SIRT1 activation with neuroheal is neuroprotective but SIRT2 inhibition with AK7 is detrimental for disconnected motoneurons

Romeo-Guitart et al. Cell Death and Disease (2018) 9:531 DOI 10.1038/s41419-018-0553-6 Cell Death & Disease ARTICLE Open Access SIRT1 activation with neuroheal is neuroprotective but SIRT2 inhibition with AK7 is detrimental for disconnected motoneurons David Romeo-Guitart 1 , Tatiana Leiva-Rodríguez 1 , María Espinosa-Alcantud 2 ,NúriaSima 2 , Alejandro Vaquero 2 , Helena Domínguez- Martín 3 ,DiegoRuano 3 and Caty Casas 1 Abstract Sirtuin 1 (SIRT1) activity is neuroprotective, and we have recently demonstrated its role in the retrograde degenerative process in motoneurons (MNs) in the spinal cord of rats after peripheral nerve root avulsion (RA) injury. SIRT2 has been suggested to exert effects opposite those of SIRT1; however, its roles in neurodegeneration and neuron response after nerve injury remain unclear. Here we compared the neuroprotective potentials of SIRT1 activation and SIRT2 inhibition in a mouse model of hypoglossal nerve axotomy. This injury induced a reduction of around half MN population within the hypoglossal nucleus by a non-apoptotic neurodegenerative process triggered by endoplasmic reticulum (ER) stress that resulted in activation of the unfolded protein response mediated by IRE1αand XBP1 by 21 days post injury. Both SIRT1 activation with NeuroHeal and SIRT2 inhibition with AK7 protected NSC-34 motor neuron-like cells against ER stress in vitro. In agreement with the in vitro results, NeuroHeal treatment or SIRT1 overexpression was neuroprotective of axotomized hypoglossal MNs in a transgenic mouse model. In contrast, AK7 treatment or SIRT2 genetic depletion in mice inhibited damaged MN survival. To resolve the in vitro/in vivo discrepancies, we used an organotypic spinal cord culture system that preserves glial cells. In this system, AK7 treatment of ER-stressed organotypic cultures was detrimental for MNs and increased microglial nuclear factor-κB and the consequent transcription of cytotoxic pro-inflammatory factors similarly. The results highlight the importance of glial cells in determining the neuroprotective impact of any treatment. Introduction Axonopathy is a common early characteristic of the neurodegenerative processes in the central nervous system 1 . Axon degeneration often leads to retrograde neuronal cell death or atrophy and the progressive and permanent loss of vital neuronal functions. Deciphering the signaling involved is necessary for the development of effective neuroprotectants that are greatly needed in the clinics. Transgenic models of neurodegenerative diseases are broadly used, but shed light on only single pieces of the puzzle. Non-transgenic models of neurodegeneration also yield valuable information regarding the “naturally occurring processes”after neuronal soma–axon disconnection. By exploiting the anatomical and technical advantages of several axotomized models, it has been shown that multiple signaling programs operate in parallel in neuronal soma during the © The Author(s) 2018 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a linktotheCreativeCommons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. Correspondence: Caty Casas ([email protected]) 1 Institut de Neurociències (INc) and Department of Cell Biology, Physiology and Immunology, Centro de Investigación Biomédica en Red sobre Enfermedades Neurodegenerativas (CIBERNED), Universitat Autònoma de Barcelona (UAB), Barcelona, Spain 2 Chromatin Biology Laboratory, Cancer Epigenetics and Biology Program (PEBC), Bellvitge Biomedical Research Institute (IDIBELL), L´Hospitalet de Llobregat, Barcelona, Spain Full list of author information is available at the end of the article. Edited by A. Verkhrtasky Official journal of the Cell Death Differentiation Association 1234567890():,; 1234567890():,; retrograde neurodegenerative process that does not end in apoptosis 2,3 . This knowledge has been useful in the discovery of effective neuroprotectants such as the new discovered drug combination NeuroHeal (NH) 4,5 . NH was discovered using unbiased proteomic data, systems biology, and artificial neural networks from two models that represented pure regenerative and pure neurodegenerative conditions after peripheral nerve distal axotomy or root avulsion (RA) injuries, respectively. The data were used to build up bona fide state-specific molecular maps, and mathematical models were used to screen databases of drugs to identify putatively neuroprotective combinations. NH is a combination of acamprosate and ribavirin, two repurposed drugs with well-described pharmacokinetics and pharmacodynamics profiles that have no serious adverse effects. Activation of sirtuin 1 (SIRT1) was shown to be necessary for the neuroprotective action of NH after RA 5 . Sirtuins constitute a highly conserved family of deacetylases that regulate cellular homeostasis 6 . Seven homologs of yeast Sir2 (SIRT1–7) that have a conserved catalytic domain have been identified in mammals 7 . Among them, SIRT1 and SIRT2 have been targeted in the design of neuroprotectants for the treatment of degenerative conditions and traumatic injuries 8 . SIRT1 regulates a number of pathways associated with normal metabolism and functioning of individual organs in mammals 9 , including those involving histone 3, p53, and nuclear factor-κB (NF- κB) among others 10 . SIRT2 regulates cell cycle and genome instability in part acting as a histone deacetylase with a preference for histone H4 lysine 16 11 . It is also the most abundant sirtuin in the brain where it deacetylates αtubulin at lysine 40 contributing in microtubule dynamic regulation 12 . Both SIRT1 activation and SIRT2 inhibition have been identified as good strategies for neurodegenerative diseases. In the present work, we aimed to validate whether these strategies could be extended to disconnected neurons using a model of axotomy of the hypoglossal nuclei, which contain highly vulnerable cranial MNs 13–15 that differ in their properties from the spinal MNs 16 . Results In order to evaluate how SIRT1 and SIRT2 function in neuroprotection of MNs after proximal nerve injury we chose to use the hypoglossal axotomy model 17 (Fig. 1a). We first characterized the pace of MN cell death at the hypoglossal nucleus over time post injury. We observed that the process slowly proceeded with the number of MNs reduced by 48.72 ± 2.2% after 21 days post injury (dpi) (Fig. 1a). Apoptotic hallmarks, such as activation of caspase-3 or apoptosis-inducing factor (AIF) nuclearization, were not observed, although MNs were degenerating as revealed by Fluoro-Jade C staining at 7 dpi (Fig. 1b). These observations were similar to those in our previously reported model of nerve RA in rat 18 . To explore whether there was activation of endoplasmic reticulum (ER) stress in the hypoglossal axotomy model, we analyzed the expression of target genes of the unfolded protein response (UPR) activated after ER stress, as Chop,Grp78, Xbp1, and its spliced version (Xbp1s), as well as the activation of three main branches of UPR, PKR-like ER kinase (PERK), inositol-requiring enzyme 1α(IRE1α), and activating transcription factor 6 (ATF6), by western blot. The only gene found overexpressed was the spliced form of Xbp1 at 3 dpi (Fig. 1c). The apparition of this form after ER stress depends on the activation of IRE1α, which autophosphorylates, dimerizes, and acquires ribonuclease activity, leading to splicing of the mRNA encoding the X- box binding protein XBP1 19 . In agreement, we observed that the phosphorylated dimer of IRE1αwas more abundant and the ratio of spliced to unspliced XBP1 was higher in animals subjected to hypoglossal axotomy than controls by western blot (Fig. 1d). However, there were not activation of the UPR branches ATF6 or PERK after 3 dpi characterized by the presence of the 50-kDa-cleaved fragment of ATF6 and the increase of the phosphorylated isoform of PERK, respectively (Fig. 1d). These results suggest that the neurodegenerative pathway involved in cranial MN death is non-apoptotic and presents some hallmarks for the presence of ER stress, similar to the process that occurs after RA in rats were only IRE1α activation was also found. Activation of SIRT1 or inhibition of SIRT2 promotes neuroprotection in some neurodegenerative mouse models 20 . Thus, we analyzed the effect of axotomy of the hypoglossal nerve on the expression and activity of SIRT1 and SIRT2. SIRT1 was overexpressed in the nuclei of axotomized MNs at 7 dpi (Fig. 2a). As a measure of its deacetylase activity, we evaluated levels of histone 3, which is acetylated at lysine 9 (Ac -H3K9), and p53, which is deacetylated at lysine 373 (Ac-p53K373), by SIRT1 as previously reported 21,22 . Both acetylated substrates significantly accumulated in the nuclei of damaged MNs at 7 dpi when compared to sham-operated control animals (Fig. 2a). SIRT2 appeared at a characteristic location probably corresponding to the ER–Golgi intermediate compartment 23 , and it was present at lower levels in axotomized MNs at 7 dpi with respect to the control (Fig. 2b). Although the total abundance of α-tubulin was diminished after injury, relative levels of the acetylated form of α-tubulin were higher than in controls (Fig. 2b). These results suggest that activities of both SIRT1 and SIRT2 are reduced early after injury. We used an in vitro model of ER stress to test whether an SIRT1 activation and SIRT2 inhibition were neuroprotective. We used tunicamycin (TUN) as stressor agent Romeo-Guitart et al. Cell Death and Disease (2018) 9:531 Page 2 of 14 Official journal of the Cell Death Differentiation Association Fig. 1 (See legend on next page.) Romeo-Guitart et al. Cell Death and Disease (2018) 9:531 Page 3 of 14 Official journal of the Cell Death Differentiation Association on NSC-34 cells, which are MN like. After 24 h, TUN treatment had induced death of 59.58 ± 1.82% of cells (Fig. 3a). Overexpression of SIRT1 promoted survival of ER-stressed cells compared to GFP-overexpressing insulted cells with only about 20% of cells dead after 24 h of TUN treatment (Fig. 3a). Treatment of ER-stressed cells with NH, which activates SIRT1 (Fig. S1), was also protective (Fig. 3a). To analyze effects of SIRT2, we used AK7, a SIRT2 inhibitor previously reported to exert neuroprotection in some models of neurodegenerative diseases. Treatment of the stressed cells with AK7 was also neuroprotective. We monitored AK7 activity by analyzing the levels of acetylated H4 at lysine 16 (Ac- H4K16) (Fig. S2). We then evaluated effects of administration of these drugs in vivo. Oral NH treatment in the hypoglossal axotomy model led to a decrease of Ac-H3K9 levels within the nucleus of MNs compared to vehicle treatment (Fig. 3b). Similarly, intraperitoneal (i.p.) AK7 treatment increased acetylated α-tubulin levels in the hypoglossal nuclei at 7 dpi with respect to control, as expected for a SIRT2 inhibitor (Fig. 3c). These results suggest that NH and AK7 are neuroprotective; we used both pharmacological and genetic approaches to test these hypotheses. Injured animals treated with NH had an increased percentage of MN survival than vehicle-treated injured mice (Fig. 4a). Similarly, a higher percentage of MN survival was observed post injury in transgenic animals overexpressing SIRT1 (tg- SIRT1) 24 compared to injured wild-type (WT) littermates (Fig. 4a). We also compared survival of MNs in injured mice that lack SIRT2 (SIRT2 knockout or KO-SIRT2 mice) 25 compared to WT littermates with and without treatment with vehicle or AK7 at 21 dpi (Fig. 4b). Unexpectedly, we observed a reduction in the number of surviving MNs in injured KO-SIRT2 mice relative to WT controls and in injured KO-SIRT2 mice treated with AK7 compared to vehicle-treated mice. Since, SIRT1 activation or overexpression resulted neuroprotective, we explored the effect on IRE1α-neuroprotective and ATF6- neuroprotective UPR branches and found a marked increase in cleaved ATF6 and its downstream target GRP78 (Fig. 4c). No significant modifications were observed regarding IRE1αactivation and downstream XBP1 splicing. These results suggest that inhibition of SIRT2 is detrimental in vivo, although not in vitro, whereas fostering SIRT1 activity is beneficial for MN survival after axotomy of the hypoglossal nerve. Furthermore, SIRT1 activation or overexpression reinforce endogenous mechanisms of neuroprotection related to UPR. Due to differences between in vitro and in vivo results regarding neuroprotection provided by AK7 treatment, we evaluated this drug in a spinal cord organotypic culture (SCOC) model 26 as we speculated that lack of glial cells in the culture might be the cause of this discrepancy. In the SCOC, TUN treatment produced a decline in SMI- 32-positive MN cells compared to untreated or vehicletreated SCOCs (Fig. 5a). We observed that TUN treatment drastically reduced the activity of SIRT1 as was the case for nerve injury in vivo, but in the SCOC, this stress enhanced the activity of SIRT2 (Figs. S1 and 2). NH treatment of TUN-treated SCOCs enhanced survival of MNs, whereas AK7 treatment yielded similar results to TUN treatment alone (Fig. 5a). Hence, for AK7, the results obtained in SCOCs were unlike those obtained in the NSC-34 model but similar to those obtained in the in vivo model. Considering that one of the main differences between the in vitro models is whether or not glia cells are present, we hypothesized that AK7 influences this particular cell type to alter MN destiny. One of the possibilities to be explored was that AK7 treatment renders microglia more reactive 27 . As the transcription factor NF-κB is involved in the production of pro-inflammatory cytokines 27 ,we investigated its protein levels in the nuclei of microglia in SCOCs. We did not observe a significant increase of NF- κB in TUN-treated SCOCs compared to levels in vehicletreated or control cultures (Fig. 5b). In contrast, there was a considerable decrease in nuclear NF-κB in the microglia of NH-treated stressed SCOCs, whereas AK7 treatment significantly increased NF-κB levels in these cultures (Fig. 5b). Accordingly, levels of interleukin-1β(IL-1β) and (see figure on previous page) Fig. 1 Hypoglossal nerve injury triggers non-apoptotic MN death with activation of IRE1αand XBP1. a (Top) Microphotographs of Fluoro- Nissl-stained MNs from the ipsilateral hypoglossal nucleus (dotted circled) from HA-injured animals at 7, 14, and 21 dpi. (Bottom left) Drawing highlighting the hypoglossal nuclei in the brainstem of the mice. (Bottom right) Graphic showing the average percentages ± SEMs of MN survival at the ipsilateral side with respect to the contralateral after HA at indicated dpi (n=4 for 14 dpi and n=5 for other groups; analysis of variances (ANOVA), post hoc Bonferroni; # p< 0.05 vs. 7 dpi and *p< 0.05 vs. 14 dpi). b(Top panels) Representative images of positive control (C+) from spinal cord samples of neonatal rats submitted to sciatic crush injury stained for active cleaved caspase-3 (Casp3.Act) merged with Fluoro-Nissl (green) and DAPI (blue). (Middle and bottom panels) Representative confocal images of MNs from hypoglossal nuclei immunolabeled for Casp3.Act or AIF1m (red) merged with Fluoro-Jade C-stained hypoglossal nuclei (green) counterstained with DAPI from control (Ctrl) non-injured mice or HA-injured mice at 7 dpi. Scale bars =25 μm. cBar graphs of the expression levels of different UPR-target genes. dWestern blots and bar graphs showing the analysis of phosphorylated IRE1αand PERK, the cleaved fragment of activated ATF6, and spliced (Xbp1s) and unspliced (Xbp1u) XBP1 protein levels in Ctrl and HA-injured animals at 3 dpi Romeo-Guitart et al. Cell Death and Disease (2018) 9:531 Page 4 of 14 Official journal of the Cell Death Differentiation Association Fig. 2 Deacetylase activities of SIRT1 and SIRT2 are reduced after nerve injury. a (Top) Confocal images of MNs immunolabeled for SIRT1, Ac- H3K9, and Ac-p53K373 (red) and counterstained with Fluoro-Nissl (green) and DAPI (blue) at hypoglossal nuclei of non-injured control (Ctrl) and HA- injured animals at 7 dpi. Scale bar =25 µm. (Bottom) Histograms of the means of immunofluorescence densities for each marker inside nuclei of MNs (n=4 animals, ttest, *p< 0.05 vs. Ctrl). b(Top) Representative confocal microphotographs of SIRT2, α-tubulin (α-Tub), and acetylated α-tubulin (Ac α- Tub) in red, counterstained with green Fluoro-Nissl and DAPI from hypoglossal nuclei of non-injured or injured animals at 7 dpi. Scale bar =25 µm. (Bottom) Histograms of the means of the immunofluorescence densities for each marker inside nuclei of MNs (n=4 animals, ttest, *p< 0.05 vs. Ctrl). Scale bar =25 µm and 10 µm for α-Tub and acetyl α-Tub staining Romeo-Guitart et al. Cell Death and Disease (2018) 9:531 Page 5 of 14 Official journal of the Cell Death Differentiation Association tumor necrosis factor-α(TNFα) expression were increased in AK7-treated SCOCs with respect to TUN- treated or control samples (Fig. 5c). To confirm that glial cells are critical to MN survival in vivo, we analyzed AK7-treated injured animals, and we observed an increase in CD86, a marker of the M1 phenotype, in microglial cells treated with AK7 at 3 dpi (Fig. 6a). We observed significantly more NF-κB in the nucleus of microglia of AK7-treated injured animals than in controls (Fig. 6b). In addition, we observed an increased in the expression of IL-1βand TNFαgenes upon AK7 treatment (Fig. 6c). Thus, although SIRT2 inhibition might be beneficial for the neuronal survival itself, effects on glial cells must also be considered, since the overproduction of pro-inflammatory cytokines by glia upon AK7 treatment might override all benefits to the neurons. Discussion Communication between nerve cells or nerve cells and their target tissue is dysfunctional in neurodegeneration. Fig. 3 SIRT1 activation by NH and SIRT2 inhibition by AK7 are neuroprotective in ER-stressed NSC-34 MN-like cell culture. a Bar graph showing the percentages of cells ± SEMs that survived after the treatment with vehicle (Veh) 1 μg/ml tunicamycin (TUN) alone or in combination with the NeuroHeal or AK7. Percent cell survival was also determined for CMV-SIRT1 and CMV-GFP cells treated with TUN (n=4–10, analysis of variances (ANOVA), post hoc Bonferroni # p< 0.05 vs. Veh, *p< 0.05 vs. TUN, $ p< 0.05 vs. TUN-GFP). b(Left) Confocal images of H3K9ac (red) in samples from mice subjected to hypoglossal axotomy and treated with vehicle or NH, counterstained with Fluoro-Nissl (green) and DAPI (blue) at 21 dpi. Scale bar =25 µm. (Right) Histogram showing the amount of fluorescent density of nuclear H3K9ac in injured MNs in vehicle or NH-treated groups at 21 dpi (n=4, ttest, *p< 0.05 vs. Veh). c(Left) Microphotographs showing acetylated and total α-tubulin (red) staining with Fluoro-Nissl (green) and DAPI (blue) counterstaining in animals treated with vehicle or AK7 at 21 dpi. Scale bar =10 µm. (Right) Histogram showing the ratio of the acetylated vs. total α-tubulin within the MNs in both groups (n=4, ttest, *p< 0.05 vs. Veh) Romeo-Guitart et al. Cell Death and Disease (2018) 9:531 Page 6 of 14 Official journal of the Cell Death Differentiation Association In addition to their particular protein hallmarks, all neurodegenerative diseases are characterized by neuronal cell death related to soma isolation due to neurite retraction. Deciphering the programs activated by neurons to cope with this damage may guide development of neuroprotective therapies. The purpose of the present study was to assess the roles of SIRT1 and SIRT2 in neuroprotection of disconnected central neurons using the hypoglossal nerve Fig. 4 SIRT1 and SIRT2 activities are essential for survival of MNs after HA. a (Left) Representative microphotographs of Fluoro-Nissl-stained ipsilateral hypoglossal nuclei MNs from injured WT mice, tg-SIRT1 mice, or NH-treated WT mice at 21 dpi. Scale bar =100 µm. (Right) The histogram represents the mean percentages ± SEMs of surviving MNs in the ipsilateral side respect to the contralateral side (n=4–5, analysis of variances (ANOVA), post hoc Bonferroni, *p< 0.05 vs. WT). b(Left) Hypoglossal nuclei stained with Fluoro-Nissl from WT mice, SIRT2-KO mice, and AK7-treated WT mice (20 mg/kg i.p, daily) at 21 dpi. Scale bar =100 µm. (Right) Plots of the percentages ± SEMs of the counted MNs at the ipsilateral side respect the contralateral for different experimental groups at 21 dpi (n=4, ANOVA, post hoc Bonferroni, *p< 0.05 vs. WT). cWestern blots and bar graphs showing the analysis of phosphorylated IRE1α, the cleaved fragment of activated ATF6, spliced (Xbp1s) and unspliced (Xbp1u) XBP1 and GRP78 protein levels in Ctrl, HA-injured, and NH-treated HA-injured wild-type animals and HA-injured tg-SIRT1 animals at 3 dpi Romeo-Guitart et al. Cell Death and Disease (2018) 9:531 Page 7 of 14 Official journal of the Cell Death Differentiation Association axotomy model. Axotomized cranial MNs suffered nonapoptotic cell death characterized by the activation of the IRE1α-mediated branch of the UPR that is triggered during ER stress. We demonstrated that pharmacological activation of SIRT1 with NH or overexpression of SIRT1 is beneficial, whereas pharmacological or genetic inhibition of SIRT2 is detrimental for axotomized neuron survival in vivo. Interestingly, in the face of ER stress, SIRT2 inhibition with AK7 exerted neuroprotection in a motoneuron (MN)-like cell line but not in MNs in organotypic culture. The difference in these two systems is the presence of glial cells in the latter. AK7 treatment in the organotypic culture and in the in vivo model promoted overproduction of pro-inflammatory cytokines by microglia that can explain its deleterious effect. In this regard, neuroinflammation disrupted UPR activation under proteasome stress situation 28 . We demonstrated that activation of SIRT1 with NH is important for neuroprotection of disconnected cranial MNs in mice as it was for root avulsed spinal MNs in rats 5 . We recently reported that treatment with NH produces long-lasting neuroprotection and also accelerates nerve regeneration and reduces muscle atrophy in a pre-clinical model of RA 4 . We previously showed that SIRT1 activation was necessary for NH neuroprotection, although this drug composition targets multiple proteins 5 . Fig. 5 SIRT1 and SIRT2 activities maintain neuron survival in an in vitro model of ER stress in SCOCs. a (Left) Representative microphotographs of ventral horn of the SCOCs stained for SMI-32 after 2 days of treatment with TUN with or without the NH or AK7 co-treatment. Scale bar =100 µm. (Right) Bar graphs showing the numbers ± SEMs of SMI-32-positive cells in the ventral horns of each hemisection of the spinal cord slice (n=5–10 complete spinal cord slices, analysis of variances (ANOVA), post hoc Bonferroni *p< 0.05 vs. Veh, # p< 0.05 vs. TUN, $ p< 0.05 vs. AK7). b(Left) Confocal microphotographs stained for the p65 subunit of NF-κB (red) and Iba1, a microglia marker (green). Samples were counterstained with DAPI (blue). Samples from different experimental conditions were stained 6 h after TUN addition. Scale bar =10 µm. (Right) Histogram showing the percentages of microglial cells with positive nuclear labeling of NF-κB for each experimental group (n=4, ANOVA post hoc Dunnett's test, *p< 0.05 vs. TUN). cHistogram of mean values obtained by quantitative real-time PCR for IL-1βand TNFαmRNA in the control (Ctrl), TUN, or TUN plus AK7 SCOCs at 6 h post treatment Romeo-Guitart et al. Cell Death and Disease (2018) 9:531 Page 8 of 14 Official journal of the Cell Death Differentiation Association SIRT1 has been shown to be important for neuroprotection in several models of neurodegenerative diseases (e.g., Alzheimer’s disease, Huntington’s disease) 29–31 and is crucial in cerebral ischemia and preconditioning 32 .We demonstrated herein that either NH-induced activation or overexpression of SIRT1 boost endogenous mechanisms of neuroprotection after injury, as it is the ATF6-GRP78 branch of UPR 5 . Further future investigations would be worth doing to unravel the mechanisms underlying this activation done by SIRT1. Nevertheless, the data shown here confirm that activation of SIRT1, in particular through the treatment with NH, is a promising therapeutic option for disconnected neurons. We expected that AK7 and deletion of SIRT2 would be neuroprotective since SIRT2 inhibition is reportedly beneficial in some age-related disorders such as Parkinson’s disease 12,33–36 and Huntington disease models 37,38 . In contrast, our results indicated that AK7 treatment was deleterious for disconnected MN survival in the mouse model. Although in cell lines the treatment was beneficial, in the SCOC model, which more accurately reflects the complex microenvironment of the nervous system, AK7 did not enhance MN survival. In our organotypic and in vivo models, AK7 treatment induced microglia cells to generate pro-inflammatory cytokines, which might override the treatment-induced protection of neurons. After axotomy, the primary injury induces inflammatory responses with cytokines and chemokines released into the peri-lesional areas. Microglial activation, which propagates inflammation into neighboring tissue, may lead to impaired neuronal survival in the peri-lesional regions as a result of the generation of cytokines and reactive oxygen species. Our data indicate that AK7 treatment results in enhanced microglial activation at the injury site, and this suggests that SIRT2 inhibition might aggravate the acute inflammatory response. SIRT2 interacts with p65, an NF- κB subunit, in the cytoplasm and deacetylates lysine 310 in vitro and in vivo 39 . After TNFαstimulation, p65 is hyper-acetylated in SIRT2-deficient cells, which results in increased expression of a subset of p65-target genes 39 . This is in agreement with recent reports that indicate that SIRT2 acetylates p65 to influence pro-inflammatory gene transcription in microglia 40,26,8 . Together, these data indicate that post-translational deacetylation of p65 by SIRT2 might be a mechanism that contributes to its antiinflammatory properties. Our results are in agreement with other reports that AK7 is not beneficial under conditions where the control of the microglial response is crucial for neuronal survival such as in amyotrophic lateral sclerosis 12 and traumatic brain injury 41 . Another study using SIRT2 knockout mice revealed an increase in microglial activation and proinflammatory cytokines upon intracortical injection of lipopolysaccharide 27 .SIRT2 deletion also promotes Fig. 6 SIRT2 inhibition exacerbates microglial reaction and compromises MN survival after HA. a (Left) Microphotographs of samples stained with CD86 (red), a marker of the M1 phenotype, and Iba (green), a microglia marker, and counterstained with DAPI and Fluoro-Nissl (blues) from vehicle or AK7-treated HA-injured animals at 3 dpi. Scale bar =10 μm. (Right) Histogram showing the means ± SEMs of the percentages of microglial cells with positive CD86 labeling (n=3 for Veh and n=4 for AK7, ttest, *p< 0.05 vs. Veh). bHistogram showing the means ± SEM of the percentages of microglial cells with positive nuclear NF-κB labeling (n=3 for Veh and n=4 for AK7, ttest, *p< 0.05 vs. Veh). cHistogram of mean values obtained by quantitative real-time PCR for IL-1βand TNFαmRNAs in the hypoglossal nuclei from Veh or AK7-treated animals at 3 dpi Romeo-Guitart et al. Cell Death and Disease (2018) 9:531 Page 9 of 14 Official journal of the Cell Death Differentiation Association