Von Hippel-Lindau regulates interleukin-32β stability in ovarian cancer cells

Hypoxia-induced interleukin-32β (IL-32β) shifts the metabolic program to the enhanced glycolytic pathway. In the present study, the underlying mechanism by which hypoxia-induced IL-32β stability is regulated was investigated in ovarian cancer cells. IL-32β expression increased under hypoxic conditions in ovarian cancer cells as it did in breast cancer cells. The amount of IL-32β was regulated by post-translational control rather than by transcriptional activation. Under normoxic conditions, IL-32β was continuously eliminated through ubiquitin-dependent degradation by the von-Hippel Lindau (VHL) E3 ligase complex. Oxygen deficiency or reactive oxygen species (ROS) disrupted the interaction between IL-32β and VHL, leading to the accumulation of the cytokine. The fact that IL-32β is regulated by the energy-consuming ubiquitination system implies that it plays an important role in oxidative stress. We found that IL-32β reduced protein kinase Cδ (PKCδ)-induced apoptosis under oxidative stress. This implies that the hypoxia- and ROS-stabilized IL-32β contributes to sustain survival against PKCδ-induced apoptosis.


INTRODUCTION
Interleukin-32 (IL-32) was originally identified as a secretory protein that enhances the production of proinflammatory cytokines, including IL-1, IL-6, and tumor necrosis factorα (TNF-α). Later, it was unveiled that IL-32 exists intracellularly in nonimmune cells such as epithelial and tumor cells [1][2][3]. IL-32 has nine isoforms that are generated by the alternative splicing of seven exons [4]. Functional studies of IL-32 in cancer cells have focused mainly on the IL-32α, IL-32β, and IL-32γ isoforms because of their abundance [5]. IL-32α and IL-32β are highly expressed in gastric, lung, pancreatic, breast, brain, and liver tumors [6][7][8], but they show context-dependent function in different tumors. For example, IL-32α shows anti-proliferative effects in human colon cancer cells [9][10][11], whereas it enhances migration in human melanoma cells through the downregulation of E-cadherin [12] and promotes proliferation in pancreatic cells [6]. In addition, IL-32β is positively correlated with tumor size, number of lymph node metastases and tumor stage in primary breast cancer tissues; increases migration and invasion through signal transducer and activator of transcription 3 (STAT3) activation; and enhances glycolysis in human breast cancer cells [13,14]. On the other hand, it shows anti-tumor activity through the activation of lymphocytes including cytotoxic T (CD8+) and natural killer (NK) cells, and inactivation of the nuclear factor-κB (NF-κB) and STAT3 signaling pathways [15].
The von Hippel-Lindau (VHL) tumor suppressor protein is an E3 ligase that ubiquitinates hypoxia-inducible factor-1α (HIF-1α) and causes its degradation by the proteasome [16,17]. When oxygen is abundant, VHL binds to HIF-1α by recognizing its two post-translationally hydroxylated proline residues (P402 and P564) [18]. On the other hand, a lack of oxygen due to the rapidly growing tumor mass cause the inactivation of prolyl hydroxylase (PHD) preventing VHL from recognizing and binding to HIF-1α and thus leading to HIF-1α accumulation, which induces the transcription of genes involved in adaptation to hypoxia [19,20]. In addition to prolyl hydroxylation, asparaginyl hydroxylation has also been identified as an oxygen-regulated signal that determines the stability of HIF-1α. Both reactions are catalyzed by members of the 2-oxoglutarate (2OG)-dependent oxygenase superfamily: HIF-1α prolyl hydroxylation by PHD domains 1-3, and HIF-1α asparaginyl hydroxylation by factor inhibiting HIF-1 α (FIH). Since hydroxylation is one of the important post-translational modifications, the assignment of molecular functions for all human 2OG oxygenases by combined biochemical and cellular approaches is required. As an effort, accumulating data on newly identified hydroxylated proteins are expanding [21,22].
The protein kinase C (PKC) family comprises serine/threonine kinases that regulate a diverse set of cellular processes including proliferation, apoptosis, survival, and migration, and there is a substantial amount of evidence linking PKC to tumorigenesis [23]. According to their domain structure and respective activators, they have been divided into three major groups: the classical PKCs, including the α, βI, βII, and γ isoforms; the novel PKCs, including the θ, η, ε, and δ isoforms; and the atypical PKCs, including ζ and ι/λ [24]. Among various isoforms of PKC, PKCε and PKCδ have opposing roles in regulating apoptosis, survival and proliferation. PKCε promotes cell survival in many cell types, whereas PKCδ performs a growth inhibitory or pro-apoptotic role [23,24]. When PKCδ is activated by oxidative and hypoxic stress, PKCδ forms complex with Abl in the endoplasmic reticulum (ER), and then PKCδ-Abl complex translocates from ER to the mitochondria, leading to caspase-3 cleavage and cytochrome c release to trigger apoptosis [24][25][26]. The functional connection between IL-32β and PKCδ has recently been unveiled by the report that IL-32β upregulates IL-10 production through its association with PKCδ around the nuclear membrane [27,28].
In the present study, we investigated the function of IL-32β in human ovarian cancer cells which is one of the representative female cancers. Our results revealed that IL-32β is degraded by VHL-mediated ubiquitination, a process that is prevented by hypoxia-induced reactive oxygen species (ROS) production. PKCδ forms trimers with IL-32β and VHL. Under hypoxic conditions, the interaction between IL-32β and PKCδ is sustained, whereas that with VHL is disrupted. In turn, the increased IL-32β level reduces PKCδ-mediated apoptosis under oxidative stress.

IL-32β level increases under hypoxic conditions in ovarian cancer cells
We had previously shown that IL-32β level is increased and enhances glycolysis in breast cancer cells to survive under hypoxic conditions [14]. In the present study, we determined the molecular mechanism by which IL-32β stability is increased under hypoxic conditions in ovarian cancer cells. The levels of IL-32 protein and mRNA were examined in three different human ovarian cancer cell lines: IGROV1, SKOV3, and OVCAR8 cells. Although IGROV1 cells expressed both IL-32β and IL-32γ mRNA, the proteins were not detected, even with longer exposure. SKOV3 cells showed lower levels of IL-32β and IL-32γ mRNA expression than OVCAR8 cells. On the other hand, the IL-32β and IL-32δ proteins were highly expressed in OVCAR8 cells ( Figure 1A). To determine whether IL-32β expression is increased by hypoxic conditions, the ovarian cancer cells were treated with CoCl 2 , a hypoxia mimetic chemical. Whereas vascular endothelial growth factor (VEGF) mRNA expression (as a positive control) was markedly increased in all three cell lines, IL-32β mRNA expression was only modestly increased in SKOV3 and OVCAR8 cells ( Figure 1B). However, the IL-32β protein was highly increased in SKOV3 and OVCAR8 cells 3 h after CoCl 2 treatment, but IL-32β proteins were not detected even in the ROS-stimulated in IGROV1 cells ( Figure 1C). One possibility is that IGROV1 cells may have epigenetically suppressed IL-32β. Thus, we determined whether IL-32β translation was increased. When cells were co-treated with cyclohexamide (CHX) and CoCl 2 for the indicated time intervals, the CoCl 2 -induced increase in IL-32β level was not observed ( Figure 1D). If IL-32β is regulated by proteasomal degradation, the pre-existing IL-32β should be detected after the CHX treatment. However, if the steady state level of IL-32β is very low, it is possible that IL-32β level is low in the presence of CHX.

Hypoxic conditions inhibit IL-32β degradation
There are two mechanisms to explain the increase of protein amount: increase of the translation rate, and inhibition of protein degradation. To examine the possibility of proteasomal degradation, SKOV3 and OVCAR8 cells were treated with MG-132 for the www.impactjournals.com/oncotarget indicated time intervals. IL-32β was clearly increased 3 h after the treatment (Figure 2A). However, when cells were treated with pepstatin A, an inhibitor of lysosomal degradation, the increase in IL-32β level was not observed ( Figure 2B). These results imply that the translated IL-32β had been continuously degraded by ubiquitination. In the previous study, we showed that the CoCl 2 -induced increase in IL-32β level was mediated by ROS production in breast cancer cells. The ovarian cancer cells also showed the CoCl 2 -mediated increase in IL-32β production, which was prevented by treatment with N-acetyl-cysteine (NAC), an anti-ROS reagent, in a dose-dependent manner ( Figure 2C), and co-treatment of MG-132 with NAC reverse the latter's effect ( Figure   2D). This implies that ROS stabilize IL-32β levels through the inhibition of proteasomal degradation. To further confirm that IL-32β is continuously degraded under normal conditions, we examined whether it is ubiquitinated. Cells were transfected with Myc-IL-32β and Hemagglutinin (HA)-ubiquitin (Ub) and then treated with CoCl 2 or MG-132. IL-32β was immunoprecipitated with anti-Myc antibody and immunoblotted with anti-HA antibody to examine whether IL-32β is ubiquitinated. MG-132 treatment increased the level of ubiquitinated IL-32β, whereas CoCl 2 treatment inhibited it ( Figure 2E). Collectively, these results imply that hypoxia-induced ROS production increases IL-32β levels by preventing its ubiquitination. The VHL E3 ligase complex is responsible for IL-32β degradation VHL is a well-known E3 ligase involved in HIF-1α degradation under normoxic conditions. Kelchlike ECH-associated protein 1 (Keap1) is also an E3 ligase involved in nuclear factor erythroid 2-related factor 2 (Nrf-2) degradation in the absence of ROS, and Carboxyl terminus of HSP70-interacting protein (CHIP) is a chaperone-dependent E3 ligase that ubiquitylates unfolded protein under ROS stress [29][30][31][32]. To find out which E3 ligase is responsible for IL-32β ubiquitination, SKOV3 and OVCAR8 cells were transfected with VHL small interfering RNA (siRNA), CHIP siRNA, and Keap1 siRNA. The IL-32β was increased after VHL siRNA transfection ( Figure 3A). To further confirm this, SKOV3 cells were induced to express IL-32β and HA-VHL, whereupon the IL-32β levels were decreased by HA-VHL in a dose-dependent manner ( Figure 3B). Next, OVCAR8 cells highly expressing IL-32β were transfected with increasing amounts of HA-VHL, whereupon the endogenous IL-32β decreased ( Figure 3C). These results imply that VHL is the E3 ligase responsible for IL-32β ubiquitination.

ROS disrupt the interaction between VHL and the exon 7 coding region of IL-32β
To investigate the underlying molecular mechanisms by which IL-32β is regulated by VHL E3 ligase, cells were induced to express Myc-IL-32β and HA-VHL and then treated with H 2 O 2 . The immunoprecipitation assay revealed that the VHL and IL-32β interaction was disrupted after H 2 O 2 treatment ( Figure 4A), implying that IL-32β is degraded by VHL-mediated ubiquitination and this degradation is prevented by hypoxia-induced ROS production. To determine which domain of IL-32 binds to VHL, the interaction between different IL-32 isoforms and VHL was examined. All isoforms of IL-32 [33] bound to VHL ( Figure 4B). Since only the exon 7-coded region is common to all isoforms of IL-32, it was the only regions shown to interact with VHL ( Figure 4C). To determine which domain of VHL binds to IL-32β, the interaction between IL-32β and VHL domain using VHL19, which is absent in the acidic domain of first 53 amino acid residues [34], was examined. The interaction between IL-32β and VHL19 was increased after H 2 O 2 treatment, implying that acidic domain of VHL may play an important role in disruption of the interaction between IL-32β and VHL under hypoxic conditions ( Figure 4D).
Next, we examined whether a proline residue of IL-32β is hydroxylated by PHD, since VHL binds only to such proline residues in HIF-1α [19,20]. SKOV3 and OVCAR8 cells were treated with a PHD inhibitor, dimethyloxaloylglycine (DMOG) or desferrioxamine (DFO), and IL-32β expression was measured. Contrary to our expectation, DMOG and DFO did not affect the IL-32β level in OVCAR8 cells, although SKOV3 cells showed increase in IL-32β level ( Figure 4E, 4F). In addition, SKOV3 and OVCAR8 cells were also transfected with PHD2 siRNA and the IL-32β was examined. No difference was observed ( Figure 4G). To further confirm this result, we examined the interaction of IL-32β with VHL mutants that are not able to bind to hydroxyl proline residues of HIF-1α [19,34]. All VHL mutants still bound to IL-32β ( Figure 4H), implying that the interaction between IL-32β and VHL is not dependent upon proline hydroxylation.

PKCδ is associated with the interaction between IL-32β and VHL
Since it is known that ROS activates PKC, which can interact with IL-32 and VHL [27,[35][36][37], PKC was examined for its possible association with the interaction between IL-32β and VHL. Cells were induced to express HA-VHL, Myc-IL-32β and Flag-PKCδ, and then immunoprecipitated with anti-Flag antibody. PKCδ precipitated with Myc-IL-32β and HA-VHL ( Figure  5A). In addition, when cells were placed in the hypoxic chamber, the co-immunoprecipitation of IL-32β and PKCδ with VHL was reduced ( Figure 5B). These results imply that the interaction of VHL with either IL-32β or PKCδ is reduced under hypoxia. However, the interaction between IL-32β and PKCδ was constant regardless of the oxygen conditions ( Figure 5C). To further confirm this interaction in human ovarian cancer cells, OVCAR8 cells were induced to express HA-VHL, Myc-IL-32β and Flag-PKCδ, and then immunoprecipitated with anti-HA antibody. VHL precipitated with Myc-IL-32β and Flag-PKCδ. In addition, when cells were treated with CoCl 2 , the co-immunoprecipitation of IL-32β and PKCδ with VHL was reduced ( Figure 5D). These results collectively suggest that IL-32β, VHL, and PKCδ form a complex. Under the hypoxic condition, the interaction between VHL and IL-32β or PKCδ is disrupted, not affecting the interaction between IL-32β and PKCδ.

IL-32β attenuates PKCδ-induced apoptosis in oxidative stress
To find out effect of IL-32β and PKCδ co-expression on apoptosis, proliferation, and migration, SKOV3 cells expressing a low level of IL-32β were induced to express IL-32β and PKCδ, and OVCAR8 cells expressing a high level of IL-32β were transfected with IL-32 siRNA and PKCδ siRNA, and then cell apoptosis was examined since PKCδ is known to be involved in those functions [23]. PKCδ increased the level of cleaved-poly (ADP-ribose) polymerase (PARP) and the activity of caspase-3/7 after H 2 O 2 treatment. However, co-expression of IL-32β and PKCδ reduced the H 2 O 2 -induced apoptosis in the presence of PKCδ, implying that IL-32β attenuates PKCδ-induced  Figure 6A, 6B). Since PKCδ also has an anti-proliferative effect on cancer cells, its effect was examined in the presence of IL-32β. The proliferation of SKOV3 and OVCAR8 cells was greatly reduced after CoCl 2 treatment, but this was not affected by the co-presence of IL-32β and PKCδ ( Figure 6C). Next, we examined effect of IL-32β and PKCδ on migration under CoCl 2 -induced hypoxic conditions. SKOV3 and OVCAR8 cell migration was considerably decreased after CoCl 2 treatment, and this effect was not affected by the co-expression of IL-32β and PKCδ ( Figure 6D). These findings imply that the accumulated IL-32β by hypoxiainduced ROS binds to PKCδ, leads to inhibit PKCδinduced apoptosis under oxidative stress without affecting other functions, including proliferation and migration.

DISCUSSION
Hypoxia-stabilized HIF-1α binds to hypoxia response elements, leading to the upregulation of several genes to promote survival under low-oxygen conditions. These genes code for enzymes that enhance glycolysis, which allows for ATP synthesis in an oxygen-independent manner [38,39]. VEGF is a product of one of the HIF-1α-induced genes, which promotes angiogenesis and supplies nutrients and oxygen to the tumor mass [40]. In the present study, we identified one more protein (i.e., IL-32β) that is as urgently needed as HIF-1α under hypoxic and ROS conditions. IL-32β stability was regulated by the post-translational mechanism in SKOV3 and OVCAR8 human ovarian cancer cells. This cytokine was The migrated SKOV3 and OVCAR8 cells were counted after 24 or 48 h incubation, respectively. Scale bars = 10 μm. www.impactjournals.com/oncotarget continuously synthesized and degraded by ubiquitination under normoxic conditions, but accumulated rapidly following exposure to low oxygen tension or ROS. Since the regulation of IL-32β in some of the ovarian tumor cells is energy-consuming, we infer that tumor cells urgently need IL-32β under the hypoxic and ROS conditions. We revealed that hypoxia-and ROS-stabilized IL-32β prevents PKCδ-mediated apoptosis, and this would rescue tumor cells from sudden death caused by environmental changes (Figure 7).
Various isoforms of IL-32 are produced by alternative splicing mechanism and there are conflicting observations of either an oncogenic or a tumor suppressive role for IL-32 [12,41]. These contradictory results might originate from differences in the IL-32 isoform expressed in tumors analyzed. It is reported that IL-32γ and IL-32β induce caspase-8-dependent cell death in HEK293 cells whereas IL-32α does not, and that the treatment of alternative splicing inhibitor results in the predominant expression of IL-32γ splice variants and cell death in thyroid cancer cell lines [42]. On the other hand, IL-32β itself shows either an oncogenic or a tumor suppressive role. In this study, ROS-induced IL-32β reduced PKCδinduced apoptosis, however, IL-32β transgenic mice have the increased cytotoxic T lymphocytes and NK cells activity, leads to the inhibition to tumor growth [15]. A plausible explanation about contrary IL-32β function could be that physiological environments such as ROS condition and change of immune system affects IL-32β role.
To find out which E3 ligases are responsible for IL-32β ubiquitination, the hypoxia-or ROS-related E3 ligases VHL, CHIP, and Keap1 [20,[29][30][31][32] were examined. Among these, only VHL could bind to and ubiquitinate IL-32β. Whereas VHL binding to HIF-1α is dependent on prolyl hydroxylation, the interaction between IL-32β and VHL was not dependent on prolyl hydroxylation of IL-32β. As with IL-32β, it is known that Aurora kinase A is also regulated by VHL in a prolyl hydroxylation independent manner [43]. We have therefore provided one more protein that is a VHL target, but for which binding is hydroxylation independent. On the other hand, Stoehr et al. showed that the inhibition of prolyl hydroxylation by DMOG stabilizes many proteins that have potential function under the hypoxic conditions [44]. Since DMOG treatment stabilized IL-32β ( Figure  4D), the prolyl hydroxylation of IL-32β itself could stimulate its degradation under normoxic conditions, but it would not depend on VHL interaction, since we showed that the interaction between VHL and IL-32β was not dependent on IL-32β prolyl hydroxylation. Therefore, we can infer that IL-32β is degraded by two ways under normoxic conditions: prolyl hydroxylation-induced IL-32β degradation, and VHL-mediated degradation in a prolyl hydroxylation-independent manner. On the other hand, since prolyl hydroxylation-induced IL-32β degradation was not observed in OVCAR8 cells ( Figure 4D, 4E), this process may need another factor that is absent in OVCAR8 cells.
It is known that PKCδ interacts with the β domain of VHL, but its overall protein levels are not affected by the interaction with VHL [45,46]. The interaction of PKCδ with VHL retains PKCδ in the cytoplasm, which in turn impairs the association between PKCδ and the insulin-like growth factor 1 (IGF-I) receptor [37]. Therefore, besides the degradation of binding partners, another effect of the interaction with VHL is the limit of the localization of binding partners. In contrast to inactive PKCs, which are present mainly in the cytosol, activated PKCs are localized at the plasma membrane, the nucleus, and other subcellular compartments [47]. Hypoxic conditions in particular trigger the translocation of PKCδ to the endoplasmic reticulum or mitochondria, resulting in oxidative stressinduced apoptosis [24][25][26]48]. Since the hypoxic stimulus translocates IL-32β to the mitochondria in breast cancer cells [14], IL-32β could inhibit PKCδ-mediated apoptosis. In turn, hypoxia-and ROS-stabilized IL-32β contributes to cancer cell survival. On the other hand, IL-32β interacts with PKCδ in the human monocytic cell line U-937, and its interaction shifts the immune reaction from an inflammatory to an anti-inflammatory one by enhancing anti-inflammatory cytokine IL-10 production and inhibiting the secretion of IL-1β and TNF-α [27,28]. These findings suggest that IL-32β plays a favorable role in tumor cell survival by affecting the function of both tumor and immune cells.
In conclusion, hypoxia releases IL-32β from ubiquitination-mediated degradation. Subsequently, the increased IL-32β interacts with PKCδ and results in the inhibition of apoptosis, which can be induced by ROSactivated PKCδ. In turn, the hypoxia-induced IL-32β is beneficial for tumor cell survival under hypoxic and ROS environments. The limitation of our study is that the effect of IL-32β could not be validated using knockout mice, because this cytokine is not expressed in rodents. Nonetheless, the functions of IL-32β have been and continued to be unveiled with the use of various cancer models.

Plasmids
The green fluorescent protein-tagged IL-32β (GFP-IL-32β), Flag-PKCδ, and Myc-IL-32 isoforms and IL-32βexpressing plasmids were obtained from Prof. Do-Young Yoon at Konkuk University. HA-Ub was obtained from Prof. Keun Il Kim at Sookmyung Women's University. HA-VHL and HA-VHL19 were obtained from Prof. Jae Whan Song at Yeonsei University. The HA-VHL Y98N and W117R were purchased from Addgene (Cambridge, MA, USA). Truncated mutants of IL-32 were prepared by PCR and subcloned into the EcoRI-SalI sites of pEGFPN2, or the EcoRI-XhoI sites of the Myc-pcDNA3.1 mammalian expression vector. VHL S111D was prepared by PCR and subcloned into the EcoRI-XhoI sites of the HA-pcDNA3 mammalian expression vector.

Immunoblot analyses
The cell lysates were prepared after transfection with plasmids or siRNA and mixed with 5× sodium dodecyl sulfate (SDS) sample buffer. The mixed samples were heated at 99°C for 10 min and separated electrophoretically on a SDS-polyacrylamide gel. Subsequently, the proteins were transferred onto a 0.45-μm nitrocellulose membrane (GE Healthcare, Buckinghamshire, UK) for 2 h. The membrane was blocked for 30 min at room temperature with 5% skim milk (Invitrogen) and subsequently incubated with primary antibody overnight at 4°C. Then, the membrane was incubated with an anti-mouse or anti-rabbit IgG antibody conjugated to horseradish peroxidase (Assay Designs, Ann Arbor, MI, USA) at room temperature for 2 h. The proteins were visualized using an enhanced chemiluminescent substrate (Thermo Fisher Scientific, Logan, MA, USA) and analyzed using the LAS3000 luminescent image analyzer (Fuji Film, Tokyo, Japan). The protein bands were quantified using ImageJ software (National Institutes of Health, Bethesda, MD, USA).

Immunoprecipitation assay
The HEK 293T and OVCAR8 cells were transfected with plasmids for 24 h and then treated with chemicals. The cell lysates were prepared using cell immunoprecipitation lysis buffer (50 mM Tris-HCl (pH 8.0), 150 mM NaCl, 5 mM ethylenediaminetetraacetic acid, 250 mM phenylmethylsulfonyl fluoride, 1% NP-40) containing a proteinase inhibitor cocktail. The cell lysates were incubated with antibody for 2 h at room temperature, and then protein-G agarose beads were added. The entire lysates were centrifuged after 2 h incubation, and then washed four times with washing buffer.

Activity assay
The Caspase-Glo 3/7 assay (Promega, Madision, WI, USA) was carried out according to the manufacturer's instruction. The assay is based on the release of aminoluciferin as a result of active caspase-3/7 cleaving the luminogenic substrate containing the DEVD amino acid sequence.

Proliferation assay
Cell proliferation was evaluated using the CellTiter-Blue assay (Promega). Cells were seeded on 48-well plates and incubated for 24 h. Then, the cells were treated with CoCl 2 for 24 h. After washing with PBS, diluted CellTiter-Blue reagent was added directly to each well and the plates were incubated at 37°C for 1 h. The reagent was transferred to 96-well plates and, the fluorescence was measured at 560/590 nm.

In vitro migration assay
Cell migration was assessed in a 24-well plate Transwell system (Corning, Corning, MD, USA). Each 8-μm Transwell insert was seeded with 100 μL of SKOV3 or OVCAR8 cells in serum-free RPMI-1640 medium with or without CoCl 2 , and the lower chamber was filled with 500 μL RPMI-1640 medium containing 10% FBS with or without CoCl 2 . The cells were incubated for 24 or 48 h, respectively. Images of the membrane were taken in three random fields per chamber, and the total number of migrated cells was counted.

Statistical analyses
Statistical analyses were performed by a paired Student's t-test and one-way factorial analysis of variance. The Scheffe and Bonferroni tests were also performed. Values of *p < 0.05, **p < 0.01, and ***p < 0.001 were considered benchmarks of significant differences. Data are presented as the means ± standard deviation.