Activation of AMPK by OSU53 protects spinal cord neurons from oxidative stress

The present study tested the potential effect of OSU53, a novel AMPK activator, against hydrogen peroxide (H2O2)-induced spinal cord neuron damages. Treatment with OSU53 attenuated H2O2-induced death and apoptosis of primary murine spinal cord neurons. OSU53 activated AMPK signaling, which is required for its actions in spinal cord neurons. The AMPK inhibitor Compound C or AMPKα1 siRNA almost abolished OSU53-mediated neuroprotection against H2O2. On the other hand, sustained-activation of AMPK by introducing the constitutive-active AMPKα1 mimicked OSU53's actions, and protected spinal cord neurons from oxidative stress. OSU53 significantly attenuated H2O2-induced reactive oxygen species production, lipid peroxidation and DNA damages in spinal cord neurons. Additionally, OSU53 increased NADPH content and heme oxygenase-1 mRNA expression in H2O2-treated spinal cord neurons. Together, we indicate that targeted-activation of AMPK by OSU53 protects spinal cord neurons from oxidative stress.


INTRODUCTION
Oxidative stress following the spinal cord injury is a main cause of secondary damages to neurons [1][2][3]. The levels of different reactive oxygen species (ROS), i.e. superoxide anion, hydroxyl free radicals and hydrogen peroxide (H 2 O 2 ), are significantly increased in the spinal cord following traumatic and non-traumatic injuries [4]. ROS production shall led to a number of harmful effects, including lipid peroxidation, protein oxidation, DNA damage, and eventually spinal cord neuronal death and apoptosis [1][2][3][4].

OSU53 protects murine spinal cord neurons from H 2 O 2
To mimic oxidative stress in vitro, primary cultured murine spinal cord neurons were treated with hydrogen peroxide (H 2 O 2 ) at different concentrations, from 50-
Decreased cell survival could be due to cell death. The release of lactate dehydrogenase (LDH) to the conditional medium is the well-established marker of cell death. We here demonstrated that treatment with 100-400 μM of H 2 O 2 significantly increased medium LDH release, suggesting spinal cord neuronal death ( Figure 1B). Such cytotoxic effect was again largely attenuated with co-treatment of OSU53 (10 μM) ( Figure 1B). The potential dose response of OSU53 was also tested. As demonstrated in Figure 1C, co-treatment with OSU53 at 1-10 μM significantly attenuated H 2 O 2 (200 μM)-induced viability reduction of the spinal cord neurons. At a lower concentration (0.5 μM), this novel AMPK activator was in-effective against H 2 O 2 ( Figure 1C), thus confirming a dose-dependent response. Collectively, these results suggest that OSU53 protects murine spinal cord neurons from H 2 O 2 .

OSU53 inhibits H 2 O 2 -induced spinal cord neuron apoptosis
Next, the potential effect of OSU53 on H 2 O 2 -induced neuronal apoptosis was analyzed. As shown in Figure 2A, treatment with H 2 O 2 (200 μM) in cultured murine spinal cord neurons induced cleavage of both caspase-9 and PARP [poly (ADP-ribose) polymerase]. Meanwhile, the caspase-9 activity was also increased in H 2 O 2 -treated neurons ( Figure 2B). Such actions by H 2 O 2 were largely attenuated with co-treatment of OSU53 (Figure 2A-2B). Additionally, H 2 O 2 stimulation in the neurons also induced production of single strand DNA ("ssDNA") ( Figure 2C), mitochondrial depolarization ( Figure 2D, JC-1 OD increase [20]) and TUNEL (terminal transferase uridyl nick end labeling) nuclei increase ( Figure 2E), which all largely attenuated by OSU53. Therefore, these results suggest that OSU53 inhibits H 2 O 2 -induced spinal cord neuronal apoptosis.
To study the link between AMPK activation and OSU53-mediated neuroprotection, pharmacological and genetic strategies were employed to block AMPK activation. Compound C is a well-established AMPK inhibitor [23]. We here demonstrated that compound C blocked OSU53-induced AMPK activation, or AMPKα1/ ACC phosphorylations, in H 2 O 2 -treated spinal cord neurons ( Figure 3B). Further, siRNA strategy was utilized to knockdown AMPKα1. The applied AMPKα siRNA (Santa Cruz Biotech Co) efficiently downregulated AMPKα1 in the spine cord neurons ( Figure 3B). Subsequently, OSU53induced AMPK activation was inhibited ( Figure 3B). Remarkably, OSU53-mediated neuroprotection against H 2 O 2 was almost abolished by Compound C or AMPKα1 siRNA ( Figure 3C and 3D). OSU53 was ineffective to protect spine cord neurons when AMPK was inhibited or silenced ( Figure 3C and 3D). These results suggest that AMPK activation is required for OSU53-mediated neuroprotection. It should be noted that Compound C or AMPKα1 siRNA also exacerbated H 2 O 2 -induced cytotoxicity ( Figure 3C and 3D), arguing that AMPK activation by H 2 O 2 itself is also neuroprotective.

Constitutive activation of AMPK protects spinal cord neurons from H 2 O 2
Based on the above-mentioned results, we can speculate that forced-activation of AMPK shall also protect spinal cord neurons from oxidative stress. To test this hypothesis, a constitutive-active AMPKα1 (T172D, caAMPKα1, by Dr. Wang [9]) was transfected to the spinal cord neurons. The caAMPKα1 construct had the constitutive phosphorylation of AMPKα1 at Thr-172, leading to sustained AMPK activation [9,24]. Western blotting assay results confirmed expression of caAMPKα1 (tagged with Flag) in the spine cord neurons ( Figure 4A). AMPK activation, tested by p-ACC, was significantly increased in caAMPKα1-expressing neurons, even in the presence of H 2 O 2 ( Figure 4A). Remarkably, forced-expression of caAMPKα1 protected the spinal cord neurons from H 2 O 2 ( Figure 4B and 4C). Thus, sustained activation of AMPK by caAMPKα1 mimicked OSU53's actions and protected neurons from oxidative stress. It should be noted that adding OSU53 in caAMPKα1-expressing neurons failed to further inhibit H 2 O 2 -induce damages ( Figure 4B and 4C). Thus, OSU53 www.impactjournals.com/oncotarget was ineffective in AMPK-pre-activated neurons, again confirming that AMPK activation is required for OSU53mediated neuroprotection against H 2 O 2 .

OSU53 inhibits H 2 O 2 -induced oxidative stress in spinal cord neurons
Recent studies have confirmed a cytoprotective function of AMPK under various stress conditions. In particularly, activated AMPK is capable of fighting oxidative stress [25,26]. AMPK inhibits reactive oxygen species (ROS) production, thus protecting cells from oxidative stress [27]. Here, we show that treatment with OSU53 also significantly attenuated H 2 O 2 -induced ROS production (DCFH-DA fluorescent dye intensity) in spinal cord neurons ( Figure 5A). Consequently, H 2 O 2 -induced lipid peroxidation ( Figure 5B) and DNA damages (tested by γ-H2AX percentage, Figure 5C) were also dramatically alleviated by OSU53 pre-treatment. These results suggest that targeted-activation of AMPK by OSU53  Oncotarget 112480 www.impactjournals.com/oncotarget significantly inhibited H 2 O 2 -oxidative stress in spinal cord neurons, which could be the major reason to explain the neuroprotection activity by this novel AMPK activator.
There are proposed mechanisms responsible for AMPK-mediated anti-oxidant activity. As discussed, AMPK maintains NADPH homeostasis to promote cell survival under stress conditions [7-9, 19, 25]. AMPK is also positively involved in activation of Nrf2 (nuclear factor erythroid 2-related factor 2) [28,29], the latter is a major anti-oxidant transcription factor responsible for transcription of heme oxygenase (HO)-1 and many other key anti-oxidant genes [30][31][32][33][34]. Here, we show that OSU53 treatment also induced NADPH synthesis in the spinal cord neurons ( Figure 5D). More importantly, although H 2 O 2 decreased NADPH content in the neurons ( Figure 5D), after co-treatment OSU53, its level was restored ( Figure 5D). Further studies showed that OSU53 treatment in spinal cord neurons also induced transcription of HO-1 mRNA ( Figure 5E). Additionally, H 2 O 2 -induced HO-1 mRNA expression was further enhanced with OSU53 co-treatment ( Figure 5E). These results suggest that OSU53-mediated anti-oxidant response could be due to AMPK-dependent NADPH production and/or Nrf2-HO-1 signaling activation.

DISCUSSION
In the current study, we show that OSU53 activated AMPK signaling, which is required for its activity in spinal cord neurons. Compound C, the AMPK inhibitor, or AMPKα1 siRNA almost abolished OSU53-mediated neuroprotection against H 2 O 2 . On the other hand, sustainedactivation of AMPK by introducing the ca-AMPKα1 (T172D) mimicked OSU53's actions and protected spinal cord neurons from oxidative stress. More importantly, adding OSU53 was ineffective in ca-AMPKα1-expressing Oncotarget 112481 www.impactjournals.com/oncotarget neurons against H 2 O 2 . Therefore, targeted-activation of AMPK by OSU53 protects the spinal cord neurons from oxidative stress.
A number of recent studies have proposed the anti-oxidant function of activated AMPK under stress conditions [7,8,35,36]. AMPK was initially found to  Oncotarget 112482 www.impactjournals.com/oncotarget be vital for maintaining cellular NADPH content [7]. AMPK phosphorylates and in-activates its major substrate protein ACC, thus reducing NADPH consumption [7]. Additionally, AMPK activation shall promote NADPH synthesis and production [25,37,38] through the mechanisms involving fatty-acid oxidation [7]. The study by Guo's group demonstrated that targeted-activation of AMPK by Compound 13, the α1 specific AMPK activator [39,40], increased NADPH production to inhibit dexamethasone (Dex)-induced oxidative stress, thus offering protection to human osteoblastic cells [36]. Similarly, miR-135b downregulated Ppm1e, the AMPK phosphatase [41][42][43], to activate AMPK and protect cells from oxidative stress [25]. Here, we show that targetedactivation of AMPK by OSU53 also increased NADPH content in the spinal cord neurons. More importantly, co-treatment with the novel AMPK activator also restored NADPH content in H 2 O 2 -treated neurons. Thus, AMPKmediated NADPH axis could be one key mechanism responsible for OSU53-mediated neuroprotection against oxidative stress.
It is estimated that there will over 150,000-200,000 cases of traumatic spinal cord injury annually due to accidents and violence [1,2,4]. The clinical and bench results have indicated that severity of neurological deficit is not solely dependent on the degree of the mechanical insults [1,2,4]. It is, interestingly, very closely associated with secondary damaging factors, including ischaemia, oedema, inflammatory response and most importantly oxidative stress [1,2,4]. The results of this study suggest that targeted-activation of AMPK by OSU53 efficiently protects the spinal cord neurons from oxidative stress, suggesting that this novel AMPK activator might be further tested in animal and clinical models of spinal cord injury.

Murine spinal cord neurons
The detailed protocol of culture of primary murine spinal cord neurons was described in detail previously [48]. In brief, the spinal cord of C57BL/6J mouse embryos (E13- 14) were initially plated at 200,000 cells/mL onto the tissue-culture plate, pre-coated with poly-l-lysine (Sigma). The neuronal feeding medium, 90% MEM, 5% horse serum, 5% fetal bovine serum (FBS) (GIBCO), and a mixture of nutrient supplements [48], was replaced with feeding medium every 2-3 days.

Survival assay
The Cell Counting Kit-8 (CCK-8) kit, purchased from the Dojindo Laboratories (Kumamoto, Japan), was employed to examine the viability of spinal cord neurons. CCK-8 optic density (OD, at 450 nm) of treatment group was always normalized to that of untreated control group.

Death assay
The release of lactate dehydrogenase (LDH) to the conditional medium is a well-established marker of neuronal cell death, which was tested by a commercial available twostep LDH assay kit (Takara, Tokyo, Japan) [49].

Enzyme-linked immunosorbent assay (ELISA) of cell apoptosis
The increased production of single stream DNA (ssDNA) is a characteristic marker of cell apoptosis [50]. ssDNA content in spinal cord neurons was tested by a commercial available ssDNA ELISA kit (Chemicon International, Temecula, CA). ELISA OD (at 405 nm) was recorded to reflect cell apoptosis.

TUNEL assay
Spinal cord neurons with described treatment were stained with TUNEL and DAPI florescence dyes (All purchased from Sigma). TUNEL staining ratio (vs. DAPI nuclei number) was recorded. www.impactjournals.com/oncotarget Caspase-9 activity assay and mitochondrial depolarization assay The caspase-9 activity assay by the caspase-Glo 9 kit (Promega, Shanghai, China) and mitochondrial membrane potential (MMP) reduction assay by the JC-1 dye (Invitrogen, Shanghai, China) were described in detail in previous studies [20,51].

Western blotting assay
Spinal cord neurons were first trypsinized and washed. Neurons were then incubated with the cell lysis buffer (Biyuntian, Wuxi, China). Protein concentration was determined using the Bio-Rad protein assay kit (Shanghai, China). For each single treatment, exact 30-40 μg total lysate proteins (per lane) were separated by the SDS-PAGE gels (7.5-10%) [52,53], which were then transferred to the PVDF blots (Millipore, Suzhou, China). After blocking, the blots were incubated with designated primary and secondary antibodies. To visualize the targeted bands, the enhanced chemiluminescence (ECL) reagents (Pierce) were added, and X-Ray film development was employed.

qRT-PCR assay
Spinal cord neurons were incubated with TRIzol reagents (Invitrogen) to achieve total cellular RNA. Quantitative Real-time PCR ("qRT-PCR") assay was performed BY the SYBR green kit on the ABI-7600 fast PCR system (Applied Biosystems, Shanghai, China) [33]. mRNA primers for murine HO-1 and GAPDH were described previously [47,54]. GAPDH mRNA was always tested as the internal control and reference gene. The 2 -ΔΔCt method was employed to calculate relative HO-1 mRNA expression (vs. GAPDH mRNA).

siRNA
The siRNA of murine AMPKα1 was purchased from Santa Cruz Biotech. The AMPKα1 siRNA or the scramble control siRNA (200 nM, 12 hours × two rounds) transfection to spinal cord neurons was performed using the Lipofectamine 2000 reagent (Invitrogen). Afterwards, AMPKα1 knockdown in the resulting neurons was verified by Western blotting assay.

NADPH assay
The NADPH content was tested via the methods described previously [8,56]. Spinal cord neurons with the described treatment were trypsinized and washed with PBS, and incubated with the lysis buffer (Biyuntian, Wuxi, China). The protein lysates were then incubated with NADP-cycling buffer along with glucose-6-phosphate dehydrogenase (G6PD, Sigma) at 60 °C for 30 min [8], which was followed by addition of glucose 6-phosphate (G6P, Sigma). Thereafter, G6P's absorbance change at 570 nm was tested every 30 seconds for 4 min. NADPH content was then recorded and normalized to that of untreated control group [56].

Reactive oxygen species (ROS) assay
The dichlorofluorescin diacetate (DCFH-DA) fluorescent dye (Invitrogen) was employed to quantify cellular ROS intensity [25]. In brief, 1 μM of DCFH-DA were added to the cultured spinal cord neurons for 45 min. Neurons were then washedS, and subjected to examination of fluorescence intensity under a Fluorescence Microplate Reader (Synergy 2, BioTek, Winooski, VT). ROS intensity OD was expressed as fold change of untreated control group.

Lipid peroxidation assay
The thiobarbituric acid reactive substances (TBAR) assay was employed to test the cellular lipid peroxidation intensity. The detailed protocol was described previously [57,58].

DNA damage assay
In the current study, γ-H2AX intensity was tested to reflect cellular DNA damages [59]. In short, spinal cord neurons with the applied treatment were trypsinized and fixed. Afterwards, anti-γ-H2AX antibody (Cellular Signaling Tech) was added to the neurons for additional 4 hours. The FITC-conjugated secondary antibody was then added to theγ-H2AX-labbled neurons, which were then subjected to FACS assay. γ-H2AX percentage was recorded [59].

Statistical analysis
The results were expressed as mean ± standard deviation (SD). The statistical analysis among different groups was done using one-way ANOVA with Scheffe's test. Experiments were repeated at least three times and consistent results were always obtained.

Author contributions
All listed authors have contributed significantly to the study.