hnRNPK inhibits GSK3β Ser9 phosphorylation, thereby stabilizing c-FLIP and contributes to TRAIL resistance in H1299 lung adenocarcinoma cells

c-FLIP (cellular FLICE-inhibitory protein) is the pivotal regulator of TRAIL resistance in cancer cells, It is a short-lived protein degraded through the ubiquitin/proteasome pathway. The discovery of factors and mechanisms regulating its protein stability is important for the comprehension of TRAIL resistance by tumor cells. In this study, we show that, when H1299 lung adenocarcinoma cells are treated with TRAIL, hnRNPK is translocated from nucleus to cytoplasm where it interacts and co-localizes with GSK3β. We find that hnRNPK is able to inhibit the Ser9 phosphorylation of GSK3β by PKC. This has the effect of activating GSK3β and thereby stabilizing c-FLIP protein which contributes to the resistance to TRAIL in H1299 cells. Our immunohistochemical analysis using tissue microarray provides the clinical evidence of this finding by establishing a negative correlation between the level of hnRNPK expression and the Ser9 phosphorylation of GSK3β in both lung adenocarcinoma tissues and normal tissues. Moreover, in all cancer tissues examined, hnRNPK was found in the cytoplasm whereas it is exclusively nuclear in the normal tissues. Our study sheds new insights on the molecular mechanisms governing the resistance to TRAIL in tumor cells, and provides new clues for the combinatorial chemotherapeutic interventions with TRAIL.


Results
Interaction and co-localization of GSK3β with hnRNPK in H1299 lung adenocarcinoma cells. We have previously demonstrated that GSK3β interacts with hnRNPK in HepG2 hepatocellular carcinoma cells 34 . Here, we firstly used GST pull-down assays to demonstrate this interaction using the cellular lysate of lung adenocarcinoma cancer H1299 cells. As shown in Fig. 1a, purified GST-hnRNPK was capable to specifically pull down the endogenous GSK3β in H1299 cell lysate. Interestingly, both anti-GSK3β antibody and anti-p9-GSK3β antibody were able to reveal the GSK3β trapped by the purified GST-hnRNPK. This raised the uncertainty about the phosphorylation state of the trapped GSK3β . Indeed, the GSK3β molecules interacting with GST-hnRNPK could be both Ser9 phosphorylated and unphosphorylated, or they could be only Ser9 phosphorylated and recognized by both the anti-GSK3β and anti-p9-GSK3β antibodies. These two different situations might respectively reflect the two possibilities where the Ser9 phosphorylated and unphosphorylated GSK3β interact equally or differentially with hnRNPK. To clarify this point, we have performed GST-pull down assays using H1299 cell lysates incubated respectively with either GST-GSK3β wt or GST-GSK3β S9A, a Ser9phospho-defective mutant, and compared the quantity of hnRNPK trapped by these two forms of GSK3β . Our results shown in Fig. S1 indicate that GST-GSK3β wt and GST-GSK3β S9A displayed similar affinities vis-à-vis hnRNPK in H1299 cell lysates. Therefore, Ser9 phosphorylation per se might not be a factor directly affecting the affinity between GST-GSK3β and hnRNPK. The interaction between GSK3β and hnRNPK was then confirmed by co-immunoprecipitation experiments using the H1299 cells transfected by Flag-hnRNPK. As shown in Fig. 1b, Flag-hnRNPK co-immunoprecipited with GSK3β by anti-GSK3β antibody in the cell lysate of H1299 cells either treated or not by TRAIL. Confocal immunofluorescence assays were then performed to confirm this result by showing the co-localization of GSK3β and hnRNPK in the cytoplasm of H1299 cells. Figure 1c,d demonstrate that both endogenous and ectopically expressed hnRNPK could be partially translocated from nucleus to cytoplasm upon TRAIL treatment, and co-localized with GSK3β that was mainly cytoplasmic.
Both GSK3β and hnRNPK antagonized TRAIL-induced apoptosis. In order to investigate the functional importance of GSK3-hnRNPK interaction in the TRAIL-induced apoptosis of H1299 cells, we searched at first to determine the optimal concentration and timing of TRAIL treatment on H1299 cells. A TRAIL dose-response experiment has been performed. H1299 cells were treated during 24 hours with increasing doses of TRAIL, and then analyzed by MTT assays for cellular viability and by Western blotting for the level of cleaved Caspase 3, both indicators of the apoptotic degree of cells. Our results in Fig. 2a,b showed that, when used within the range of 2-20 ng/ml, TRAIL induced apoptosis of H1299 cells in a dose-dependent manner, with a marked effect observed at 20 ng/ml, as judged by both the cleaved caspase 3 level and the cell death rate. This concentration was then used for another experiment destined to determine the timing of the treatment. Cells were harvested at different time points after TRAIL stimulation and analyzed as above for the apoptotic state of the cells.
Our experiment suggested that 8-12 hours of TRAIL treatment were sufficient to induce significant apoptosis in H1299 cells (Fig. 2c,d). Subsequently, we attempted to assess the role of GSK3β in the TRAIL-induced apoptosis of H12199 cells by using LiCl that is a specific inhibitor of GSK3β kinase activity and inducer of the Ser9 phosphorylation of GSK3β 42 . In our experiment, increasing doses of LiCl, ranging from 5 mM to 50 mM, indeed resulted in the increasing level of GSK3β Ser9 phosphorylation, which was also confirmed by the increasing level of β -catenin whose degradation is an indicator of the kinase activity of GSK3β (Fig. 2e). The concentration of 20 mM was then chosen for the subsequent experiments of LiCl treatment. As described in Materials and Methods, we have treated H1299 cells separately or jointly with 20 mM LiCl and 20 ng/mL TRAIL, then examined the level of total GSK3β , Ser9 phosphorylated GSK3β , and cleaved Caspase 3 in these cells by Western blotting using specific antibodies (Fig. 2f). Our experiments showed that LiCl was capable to enhance the effect of TRAIL to induce apoptosis, reflected by the increased level of cleaved Caspase 3. MTT assays were then performed with This preliminary result conforming the pro-survival role of active GSK3β under the treatment of TRAIL was consistent with other reports 31,43 .
To further investigate the functions of both GSK3β and hnRNPK in TRAIL-induced tumor cell apoptosis, we have either knocked down or overexpressed GSK3β or hnRNPK respectively by transient transfection of H1299 cells with corresponding siRNA or Flag-tagged expression plasmids. Cells were then examined by Western blotting using specific antibodies against Flag tag, GSK3β , hnRNPK and cleaved Caspase 3, or by FACS with Annexin V/PI double staining (Fig. 3a-h). As shown, knockdown of either GSK3β or hnRNPK significantly increased the level of the cleaved Caspase 3 in Western blotting analyses and also the cell apoptotic rates in FACS assays, whereas the opposing effects were observed when cells were transfected by Flag-GSK3β or Flag-hnRNPK ( Fig. 3a-h). These results indicated that both GSK3β and hnRNPK antagonized TRAIL-induced apoptosis in H1299 cells.
hnRNPK down-regulates the Ser9 phosphorylation of GSK3β by PKC in H1299 cells. In our preliminary studies, we have observed that the overexpression of Flag-hnRNPK dramatically reduced the Ser9 phosphorylation of GSK3β in several kinds of cells, including H1299 cells (data not shown). This suggests that hnRNPK might inhibit the Ser9 phosphorylation of GSK3β and thereby regulate the TRAIL-induced apoptosis in   H1299 cells. To confirm this hypothesis, we have either overexpressed Flag-hnRNPK or knocked down hnRNPK by siRNA in H1299 cells with or without TRAIL treatment, and examined the levels of total GSK3β and Ser9 phosphorylated GSK3β . As shown in Fig. 4a,b, the overexpression and depletion of hnRNPK respectively resulted in sharp inhibition and stimulation of GSK3β Ser9 phosphorylation, regardless the cells were treated or not with TRAIL, while the level of total GSK3β remained unaffected. This result clearly indicates that hnRNPK is capable of down-regulating the Ser9 phosphorylation of GSK3β .
Firstly, we confirmed that the Ser9 phosphorylation of GSK3β in H1299 cells was mainly the effect of PKC kinase activity, since Gö6983, an inhibitor against the classic PKC α -ξ subtypes, was capable to inhibit the quasi totality of Ser9 phosphorylation of GSK3β in these cells (Fig. 4c). And more interestingly, with Gö6983 treatment, the knocking down of hnRNPK by siRNA had no longer obvious effect on the Ser9 phosphorylation of GSK3β . As contrast, when using rottlerin that is an inhibitor against the non-classic PKC δ subtype, hnRNPK siRNA still significantly stimulated the Ser9 phosphorylation of GSK3β (Fig. 4d). These results suggest that hnRNPK inhibits the Ser9 phosphorylation of GSK3β by PKC α -ξ but not PKC δ in H1299 cells. However, since the Ser9 phosphorylation state of GSK3β was regulated both by its phosphorylation by PKC and its dephosphorylation by PP1 50 , it remained to verify if hnRNPK might also stimulate the dephosphorylation of GSK3β by PP1. To address this issue, we have treated H1299 cells with 100 nM okadaic acid which inhibited the dephosphorylation of GSK3β by PP1, and examined the effect of hnRNPK expression on the phosphorylation state of GSK3β . Our result showed that the treatment by okadaic acid again did not alter the inhibitory effect of hnRNPK overexpression, excluding the possible regulation of GSK3β dephosphorylation by hnRNPK (Fig. 4e).
Taken together, these results suggest that hnRNPK is capable to inhibit the Ser9 phosphorylation of GSK3β by PKC, probably via steric hindrance caused by its interaction with GSK3β . hnRNPK inhibits GSK3β Ser9 phosphorylation, leading to the stabilization of c-FLIP protein and inactivation of Caspase 8 during the TRAIL-induced apoptosis of H1299 cells. Based on the regulation of GSK3β Ser9 phosphorylation by hnRNPK as described hereinabove, and the stabilization of c-FLIP by GSK3β reported by others 33 , we searched to know if there was any link between these two events which might be part of the mechanisms involving these three proteins in the resistance of the TRAIL-induced apoptosis. We have firstly confirmed the regulatory role of GSK3β on the protein level of c-FLIP in H1299 cells treated with TRAIL. As shown in Fig. 5a, LiCl treatment of cells resulted in the increased level of GSK3β Ser9 phosphorylation and the down-regulation of both p55 and p43 c-FLIP protein levels, and as a consequence, the augmentation of the cleaved Caspase 3 amount. Next, we examined the effect of hnRNPK overexpression on c-FLIP protein stability in H1299 cells by performing cycloheximide chase experiments. H1299 cells transfected by either 2 μg Flag-hnRNPK or empty Flag plasmids were treated with 10 μg/ml cycloheximide (CHX). After various durations of treatment, cells were harvested and subjected to Western blotting analysis to assess the protein level of c-FLIP. Our results in Fig. 5b shows that in cells transfected with the empty Flag vector, most of the c-FLIP protein is degraded as rapidly as 2 hours after the treatment with CHX, while the overexpression of Flag-hnRNPK significantly slows down the c-FLIP degradation (Fig. 5b). The stabilizing effect of hnRNPK on c-FLIP protein was then further confirmed by a dose-response experiment. Cells transfected by increasing amount of either Flag-hnRNPK or empty Flag plasmids and treated with 10 μg/ml CHX during 2 hours were harvested. The protein level of c-FLIP was then assessed by Western blotting as described. As shown in Fig. 5c, the increasing concentration of overexpressed Flag-hnRNPK indeed resulted in the augmentation of c-FLIP in the cells as compared with the empty vector-transfected cells. And, as expected, the effect of hnRNPK was dependent on the Ser9 phosphorylation state of GSK3β , since LiCl treatment of the cells cancelled this effect. Moreover, c-FLIP expression levels are inversely correlated with those of the cleaved caspase 8, further confirming the functional effectiveness of this regulatory pathway (Fig. 5d). These results suggest that, in addition to the possible modulation of c-FLIP expression by hnRNPK at the transcriptional level as described in other report 41 , hnRNPK might also stabilize c-FLIP protein through regulation of GSK3β Ser9 phosphorylation. hnRNPK expression negatively correlates with Ser9 phosphorylated GSK3β in clinical lung adenocarcinoma tissues. In order to provide the clinical evidence of the regulation of GSK3β Ser9 phosphorylation by hnRNPK in lung cancer, we have performed the immunohistochemical essays using two commercialized tissue microarrays. 52 paired lung adenocarcinoma and adjacent normal samples were stained with either anti-hnRNPK antibody or anti-Ser9 phosphorylated GSK3β antibody, evaluated and scored as described in Materials and methods section. The results showed that hnRNPK was highly overexpressed in both the nucleus and cytoplasm of lung adenocarcinoma cells as compared with normal tissues (Fig. 6a-c), whereas the phospho-GSK3β was mainly located in the cytoplasm which was significantly downregulated in the tumor tissues (Fig. 6d). Furthermore, negative correlation between the level of hnRNPK and the Ser9 phosphorylated GSK3β could be established in both lung cancer and normal tissues (Fig. 6e). Strikingly, in the same histological section of lung cancer, we were able to simultaneously observe the low hnRNPK expression accompanied with the high level of phospho-GSK3β in a region (the arrows, Fig. 6e), and the opposite pattern in a neighbouring zone (the stars, Fig. 6e). Statistical analysis demonstrated that the expression levels of hnRNPK and phospho-GSK3β were inversely correlated in lung tumor tissues (p < 0.05, two-tailed). Finally, we found that in 52 (100%) carcinoma tissues examined, hnRNPK could be found in both the cytoplasm and nucleus whereas it was exclusively nuclear in normal tissues (Fig. 6f). This is consistent with another study performed by Pino and colleagues showing the cytoplasmic staining of hnRNPK in half of the 32 lung carcinoma samples examined 51 . This result strongly suggests the importance of the cytoplasmic hnRNPK in the genesis of lung adenocarcinoma, and its role in the regulation of the GSK3β Ser9-phosphorylation.

Discussion
Owing to its capacity to specifically induce the apoptosis in tumor cells but not in normal cells, TRAIL is predicted to be one of the most effective anticancer therapeutic agents in the future 7,8,52 . However, resistance to TRAIL has turned out to be a general phenomenon and become a serious obstacle for the prospects of TRAIL therapy. The comprehensive understanding of the molecular mechanisms of TRAIL's action thus reveals to be important and necessary for the development of combinatorial treatments involving TRAIL and others synergizing or sensitizing agents. Previous studies have established that the effects of TRAIL pass through the membrane death receptors 4 and 5 (DR4 and DR5) with their intracellular death domains, which activates dual opposing pathways leading respectively to the apoptosis but also the survival of the cell 52 . These two so called "apoptotic pathway" and "resistance pathway" are thought to be respectively responsible for the apoptotic effect of TRAIL and resistance to TRAIL in the cells. c-FLIP is regarded as the master anti-apoptotic regulator and resistance factor in TRAIL-induced apoptosis due to its effect of shifting the apoptotic pathway to the resistance pathway by facilitating the formation of the so called "secondary complex". The observation that chemotherapeutic agents like celecoxib, camptothecin, and cisplatin could down regulate c-FLIP expression and sensitize the resistant cells to TRAIL treatment further demonstrated that c-FLIP plays a key role in the resistance to TRAIL-induced apoptosis and represents a potential target candidate for the combinatorial chemotherapeutic interventions with TRAIL 33,53 . c-FLIP has been shown to be a short-lived protein that is degraded via the proteasome pathway 17,23,54 . Therefore, the regulation of its stability might represent an interesting way to overcome the resistance to TRAIL. Our present study describes a novel mechanism of action of hnRNPK to antagonize TRAIL-induced apoptosis. hnRNPK is a DNA/RNA binding protein whose function has been so far associated with its nuclear localization and/or its regulation of the RNA expression and metabolism. We show here for the first time that the cytoplasmic hnRNPK interacts with GSK3β , and regulates its function to stabilize c-FLIP protein. Our finding is corroborated by our immunohistochemical analysis with tissue microarray demonstrating the negative correlation between hnRNPK expression level and that of the Ser9-phsphorylated GSK3β in lung adenocarcinoma tissues. Interestingly, hnRNPK has been previously reported to transcriptionally activate c-FLIP and antagonize TRAIL-induced apoptosis in nasopharyngeal carcinoma cells 41 . Taken together, these studies suggest that hnRNPK may up-regulate the c-FLIP protein level in cancer cells through two different mechanisms and at two different subcellular locations to antagonize the apoptotic effect of TRAIL, which would make this protein an interesting target for the combinatorial therapy with TRAIL.
GSK3β is a ubiquitous serine/threonine kinase with multiple functions involved in a wide range of cellular functions, including differentiation, growth, apoptosis, cell cycle, embryonic development, and insulin response [55][56][57] . The phosphorylation on the serine 9 residue of GSK3β is critical for its multiple functions. Our present study has permitted to identify hnRNPK protein as an efficient activator of GSK3β through inhibiting its Ser9 phosphorylation, one of the most active and versatile family of kinases responding to a variety of extracellular signals to regulate multiple signaling pathways in many types of cells 58 . Thus, in addition to its classic role of a DNA/RNA binding protein involved in the metabolism and expression of RNA, the discovery of its role as a regulator of GSK3β playing key roles in many signaling pathways and cellular processes turns out to be very exciting. The investigation of the possible regulation of GSK3β activity by hnRNPK in various physiological and pathological contexts reveals to be, in this regard, very interesting and important.
How could hnRNPK inhibit the phosphorylation state of GSK3β ? Two possibilities could be taken into consideration. hnRNPK might either inhibit the phosphorylating activity of PKC, or facilitate the action of PP1 which dephosphorylates Ser9 phosphorylated GSK3β . The results of our experiments using the inhibitors respectively against PKC or PP1 have shown that hnRNPK acts more likely through the regulation of PKC rather than that of PP1 (Fig. 4c,e). The most probable scenario should be the stoichiometric hindrance caused by the hnRNPK-GSK3β interaction which renders inaccessible the Ser9 residue to PKC. We have attempted to Scientific RepoRts | 6:22999 | DOI: 10.1038/srep22999 verify this hypothesis by trying to assess the effect of hnRNPK overexpression on PKC-GSK3β interaction using co-immunoprecipitation assays. However, it turned out to be impossible to detect PKC-GSK3β interaction by this method (data not shown). Consistently, whilst the Ser9 phosphorylation of GSK3β by PKC has been well documented in the literature 33,[47][48][49] , no report concerning the physical interaction between PKC and GSK3β could be found. This might probably due to the extremely dynamic interaction between these two proteins, as often observed between the kinases and their substrates. The verification of such hypothesis with more elaborated methodology would be of interest, and will be part of our future study.  Confocal microscopy assay. H1299 lung adenocarcinoma cells transfected with or without Flag-hnRNPK were treated with TRAIL (20 ng/ml). Immunofluorescence was performed as described previously 34,60 . The primary antibodies (1:100 dilution) of hnRNPK, Flag and GSK3β , and the Alexa Fluor 488 and Alexa Fluor 594 conjugated secondary antibodies (ZSGB-BIO, China) were used for observation. DAPI staining was used to determine the morphology of cell nuclei.
The imaging experiments were carried out on laser scanning confocal microscopes (LSM700, Zeiss, Jena, Germany) equipped with a Zeiss Plan-Neofluar 40× /1.3 NA Oil Dic objective as described previously 34,60 .
Scientific RepoRts | 6:22999 | DOI: 10.1038/srep22999 Tissue microarrays and Statistics analysis. The expression of hnRNPK and the Ser9 phosphorylation of GSK3β was analyzed using two consecutively numbered commercial tissue microarrays (TMAs) of lung adenocarcinoma (catalog no. OD-CT-RsLug04-003, Shanghai Outdo Biotech, Shanghai, China), consisting of 53 paired lung adenocarcinoma and adjacent normal samples, 2 unpaired lung adenocarcinoma tissues. The TMAs came from 30 male and 25 female patients, and were mainly grade II in the pathological grading. During the experimental process, one sample of lung adenocarcinoma tissue was destroyed, the final 52 paired samples of adenocarcinoma and normal tissues were analyzed.
The immunohistochemical staining of TMAs were performed in Shanghai Outdo Biotech. Briefly, tissue sections (4 μm) were incubated in a 63 °C oven for one hour. After deparaffnization and rehydration, antigen retrieval was carried out by boiling sections in citrate buffer (10 mM, pH 6.0) for 5 min. Then, endogenous peroxidase activity was blocked with 3% H 2 O 2 for 15 min at room temperature. The slides were blocked with 3% BSA in PBS for 30 min and subsequently cultured with primary antibodies against phospho-GSK3β (Ser9) (1:100 diluion) and hnRNPK (1:500 diluion), respectively. The immune complex was detected using a Dako EnVision ™ Detection system (Dako Japan Ltd.). Nuclei were counterstained with hematoxylin. Images of TMAs were captured by Aperio Scanscope XT (Leica, Germany).
The results were evaluated independently by 2 pathologists according to both the intensity of protein staining and the percentage of positive cells. The intensity of protein staining was scored as 1 (negative), 2 (weak), 3 (moderate) or 4 (strong). Evaluation of the percentage of positive cells was scored as: 1 (0%), 2 (≤ 30%), 3 (31-60%), and 4 (> 60%). The total score of each case was consisting of the score of the intensity of protein staining and the score of percentage of positive cells. The paired Student's t-test was employed to analyze the significance of phospho-GSK3β or hnRNPK with tumor or pericancerous normal tissue. Correlation between the expression of hnRNPK and phospho-GSK3β (Ser9) were estimated using the Spearman's rank correlation analysis.