Nmnat2 attenuates amyloidogenesis and up-regulates ADAM10 in AMPK activity-dependent manner

Amyloid-β (Aβ) accumulating is considered as a causative factor for formation of senile plaque in Alzheimer’s disease (AD), but its mechanism is still elusive. The Nicotinamide mononucleotide adenylyltransferase 2 (Nmnat2), a key redox cofactor for energy metabolism, is reduced in AD. Accumulative evidence has shown that the decrease of α-secretase activity, a disintegrin and metalloprotease domain 10 (ADAM10), is responsible for the increase of Aβ productions in AD patient’s brain. Here, we observe that the activity of α-secretase ADAM10 and levels of Nmnat2 are significantly decreased, meanwhile there is a simultaneous elevation of Aβ in Tg2576 mice. Over-expression of Nmnat2 increases the mRNA expression of α-secretase ADAM10 and its activity and inhibits Aβ production in N2a/APPswe cells, which can be abolished by Compound C, an AMPK antagonist, suggesting that AMPK is involved in over-expression of Nmnat2 against Aβ production. The further assays demonstrate that Nmnat2 activates AMPK by up-regulating the ratio of NAD+/NADH, moreover AMPK agonist AICAR can also increase ADAM10 activity and reduces Aβ1-40/1-42. Taken together, Nmnat2 suppresses Aβ production and up-regulates ADAM10 in AMPK activity-dependent manner, suggesting that Nmnat2 may serve as a new potential target in arresting AD.


INTRODUCTION
Alzheimer's disease (AD) is the most prevalent neurodegenerative disorder affecting more than 44 million patients worldwide [1]. It is characterized as a slow loss of cognitive function, leading to dementia and death [2]. The neuropathologic hallmarks of AD include extracellular protein deposits of amyloid plaques and intracellular phosphorylation and tangles of neurofibrillar tau protein [3,4]. Amyloid-beta (Aβ), a 38-42 amino acid long peptide, is prone to aggregation and forms the major constituent of plaques [5,6]. Aβ peptides and its pathogenic aggregations initiate a cascade of widespread protein depositions and induce neurodegeneration diseases, such as AD [7,8]. Aβ is generated from the sequential proteolytic processing of amyloid precursor protein (APP) by the enzymes βand γ-secretase [9,10]. Indeed, Aβ production can be avoided by an alternate APP cleavage pathway mediated by the α-secretases in the normal brain [11]. Interestingly, α-secretase cleavage of APP generates a secreted form of APP (sAPP), which has been shown to play a role of neurotrophy and neuroprotection in AD models [12,13].
A disintegrin and metalloprotease (ADAM) is a multifunctional gene family including ADAM9, ADAM10 and ADAM17, which can act as α-secretase in various cell lines [14,15]. ADAM10 is particularly expressed in mouse and human brain, and is the physiologically essential constitutive α-secretase cleavage of APP, but not ADAM9 or 17 [16][17][18]. The up-regulation of ADAM10 in AD model mice prevents formation of Aβ plaques and cognitive deficits [19,20]. It is reported that ADAM10 is reduced in AD brain or in cerebrospinal fluid (CSF) and platelets from AD patients [21][22][23], but the upstream effectors of regulating α-secretase ADAM10 are still not fully ascertained. Therefore, to explore upstream targets of regulating α-secretase ADAM10 are valuable for developing drugs to reduce Aβ and cure AD.
Nicotinamide mononucleotide adenylyltransferase (Nmnat) is a vital enzyme in catalyzing the synthesis of Nicotinamide adenine dinucleotide (NAD + ) from NMN [24,25]. There are three Nmnat isoforms including Nmnat1, Nmnat2 and Nmnat3, which have been identified in human and mouse brains [26][27][28]. Compared with Nmnat1 and Nmnat3, Nmnat2 is specifically and highly expressed in brain [29,30], and over-expression of Nmnat2 shows a neuroprotective effect in AD models of Drosophila and rodents [31][32][33][34]. Indeed, accumulative evidence has shown that mRNA of Nmnat2 is reduced in the brains of AD patients and mouse models of AD [35][36][37]. All of previous research suggest that Nmnat2 is involved in AD development, but its underlying mechanism is not fully elucidated.
Here, we find that the accumulation of Aβ increases in parallel with reduction of ADAM10 in Tg2576 mice. The over-expression of Nmnat2 suppresses amyloidogenesis and activates ADAM10 in AMPK activity-dependent manner in N2a/APPswe cells, and Nmnat2 activates AMPK through increasing the ratio of NAD + /NADH.

Immunofluorescence and ELISA
Immunofluorescence and ELISA were carried out according to the procedure described previously [38][39][40].
For cell studies, N2a/APPswe cells were cultured on coverslips and fixed with 4% paraformaldehyde.
The levels of Aβ in the medium of N2a/APPswe cells transfected with Nmnat2 or CC treatment were measured by a sandwich ELISA kit using an anti-Aβ N-terminal antibody and an anti-Aβ1-40 or Aβ1-42C-terminal antibody, according to the manufacturer's instructions (Biosource International, Camarillo, CA, USA).

Western blot
Western blot was performed according to the established methods in our laboratory [41,42]. In brief, the cortex was rapidly removed from brain and homogenized on ice using a Teflon glass homogenizer in the buffer containing 50 mM Tris-HCl, pH 7.4, 1% NP-40, 0.25% Na-deoxycholate, 150 mM NaCl, 1 mM Na3VO4, 1 mM EDTA, 50 mM NaF, 1 mM PMSF and protease inhibitors mixture (2 mg/l each of aprotinin, leupeptin and pepstain A). The homogenates were mixed with loading buffer (3:1, v/v) containing Tris-HCl (pH 7.6) 200 mM, 8% sodium dodecyl sulfate (SDS), 40% glycerol, 40 mM dithiothreitol (DTT) and boiled for ten minutes. For cell experiment, the cells were rinsed twice with ice-cold phosphate-buffered saline (PBS, pH 7.4), then lysed in a cooled buffer containing 50 mM Tris-Cl, pH 8.0, 150 mM NaCl, 1% NP-40, 0.5% Na-deoxycholate, 0.1% SDS, 0.02% NaN3, 100 μg/ml PMSF, and 10 μg/ml protease inhibitors (leupeptin, aprotinin, and pepstatin) followed by sonication for five seconds on ice. The cell lysates AGING were added to one-third volume of sample buffer containing Tris-HCl (pH 7.6) 200 mM, 8% sodium dodecyl sulfate (SDS), 40% glycerol, 40 mM dithiothreitol (DTT) and boiled in a water bath for ten minutes. The protein concentrations of all samples were measured by BCA kit (Pierce, Rockford, IL, USA) according to the manufacturer's instruction. For sAPPα detection, the medium of N2a/APPswe cells was collected, cleared by centrifugation, lyophilized, and finally resuspended in the SDS-sample buffer. The equal amounts of proteins were separated by 10% SDSpolyacrylamide gel electrophoresis and transferred to nitrocellulose (NC) membranes. The membranes were blocked with 5% non-fat milk dissolved in TBSTween-20 (50 mM Tris HCl, pH 7.6, 150 mM NaCl, 0. at 4° C overnight. The blots were detected using antirabbit or anti-mouse IgG secondary antibody conjugated to IRDyeTM (800CW, Licor Biosciences, Lincoln, NE, USA) at room temperature for one hour and visualized by infrared fluorescence imaging. The optical density of bands was quantified by Odyssey system (Li-Cor Bioscience, Lincoln, NE, USA). Proteins were normalized against DM1A, see Table 1 for more details.

Real-time quantitative PCR
Total cortex RNA was extracted using TRIzol reagents according to the instructions (Invitrogen, Carlsbad, CA, USA) and reverse-transcribed to cDNA using reverse transcription reagents kit (Takara, Dalian, China). Fifty nanograms of cDNA were used for real-time PCR. Primers for ADAM10 were 5'-CTGGCCAACCTATTT GTGGAA-3' and 5'-GACCTTGACTTGGACTGCACT G-3', and primers for GAPDH were 5'-AACGACCCC TTCATTGAC-3' and 5'-TCCACGACATACTCAGCA C -3', respectively. The parameters of PCR cycle were 95° C/10 min, and 40 cycles of 95° C/10 s, 60° C/30 s, and 72° C/30 s. The amplification and analysis were performed using a StepOne Plus Real-Time PCR Detection System (Life Technologies, Grand Island, NY, USA). Samples were compared using the relative CT method.

NAD + /NADH quantification assay
NAD + /NADH ratio was calculated from the cortex and N2a/APPswe cells using NAD + /NADH assay kit provided by the Abcam. In brief, the cortex from 10 month old Tg2576 mice and N2a/APPswe cells transiently transfected with Nmnat2 plasmid cDNAs were homogenized in ice cold NAD + /NADH extraction buffer (freeze/thaw two cycles for cells) after the samples washed with cold PBS, and then centrifuged at 14000 rpm for 5 min. The supernatants were filtered through 10 kD molecular weight cutoff filters at 4° C to remove the enzymes that consume NADH rapidly, separately. 50 μl of ultra filtrates were heated for 30 min at 60° C in a water bath to decompose NAD + for NADH measurement because NADH will still be intact under these conditions. The heated and unheated samples were mixed separately with NAD + /NADH cycling mix assay for 5 min in a labeled 96-well plate in duplicates. The color was developing with NADH developer solution, and the absorbance was measured at 450 nm after 2 hours. We measured the protein content of an aliquot of homogenized samples before ultrafiltration by using standard Biorad procedure. The concentrations of NAD + and NADH were expressed in pmol/10 6 cells or ng/mg protein based on standard NADH readings.

Statistical analysis
All data were presented with the mean ± S.D. and analyzed using SPSS 12.0 statistical software (SPSS Inc., Chicago, IL, USA). The one-way analysis of variance (ANOVA) procedure followed by LSD's post hoc tests and student t test were used to determine the differences among groups. P < 0.05 was accepted as statistically significant.

Decrease of Nmnat2 and ADAM10 is associated with amyloidogenesis in Tg2576 AD mice model
Studies have demonstrated that Nmnats play an important role in maintaining axonal, dendritic and neuronal integrity [43][44][45], and over-expression of Nmnats offer several protective effects against neurodegeneration and axonal degeneration [31][32][33]. Our previous results demonstrated that Nmnat2 attenuates tau phosphorylation through activation of PP2A [35]. ADAM10 is the valid active component of α-secretase [15,16]. Up-regulation of ADAM10 in AD mouse models prevents the formation of senile plaques and cognitive deficits [11,46,47]. The levels of ADAM10 and Aβ were detected by western blot and immunohistochemistry in 10-month old Tg2576 mice and the age-matched wild type littermates. We observed that the ADAM10 levels ( Figure 1B, 1C) and cleavage of APP generating α-CTF levels ( Figure  1A-1C) were decreased significantly, suggesting that activity of α-secretase reduced in Tg2576 mice.
Meanwhile the Aβ productions were increased obviously in particular soluble Aβ and its oligomers ( Figure 1D-1F). Importantly, we found that a 56-kDa soluble dodecameric Aβ (Aβ * 56-kDa) is significantly enhanced, which is supposed to disrupt memory in middle-aged AGING Tg2576 mice [48], (Figure 1D-1F). The soluble Aβ oligomers, containing Aβ * 56-kDa, impair synapse structure and function and cause memory deficits in Tg2576 mice and AD brains [7,8,49,50]. Our previous study found that Nmnat2 levels are obviously decreased in the cortex and hippocampus in 10-month old Tg2576 mice compared with the age-matched wild type littermates [35]. All data indicated that the decrease of Nmnat2 and ADAM10 is associated with amyloidogenesis in Tg2576 mice.

Over-expression of Nmnat2 suppresses Aβ production and up-regulates ADAM10
To further determine whether Nmnat2 over-expression causes the reduction of Aβ productions and increases ADAM10 expression, we transiently transfected the Nmnat2 plasmid in N2a/APPswe cells for 48 hours. Strikingly, the mRNA ( Figure 2A) and protein levels of ADAM10 substantially increased in Nmnat2 overexpressing N2a/APPswe cells (Figure 2A-2C). In addition, we detected the productions of APP cleavage by α-secretase, a small fragment termed α-CTF and the large α-fragment ectodomain of APP (sAPPα), which are shown in Figure 2B, 2C and (Supplementary Figure  1A, 1B). The protein levels of α-CTF and sAPPα were significantly increased in Nmnat2 over-expressing cells compared to control. To eliminate the impact of overexposure of DM1A on data analysis, so we detected β-actin by western blot as the loading control. The images showed that the expression of DM1A was same as expression of β-actin in the same cell samples ( Supplementary Figure 2A, 2B). Together, it is indicated that α-secretase is activated by Nmnat2.
We next measured Aβ productions including soluble Aβ and its oligomers. Our results showed that soluble Aβ, its oligomers and Aβ * 56-kDa were markedly reduced in N2a/APPswe cells with over-expression of Nmnat2 compared with control cells (Figure 2D-2F). To further determine whether over-expression of Nmnat2 suppresses Aβ production in N2a/APPswe cells, the mediums of N2a/APPswe cells transiently transfected the Nmnat2 plasmid or empty vector were collected and measured the levels of Aβ1-40 and Aβ1-42 by ELISA. The Aβ levels were obviously decreased in the medium of N2a/APPswe cells with Nmnat2 plasmid cDNAs compared with the medium of N2a/APPswe cells with Nmnat2 empty vector ( Figure 3A-3C). In short, the above findings suggested that over-expression of Nmnat2 reduces Aβ production and up-regulates the expression of ADAM10.

Nmnat2 activates AMPK by elevating the ratio of NAD + /NADH
AMP-activated protein kinase (AMPK), a master regulator of cellular energy homeostasis [51], is potentially implicated in the pathogenesis of AD [52,53]. The levels of AMPK and phosphorylated AMPK at Thr172 (p-AMPK, an active form) were measured by western blot in 10-month old Tg2576 mice and the agematched wild type littermates. We found that the reduction of AMPK and p-AMPK ( Figure 4A, 4B) was accompanied with the decrease of Nmnat2 in 10-month old Tg2576 mice. However, how Nmnat2 to regulate activity of AMPK needs to be further determined. So the Nmnat2 plasmid and empty vector were transfected into N2a/APPswe cells for 48 hours respectively, then the levels AMPK and p-AMPK were checked by western blot. The results showed that levels of AMPK and p-AMPK had a remarkable elevation in N2a/APPswe cells of over-expressing Nmnat2 compared to control cells ( Figure 4C, 4D). To further confirm that the Nmnat2 is required for AMPK activity, we constructed shRNA plasmid of Nmnat2 (siNmnat2), and then the shRNA plasmid of Nmnat2 or the scrambled shRNA control  AGING (Vector) was transfected into N2a/APPswe cells for 48 hours. We found that the expression of endogenous Nmnat2 protein was efficaciously inhibited by siNmnat2 (reduction of 52.8%) ( Figure 4E, 4F). Simultaneously, the levels of AMPK and p-AMPK were obviously decreased after normalization against DM1A ( Figure 4E, 4F). In addition, we also employed immunofluorescence assays to further investigate the relationship between Nmnat2 and AMPK in N2a/APPswe cells through transiently transfecting the Nmnat2 and siNmnat2 plasmid cDNAs, respectively. We observed that the fluorescence intensity of AMPK was significantly elevated in N2a/APPswe cells with over-expressing Nmnat2 ( Figure 5A1ʹ-5A4ʹ) compared with control cells (Figure 5A1-5A4). Conversely, the AMPK fluorescence intensity was substantially lower in N2a/APPswe cell transfected by siNmnat2 ( Figure 5B1ʹ-5B4ʹ) than that in control cells ( Figure 5B1-5B4). These results indicate that Nmnat2 up-regulates the expression and activity of AMPK in vitro and in vivo.
The mechanism of Nmnat2 regulating AMPK needs to be further determined. Studies have shown that a high intracellular NAD + /NADH ratio affects AMPK activity [54,55]. Nmnat2 is a key enzyme to synthesize NAD + using NMN and ATP [56,57], and NAD + is a cofactor required for glycolysis and tricarboxylic acid cycle, and involved in energy metabolism [58]. So we speculated that Nmnat2 activates AMPK via the ratio of NAD + / NADH. We detected the intracellular NAD + /NADH ratio in the cortex in 10-month old Tg2576 mice. There was a significant decrease of NAD + /NADH ratio in 10month old Tg2576 mice compared to the age-matched wild type littermates ( Figure 6A), which was positive correlation with the activity of AMPK. In contrast, we also measured the ratio of NAD + /NADH in N2a/APPswe cells transiently transfected by the Nmnat2 plasmid. The levels of AMPK and p-AMPK were clearly enhanced ( Figures 4C, 4D, 5A1ʹ-5A4ʹ), and marked increased in the ratio of NAD + /NADH compared to control cells ( Figure 6B), suggesting that Nmnat2 activating AMPK is dependent on ratio of NAD + /NADH.

Nmnat2 attenuates Aβ production and up-regulates ADAM10 in AMPK activity-dependent manner
Several studies have demonstrated that activation of AMPK alleviates Aβ pathogenesis and rescues synapse damage [59,60]. We assumed that AMPK involves Nmnat2-mediated reduction of Aβ. To address this question, N2a/APPswe cells were treated with or without AICAR (2 mM) [61][62][63], an AMPK specific agonist for 24 hours. We observed that AICAR induced the expression of ADAM10 ( Figure 7A, 7B) and reduced the productions of Aβ1-40 and Aβ1-42 ( Figure 7C), suggesting that activating AMPK inhibits Aβ production through the up-regulation of ADAM10.
To further determine that Nmnat2 attenuates Aβ production and up-regulates ADAM10 in AMPK AGING activity-dependent manner. N2a/APPswe cells were transiently transfected by the Nmnat2 plasmid and then were treated with or without Compound C (20 μM, an AMPK specific inhibitor) [64][65][66] for another 24 hours. The results of western blot and ELISA assays showed that inhibiting AMPK activity with Compound C abolished the Nmnat2-induced upregulation of ADAM10 levels ( Figure 7D, 7E) and reduction of Aβ1-40 and Aβ1-42 ( Figure 7F). These data provide the supporting evidence that Nmnat2 attenuates Aβ production and up-regulates the level of ADAM10 in AMPK activity-dependent manner.

DISCUSSION
Our findings clearly indicated that the levels of αsecretase ADAM10, α-CTF and sAPPα of α-secretase cleavage are obviously decreased in Tg2576 mice. More interestingly, the Aβ productions including Aβ*56 (an Aβ dodecameric) and monomer, are also dramatically enhanced in the AD models, which have strong neurotoxicity and mainly cause damage of the ability of learning and recognition in the Tg2576 mice [48][49][50].
Our previous study has shown that expression of Nmnat2 is significantly reduced in Tg2576 mice [35], which the deficit of Nmnat2 is simultaneous with the increase of Aβ productions and reduction of ADAM10 in our present study. Moreover, the over-expression of Nmnat2 upregulates ADAM10 and suppresses Aβ productions in N2a/APPswe cells. Nmnat2 appears to direct APP processing toward the α-secretase and away from the βsecretase, which results in a reduction at the productions of toxic Aβ peptides [67,68]. Indeed, our data showed that Nmnat2 attenuates Aβ productions (including Aβ*56 and monomer) and up-regulates ADAM10 in AMPK activity-dependent manner. It is obvious that these data support an opinion, which Nmnat2 has neuroprotective effects on brain of AD patients and AD models [32,33,36,69]. We demonstrated that Nmnat2 exerts its enzymatic function in AMPK activity-dependent manner to attenuate Aβ productions and up-regulate ADAM10 in this study.
Nmnat2 is a nicotinamide adenine dinucleotide (NAD + ) synthesizing enzyme. Increasing evidence has shown that the loss of Nmnat2 leads to mitochondrial impairment as well as the fall in its substrate product NAD + [24,70,71]. NAD + is a crucial redox cofactor for metabolism and ATP production, and it is also a key substrate for several families of enzymes in health span and longevity [58,[72][73][74]. In current study, the decrease of NAD + /NADH ratio is related to the reduction of Nmnat2 expression in Tg2576 mice. Over-expression of Nmnat2 increases the NAD + /NADH ratio in N2a/APPswe cells, which is positive correlation with the NAD + /NADH ratio. However, the mechanism of Nmnat2 and its substrate product NAD + regulating APP protein is unexplained. AGING AMPK, acting as an energy and nutrient sensor, coordinates an integrated signaling network, which constitutes metabolic pathways and involves AD pathogenesis [53,75,76]. Studies have found that a high intracellular NAD + /NADH ratio affects AMPK activity [54,55]. In this study, we found that levels of AMPK and p-AMPK (Thr 172, active form) are lower in Tg2576 mice than in that of WT mice, which variation tendency of AMPK and p-AMPK are the same as Nmnat2 and ADAM10 in Tg2576 mice. And Nmnat2 positively regulates AMPK activity in N2a/APPswe cells with western blot and immunofluorescence by genetic manipulation. Therefore, we think that Nmnat2 adjusting AMPK activity is related to the increase of NAD + /NADH ratio, which is similar to previous results in vascular smooth [54,55]. Growing evidence has shown that increase of ADAM10 levels will benefit for reduction of Aβ productions in AD-like models [13,21,77], which is similar to our study. Interestingly, we found that over-expression of Nmnat2 also enhances expression of ADAM10 and reduces Aβ1-40/1-42 in N2a/APPswe cells. The above data indicate that both of Nmnat2 and AMPK can activate ADAM10 and reduce Aβ production.
Recent studies have revealed a key neuronal maintenance and protective function for Nmnat [26][27][28][29]. Nmnat protects neurons through multiple functions such as NAD + synthesizing enzymes and chaperones etc [24,29,31]. Loss and mutations in Nmnat are associated with human disease [78,79]. Some studies suggest that AMPK may increase the elimination of the Aβ peptide via the induction of autophagy and neuronal cholesterol and sphingomyelin levels or other mechanisms [59,63,80,81]. Our study demonstrate that activating AMPK with its specific agonist (AICAR) up-regulates the level of ADAM10 (a form of αsecretase activity) and reduced Aβ1-40/1-42 in N2a/APPswe cells. It is suggested that AMPK as a upstream molecule of ADAM10 activates its enzyme activity and inhibits Aβ productions. The reduction of Aβ1-40/1-42 and increase of ADAM10 could be induced by the over-expression of Nmnat2, which was blocked by Compound C (an AMPK specific antagonist) in N2a/APPswe cells. These data indicated that Nmnat2 activates α-secretase ADAM10 and inhibits Aβ productions in AMPK activity-dependent manner, which suggests that a new mechanism of Nmnat2 plays neuroprotective role in brain of AD models. An unavoidable issue rises. How does AMPK regulate ADAM10? Studies show that the intervention of AMPK/Sirt1 signaling pathway may improve neuropathological defects in Alzheimer's disease [82,83]. Previous reports have suggested that PKC activation may cause alterations in subcellular localization of ADAM10 and regulate secretion of amyloid precursor protein [84,85]. Whether AMPK acts on ADAM10 through its direct kinase effect or AMPK-PKC pathway should be explored in future.
In summary, the present study showed that Nmnat2 attenuates amyloidogenesis and up-regulates ADAM10 by increasing NAD + /NADH ratio in AMPK activitydependent manner (Figure 8). At the same time, it is AGING also revealed that Nmnat2 might become a new target by increasing the activity of AMPK and ADAM10 to prevent Aβ generation in AD.