essential and sufficient transcriptional regulator needed to differentiate adipocytes from their precursor cells [7, 8]. When PPARγ is activated, the expression of lipid synthesis- and differentiation-related genes is increased, including the expression of CCAAT/enhancer-binding protein α (C/EBPα), fatty acid synthase, lipoprotein lipase, and fatty acid-binding protein [9, 10]. Pre-adipocyte factor-1, a pre-adipocyte-specific inhibitor of adipogenesis, is transcriptionally suppressed by activation of PPARγ [11].

PPARγ is subjected to various post-translational modifications [12]. Phosphorylation, sumoylation, and poly-ubiquitination on specific regions of PPARγ are related to its transcriptional activity and stability [13, 14]. It has recently been reported that PPARγ acts as an E3 ubiquitin ligase through its cysteine (Cys) residue at the 139th position [15]. Its function as an E3 ubiquitin ligase, that induces the ubiquitination of substrates, is independent of its transcriptional activity. So far, nuclear factor-κB/p65 and mucin1 C-terminal subunit are known to be ubiquitinated and degraded by PPARγ [15, 16].

It has been reported that endoplasmic reticulum (ER) stress and ER resident selenoproteins are associated with adipocyte differentiation and PPARγ activation [17, 18]. When ER homeostasis is disrupted, misfolded proteins are accumulated in the ER, which subsequently activates the unfolded protein response (UPR) [19]. The expression of UPR-related-proteins, such as glucose-regulated protein 78 (GRP78), X-box binding protein 1 (XBP1), and ER-resident selenoproteins [selenoprotein S (SelS) and selenoprotein K (SelK)] are induced by ER stress. These ER stress response proteins are involved in restoration of ER stress through proper protein folding, ER-associated protein degradation (ERAD), or ER stress-mediated apoptosis [20,21,22]. XBP1, a transcription factor of UPR, is involved in the signaling pathway of adipocyte differentiation by up-regulating PPARγ transcription in the early phase of adipogenesis, which contributes to enhanced adipocyte differentiation [23]. SelS, also known as a valosin containing protein (VCP)-interacting membrane protein, is an ER membrane selenoprotein that contains a single selenocysteine (Sec, U) at position 188, which is the penultimate C-terminal residue. SelK is another ER membrane selenoprotein that also contains a single Sec at position 92 as the penultimate C-terminal residue [24]. Both SelS and SelK are important components of the ERAD complex and interact with the p97/VCP. It is known that both SelS and SelK regulate ERAD and play a role in cell survival [21, 22].

Although the mechanism is still unknown, a recent study suggests that SelS inhibits adipocyte differentiation and that SelS is degraded during conditioned medium induced-adipocyte differentiation from 3T3-L1 cells [18]. However, the role of SelK in adipocyte differentiation has not yet been studied. In this study, we investigated the role of both SelS and SelK and their degradation mechanism during adipocyte differentiation. The protein levels of these two ER-resident selenoproteins were dramatically decreased in the early phase of 3T3-L1 cell differentiation, while the level of PPARγ was increased. Thus, there was an inverse relationship between these two selenoproteins and PPARγ. We also found that SelS and SelK were ubiquitinated by the E3 ligase activity of PPARγ, and then degraded during adipogenesis. The degradation of both SelS and SelK was essential for adipogenesis. These results indicate that PPARγ-mediated ubiquitination and degradation of SelS and SelK is required for adipocyte differentiation.

Results

Protein levels of SelS and SelK are reduced during adipogenesis of 3T3-L1 pre-adipocytes

It has been reported that expression of SelS and SelK is increased by ER stress in various cell types [25,26,27]. Like GRP78, the promoter region of both SelS and SelK contains an ER stress response element, which is a consensus binding site for regulating the ER stress response [28, 29]. We previously reported that both SelS and SelK, components of the ERAD complex, interact with the p97/VCP protein and are involved in cell survival during ER stress [21, 22]. It has been suggested that ER stress-related proteins, such as XBP1 [23], protein kinase RNA-like ER kinase, inositol-requiring enzyme 1 alpha [20], and SelS are associated with adipogenesis and obesity [30]. During differentiation of 3T3-L1 pre-adipocytes induced by adipogenic cocktail containing dexamethasone (Dex) treatment, SelS is degraded; this degradation is necessary for adipogenesis [18]. However, the role of the ER-resident selenoproteins or their regulatory mechanism in adipocyte differentiation remains unclear. To investigate the role of SelS and SelK, we first measured their expression at both transcriptional and translational levels during adipogenesis. Adipocyte differentiation from 3T3-L1 cells was determined by monitoring intracellular lipid accumulation after staining the cells with oil red O (ORO) (Fig. 1a). We observed that the mRNA levels of PPARγ and C/EBPα, which are key regulators of adipogenesis were increased during adipocyte differentiation (Fig. 1b, c). The mRNA levels of ER stress-related genes, such as GRP78, spliced XBP1, SelS, and SelK were also increased (Fig. 1d–g). The protein level of GRP78 gradually increased from day 1 to day 8. Consistent with a previous report [31], PPARγ expression increased from day 1 to day 4, and then decreased (Fig. 1h). However, levels of both SelS and SelK significantly decreased from day 1 to 6, and then recovered on day 8 (Fig. 1h). These results show that even though the mRNA levels of SelS and SelK are increased during adipocyte differentiation, their protein levels are decreased; this decrease shows an inverse relationship with the levels of PPARγ. These results also suggest that both SelS and SelK are post-translationally regulated during adipogenesis.

Fig. 1
figure 1

Both SelS and SelK protein levels are decreased during 3T3-L1 early differentiation. a Differentiating cells were fixed at days 0, 1, 2, 4, 6, and 8. Intracellular lipid droplets in differentiated adipocytes were stained with ORO for image analysis (left panel). ORO-stained lipid droplets were extracted for quantification of lipid accumulation as described in the Materials and methods section (right panel). Differentiating cells were also harvested at days 0, 1, 2, 4, 6, and 8 for analysis of mRNA levels of PPARγ (b), C/EBPα (c), GRP78 (d), spliced XBP1 (sXBP1) (e), SelS (f), and SelK (g) by real-time RT-PCR. β-actin was used for normalization. h Differentiating cells were harvested at indicated days, and then cells were analyzed by western blotting (left panel). Whole cell lysates were detected by indicated antibodies. PPARγ expression is represented as a fold of induction of control (right panel). All graphs indicated the results from three independent experiments. The error bars represent ± S.D., and the p-values represent comparisons with the control (***p < 0.001 **p < 0.005, *p < 0.05, compared to day 0)

Full size image

Degradation of both SelS and SelK proteins is dependent on expression of PPARγ

Dex in an adipogenic cocktail is an essential requirement for differentiation of 3T3-L1 preadipocytes, and induces the degradation of SelS during adipogenesis [18]. To investigate how SelS is degraded, and whether SelK is also degraded during adipogenesis, differentiation from 3T3-L1 cells was induced using adipogenic medium in the presence or absence of Dex. As shown in Fig. 2a, both SelS and SelK were degraded in the presence of Dex. To test how the proteins were degraded, we first determined ubiquitination of the selenoproteins in the presence or absence of a proteasome inhibitor (MG132). Ubiquitination of both SelS and SelK was observed after treatment with MG132 (Fig. 2b, d). In the absence of MG132, their protein levels were significantly decreased. Although cells were treated with MG132, when cells were cultured without Dex, SelS and SelK, they were neither ubiquitinated nor showed any decrease in their protein levels (Fig. 2c, e). As shown in Fig. 2a–e, PPARγ was expressed in the presence of Dex, and the protein level of PPARγ was inversely related to the amount of SelS and SelK during adipogenesis (Fig. 1h). Recently, E3 ligase activity of PPARγ has been reported [15, 16]. Hence, we hypothesized that expression of PPARγ might promote ubiquitin-dependent protein degradation of both SelS and SelK, regardless of the presence of Dex. To examine this, the human colon cancer cell line (HT29) and the human breast adenocarcinoma cell line (MCF7) were used. It has been known that HT29 cells constitutively express relatively high amounts of PPARγ, whereas MCF7 cells express lower amounts than other cells [15, 32]. As shown in Fig. 2f, the protein level of PPARγ in HT29 cells was higher than that in MCF7 cells, whereas the mRNA levels of PPARγ were not different. When HT29 cells were transfected with siRNA of PPARγ (siPPARγ), the protein levels of both SelS and SelK were increased compared with control cells, whereas their mRNA levels remained unchanged (Fig. 2g). In addition, we also found that Flag-PPARγ-transfected MCF7 cells showed decreased protein levels of both SelS and SelK without change in their mRNA levels (Fig. 2h). Protein levels of SelS and SelK in MCF7 cells were unchanged in both untreated cells and in cells treated with MG132 for 12 h. However, their levels in HT29 cells were increased by MG132 treatment (Fig. 2i). To understand the relationship between the levels of these two selenoproteins and expression of PPARγ, the levels of SelS and SelK in MCF7 and HT29 cells were determined using a cycloheximide (CHX) chase assay. As shown in Fig. 2j, both SelS and SelK were not degraded in MCF7 cells, but were degraded in HT29 cells within 2 h. Since PPARγ is well known as a ligand-activated nuclear receptor [33], we investigated the effect of PPARγ ligand on the degradation of SelS and SelK. When HT29 cells were treated with rosiglitazone, a thiazolidinedione class PPARγ ligand, the mRNA level of C/EBPα was increased (Supplementary fig. S1a). However, protein levels of both SelS and SelK were not affected by the ligand (Supplementary fig. S1b and c). These results suggest that degradation of both SelS and SelK during adipogenesis is dependent on PPARγ expression regardless of its activation by ligand.

Fig. 2
figure 2

Degradation of both SelS and SelK proteins is dependent on PPARγ expression. a 3T3-L1 pre-adipocytes were incubated in adipogenic medium containing IBMX and insulin, in the presence and absence of Dex for indicated days. Whole cell lysates were detected by indicated antibodies. b, d 3T3-L1 pre-adipocytes were incubated in adipogenic medium containing IBMX, insulin and Dex for 1 day. These cells were then treated with or without 2 μM MG132 for 12 h. c, e 3T3-L1 pre-adipocytes were incubated in adipogenic medium containing IBMX, insulin, in the presence and absence of Dex for 1 day. These cells were treated with 2 μM MG132 for 12 h. Lysates and immunoprecipitation samples [anti-SelS (c), anti-SelK (e)] were analyzed by western blotting with the indicated antibodies. HC heavy chain, IP immunoprecipitation, Ub ubiquitin. f The indicated cells were lysed, and then whole cell lysates were analyzed by western blotting using indicated antibodies (left panel). Total RNAs isolated from indicated cells were subjected to RT-PCR analysis of PPARγ, SelS, SelK, and GAPDH mRNA (right panel). g HT29 cells were transfected with siPPARγ or control siRNA. Whole cell lysates were assessed by western blotting using indicated antibodies (left panel). Total RNAs isolated from HT29 cells transfected with siPPARγ or control siRNA were subjected to RT-PCR analysis of SelS, SelK, PPARγ, and GAPDH mRNA (right panel). h MCF7 cells were transfected with siPPARγ or control siRNA. Whole cell lysates were assessed by western blotting using indicated antibodies (left panel). Total RNA isolated from MCF7 cells transfected with Flag-PPARγ or pCMV-tag2b vector was subjected to RT-PCR analysis of SelS, SelK, PPARγ, and GAPDH mRNA (right panel). i The indicated cells were treated with 2 μM MG132 for 12 h. These cells were analyzed by western blotting. Whole cell lysates were detected by indicated antibodies. j MCF7 and HT29 cells were treated with 10 μg/mL CHX for the indicated times. Whole cell lysates were analyzed by western blotting with the indicated antibodies

Full size image

SelS and SelK are ubiquitinated and degraded by the E3 ligase activity of PPARγ

We investigated whether the reduced intracellular level of SelS and SelK proteins during adipogenesis was due to ubiquitination and degradation induced by the E3 ligase activity of PPARγ. To do this, the interactions between both selenoproteins and PPARγ were first determined using plasmids containing PPARγ, SelS, or SelK, as shown in Fig. 3a, b. Since the Cys residue at position 139 on PPARγ is an essential amino acid for E3 ligase activity [15], a mutant in which Cys was changed to alanine (Ala) was also constructed. Under basal conditions, Flag-PPARγ and Flag-C139A were bound to both endogenous SelS and SelK (Fig. 3c). However, ligand-binding domain (LBD) deletion mutants of PPARγ, Flag-PPARγ(ΔLBD) and Flag-C139A(ΔLBD), did not bind to either endogenous SelS or SelK. LBD is the essential domain for PPARγ to bind its substrate. The interaction between endogenous PPARγ and the selenoproteins was not impacted by PPARγ ligand-activation (Supplementary fig. S1c). To determine whether PPARγ interacted with SelS and SelK directly, a glutathione S-transferase (GST) pull-down assay was performed with purified GST, GST-tagged SelS and SelK. To increase the efficiency of purification, we deleted the transmembrane (TM) region on both SelS and SelK, generating GST-SelS(ΔTM) and GST-SelK(ΔTM), respectively (Fig. 3b). These purified proteins were expressed in Escherichia coli (Fig. 3d). After pre-incubating the purified proteins with MCF7 cell lysates containing overexpressed Flag-PPARγ or indicated mutants of PPARγ, each mixture was incubated with glutathione beads, followed by elution with reduced glutathione. Protein interaction was then determined by western blot analysis. As shown in Fig. 3e, f, GST-SelS(ΔTM) and GST-SelK(ΔTM) interacted with Flag-PPARγ, as well as Flag-C139A. However, neither GST-SelS(ΔTM) nor GST-SelK(ΔTM) interacted with the LBD deletion mutants of PPARγ. These results suggest that LBD is critical for direct interaction of PPARγ with these two selenoproteins, regardless of the E3 ligase active site of PPARγ.

Fig. 3
figure 3

PPARγ directly binds to SelS and SelK. a Organization of PPARγ is schematically represented. Flag is the N-terminal-tagged Flag. LBD is the ligand-binding domain region. The location of the Cys139 residue is indicated by a vertical bar, and the location of Cys139 changed to Ala is indicated by a dotted-vertical bar. b The above two constructs display the schematic representation of SelK organization, and the two constructs below represent SelS. TM is the transmembrane region. HA-TEV is the N-terminal-tagged HA-TEV, and GST is the N-terminal-tagged GST. The location of the Sec or the Sec changed to Cys residue is indicated by vertical bar. Coiled-coil is the coiled-coil domain, His is the N-terminal-tagged His. c MCF7 cells were transfected with Flag-PPARγ, Flag-C139A, Flag-PPARγ(ΔLBD), or Flag-C139A(ΔLBD) for 24 h. Lysates and immunoprecipitation samples (anti-Flag and anti-IgG) were analyzed by the indicated antibodies. Cell lysates (5% input) were used as positive control. IP immunoprecipitation, Ub ubiquitin. d GST and GST fusion proteins [GST-SelS(ΔTM), GST-SelK(ΔTM)] were purified and visualized by Coomassie Blue staining and western blotting (anti-GST, anti-SelS, and anti-SelK). CB coomassie blue staining. e, f Purified proteins were incubated with MCF7 cell lysates in which Flag-PPARγ, Flag-C139A, Flag-PPARγ(ΔLBD), or Flag-C139A(ΔLBD) was overexpressed, followed by GST pull-down using glutathione beads. Each mixture and lysate was analyzed by Western blotting with the indicated antibodies. PD GST pull-down

Full size image

Next, we determined whether the E3 ligase activity of PPARγ could induce ubiquitin/proteasome-dependent degradation of both SelS and SelK. As described above, HT29 cells produce relatively more PPARγ, and MCF7 cells produce less PPARγ compared to other cells (Fig. 2f). Both SelS and SelK proteins were not ubiquitinated and showed decreased degradation rate in the absence of PPARγ in HT29 cells (Fig. 4a, b). Ubiquitination of His-SelS or HA-SelK was observed in MCF7 cells transfected with Flag-PPARγ, but not in cells transfected with Flag-C139A (Fig. 4c, d). His-SelS or HA-SelK was significantly degraded in MCF7 cells expressing Flag-PPARγ. However, degradation of His-SelS or HA-SelK was not observed in MCF7 cells transfected with Flag-C139A (Fig. 4e, f). These results indicate that the E3 ligase-active form of PPARγ containing the Cys residue at the 139th position is necessary for ubiquitination and degradation of both SelS and SelK.

Fig. 4
figure 4

Both SelS and SelK are ubiquitinated by PPARγ. a HT29 cells were transfected with siPPARγ or control siRNA for 48 h, and then treated with 2 μM MG132 for 12 h. Lysates and immunoprecipitation samples (anti-SelS and anti-SelK) were analyzed by western blotting with the indicated antibodies. b HT29 cells were transfected with siPPARγ or control siRNA for 48 h. These cells were treated with 10 μg/mL CHX for indicated times. Whole cell lysates were analyzed by western blotting with the indicated antibodies. c MCF7 cells were co-transfected with His-SelS and Flag-PPARγ/Flag-C139A or His-SelS alone for 48 h. d MCF7 cells were co-transfected with HA-SelK and Flag-PPARγ/Flag-C139A or HA-SelK alone for 48 h. c, d These cells were treated with 2 μM MG132 for 12 h. Lysates and immunoprecipitation samples [anti-His (c), anti-HA (d)] were analyzed by western blotting with the indicated antibodies. HC heavy chain, IP immunoprecipitation, UB ubiquitin. e MCF7 cells were transfected with His-SelS for 16 h and then transfected with Flag-PPARγ or Flag-C139A. f MCF7 cells were transfected with HA-SelK for 16 h and then transfected with Flag-PPARγ or Flag-C139A. e, f At 32 h after the 1st transfection, cells were treated with 10 μg/mL CHX for indicated times. Whole cell lysates were analyzed by western blotting with the indicated antibodies

Full size image

To further confirm the E3 ligase activity of PPARγ on both SelS and SelK, ubiquitination analysis was carried out using an in vitro ubiquitin assay, as described in the Materials and methods section. Poly-ubiquitination of GST-SelS(ΔTM) or GST-SelK(ΔTM) was induced by Flag-PPARγ, but not by Flag-C139A even in the presence of ATP and ubiquitin (Ub) (Fig. 5a, b). Poly-ubiquitination of GST-SelS(ΔTM) or GST-SelK(ΔTM) was observed during PPARγ expression. However, poly-ubiquitination was not detected in the PPARγ knockdown condition (Fig. 5c, d). Taken together we suggest that PPARγ interacts with both SelS and SelK and that these selenoproteins are ubiquitinated and degraded by the E3 ligase activity of PPARγ.

Fig. 5
figure 5

E3 ligase activity of PPARγ is required for SelS and SelK ubiquitination in vitro. a, b MCF7 cells were transfected with Flag-PPARγ or Flag-C139A. Lysates and purified GST fusion proteins [GST-SelS(ΔTM) (a), GST-SelK(ΔTM) (b)] were incubated with 2 mM ATP and 10 μg ubiquitin in the absence or presence of Flag-PPARγ or Flag-C139A, as indicated. Reaction mixtures were boiled in 0.5% SDS to disrupt protein interaction, followed by GST pull-down using glutathione beads. Lysates and GST pull-down samples were analyzed by western blotting using indicated antibodies. c, d HT29 cells were transfected with siPPARγ or control siRNA. Lysates and purified GST fusion protein [GST-SelS(ΔTM) (c), GST-SelK(ΔTM) (d)] were incubated with 2 mM ATP and 10 μg ubiquitin in the absence and presence of siPPARγ or control siRNA, as indicated. Reaction mixtures were boiled in 0.5% SDS to disrupt protein interaction, followed by GST pull-down using glutathione beads. Lysates and GST-pull-down samples were analyzed by western blotting using indicated antibodies. PD GST pull-down, IB immunoblot, -(Ub)n conjugated poly-ubiquitin

Full size image

Lys150 of SelS is required for PPARγ-mediated ubiquitination

Next, we tried to identify the ubiquitination site of SelS mediated by PPARγ. To do this, we determined the region(s) of SelS that bind to PPARγ using three truncation mutants of His-SelS as shown in Fig. 6a. All mutants could interact with Flag-PPARγ (Fig. 6b). However, ubiquitination only occurred in MCF7 cells transfected with His-SelS or His-SelS[1–167] (Fig. 6c). This result indicates that lysine (Lys) residues at 150th, 152nd or 156th position on the cytosolic tail of SelS may be ubiquitinated by PPARγ. To identify the specific site for ubiquitination of SelS, each Lys residue was changed to arginine (Arg). Flag-PPARγ-mediated degradation and ubiquitination levels of His-SelS[K152R] and His-SelS[K156R] were same as His-SelS; however those levels of His-SelS[K150R] and His-SelS[K150, 152R] were significantly decreased (Fig. 6d, e). These data suggest that the Lys residue at the 150th position of SelS is poly-ubiquitinated by PPARγ and this ubiquitination is required for protein degradation.

Fig. 6
figure 6

Lys150 of SelS is required for PPARγ-mediated ubiquitination. a Amino acid sequence of SelS is represented. The following elements of residues and domains are marked: italic letters (Lys residues), under lined letters (TM region). The construction of SelS mutants, His-SelS(1–122), His-SelS(1–144), and His-SelS(1–167) has been described in the Materials and methods section. b MCF7 cells were co-transfected with Flag-PPARγ and His-SelS or the truncation mutants of SelS for 48 h. Lysates and immunoprecipitation samples (anti-His and anti-IgG) were analyzed using the indicated antibodies. Cell lysates (5% input) were used as positive control. c Transfected MCF7 cells, under same conditions as (b), were treated with 2 μM MG132 for 12 h. Lysates and immunoprecipitation samples (anti-His) were analyzed by western blotting with the indicated antibodies. d MCF7 cells were transfected with Flag-PPARγ, His-SelS or Lys-to-Arg mutants of SelS for 48 h. Whole cell lysates were analyzed by western blotting using indicated antibodies. e Transfected MCF7 cells, under same conditions as (d), were treated with 2 μM MG132 for 12 h. Lysates and immunoprecipitation samples (anti-His) were analyzed by western blotting with the indicated antibodies. HC heavy chain, IP immunoprecipitation

Full size image

Both Lys47 and Lys48 of SelK are required for PPARγ-mediated ubiquitination

In addition to SelS, we also tried to identify the ubiquitination sites of SelK induced by the E3 ligase activity of PPARγ. To do this, we constructed two truncation mutants of GFP-SelK as shown in Fig. 7a. Both SelK truncation mutants interacted with Flag-PPARγ (Fig. 7b) and were ubiquitinated in MCF7 cells transfected with Flag-PPARγ (Fig. 7c). These data suggest that a region in the N-terminal 55 amino acids of SelK is required for interaction with PPARγ, and that Lys residues at 47th or 48th positions in the cytosolic tail of SelK may be the putative ubiquitination site. To identify the specific site for ubiquitination of SelK, each Lys residue was mutated into Arg. Like wild-type SelK, both the mutants, HA-SelK[K47R] and HA-SelK[K48R], were ubiquitinated and degraded by Flag-PPARγ. However, Flag-PPARγ-mediated ubiquitination and degradation of HA-SelK[K47, 48R] in which both Lys residues were mutated into Arg were significantly decreased (Fig. 7d, e). These data suggest that both Lys47 and Lys48 of SelK are poly-ubiquitinated by PPARγ, and this ubiquitination is required for protein degradation.

Fig. 7
figure 7

Both Lys47 and Lys48 of SelK are required for PPARγ-mediated ubiquitination. a Amino acid sequence of SelK is represented. The following elements of residues and domains are marked: italic letters (Lys residues), under lined letters (TM region). The construction of SelK mutants, GFP-SelK[1–55] and GFP-SelK[1–75] has been described in the Materials and methods section. b MCF7 cells were co-transfected with Flag-PPARγ and GFP-SelK or the truncation mutants of SelK for 48 h. Lysates and immunoprecipitation samples (anti-GFP and anti-IgG) were analyzed using the indicated antibodies. Cell lysates (5% input) were used as positive control. c Transfected MCF7 cells, under same conditions as (b), were treated with 2 μM MG132 for 12 h. Lysates and immunoprecipitation samples (anti-GFP) were analyzed by western blotting with the indicated antibodies. d MCF7 cells were co-transfected with Flag-PPARγ and HA-SelK or Lys-to-Arg mutants of SelK for 48 h. Whole cell lysates were analyzed by western blotting using indicated antibodies. e Transfected MCF7 cells, under same conditions as (d), were treated with 2 μM MG132 for 12 h. Lysates and immunoprecipitation samples (anti-HA) were analyzed by western blotting with the indicated antibodies. HC heavy chain, IP immunoprecipitation

Full size image

PPARγ-mediated ubiquitination and degradation of both SelS and SelK are required for 3T3-L1 differentiation

The effect of ubiquitination and degradation of the selenoproteins induced by PPARγ on adipogenesis was examined using 3T3-L1 pre-adipocytes transfected with siPPARγ or control siRNA. Compared to control cells, protein levels of both SelS and SelK in PPARγ knockdown 3T3-L1 cells were high, whereas intracellular lipid accumulation was very low (Fig. 8a, b). We also found that mRNA levels of SelS and SelK were unchanged in siPPARγ-transfected 3T3-L1 cells, but those of adipogenesis- and lipid synthesis-related genes encoding PPARγ and fatty acid synthase were decreased (Fig. 8c). These results indicate that PPARγ is an essential protein for adipocyte differentiation, as previously reported [34], and that it has the ability to degrade both SelS and SelK proteins during adipogenesis. To further investigate the effect of PPARγ-mediated SelS and SelK degradation on adipocyte differentiation, we constructed lentivirus plasmids (Fig. 8d). Compared to control cells, intracellular lipid accumulation in His-SelS- or His-SelS[K150, 152R]- overexpressing cells was decreased. The intracellular lipid accumulation was restored by lentivirus-mediated Flag-PPARγ expression in pLenti-His-SelS-transduced cells but not in His-SelS[K150, 152R]-overexpressing cells (Fig. 8e and Supplementary fig. S2a). Protein levels of both endogenous SelS and overexpressed His-SelS were decreased by expression of Flag-PPARγ during adipocyte differentiation. However, the level of His-SelS[K150, 152R] protein was not decreased by expression of Flag-PPARγ (Fig. 8f). And, both pLenti-His-SelK- and pLenti-His-SelK[K47, 48R]-transduced cells showed decreased intracellular lipid accumulation. When Flag-PPARγ was expressed, the decreased intracellular lipid accumulation was recovered more in lentivirus-mediated His-SelK-overexpressing cells than in His-SelK[K47, 48R]-overexpressing cells (Fig. 8g and Supplementary fig. S2b). Endogenous SelK and overexpressed His-SelK were decreased by expression of Flag-PPARγ. However, this decrease did not occur in pLenti-His-SelK[K47, 48R]-transduced cells (Fig. 8h). We also observed that mRNA levels of PPARγ were decreased in 3T3-L1 cells overexpressing His-SelS or His-SelK during adipogenesis, as well as adiponectin which is a lipid synthesis marker (Supplementary fig. S2c and d). These results suggest that accumulation of either SelS or SelK inhibits adipocyte differentiation, and PPARγ-mediated degradation of SelS and SelK is an essential requirement for adipocyte differentiation.

Fig. 8
figure 8

PPARγ-mediated degradation of both SelS and SelK is essentially required for 3T3-L1 differentiation. a 3T3-L1 cells were transfected with siPPARγ or control siRNA. These pre-adipocytes were differentiated for 2 days. Whole cell lysates were analyzed by western blotting using indicated antibodies. b 3T3-L1 cells were transfected with siPPARγ or control siRNA. These pre-adipocytes were differentiated for 6 days, and then cells were fixed. Intracellular lipid droplets in differentiated adipocytes were stained with ORO for image analysis (bottom panel). Quantification of lipid accumulation in the ORO-stained lipid droplets was performed as described in the Materials and methods section (upper panel). c The cells differentiated under same conditions as (b) were also harvested at day 6 for analyzing the mRNA levels of PPARγ, SelS, SelK, and fatty acid synthase by quantitative RT-PCR. β-actin was used for normalization. FAS fatty acid synthase. d Organization of cloned lentiviral plasmids is schematically represented. e, f 3T3-L1 pre-adipocytes were co-transduced with lentivirus-containing control vector (pLenti-EGFP-neo) or Flag-PPARγ (pLenti-Flag-PPARγ), and His-SelS (pLenti-His-SelS) or His-SelS[K150, 152R] (pLenti-His-SelS[K150, 152R]). g, h 3T3-L1 pre-adipocytes were co-transduced with lentivirus-containing control vector (pLenti-EGFP-neo) or Flag-PPARγ (pLenti-Flag-PPARγ), and His-SelK (pLenti-His-SelK) or His-SelK[K47, 48R] (pLenti-His-SelK[K47, 48R]). These cells were differentiated for 2 (f, h) or 6 (e, g) days. e, g Intracellular lipid droplets in differentiated adipocytes were stained with ORO for image analysis (left panels). Quantification of lipid accumulation in the ORO-stained lipid droplets was performed as described in the Materials and methods section (right panels). f, h Whole cell lysates were analyzed by western blotting using indicated antibodies. Asterisk (*) indicates endogenous SelS (f) or SelK (h). All graphs indicate the results from three independent experiments. The error bars represent ± S.D., and the p-values represent comparisons with each control (***p < 0.001 **p < 0.005, *p < 0.05)

Full size image

Discussion

It has been reported that some selenoproteins, such as selenoprotein W, selenoprotein P, and selenoprotein O have a function in various cellular processes, such as myogenesis [35, 36], bone remodeling and adipogenesis [37, 38], chondrogenesis [39]. It has also been shown that some selenoproteins have a function in maintaining cellular redox homeostasis [24]. In this study, we investigated the function of two selenoproteins localized on the ER membrane, SelS and SelK, in adipogenesis. These selenoproteins contain a single TM domain and a C-terminal end that faces the cytosol (Fig. 3b). The cytosolic tails of the proteins are highly unstructured. Both SelS and SelK are involved in ER stress signaling through p97/VCP-dependent ERAD, and proline residues in the C-terminal region of SelS or SelK are essential for their interaction with p97/VCP [21, 22, 29]. ER stress promotes adipocyte differentiation and development of obesity [18, 20, 23]. A recent study suggested that SelS might play an inhibitory role in adipogenesis [18]. However, the inhibitory mechanism of SelS is still unknown, and the role of SelK in adipogenesis has not been studied at all. In this study, we found that the protein expression of both SelS and SelK was inversely proportional to that of PPARγ (Fig. 1h). We also found that overexpression of SelS or SelK in pre-adipocytes resulted in impaired adipogenesis (Fig. 8e, g, and Supplementary fig. S2c and d). These results imply that SelS and SelK are novel proteins that may play a role in inhibiting adipocyte differentiation. Both SelS and SelK are known as negative regulators of the ER stress signaling pathway [22]. It has been demonstrated that expression of ER stress-related selenoproteins plays a critical role in the expression of XBP1 [21, 26], which is an ER stress-induced transcription factor required for initiation of the early phase of adipogenesis [23, 40]. Transcription of PPARγ is regulated by ER stress-inducible spliced XBP1, which stimulates adipogenesis by up-regulating PPARγ expression [23]. Therefore, it would be plausible to assume that the anti-adipogenesis functions of SelS and SelK are likely to act through modulation of ER stress and UPR during adipocyte differentiation.

PPARγ is the master regulator of adipogenesis and is abundantly expressed in adipose tissue. Two isoforms of PPARγ have been identified, PPARγ1 and PPARγ2, which are generated by alternative splicing and differential promoter usage. Expression of PPARγ2 is specific to adipose tissue, whereas PPARγ1 is expressed in many tissues such as intestinal tract, liver, lung, and the immune system under physiological conditions [14, 41,42,43]. Expression of PPARγ is controlled at the transcriptional and post-translational levels [5, 44]. PPARγ is a short-lived protein, which is controlled by post-translational modifications. It has been reported that PPARγ is poly-ubiquitinated and degraded in a proteasome-dependent manner [13, 45]. Moreover, it has also been reported that PPARγ functions as an E3 ligase to induce ubiquitination of substrates, which is independent of its transcriptional activity [15, 16]. In this study, we first report that PPARγ functions as an E3 ligase to degrade SelS and SelK through poly-ubiquitination at Lys150 and Lys47/Lys48, respectively (Figs. 5, 6, 7). We also found that ubiquitination and degradation of both SelS and SelK through the E3 ligase activity of PPARγ are independent of its activation by ligand (Supplementary fig. S1), and are required for adipocyte differentiation (Fig. 8). It has been known that Lys48-linked poly-ubiquitination are the most abundant in cells [46, 47] and thought to be the major signal for proteasome-mediated degradation [48, 49]. PPARγ also induced Lys-48-linked ubiquitination and degradation of p65 [15]. It has been reported that E3 enzymes regulate conjugate type of their substrates specifically [50]. Therefore, it is presumed that the Lys48-linked poly-ubiquitination of SelS and SelK may be induced by PPARγ during adipogenesis.

Recently, SelS has been suggested to be involved in neurodegenerative disorders, such as Alzheimer disease, by regulating ER stress [27]. It has also been shown that SelK regulates Ca2+ flux during immune responses [51]. In this study, we report a new function of SelS and SelK in inhibiting adipogenesis, and suggest that these two ER-resident selenoproteins are potential novel therapeutic targets for obesity (Fig. 8e, g). It has been reported that obesity is linked to cancer [52]. Using the leptin-deficient (ob/ob) mouse which is the model of type 2 diabetes with obesity [53], we observed the protein level of PPARγ was significantly increased in liver of ob/ob mice compared to wild type, whereas those of both SelS and SelK were decreased (Supplementary fig. 3a). To understand the effect of SelS and SelK expression on human cancer, we performed Kaplan–Meier survival analysis based upon data generated by the Gene Expression Omnibus on an online platform [54, 55]. The results showed that lower expression of SelS or SelK was associated with the poor prognosis of gastric cancer patients (supplementary Fig. S3b, c).

While we report for the first time the degradation of SelS and SelK by E3 ligase activity of PPARγ during adipogenesis, the inhibitory mechanism of SelS and SelK during adipogenesis still remains obscure. Further studies on the inhibitory mechanism of SelS and SelK during adipogenesis may provide meaningful candidates for therapeutic intervention in obesity.

Materials and methods

Cell culture and adipocyte differentiation

The human colon cancer cell line (HT29) and human breast adenocarcinoma cell line (MCF7) were cultured in RPMI containing 10% fetal bovine serum (FBS, Hyclone, GE healthcare Life Sciences, UT, USA) at 37 °C in a humidified 5% CO2 atmosphere. Adipocytes were differentiated from pre-adipocytes (3T3-L1) [18]. 3T3-L1 cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) with 10% fetal calf serum (Gibco, Invitrogen, Carlsbad, CA, USA) at 37 °C in a humidified 5% CO2 atmosphere. At day 0, when 3T3-L1 pre-adipocytes were at two-days post-confluence, differentiation was induced with DMEM containing 10% FBS, 2 μg/mL insulin (Sigma-Aldrich, St. Louis, MO, USA), 115 μg/mL 3-isobutyl-1methylxanthine (IBMX, Sigma-Aldrich) and 5 μM Dex (Sigma-Aldrich) for 2 days. Medium was then changed to DMEM containing 10% FBS and 2 μg/mL insulin (Sigma-Aldrich), and incubated for 2 additional days. The cells were then cultured in DMEM with 10% FBS until indicated days. For ligand-dependent activation of PPARγ, rosiglitazone (Sigma-Aldrich) was treated to 3T3-L1 or HT29 or MCF7 cells for 24 h.

Oil red O staining

To determine the intracellular lipid accumulation in differentiated 3T3-L1 cells, ORO staining was performed according to the method described by Kim et al. [18], with a slight modification. Adipocytes were fixed with 4% para-formaldehyde for 24 h at 24 °C and then incubated with fresh ORO solution (Sigma-Aldrich) for 15 min. The ORO-stained cells were scanned using EPSON PERPECTION 3200 scanner. Quantification of ORO-stained intracellular lipid was done by measuring the optical density of isopropanol-extracted ORO dye from ORO-stained cells at 500 nm.

Isolation of RNA, RT-PCR, and real-time RT-PCR

Total RNA was extracted from the cells with TRIzol (Invitrogen) according to the manufacturer’s instructions. Reverse transcription was performed to generate cDNA using 1 μg of isolated RNA and a SuperScript III reverse transcriptase kit (Invitrogen). RT-PCR analysis of the resulting cDNA preparation was performed using PCR Premix, Sapphire (Super Bio Co., Seoul, Korea) [56]. The PCR products were loaded on 2% agarose gels and visualized by ethidium bromide staining. Real-time RT-PCR analysis of the resulting cDNA preparation was performed using the LightCycler 480 SYBR Green I Master (Roche Diagnostics, Manheim, Germany) [36]. The sequences of primers used in RT-PCR and real-time RT-PCR were as follows (forward and reverse, respectively): mouse SelS, 5′-GAG GCT TTA GCA GCT GCT CG-3′, 5′-GTC AGA GGG TTA TAA CCA CCT CC-3′; human SelS, 5′-GAA GAA CTA AAT GCG CAA GTT G-3′ and 5′-GCG TCC AGG TCT CCA GGA GCA AGC-3′; mouse SelK, 5′-GTT TAC ATC TCG AAT GGT CAG G-3′ and 5′-CC ACC AGC CAT TGG AGG AGG GC-3′; human SelK, 5′-GTT TAC ATC TCG AAC GGA CAA G-3′ and 5′-CC ACC AGC CAT TGG AGG GGG AC-3′; mouse PPARγ, 5′-CCC AAT GGT TGC TGA TTA CAA AT-3′ and 5′-CTA CTT TGA TCG CAC TTT GGT ATT CT-3′; human PPARγ, 5′-GGG TGA AAC TCT GGG AGA TTC-3′ and 5′-CT TGT GAA TGG AAT GTC TTC-3′; mouse C/EBPα, 5′-CCC ACT TGC AGT TCC AGA T-3′ and 5′-CTG TTC TTG TCC ACC GAC TT-3′; mouse GRP78; 5′-CGG ACG CAC TTG GAA TGA C-3′ and 5′-AAC CAC CTT GAA TGG CAA GAA-3′; mouse spliced-XBP1; 5′-CTG AGT CCG CAG CAG GT-3′ and 5′-TGT CAG AGT CCA TGG GAA GA-3′; mouse adiponectin, 5′-GAT GCA GGT CTC TTG GTC CTA A-3′ and 5′-GGC CCT TCA GCT CCT GTC A-3′; mouse fatty acid synthase, 5′-GGC CCC TCT GTT AAT TGG CT-3′ and 5′-GGA TCT CAG GGT TGG GGT TG-3′; human glyceraldehyde-3-phosphate dehydrogenase (GAPDH), 5′-CTG CAC CAC CAA CTG CTT AGC-3′ and 5′-CTT CAC CAC CTT CTT GAT GTC-3′; mouse β-actin, 5′-ATG CCC TGA GGC TCT TTT CC-3′ and 5′-TGC TAG GAG CCA GAG CAG TC-3′. The real-time RT-PCR data were normalized to mouse β-actin.

Antibodies and immunoblot analysis

The cells were lysed as described in our previous study [57]. Immunoblot analysis was performed as described previously [21]. Antibodies were obtained from the following sources: the anti-His and anti-HA antibodies were from Applied Biological Materials (ABM, Richmond, BC, Canada), the anti-Flag and anti-SelK antibodies were obtained from Sigma-Aldrich, the anti-PPARγ antibody was obtained from Cell Signaling Technology (Danvers, MA, USA), and the anti-GST, anti-Ub, anti-β-actin, and anti-GRP78 antibodies were obtained from Santa Cruz Biotechnology (Santa Cruz, CA, USA). The anti-SelS antibody was prepared as described previously [21].

Cycloheximide chase assay

To determine the degree of SelS, SelK, His-SelS, and HA-SelK degradation, CHX chase analysis was performed according to the method described by Ballar et al. [58], with slight modification. CHX was purchased form Sigma-Aldrich.

Construction of SelS, SelK and PPARγ mutants

The human PPARγ2 plasmid, Flag-PPARγ was a gift from Dr. Jesang Ko (Korea University) [59]. Point mutations in Flag-PPARγ at Cys139 to Ala (Flag-C139A) and truncation mutations in Flag-PPARγ and Flag-C139A at their LBD were generated by site-directed mutagenesis reactions using QuikChange site-directed mutagenesis kit according to the instructions of the manufacturer (Stratagene, CA, USA). These plasmids were named Flag-PPARγ(ΔLBD) and Flag-C139A(ΔLBD), respectively. The human SelS plasmid His-SelS (Sec188 changed to Cys) and the human SelK plasmid HA-TEV-SelK, named HA-SelK, were constructed as described previously [21, 22]. His-SelS was then used as a template DNA for the following SelS mutants. His-SelS[1–122], His-SelS[1–144], and His-SelS[1–167] were constructed as described previously [21]. Point mutation in His-SelS at Lys150, Lys152, Lys156, and double point mutation at Lys150 with Lys152 to Arg were generated by site-directed mutagenesis reactions using QuikChange site-directed mutagenesis kit according to the manufacturer’s instructions (Stratagene). These plasmids were named His-SelS[K150R], His-SelS[K152R], His-SelS[K156R], and His-SelS[K150,152R], respectively. Lys residues, positioned at 47th, 48th, and both 47th and 48th in HA-SelK, were changed to Arg following QuikChange site-directed mutagenesis protocol (Stratagene). These plasmids were named HA-SelK[K47R], HA-SelK[K48R], and HA-SelK[K47,48R], respectively. The human SelK plasmid GFP-SelK, named GFP-SelK, was constructed as described previously [22]. This plasmid was then used as a template DNA for following truncation mutants. The primers used for constructing plasmids that were designed for GFP-SelK[1–55] and GFP-SelK(1-55) were as follows: GFP-SelK[1–55] forward, 5′-GC TAT GGA AAC TCA TAA GTC GAC CCG GGC GGC CGC TTT CAT-3′ and reverse, 5′-CGG GTC GAC TTA TGA GTT TCC ATA GCT TCT TCT TTT TTT CAC-3′; GFP-SelK[1–75] forward, 5′-CGA AGA ATG GGT TAA GTC GAC CCG GGC GGC CGC TTT CAT-3′ and reverse, 5′-CGG GTC GAC TTA ACC CAT TCT TCG GGG AGG GTT TCC TGG-3′. All SelS, SelK and PPARγ mutants were confirmed by sequencing (Macrogen, Seoul, Korea).

Construction and purification of the GST fusion SelS and SelK mutants

Deletion mutations in His-SelS at the TM domain (amino acids from 26th to 48th), and in HA-SelK at the TM domain (amino acids from 20th to 42nd) and at Sec92 changed to Cys were generated by site-directed mutagenesis reactions using QuikChange site-directed mutagenesis kit according to the manufacturer’s instructions (Stratagene). TM regions and hydrophobic domains are deleted from His-SelS and HA-SelK to allow the production and purification of GST fusion proteins [60]. These TM-deleted plasmids were used as templates for GST fusion SelS and SelK. The primers used for constructing plasmids that were designed for GST-SelS(ΔTM) and GST-SelK(ΔTM) were as follows: GST-SelS(ΔTM) forward, 5′-GC GGA TCC ATG GAA CGC CAA GAG GAG TCT CTG-3′ and reverse, 5′-CG CTC GAG TTA GCC GCA TCC GCC AGA TGA CGG-3′; GST-SelK(ΔTM) forward, 5′-GC GGA TCC ATG GTT TAC ATC TCG AAC GG -3′ and reverse, 5′-GC CTC GAG TTA CCT TCC GCA TCC ACC AGC C-3′. The PCR products were cloned into the BamHI and XhoI sites of pGEX-4T-3 vector. These plasmids were designated GST-SelS(ΔTM) and GST-SelK(ΔTM). The plasmids were confirmed by sequencing (Macrogen). The GST, GST-SelS(ΔTM), and GST-SelK(ΔTM) proteins were expressed in E. coli with 1 mM isopropyl 1-thio-β-D-galactopyranoside induction for 6 h at 18 °C. The protein was lysed by sonication. The lysis buffer contained 50 mM Tris-HCl (pH 8.0), 120 mM NaCl, 0.5% Nonidet P-40, 4 μg/mL leupeptin, and 1 mM PMSF. The prepared cell lysates were incubated with glutathione beads (Invitrogen) for 2 h at 4 °C. The GST beads were washed with wash buffer containing 20 mM Tris-HCl (pH 8.0), 100 mM NaCl, 1 mM EDTA, 0.05% sodium dodecyl sulfate (SDS), and 0.5% Nonidet P-40 and then eluted with elution buffer containing 50 mM Tris-HCl (pH 8.0), 20 mM KCl, 1 mM DTT, and 20 mM glutathione for 10 min at 37 °C.

RNA interference

The siPPARγ used in this study was obtained by Ambion (Thermo Fisher Scientific, Waltham, MA, USA; Catalog number: #AM51331). The siPPARγ was a validated siRNA for human (NM_005037) and mouse (NM_011146) PPARγ. The sequence for siPPARγ was as follows: sense, 5′-GGG CGA UCU UGA CAG GAA Att-3′ and anti-sense, 5′-UUU CCU GUC AAG AUC GCC Ctc-3′. A stealth negative control siRNA was obtained from Invitrogen. HT29 and 3T3-L1 cells were transfected with siRNAs using a microporator (Neon Transfection System, Invitrogen).

Transfection

For transfection, 3 × 105 MCF7 cells were seeded in 60 mm dishes. At 16 h after seeding, these cells were transfected with plasmids using Lipofectamine 2000 transfection reagent (Invitrogen) according to the manufacturer’s instructions [21, 22].

Immunoprecipitation and ubiquitin assay

For immunoprecipitation, cells were transfected with siRNA or plasmids. To detect poly-ubiquitination, these cells were treated with 2 µM MG132 for 12 h before harvest. The lysates were mixed with antibodies for 16 h at 4 °C. Immune complexes were incubated with protein G-agarose for 90 min at 4 °C, then washed with lysis buffer three times, and boiled with SDS sample buffer for 5 min [21, 36]. The samples were loaded onto SDS-polyacrylamide gels and subjected to electrophoresis, transferred to polyvinylidene fluoride membranes, and incubated with primary antibody for 16 h at 4 °C. After further incubation with a horseradish peroxidase-conjugated secondary antibody for 1 h at room temperature, immunoreactive bands were visualized using a West Pico enhanced ECL detection kit (Thermo Fisher Scientific).

GST pull-down assay

MCF7 cells were transfected with Flag-PPARγs in 60 mm dishes. The cells were lysed with lysis buffer [22]. After the purification of GST, GST-SelS(ΔTM), and GST-SelK(ΔTM) proteins as described above, the purified GST proteins were pre-incubated with the cell lysates and rotated for 16 h at 4 °C. Glutathione beads were added to the mixtures and rotated for an additional 30 min at room temperature. The beads were washed and then eluted. The eluted products were visualized using Coomassie Blue staining or western blotting.

In vitro ubiquitin assay

For poly-ubiquitination formation assay, MCF7 or HT29 cells were transfected with Flag-PPARγ or siPPARγ in 60 mm dishes. The cells were lysed with lysis buffer. After the purification of GST, GST-SelS(ΔTM), and GST-SelK(ΔTM) proteins as described above, the cell lysates and purified proteins were incubated in reaction buffer [50 mM NaCl, 0.5 mM dithiotheritol, and 40 mM Tris-HCl (pH 7.4)] using Amicon Ultra Centrifugal filler device (Amicon Ultra-0.5, Millipore, MA, USA). GST fusion proteins were incubated with 2 mM ATP (Fermantas, Vilnius, Lithuania, R0441) and 5 μM Ubiquitin (Sigma-Aldrich, U6235) in transfected cell lysates in 500 μL of reaction buffer for 3 h at 37 °C. After incubation, reaction mixtures were boiled in 0.5% SDS to disrupt any protein interactions [13] and then glutathione beads were added to the reaction mixtures, which were then rotated for an additional 30 min at 24 °C. The beads were washed and then eluted. The eluted products were visualized using western blotting.

Construction of lentiviral plasmids, production of lentivirus, and transduction

His-SelS and His-SelS[K150, 152R] plasmids were changed from Sec188 to Cys. His-SelK and His-SelK[K47, 48R] plasmids were changed from Sec92 to Cys, and were constructed as described previously [21]. Flag-PPARγ was constructed as described above. These plasmids were used as templates for constructing the lentiviral vector (pLenti-EGFP-neo; Clontech, Takara Bio, CA, USA). The GFP expression region in pLenti-EGFP-neo vector was replaced to His-SelSs or His-SelKs or Flag-PPARγ using XbaI and SalI sites. The primers for these lentiviral plasmids were as follows (forward and reverse, respectively): His-SelSs (His-SelS, His-SelS[K150, 152R]), 5′-GCT CTA GAG ATG GGG GGT TCT CAT CAT CAT CAT CAT CAT GG-3′ and 5′-GC GTC GAC TTA GCC GCA TCC GCC AGA TGA CGG GCC TCT GCG-3′; His-SelKs (His-SelK, His-SelK[K47, 48R]), 5′-GCT CTA GAG ATG GGG GGT TCT CAT CAT CAT CAT CAT CAT GG-3′ and 5′-GC GTC GAC TTA CCT TCC GCA TCC ACC AGC CAT TGG AGG GGG-3′; Flag- PPARγ, 5′-GCT CTA GAG ATG GAT TAC AAG GAT GAC GAC GAT AAG G-3′ and 5′-GC GTC GAC CTA ATA CAA GTC CTT GTA GAT CTC CTG G-3′. The PCR products were cloned into the XbaI and SalI sites of pLenti-EGFP-neo. These plasmids were designated pLenti-His-SelS, pLenti-His-SelS[K150,152R], pLenti-His-SelK, pLenti-His-SelK[K47,48R], and pLenti-Flag-PPARγ, respectively. The cloned-pLenti-viral vectors or pLenti-EGFP-neo were transfected into HEK293T cells with pCMV-VSV-G and Δ8.9 (Clontech) using Lipofectamine 2000 transfection reagent (Invitrogen). The viruses were harvested every 24 h from 24 to 72 h post-transfection with cell culture medium. The viruses in the culture supernatants were precipitated overnight with 10% poly(ethylene glycol) 8000 (Sigma-Aldrich) and 0.4 M NaCl and then centrifuged at 10,000×g for 10 min [61]. The 3T3-L1 cells were transduced with these lentiviruses in DMEM-containing 10% fetal calf serum (Gibco, Invitrogen) and 8 μg/mL of polybrene (Sigma-Aldrich) for 24 h, and then the cells were differentiated for indicated days.

Animal tissues

Liver tissues of wild-type and ob/ob mice were kindly provided by Dr. Seung-Hoi Koo (Korea University). The liver tissues were lysed, and then subjected to immunoblot analysis [21].

Online gastric cancer tissue database and data retrieval

Overall survival of gastric cancer patients was analyzed using the online mRNA expression database developed by Gyorffy B (http://kmplot.com) [54]. The data were retrieved from the gene expression omnibus (GEO). The two patient cohorts were compared by a Kaplan–Meier survival plot, and the hazard ratio with 95% confidence intervals and log rank p value were calculated. p-Values <0.05 were considered significant.

Data analysis and statistics

All data in this study are represented as the means and standard deviations of the control value. Statistical comparisons from at least three independent experiments were determined using Student’s t-tests. p-Values <0.05 were considered significant.