A novel AMPK activator hernandezine inhibits LPS-induced TNFα production

Here, we found that hernandezine, a novel AMPK activator, inhibited LPS-induced TNFα expression/production in human macrophage cells (THP-1 and U937 lines). Activation of AMPK is required for hernandezine-induced anti-LPS response. AMPKα shRNA or dominant negative mutation (T172A) blocked hernandezine-induced AMPK activation, which almost completely reversed anti-LPS activity by hernandezine. Exogenous expression of the constitutively activate AMPKα (T172D, caAMPKα) also suppressed TNFα production by LPS. Remarkably, hernandezine was unable to further inhibit LPS-mediated TNFα production in caAMPKα-expressing cells. Hernandezine inhibited LPS-induced reactive oxygen species (ROS) production and nuclear factor kappa B (NFκB) activation. Treatment of hernandezine in ex-vivo cultured primary human peripheral blood mononuclear cells (PBMCs) also largely attenuated LPS-induced TNFα production. Together, we conclude that AMPK activation by hernandezine inhibits LPS-induced TNFα production in macrophages/monocytes.

Hernandezine is an alkaloid isolated from Chinese medicinal herb manyleaf meadowure rhizome and root [20]. A very recent study by Law et al., has characterized hernandezine as a novel AMPK activator [20]. In the current report, we show that hernandezine inhibits LPSinduced TNFα production via activating AMPK signaling.

RESULTS
The effect of hernandezine on macrophage cell survival and TNFα production First, we tested the potential effect of hernandezine on the survival of human macrophage cells. U937 cells (macrophage cell line [11]) were treated with gradually increased concentrations of hernandezine (1-100 μM) for 24 hours, trypan blue staining assay [11] was applied to test cell survival. Viable cells were trypan blue negative [11]. Results in Figure 1A demonstrated that hernandezine was not cytotoxic to U937 cells until at 100 μM, the latter induced obvious U937 cell death ( Figure 1A). Histone DNA apoptosis ELISA assay [10,11] results in Figure 1B demonstrated that only 100 μM of hernandezine induced significant U937 cell apoptosis. It was not pro-apoptotic at lower concentrations ( Figure 1B). As shown in Figure  1C, treatment with hernandezine (1-30 μM, non-cytotoxic concentrations) failed to change basal TNFα production in U937 cells. However, at 100 μM, hernandezine inhibited TNFα production ( Figure 1C), which could be probably due to cell death ( Figure 1A and 1B). The similar experiments were also performed in the other human macrophage cell line: THP-1 [11]. Results showed that hernandezine was indeed not cytotoxic ( Figure 1D) nor pro-apoptotic ( Figure 1E) to THP-1 cells until at a high concentration (100 μM). Basal TNFα production in THP-1 cells was also not changed in hernandezine-treated THP-1 cells, except at 100 μM ( Figure 1F).

Activation of AMPK is required for hernandezine-induced anti-LPS response
Our group [10,11] and others [15,16,19] have implied that activation of AMPK could inhibit LPS-induced pro-inflammatory response. The study by Law et al., has confirmed that hernandezine is a novel AMPK activator [20]. We therefore tested AMPK signaling in hernandezine-treated macrophage cells. As shown in Figure 3A, hernandezine dose-dependently induced AMPK activation in U937 cells. As phosphorylated-("p-") AMPKα and p-acetyl-CoA carboxylase (p-ACC, the main downstream target protein of AMPK [12,21]) were significantly increased after treatment of 10 and 30 μM of hernandezine (See quantified blot results in Figure  3A). On the other hand, 1 μM and 3 μM of hernandezine failed to induce significant AMPK activation (See quantified blot results in Figure 3A). Total AMPKα and ACC expression was unchanged following hernandezine treatment.

Hernandezine inhibits LPS-induced ROS production and NF-kB activation
As discussed, forced-activation of AMPK was shown to efficiently suppress LPS-induced ROS production and subsequent nuclear factor kB (NF-kB) activation [10,11,15,16,19], leading to TNFα transcription inhibition in monocytes/macrophages. Here, we showed that treatment with hernandezine (30 μM) in U937 cells largely attenuated LPS-induced ROS production ( Figure 5A). Further, NF-kB activation in LPS-treated U937 cells was also inhibited by hernandezine ( Figure 5B). Importantly, AMPKα knockdown (by targeted shRNA, shAMPKα-No.1) or mutation (by expressing dnAMPKα) almost abolished hernandezine-induced inhibition on ROS and NF-kB ( Figure 5A and 5B). These results suggest that AMPK activation is required for hernandezine-induced inhibition on ROS production and NF-kB activation in LPS-treated cells.

Hernandezine inhibits LPS-induced TNFα production in primary human peripheral blood mononuclear cells (PBMCs)
At last, we tested the potential activity of hernandezine in human monocytes. In consistent with our previous studies [9,10], primary PBMCs from COPD patients were ex-vivo cultured. Trypan blue assay results in Figure 6A showed again that treatment with 30 μM of hernandezine (or plus LPS) was non-cytotoxic to the primary PBMCs. Significantly, hernandezine remarkably inhibited LPS-induced TNFα mRNA expression ( Figure  6B) and production ( Figure 6C). Therefore, in line with the cell line data, hernandezine similarly inhibits LPS- induced TNFα production and expression in primary human PBMCs.

DISCUSSION
We have previously shown that GSK621, the novel AMPK activator [25], attenuated LPS-induced TNFα production [10]. Similarly, our very recent study has demonstrated that microRNA-135b-5p ("miR-135b-5p") inhibited LPS-induced TNFα production via activating AMPK [11]. miR-135b-5p activated AMPK signaling via silencing its phosphatase Ppm1e [11]. These results indicate that AMPK activation could be a novel and efficient strategy to inhibit LPS-induced pro-inflammatory response. Here, we showed that AMPK activation is also required for hernandezinemediated anti-LPS response. AMPK knockdown (by targeted shRNAs) or dominant negative mutation  blocked hernandezine-induced AMPK activation, and almost completely reversed its anti-LPS activity. Meanwhile, exogenous expression of caAMPKα also inhibited LPS-induced TNFα production. Importantly, hernandezine was almost invalid against LPS in the caAMPKα-expressing cells. Thus, activation of AMPK by hernandezine is responsible for its anti-LPS activity in macrophage cells.
LPS, which is sensed by CD14 and LPS-binding protein, binds to Toll-like receptor 4 (TLR-4) on macrophages/monocytes [26,27], which will activate downstream NFκB signaling to initiate pro-inflammatory response [26,27]. ROS production is known to be critical in the process. Recent studies [19,28,29] including ours [11] have implied that AMPK could be effective in suppressing oxidative stresses. For example, energy depletion-activated AMPK-ACC signaling was shown to increase intracellular nicotinamide adenine dinucleotide phosphate (NADPH) content and to inhibit oxidative stress [28]. She et al., demonstrated that activation of AMPK significantly suppressed H 2 O 2 -induced oxidative damages [29]. Similarly, AMPK activation by cordycepin inhibited LPS-induced ROS accumulation [19]. Our previous studies have shown that GSK621 [10] or miR-135b-5p [11] inhibited LPS-induced ROS production, thus blocking the downstream NFκB activation. This could be the key mechanism responsible for AMPK-induced anti-LPS activity.
In line with these findings, we show that hernandezine largely inhibited LPS-induced ROS production and NFκB activation in U937 cells. AMPK inhibition, by targeted shRNA or dominant negative mutation, almost completely reversed hernandezine's above actions. Thus, we conclude that hernandezine activates AMPK signaling to inhibit LPS-induced ROS production and subsequent NFκB activation, which then leads to decreased TNFα mRNA synthesis and production. The detailed mechanism may warrant further investigations.

Chemicals and antibodies
Hernandezine was purchased from EFE-Bio Company (Shanghai, China). LPS and puromycin were provided from Sigma Chemicals (Shanghai, China). The antibodies were all obtained from Cell Signaling Technology (Danvers, MA). Cell culture reagents were provided by Hyclone (Shanghai, China).

Cell culture
As described previously [11], the two human macrophage cell lines, U937 and THP-1, were cultured in RPMI 1640 medium supplemented with 10% FBS and 1% glutamine at 37 °C.

Ex-vivo culture of human PBMCs
As described previously [9,10], PBMCs of COPD patients (administrated at the Second Affiliated Hospital of Xi'an Jiao Tong University, Xi'an, China) were collected via lymphocyte separation medium (Sigma, Shanghai, China). The resulting PBMCs were cultured in DMEM plus 10% FBS, and necessary supplements [30]. Experiments and protocols requiring human samples were approved by the Ethics Committee and Internal Review Board of Xi'an Jiao Tong University. Written-informed consent was provided by each patient.

TNFα enzyme-linked immunosorbent assay (ELISA) assay
TNFα protein content in the conditional medium was tested by the TNFα ELISA kit (R&D Systems, Abingdon, UK), and detailed protocol was described previously [9].

Western blotting assay
As described [9][10][11], after applied treatment, cells were lysed, the protein lysates (20 μg per sample) were separated by the SDS-PAGE gel (10-12%). Protein samples were then transferred onto PVDF membranes, which were then probed with indicated primary and corresponding secondary antibodies. The indicated bands were then visualized by the enhanced chemiluminescence (ECL, Amersham, Shanghai, China) regents [9].

AMPKα mutation
The pSuper-puro construct with dominant negative AMPKα (T172A), the constitutively-active AMPKα (T172D), and the empty vector were provided by Dr. Lu's group [10,11,34]. We utilized Lipofectamine 2000 to transfect the mutant AMPKα or the empty vector to U937 cells. Stable cells were again selected by puromycin.

ROS assay
The detailed protocol for ROS assay was described previously [10,11,19]. Briefly, ROS content in cells with applied treatment was measured by dichlorofluorescin (DCF) oxidation assay. Cells were incubated with 10 μM of DCFH-DA (Invitrogen, Shanghai, China) for 30 min, and were then washed in PBS for three times. DCF fluorescence intensity was then tested [10,11,19].

Measuring NFκB (p65) DNA-binding activity
NFκB (p65) DNA-binding activity was tested as described in our previous studies [9][10][11]. Briefly, after applied treatment, 1.0 μg of cell nuclear extracts per treatment were analyzed of the NFκB (p65) DNA-binding activity, via the TransAM™ ELISA kit (Active Motif, Carlsbad, CA) according to the recommended protocol. The OD value of treatment group was normalized to that of control group to reflect relative NFκB activity.

Statistics analysis
The statistical analyses were performed via the SPSS software (18.0), with p < 0.05 taken as significant. Data were expressed as mean ± standard deviation (SD). For comparisons among multiple groups, two-way ANOVA with the Bonferroni post hoc testing was performed.