Crosstalk between JNK and SUMO Signaling Pathways: deSUMOylation Is Protective against H2O2-Induced Cell Injury

Background Oxidative stress is a key feature in the pathogenesis of several neurological disorders. Following oxidative stress stimuli a wide range of pathways are activated and contribute to cellular death. The mechanism that couples c-Jun N-terminal kinase (JNK) signaling, a key pathway in stress conditions, to the small ubiquitin-related modifier (SUMO), an emerging protein in the field, is largely unknown. Methodology/Principal Findings With this study we investigated if SUMOylation participates in the regulation of JNK activation as well as cellular death in a model of H2O2 induced-oxidative stress. Our data show that H2O2 modulates JNK activation and induces cellular death in neuroblastoma SH-SY5Y cells. Inhibition of JNK's action with the D-JNKI1 peptide rescued cells from death. Following H2O2, SUMO-1 over-expression increased phosphorylation of JNK and exacerbated cell death, although only in conditions of mild oxidative stress. Furthermore inhibition of SUMOylation, following transfection with SENP1, interfered with JNK activation and rescued cells from H2O2 induced death. Importantly, in our model, direct interaction between these proteins can occur. Conclusions/Significance Taken together our results show that SUMOylation may significantly contribute to modulation of JNK activation and contribute to cell death in oxidative stress conditions.

Oxidative stress is mediated by excessive exposure of cells to reactive oxygen species, which generate an oxidative burst of intracellular signaling cascades that induce cell death.
Among others, H 2 O 2 induced oxidative stress leads to activation of c-Jun N-terminal kinase (JNK), a kinase that is strongly associated with many different stress stimuli and cell death [12,13,14,15] as well as of the SUMOylation pathway, recently associated with ischemic events and cytoprotection [16,17,18,19]. SUMO is a family of three proteins (SUMO-1, -2, and -3) that are involved in SUMOylation process, a posttranslational modification consisting of covalent conjugation of SUMO to target proteins. The SUMOylation cascade is ATP dependent. SUMO conjugates proteins through an enzymatic cascade similar to ubiquitination. The process involves a single SUMO-activating enzyme E1 (Uba2-Aos1), a SUMO-conjugating enzyme E2 (Ubc9) and often a SUMO E3 that facilitates the conjugation. A specific isopeptidase, member of the SENP family, ensures reversibility of the SUMOylation process [20,21].
The role of SUMOylation in oxidative stress is yet to be defined. Nevertheless some very recent reports have shown a very intriguing link between JNK and SUMO signaling pathways in oxidative stress paradigm with H 2 O 2 stimulation [22]. Other links between these two pathways have been reported. JNK activates the c-jun transcription factor while SUMOylation down-regulates it [23]. Additionally, SUMO inhibits the apoptosis signal-regulating kinase 1 (ASK1) activation, an upstream activator of JNK [24].
With this study we aim to elucidate the mechanism that couples JNK, a key kinase in cellular stress, to SUMO during H 2 O 2induced oxidative stress and clarify the impact of SUMOylation on cell death.
To explore the possibility that SUMOylation modulates JNK activity and consequently cellular death following oxidative stress, we stimulated human neuroblastoma SH-SY5Y cells with H 2 O 2 and examined their activation pattern. To study the possible link between JNK and SUMO we over-expressed SUMO-1 or the de-SUMOylation enzyme catalytic sequence of SENP1 in SH-SY5Y cells.
We demonstrated a cross-talk between JNK and SUMO pathways. More specifically protein deSUMOylation prevented JNK activation and cell death. We provided also evidence for a potential interaction between P-JNK and SUMO-1.

H 2 O 2 -induced activation of JNK in SH-SY5Y cells
In the first series of studies we examined the effect of increasing doses of H 2 O 2 (10, 50, 75 and 100 mM) in cell death and JNK activation in undifferentiated human SH-SY5Y neuroblastoma cells.
Cells Over-expression of SUMO-1+UBC 9 and SENP1 plasmids in SH-SY5Y cell lines SH-SY5Y cell lines were successfully transfected with SUMO-1+UBC 9 and SENP1 plasmids (YFP-SUMO-1, Flag-Ubc 9 and GFP-SENP1). The expression level of SUMO-1+UBC 9 was detected in cell lysates by Western blot using the anti-SUMO-1 antibody. A large number of SUMOylated proteins were detected compared to control SH-SY5Y indicating that the SUMO-1+UBC 9 plasmid is functional. On the contrary a significant reduction in SUMOylated proteins is detected in cells in which SENP1 was over-expressed compared to untransfected cells indicating that the SENP1 plasmid is also functional (Fig. 3).

SUMO-1+UBC 9 over-expression increased H 2 O 2 -induced JNK activation in SH-SY5Y
It was recently shown that ROS-dependent JNK activation converges on the SUMO pathway [22] and we here tested the effect of SUMO-1+UBC 9 over-expression on H 2 O 2 -induced activation of JNK.
SH-SY5Y cells over-expressing SUMO-1+UBC 9 were stimulated with increasing concentrations of H 2 O 2 (10, 50, 75 and 100 mM) O/N and 50, 75 mM for 3 h. Western blot analysis was employed to assess the role of SUMO-1 on JNK activation.
As shown in Fig. 1, P-JNK/JNK ratio increased following O/N stimulation with H 2 O 2 . SUMO-1+UBC 9 over-expression did not affect JNK activation and did not prevent cell death ( Fig. 4A-C).
On the contrary, SUMO-1+UBC 9 over-expression led to a significant increase in JNK phosphorylation following stimulation of cells with H 2 O 2 for 3 h. Specifically the P-JNK/JNK ratio increased by 1.

SENP1 over-expression prevents H 2 O 2 -induced activation of JNK in SH-SY5Y
In light of the results obtained with SUMO-1+UBC 9 overexpression we here tested the effect of de-SUMOylation on JNK activation and cell death.
SH-SY5Y cells, over-expressing SENP1, were stimulated O/N with 10, 50, 75 and 100 mM H 2 O 2 . As shown in Figure 4 A, B, SENP1 over-expression led to a remarkable and significant dose dependent prevention of JNK activation compared to control (Fig. 5A, B). Notably, cells that over-express SENP1 were more resistant to oxidative stress compared to control (Fig. 5C).
SH-SY5Y cells over-expressing SENP1 were then stimulated with 50, 75 mM H 2 O 2 for 3h ( Fig. 5D, E). SENP1 SH-SY5Y stimulated with H 2 O 2 presented a powerful reduction of P-JNK/ JNK ratio (Fig. 5D). At 50 mM and of H 2 O 2 and in the presence of SENP1 P-JNK levels were close to control levels (0.8860. 16

P-JNK interacts with SUMO-1 in SH-SY5Y
To investigate whether P-JNK interacts with SUMO-1 we performed immunoprecipitation (IP) and immunostaining experiments.
Cells were transfected with YFP tagged SUMO-1+UBC 9 or YFP tagged SUMO-1DGG+UBC 9 and stimulated with H 2 O 2 (50 mM) for 3 hours. YFP tagged SUMO-1DGG plasmid encodes for a modified form of SUMO-1 in which the 2 glycine-residues responsible for protein SUMOylation bonds have been deleted resulting in a conjugation-deficient SUMO-1 protein. Cell extracts were immunoprecipitated with P-JNK antibody and immunoblotted with anti-GFP antibody, which recognises both SUMO-1 and SUMO-1DGG YFP tag. Association of P-JNK with SUMO-1 was detected both in control and in oxidative stress conditions. (Fig. 6A). SUMOylated proteins in SUMO-1 cells mildly increased by stimulation with 50 mM of H2O2. As expected, SUMOylation was drastically decreased in cells transfected with SUMO-1DGG both in unstimulated and stimulated conditions. Moreover, protein SUMOylation was absent in control conditions where cells were not transfected but still P-JNK is detectable. Such result denotes the specificity of the IP between P-JNK and GFP-tagged proteins. Aspecific protein immunoprecipitation by beads and antibody was verified using a control rabbit-IgG antibody for the IP (Fig. 6A, first lane Ctr Ab does not show any GFP immune reactivity).
Membranes were stripped and re-blotted for P-JNK to show the presence of active JNK in the lysates (Fig. 6A). As expected, in the presence of H 2 O 2 phosphorylation of JNK was increased. These data indicate that in SH-SY5Y cells over-expressing SUMO-1+UBC 9 , P-JNK interacts with SUMO-1. Such an interaction increases in stress conditions. Interestingly, as shown in Figure 6B, endogenous SUMOylation profile was not affected by H 2 O 2 .
Immunostaining further confirmed this result. In non transfected SH-SY5Y cells endogenous SUMO-1 (green) and P-JNK (red) are only mildly expressed (Fig. 6B, CTR). Instead H 2 O 2 for 3 h enhanced the endogenous SUMO-1 signal and led to the formation of characteristic and previously described SUMO nuclear bodies (SNB) [27,28]. Similarly, P-JNK is strongly induced by H 2 O 2 and is principally localized in the nuclei of stimulated cells. In this condition, SUMO-1 and P-JNK colocalized (see merge) within the nucleus (Fig. 6B, 50 mM of H 2 O 2 ).

Discussion
JNK is a key enzyme in the cellular response to stress. Stress signals such as NMDA stimulation, Abeta fragments, hypoxia, reactive oxygen species, ultraviolet radiation, protein synthesis inhibitors can all activate JNK.
We found that H 2 O 2 -induced injury significantly increased JNK activation in SH-SY5Y. We proved that JNK plays a pivotal role in apoptosis, since the specific JNK inhibitor peptide, D-JNKI1, by inhibiting JNK action, totally prevents H 2 O 2 -induced death. These data are in agreement with our previous reports on the importance of JNK signaling in determining cell fate in stress conditions [14,15,29].
Several indications suggested that different stress-stimuli may link the JNK and SUMO pathways [23,30,31]. Similar to JNK, the SUMO pathway is also modulated by stress including heat shock, osmotic and oxidative stress (H 2 O 2 ). Furthermore, SUMOylation is involved in the regulation of members of the JNK signaling cascade, including ASK, an upstream activator of JNK, and c-Jun, a major downstream target of JNK.
We here tried to determine whether JNK and SUMO-1 are interconnected in this model of oxidative stress.
To establish this crosstalk between JNK and SUMO-1 we overexpressed SUMO-1 in SH-SY5Y cells. Surprisingly over-expression of SUMO-1+UBC 9 exacerbated the cellular-stress responses and co-participated with JNK in the apoptotic death mechanism induced by H 2 O 2 -stimulation. More specifically, SUMO-1+UBC 9 over-expression increased JNK activation and H 2 O 2 induced cell death at 3 h. Interestingly this mechanism seems to be more important in conditions of mild oxidative stress (3 h H 2 O 2 ) since significant regulation was not observed following overnight stimulation. One could hypothesise that regulation of JNK activation by SUMO is an early event in the cell death pathway, which could explain why in extreme conditions, when cell viability is markedly decreased such a mechanism is no longer pivotal.
To better clarify the interaction between JNK and SUMO we over-expressed the catalytic portion of the deSUMOylating enzyme, SENP1, in SH-SY5Y cells and examined its role in H 2 O 2 cell injury and JNK activation/regulation. We showed that SENP1 plays an intriguing role in this condition, since it is able to prevent JNK activation as well as cell death in both O/N and 3 h paradigms. This corroborates our previous finding and supports the idea that SUMO is not playing a protective role but instead participates with JNK in the stress cascade of H 2 O 2injury in SH-SY5Y. Consequently, by preventing SUMOylation, JNK activation is inhibited and prevents the specific cellular responses to mild as well as severe oxidative stress. To determine how P-JNK-SUMO interaction influences the cellular death pathway further investigation is needed as well as more suitable model. It will be of great interest to study this interaction and SUMO role in brain/neurons both in pathological and physiological conditions.   Our findings are in contrast with some papers [32,33] suggesting that SUMO-1 may play a protective role in models of oxidative stress. It is agreed that oxidative stress is complex and the data are rather confusing since high concentrations of H 2 O 2 lead to increased SUMO conjugation while low H 2 O 2 concentrations (1 mM and below) lead to almost complete loss of SUMO- 1, -2 and -3 conjugates within the first 30 min [34]. In our model SUMO-1 conjugates remained stable, but the conditions and cells tested differ from the ones reported in the literature, thus it is very difficult to compare results.
In our model of oxidative stress inhibition of SUMOylation, with over-expression of the SENP1-de-SUMOylation enzyme, led to inhibition of JNK and to partial protection against H 2 O 2induced death. On the contrary, over-expression of SUMO-1 increased somewhat JNK activation and deteriorated cell death. This crosstalk between JNK and SUMO-1 pathways brought us to search a physical interaction connecting P-JNK to SUMO-1. Immunoprecipitation experiments convincingly proved that P-JNK interacts with SUMO-1 in both control (unstimulated) and H 2 O 2 -stimulated SH-SY5Y cells that overexpressed SUMO-1. Such interaction seems to increase in the presence of oxidative stress. These data were confirmed by immunostaining, where we could show that P-JNK colocalises with SUMO-1 in the nuclear bodies only in cells that are exposed to H 2 O 2 stimulus. Altogether, this study is the first to present evidence for a direct interaction between JNK and SUMO in cellular death processes. The implication of this interaction requires further investigation. Although more studies are required it is tempting to speculate that such an interaction occurs also in neuronal cells and that deSUMOylation can be protective in neuronal loss occurring in several neurodegenerative diseases. Such studies are currently underway.
We here proved that SUMO interacts with the active form of JNK. For this reason and for the fact that the cell specific JNK inhibitor peptide totally prevented H 2 O 2 -induced cell death we suggest that SUMO-1 co-partecipates in the stress activated process by incrementing JNK activity. This is the first report proposing that SUMO may not play a protective role but on the contrary can be implicated in death pathways.
Cells were seeded into 6-well plates at density of 4610 5 or 12well plates at density of 2610 5 cells, one day before transfection.

Transfection and hydrogen peroxide stimulation
SH-SY5Y cells were transfected in OPTI-MEM+glutaMAX medium (Invitrogen -GIBCO) using Lipofectamine 2000 (Invitrogen) according to the manufacturer's procedure. 0.5 mg (for 6 wells) or 0.2 mg (for 12 wells) of each DNA was transfected per well. Empty plasmid was used where required to maintain equal total DNA quantity. 3 hours later transfection OPTI-MEM+glutaMAX medium was replaced by complete medium.
After washes membranes were incubated for 1 h at room temperature with appropriate secondary antibody: anti-Mouse, 1: 4000 (Santa Cruz Biotechnology) or anti-Rabbit, 1:4000 (Santa Cruz Biotechnology) and anti-Rabbit Light Chain Specific (Jackson ImmunoResearch) for IP experiments. After washes immunoreactive bands were detected by enhanced chemiluminescence (ECL). In all experiments where an evaluation of phosphorylated proteins was required, antibodies were applied on the same membranes after stripping procedure (stripping buffer from Pierce). Actin was used as a loading control. The immunoreactive bands were visualized by exposure to Amersham Hyperfilm ECL (Ge Healthcare). Western blots were quantified by densitometric analysis using ImageJ software.

MTT assay
MTT (Thiazolyl Blue Tetrazolium Bromide-SIGMA) colorimetric assay was employed to test cell viability in oxidative stress conditions. A 106 stock solution was prepared by dissolving 4 mg/mL of MTT powder in PBS 16 (Invitrogen -GIBCO).
200 mL MTT solution per well were added on cells, in 6-well plates, incubated at 37uC for 4 h. Medium was removed and cells were treated with a solution prepared by diluting 1:25 HCl 1N with isopropanol (100%). Supernatants containing solubilized MTT crystals were analyzed at 540 nm to the spectrophotometer.
Nuclear staining was obtained applying Hoechst solution (Roche) 1:5000 for 5 min at room temperature. Cells were washed with 16 PBS after every step. Coverslips were mounted in Fluorsave mounting medium (Calbiochem, 345789). Staining was acquired with a Olympus microscope equipped with a Olympus Confocal scan unit (microscope BX61 and Confocal system FV500) managed by AnalySIS Fluoview software with 3 lasers line, UV-diode laser (405 nm), Ar-Kr (488 nm), He-Ne green (546 nm), respectively used to detect Hoechst staining and secondary antibody conjugated to Alexa 488 and Alexa 546. Double staining was revealed with a scanning sequential mode to eliminate possible bleed-through effect.

Immunoprecipitation
Cells were lysed in 500 mL (around 300 mg of proteins) of Lysis Buffer (LB) and a preclearing step was performed adding for each sample 15 mL of washed (TBS buffer: 50 mM Tris, 150 mM NaCl pH 7.5) magnetic protein G sepharose beads (GE Healthcare) for 1 h at 4uC on wheel. Rabbit anti-P-JNK antibody (Cell Signaling; 1:200) was then added to supernatants for 30 min at 4uC on wheel. 15 mL of washed beads where again added to each sample for O/N incubation. Beads were washed three times with TBS buffer and then 100 mL SDS-sample buffer was added. 20 mL of each sample was used for western blotting.

Statistics
Statistical analysis was performed with GraphPad PRISM 5. Values shown represent the mean 6 SEM of at least five separate experiments. One-way or Two-ways ANOVA followed by Tukey's test was carried out for intergroup comparisons were required.