NF-κB p65 directs sex-specific neuroprotection in human neurons

Protection of neurons against oxidative stress is crucial during neuronal development, maintenance and for treating neurodegenerative diseases. However, little is known about the molecular mechanisms underlying sex-specific maturation and survival of neurons. In the present study, we demonstrate NF-κB-p65 mediated neuroprotection in human glutamatergic neurons differentiated from inferior turbinate stem cells (ITSCs) in a sex-dependent manner. We successfully differentiated ITSCs into MAP-2+/NF200+/Synaptophysin+/vGlut2+-glutamatergic neurons in vitro and ex vivo and validated their functionality. TNF-α-dependent NF-κB-p65 activation was accompanied by significant neuroprotection against oxidative stress-induced neuronal death, which was surprisingly higher in neurons from female donors. Accordingly, sex-specific neuroprotection of female neurons was followed by an increased expression of special NF-κB target genes SOD2 and IGF2. Among these, SOD2 is a well known gene protecting cells against oxidative stress resulting in longevity. In addition, IGF2 is known to promote synapse formation and spine maturation, and it has antioxidant and neuroprotective effects against oxidative damage. In conclusion, we show that NF-κB-p65 is a key player in neuroprotection of human neurons, however the protective gene expression program beneath it differs between sexes. Our findings are in accordance with the increasing evidences pointing towards sex-specific differences in risk and severity of neurodegenerative diseases.

Integration and differentiation of ITSC-derived glutamatergic neurons after transplantation into ex vivo-cultivated murine organotypic hippocampal slices. In addition to their efficient neuronal differentiation in vitro, we evaluated the ability of ITSCs to integrate and differentiate within a neural environment by transplanting undifferentiated stem cells into murine organotypic hippocampal slices (Fig. 2a). Transplanted human ITSCs were able to integrate in the murine neural tissue and differentiated into MAP-2 + and Gat-1 + neurons after 14 days of co-cultivation (Fig. 2b,c). Furthermore, GFP-positive ITSCs integrated particularly into the dentate gyrus of organotypic hippocampal slices, where they exhibited a clear neuronal phenotype accompanied by expression of vGlut2 and Synaptophysin on protein level (Fig. 2d-f). These findings confirmed that ITSCs are also able to give rise to excitatory glutamatergic neurons within the proper neural environment.
ITSC-derived glutamatergic neurons show AMPA or glutamate-dependent activation of NF-κB-p65. We next investigated the capability of ITSC-derived neurons to respond to the excitatory neurotransmitter glutamate (GLU) or its agonist α-amino-3-hydroxyl-5-methyl-4-isoxazole-propionate (AMPA). Stimulation with GLU or AMPA resulted in a significant increase in nuclear translocation of NF-κB-p65 in a dose-dependent manner (5-10 µM) in comparison to untreated neurons. On the contrary, treatment with 50 µM GLU or AMPA led to a significant decline in NF-κB-p65 nuclear translocation compared to 10 µM-treatment ( Fig. 3a-d). We also observed high levels of basal NF-κB-activity ( Fig. 3a-d), in accordance to the already described constitutive activation of NF-κB particularly in glutamatergic neurons 39 . Treatment of ITSC-derived neurons with their respective inhibitors 6-cyano-7-nitroquinoxaline-2, 3-dione (CQNX) or dibenzocyclohepteneimine (MK-801) prior to application of GLU (10 µM) or AMPA (10 µM) resulted in a significantly reduced translocation of NF-κB-p65 into the nucleus compared to the stimulation approaches (Fig. 3e,f). These findings provide pharmacological evidence that both kinds of receptors were expressed in human ITSC-derived glutamatergic neurons, which in turn were observed to be fully functional after 30 days of differentiation. Stimulation with TNF-α leads to significantly increased nuclear translocation of NF-κB-p65 in ITSC-derived glutamatergic neurons. After validating human ITSC-derived neurons as a model system for studying the role of NF-κB during maturation, we investigated the potential of TNF-α to stimulate NF-κB in these neurons. Stimulation of ITSC-derived neurons with TNF-α for 30 minutes or 1 hour resulted in nuclear translocation of NF-κB-p65 (Fig. 4a, arrowheads) in comparison to untreated neurons or 24 h of  TNF-α-treatment (Fig. 4a, arrows). Quantification of the NF-κB-p65 nuclear mean integrated density clearly validated these dynamics by showing a highly significant increase in nuclear NF-κB-p65 fluorescence after 30 minutes (93,96% ± 6,04%) and 1 hour (88,00% ± 12,00%) of TNF-α-treatment compared to untreated controls (<20%) (Fig. 4b). Accordingly, stimulation of ITSC-derived neurons with TNF-α for 24 hours did not result in a significantly different nuclear NF-κB-p65 fluorescence intensity compared to control (Fig. 4b).
TNF-α-pre-treatment of human ITSC-derived glutamatergic neurons leads to increased NF-κB-p65-activity upon oxidative stress insult. We further analyzed the effect of H 2 O 2 -mediated oxidative stress insult on the activity of NF-κB-p65 in ITSC-derived neurons. Application of H 2 O 2 for 25 h on human glutamatergic neurons differentiated for 30 days led to significantly increased nuclear translocation of NF-κB-p65 in comparison to control. In order to analyze a potential neuroprotective role of NF-κB, we performed a pre-treatment using 10 ng/ml TNF-α during 2 hours prior to oxidative stress insult. Notably, TNF-α-pre-treatment of ITSC-derived glutamatergic neurons followed by H 2 O 2 -mediated oxidative stress resulted in a significant increase in nuclear translocation of NF-κB-p65 compared to the H 2 O 2 alone or control ( Fig. 5a,b, arrowheads). We further applied pyrrolidine dithiocarbamate (PDTC) as a control for guided inhibition of NF-κB. Pre-treatment of the cultivated neurons with PDTC for 1 hour followed by application of TNF-α or sole PDTC-treatment did not result in changes of nuclear translocation of NF-κB-p65 ( Fig. 5a,b, arrows). Quantification of the nuclear mean integrated density for p65 indicated a small but significant increase in nuclear NF-κB-p65 in both treatments compared to the untreated negative control (Fig. 5b).
ITSC-derived glutamatergic neurons are protected from oxidative stress-mediated cell death via TNF-α-dependent activation of NF-κB-p65. Determining the physiological consequences of TNF-α-dependent activation of NF-κB-p65 during oxidative stress, we analysed the death rate of ITSC-derived Figure 5. Treatment of ITSC-derived glutamatergic neurons with TNF-α prevents from oxidative stressmediated cell death in a sex-dependent manner. (a) Immunocytochemistry of ITSC-derived neurons after 30 days of differentiation, after treatment with H 2 O 2 alone, TNF-α-pre-treatment prior to H 2 O 2 , PDTC alone and PDTC followed by TNF-α against NF-κB-p65. (b) Quantification of immunocytochemical assays showed significantly increased nuclear translocation of NF-κB-p65 after treatment with TNF-α alone and TNF-α prior to H 2 O 2 compared to H 2 O 2 alone and untreated control. Pre-treatment of ITSC-derived neurons with PDTC for one hour followed by TNF-α-treatment did not result in significantly different amounts of nuclear NF-κB-p65 compared to PDTC alone. Mean values were normalized to the highest value. (c) Quantification of neuronal cell death showed significant death after oxidative stress insult (H 2 O 2 ) compared to TNF-α/ H 2 O 2 , TNF-α, PDTC, PDTC/TNF-α and untreated control (n = 6). (d) Quantification of neuronal cell death after oxidative stress (H 2 O 2 ), TNF-α-pre-treatment, TNF-α, PDTC, PDTC/ TNF-α and untreated control comparing sex differences (n = 3 males, n = 3 females). Data were showed not to be normally distributed using Kolmogorov-Smirnov and Shapiro-Wilk normality tests. Non-parametric Kruskal-Wallis test was further used (p ≤ 0.001), and Tukey's post-hoc test (**p ≤ 0.01, ***p ≤ 0.001). Mean ± SEM (standard error of the mean). -mediated oxidative stress led to robust and significant apoptosis of glutamatergic neurons compared to untreated control (Fig. 5c). Notably, this H 2 O 2 -mediated increase in apoptosis was significantly reduced down to a level similar to control upon TNF-α-pre-treatment prior to the oxidative stress insult (Fig. 5c). Application of TNF-α alone did not affect the survival of ITSC-derived neurons nor did the PDTC treatment, or the PDTC treatment followed by TNF-α in comparison to control (Fig. 5c).

Sensitivity of glutamatergic neurons to ROS-mediated cell death and neuroprotection via
NF-κB-p65 is dependent on the sex of the ITSC-donor. Investigating the effects of TNF-α-treatment on H 2 O 2 -mediated death of ITSC-derived neurons in more detail, we analysed the amount of apoptotic cells after oxidative stress and TNF-α-dependent neuroprotection in dependence to the sex of the ITSC-donor. We observed a significant increase in cell death of neurons differentiated from female ITSC-donors compared to their male counterparts, indicating an elevated sensitivity of human female glutamatergic neurons to oxidative stress (Fig. 5d). Pre-treatment of ITSC-derived neurons from female donors with TNF-α led to a significant and complete neuroprotection against H 2 O 2 -mediated cell death. Although neurons from male ITSC-donors were likewise protected against cell death via exposure to TNF-α, we observed a 2-fold increase in TNF-α-dependent neuroprotection in female ITSC-derived neurons compared to those differentiated from male ITSCs. These findings not only demonstrate a NF-κB-dependent neuroprotection of ITSC-derived neurons against oxidative stress-mediated cell death, but emphazise the dependence on its sensitivity to the sex of the ITSC-donor.  (Fig. 6b). Expression levels of cellular inhibitor of apoptosis protein-1 and 2 (c-IAP1 and c-IAP2) showed the tendency to be elevated in male ITSCs-derived neurons after TNF-α/H 2 O 2 -treatment compared to their female counterparts (Fig. 6c,d) however no significant alteration was detectable. Treatment with TNF-α, H 2 O 2 and TNF-α/H 2 O 2 further resulted in significantly increased expression levels of insulin-like growth factor 1 (IGF1) in ITSCs-derived neurons compared to control (Fig. 6e), although no significant sex-dependent differences were observable. Female ITSCs-derived neurons showed significantly increased expression levels of IGF2 after H 2 O 2 and TNF-α/H 2 O 2 -treatment compared to control (Fig. 6f), while no expression was detectable in male counterparts. However, sole treatment with TNF-α resulted in significantly increased expression levels of IGF2 in neurons from male and female donors compared to control (Fig. 6F). These findings strongly suggest a sex-specific NF-κB-p65 target gene expression in dependence to TNF-α-mediated neuroprotection during oxidative stress.

Discussion
The present study describes for the first time a neuroprotective role of NF-κB-p65 in human ITSC-derived glutamatergic neurons after oxidative stress insult in a sex-specific manner. We successfully differentiated human neural crest-derived inferior turbinate stem cells into MAP-2 + /NF200 + /Synaptophysin + /vGlut2 + -glutamatergic neurons by application of a directed differentiation assay or via transplantation into organotypic mouse hippocampal slice cultures. Extending our previous findings depicting vesicle recycling and calcium spiking of ITSC-derived neurons 38 , we validated their functionality by showing increased NF-κB-activity upon stimulation with the excitatory neurotransmitter glutamate or its agonist AMPA. Inhibitor controls using CQNX and MK-801 led to a decrease in glutamate or AMPA-dependent stimulation of NF-κB-activity, further validating the specificity of the respective receptors. In accordance to our findings, stimulation of ionotropic glutamate receptors was shown to activate NF-κB in primary rat cerebellar granule neurons 11,12 . Given the pivotal role of NF-κB signalling in key elements for neuronal morphology like neurite growth 14 , dendritic spine formation 15 , axonal outgrowth 16 and synaptic plasticity 17,18 , our data suggest the participation of NF-κB in the normal physiology of the human nervous system. In addition to its AMPA-and glutamate-dependent stimulation, we also observed a significant increase in NF-κB-activity in ITSC-derived neurons after treatment with TNF-α. In canonical NF-κB-signalling, recognition of stimuli like cytokines or neurotransmitters leads to phosphorylation of IκB kinases 40,41 , in turn resulting in phosphorylation, polyubiquitination and 26S-proteasome-mediated degradation of IκBs. Demasking of the nuclear translocation signal region of p50/p65 by degradation of IκBs is subsequently followed by translocation of p50/p65 into the nucleus and activation of target gene expression by binding to κB elements 18,[42][43][44][45] . TNF-α is one of the best characterized cytokines inducing this pathway, and its receptors TNFR1 and TNFR2 are widely expressed in the nervous system both in neurons and glia [46][47][48] . Besides its modulatory effects of neuronal responses to excitotoxic and hypoxic insults in the nervous system 49 , the absence of TNFR was shown to result in an increased neuronal damage following either ischemic or kainic acid induced excitotoxic damage 50 . In mouse NSCs, TNF-α-mediated NF-κB signalling was reported to be required for initial neuronal differentiation 29 . In accordance, preliminary data from our lab suggests that NF-κB-c-Rel might be the relevant subunit for glutamatergic differentiation and not NF-κB-p65, having no significant differences with respect to sex (unpublished data). Extending these findings, mature human ITSC-derived glutamatergic neurons revealed a significantly increased nuclear translocation of NF-κB-p65 after TNF-α-stimulation in the present study, indicating the crucial role of NF-κB-signalling during stem cell-based neuronal differentiation and neuroprotection in humans. Being a major cause of several neurologic diseases and brain damage 20 , oxidative stress is known to be directly caused by Alzheimer's disease via amyloid beta peptide-dependent production of hydrogen peroxide through metal ion reduction 51,52 . In Parkinson's disease, free radicals accumulate in the substantia nigra pars compacta, resulting in the formation of 6-hydroxydopamine, in turn leading to the generation of superoxide 53,54 . In the present study, H 2 O 2 -mediated oxidative stress led to cell death of human ITSC-derived glutamatergic neurons. Although NF-κB is known to be activated by oxidative stress in the nervous system 20 , several studies indicated its neuroprotective role in murine cells. Here, Heck and colleagues demonstrated an Insulin-like growth factor-1-mediated neuroprotection of rat primary cerebellar neurons against oxidative stress directly associated to activation of NF-κB 55 . Erythropoietin-mediated neuroprotection of rat cerebral cortical cell cultures from oxidative stress was also shown to occur in an NF-κB-dependent manner 56 . On the contrary, Zou and colleagues demonstrated TNF-α-treatment of rat hippocampal-entorhinal cortex slice cultures to result in increased neurotoxicity to both glutamate and oxidative stress 57 . In the present study, TNF-α-pre-treatment led to a significant decrease in H 2 O 2 -mediated cell death of ITSC-derived human neurons accompanied by a significantly increased nuclear translocation of NF-κB-p65. Our data therefore demonstrate a key role of NF-κB-p65 in protection of human stem cell-derived neurons from oxidative stress, further emphasizing the importance of NF-κB-signalling in neuroprotection 20,55,56 .
Interestingly, we further observed a significantly elevated sensitivity of ITSC-derived neurons from female donors to oxidative stress-induced cell death and to NF-κB-dependent neuroprotection compared to neurons from male donors. These findings were confirmed by a differential expression of NF-κB target genes in dependence to the sex of the ITSC-donor. Here, increased SOD2 mRNA levels observed in female but not in male ITSC-derived neurons indicated a NF-κB-associated induction of SOD2 protecting against oxidative stress-induced neuronal apoptosis. Accordingly, SOD2 expression was described to be inducible by TNF-α, having an anti-apoptotic role by directly reducing cellular ROS levels 58 . Next to SOD2, expression levels of IGF2 were significantly elevated only in female neurons after TNF-α/H 2 O 2 -treatment compared to control. IGF2 is known to promote synapse formation and spine maturation in the mouse brain 59 . Within a mouse model of Alzheimer's disease, IGF2 administration rescued spine formation and synaptic transmission in the hippocampus 60 . In accordance to the present findings, IGF2 was reported to have an antioxidant and neuroprotective effect on oxidative damage and mitochondrial function in cultured adult rat cortical neurons 61,62 . In contrast to their female counterparts, male ITSC-derived neurons showed a significant increase in the expression level of PKAcatα after TNF-α/H 2 O 2 -treatment. Interestingly, we observed no significantly altered expression levels of the antiapoptotic proteins c-IAP1 and c-IAP2, known mediators of TNF-α-dependent neuroprotection 63 . With PKAcatα being an essential regulator in learning and memory by transducing synaptic responses through CREB signalling 64,65 and controlling synaptic incorporation of AMPA receptors 66 , PKA-activity may directly contribute to the neuroprotective effects observed here. Although being a matter of debate, sex-dependencies in stem cell biology have already been shown in terms of autosomal gene expression 67 and proliferation 68 , particularly regarding mouse NSCs 69,70 . Compared to their male counterparts, female muscle-derived stem cells were reported to have higher muscle regeneration efficiency in mice 71 . In terms of neuroinflammation and neuroprotection, sex-dependent differences between patients have been likewise reported in ischemic stroke 4 , PD 2 , or AD 3 . While female AD patients were described to have an increased risk of developing AD 3,6 , PD was shown to have a greater prevalence in male patients 2 . These data are in agreement with sex-specific differences found in adult murine microglia, where female microglia exhibited a neuroprotective phenotype upon ischemic insult, which was retained after being transferred into male brains 72 . In addition sex-specific differences in the expression of iNOS and NF-κB were previously reported in human polymorphonuclear neutrophils, being higher in female than in male cells 73 .
In line with these findings, our data indicate for the first time a direct sex-dependent difference in neuroprotection of human stem cell-derived neurons against oxidative stress mediated by NF-κB-signalling. In summary, we provide here evidence that NF-κB-p65 is a key player in neuroprotection of human neurons against oxidative stress in a sex-dependent manner. We demonstrate a sex-dependent difference of stress response and TNF-α-mediated neuroprotection, with a strong increase of both H 2 O 2 -mediated cell death as well as neuroprotection against cell death in female derived neurons compared to their male counterparts (Fig. 7). These differences were emphasized by the sex-specific differential expression of NF-κB-p65 target genes SOD2 and IGF2 in TNF-α-dependent neuroprotection upon oxidative stress-insult. In line with our findings, increasing evidences pointing towards sex-specific differences in risk and severity of neurodegenerative diseases, such as Alzheimer's disease. Since oxidative stress is directly associated to neurodegenerative diseases, but little is known about the underlying molecular mechanisms of neuroprotection, NF-κB-signalling may be a crucial parameter for treatment strategies and neuronal regeneration therapies.

Methods
All methods were performed in accordance to the relevant guidelines and regulations.
Isolation and Cultivation of ITSCs. ITSCs were isolated from adult human inferior turbinate tissue obtained by biopsy during routine surgery after informed consent according to local and international guidelines and cultivated as described previously 36 . The ethics board of the medical faculty of the University of Münster Figure 7. Sex-specific response to oxidative stress insult and NF-κB-mediated neuroprotection in human NCSC-derived neurons. Female ITSC-derived neurons responded with a higher sensitivity to oxidative stressinduced neuronal death, and TNF-α-mediated neuroprotection compared to their male counterparts. TNFα-mediated neuroprotection led to an increase in NF-κB-p65 nuclear translocation, triggering differential expression of sex-specific NF-κB target genes. approved all the procedures described in this article (No. 2012-015-f-S). All experiments and methods were performed in accordance to the relevant guidelines and regulations. ITSCs were cultivated within the 3D blood plasma (BP) matrix 37 , and dissociated ITSCs were resuspended in Dulbecco's modified Eagle's medium/Ham F-12 (DMEM/F-12; Biochrom, Berlin, Germany, http://www.biochrom.de) supplemented with basic fibroblast growth factor-2 (FGF2; 40 ng/ml; Miltenyi Biotec), epidermal growth factor (EGF; 20 ng/ml; Miltenyi Biotec) and B27 supplement (Gibco) followed by supplementation with 10% of clinically accredited therapeutic human blood plasma (BP; obtained from Institut für Laboratoriums und Transfusionsmedizin, Bad Oeynhausen, Germany) and cultivated at 37 °C, 5% O 2 and 5% CO 2 .
Neuronal stimulation. After 30 days of differentiation neurons were exposed to the excitatory neurotransmitter glutamate (GLU) or its agonist α-amino-3-hydroxyl-5-methyl-4-isoxazole-propionate (AMPA), the In order to confirm NF-κB activation due to TNF-α, a pre-treatment using 100 μM PDTC for one hour was performed and samples were directly used or further treated with TNF-α for at least one hour. For Indirect immunofluorescence assay cells were fixed as described below. For Smart-seq. 2, cells were directly used after treatment, whose duration was 3 h for PDTC and TNF-α treatment as well as for the combination treatments.
Immunocytochemistry. Differentiated ITSCs were fixed for 15 min in phosphate-buffered 4% paraformaldehyde (PFA 4% pH 7.4) at room temperature (RT) followed by 3 wash steps in phosphate-buffered saline (1xPBS). The cells were permeabilized with 0.02% Triton X-100 and blocked using 5% of appropriate serum or 3% BSA for 30 minutes at RT, followed by incubation with primary antibodies for 1 hour at RT. Antibodies used were anti-neurofilament NF200 ( 78 . In parallel ITSCs were transduced with lentivirus pFUGW containing a constitutively expressed GFP-gene under control of human ubiquitin c promoter. GFP + -ITSCs and ITSCs were cultivated as neurospheres for 2 days. 7 days after slice preparation, dissociated cells (1 × 10 4 ) were dropped onto each slice following by cultivation for 14 days at 37 °C and 5% CO 2 . Slices were cut off from the membrane and free-floating fixated in PFA 4% for 1 h at 4 °C on agitation. After 3 washes with PBS, slices were incubated in PBS containing 0.1% Triton X-100 and 5% goat serum for 1 hour at RT. Slices containing transplanted GFP + -ITSCs were double immuno-labeled with anti- GFP  Cell Counting and Statistics. Quantification of immunofluorescence staining was performed for minimum 3 different donors. For each time point 6-12 pictures were analysed per donor, where the mean of the nuclear integrated density was measured by defining the region of interest with the nuclear DNA channel using ImageJ 76 . For analysis of neuronal survival the same channel was used to analyse the nuclear chromatin morphology. Nonviable neurons were recognized by nuclear condensation and/or fragmented chromatin. In phase contrast images, those neurons were irregularly shaped with shrunken cell body and/or disrupted neurites. The number of viable and nonviable neurons was counted in four to five field pictures and death rate was calculated. Data was further analysed for statistics using Past3 79 and/or GraphPad Prism 5 (GraphPad software, La Jolla, CA, http://www.graphpad.com). Normality of the data sets was refuted after analysis using Kolmogorov-Smirnov and Shapiro-Wilk normality tests. Non-parametric Kruskal-Wallis test was used to compare the medians between the different data sets for the different donors (***p ≤ 0.001). Non-parametric Mann-Whitney test was used to compare two pair of groups (***p ≤ 0.001). Further analysis was performed using Tukey's test (*p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001).