Transcription factor NF-kappa B represses ANT1 transcription and leads to mitochondrial dysfunctions

Mitochondria are intracellular organelles involved in cell survival and death, and dysfunctions of mitochondria are related to neurodegenerative diseases. As the most abundant protein in the mitochondrial inner membrane, adenine nucleotide translocator 1 (ANT1) plays a critical role in mitochondrial function, including the exchange of adenosine triphosphate/adenosine diphosphate (ATP/ADP) in mitochondria, basal proton leak and mitochondrial permeability transition pore (mPTP). Here, we show that ANT1 transcription is regulated by transcription factor NF-kappa B (NF-κB). NF-κB is bound to two NF-κB responsive elements (NREs) located at +1 to +20 bp and +41 to +61 bp in the ANT1 promoter. An NF-κB signalling stimulator, tumour necrosis factor alpha (TNFα), suppresses ANT1 mRNA and protein expression. Activation of NF-κB by TNFα impairs ATP/ADP exchange and decreases ATP production in mitochondria. Activation of NF-κB by TNFα decreases calcium induced mPTP opening, elevates mitochondrial potential and increases reactive oxygen species (ROS) production in both T98G human glioblastoma cells and rat cortical neurons. These results demonstrate that NF-κB signalling may repress ANT1 gene transcription and impair mitochondrial functions.

a non-specific pore permeable to any molecule smaller than 1.5 kDa, which opens in the inner mitochondrial membrane under conditions of elevated matrix Ca 2+ 19,20 .
Known as a family of transcription factors, nuclear factor kappa B (NF-κ B) is a critical regulator of genes involved in immuno-inflammatory responses, tumorigenesis and apoptosis 21 . Mammalian NF-κ B has five constituents: RelA/p65, RelB, c-Rel, NF-κ B1 (p50) and NF-κ B2 (p52) 22 . Inactivated NF-κ B is sequestered in cytoplasm and bounded to the Iκ B family of inhibitor proteins, including Iκ Bα , Iκ Bβ , Iκ Bγ and Iκ Bε 23 . Stimulated by a variety of inducers, such as tumour necrosis factor α (TNFα ) and lipopolysaccharide (LPS), the activated NF-κ B sub-units translocate into the nucleus to bind with target genes and regulate their transcriptions, leaving the cytoplasmic partner Iκ Bs phosphorylated and degraded 24,25 . Mitochondria are important targets of pro-inflammatory cytokines, and interrelated factors may contribute to the mitochondrial dysfunction associated with inflammation. IL-6, a pro-inflammatory cytokine released in activated glia, has been shown to stimulate ROS accumulation in the brain, contributing to other ROS production mechanisms in inflammation 26 . The inflammatory cytokine TNFα has been shown to induce deterioration of mitochondrial function through suppression of mitochondrial complexes I and IV and pyruvate dehydrogenase activities 27 . It has been shown that LPS-induced inflammation promotes strong microglial activation and induces mitochondrial dysfunction, both in vitro and in vivo 28 . NF-κ B represses mitochondrial gene expression, including cytochrome B and cytochrome C oxidase mRNA levels and NF-κ B is found to be located in mitochondria 29,30 . Previous studies also showed ANT1 expression was reduced following TNFα or H 2 O 2 treatment 31,32 .
ANT1 plays critical roles in mitochondrial functions; however, its molecular transcription is unknown. It would be interesting to investigate the internal relationship between transcriptional factor NF-κ B and ANT1 gene and potential roles of NF-κ B in mitochondrial functions related to ANT1. Our studies here elucidated the mechanism of ANT1 gene transcription. We found NF-κ B repressed ANT1 gene transcription by binding to two NREs in ANT1 promoter. Activation of NF-κ B by TNFα impaired the ATP/ADP exchange, mPTP opening and ROS production. The study provides a molecular link between inflammation and mitochondrial functions.

NF-κB represses ANT1 gene transcription and protein expression. To examine whether ANT1
transcription is regulated by NF-κ B signalling, RT-PCR was used to measure ANT1 mRNA levels in HEK293 cells transfected with IKKβ and NF-κ B/p65. ANT1 mRNA was reduced to 60.11 ± 2.307% and 69.14 ± 2.635% by transfection of IKKβ and NF-κ B/p65 (p = 0.012 and p = 0.022, Fig. 1A), while the ANT1 mRNA level was elevated to 223.9 ± 1.423% by the NF-κ B transcriptional activity inhibitor JSH-23 (p < 0.01, Bar 2 of Fig. 1B). The mRNA of ANT1 was also reduced to 39 ± 2.507% by the NF-κ B activator H 2 O 2 (p < 0.01, Bar 3 of Fig. 1B). The activation of NF-κ B signalling in cells with the transfection of plasmids or stimulators was confirmed by the sharply increased mRNA levels of IκBα (~4 fold increase in NF-κ B transfected group and ~2 fold increase with H 2 O 2 treatment) and TNFα (~2 fold increase in NF-κ B transfected group), which were regarded as the canonical target genes of NF-κ B. Consistent with changes in the mRNA level, the ANT1 protein level was decreased to 57.84 ± 8.528% in cells transfected with NF-κ B/p65 (p = 0.038, Bar 2 of Fig. 1C), and the ANT1 protein level was increased to 122.3 ± 4.273% by NF-κ B inhibitor JSH-23 treatment (p = 0.037, Bar 3 of Fig. 1C). These results demonstrate that NF-κ B signalling represses the transcription of ANT1.
To mimic the physiological activation of NF-κ B signalling in vivo, TNFα was used to stimulate NF-κ B signalling in T98G human glioblastoma cells 33 . Electrophoretic mobility shift assay (EMSA) using consensus NRE as a probe showed that nuclear NF-κ B activity was markedly elevated after 90 minutes of TNFα (10 ng/ml) treatment and decreased to normal level after 6 hours' treatment ( Fig. 1D). Concomitant with the increased NF-κ B activity, real-time PCR showed that ANT1 mRNA levels in T98G cells were reduced to 49.85 ± 4.496% and 77.83 ± 1.781% after 90 and 180 minutes' stimulation with TNFα (10 ng/ml) respectively (Fig. 1E). Similarly, ANT1 protein levels were markedly reduced to 16.48 ± 0.5473% and 31.10 ± 1.659% at 90 and 180 minutes of TNFα (10 ng/ml) treatment in T98G cells, and Iκ Bα levels were also reduced to 39.89 ± 5.541% and 16.14 ± 2.718% at 90 and 180 minutes of TNFα (10 ng/ml) treatment. To exclude that changes in ANT1 were not due to mitochondrial content change, COX-IV protein was measured and no changes were observed with TNFα treatment (Fig. 1F). To differentiate between transcriptional level and post-translational level changes in ANT1 protein, exogenous ANT1 protein was expressed by a plasmid vector, and no significant changes were observed in the Western blot (WB) of exogenous ANT1 with anti-flag antibody after simulation of TNFα , indicating that the change in ANT1 by TNFα arose from ANT1 mRNA transcription and not from post-translational modifications (Fig. 1G). These studies indicate that ANT1 mRNA and protein are repressed by NF-κ B signalling.
Scientific RepoRts | 7:44708 | DOI: 10.1038/srep44708 NF-κB suppresses ANT1 transcription by binding to NREs in ANT1 promoter. To further clarify the molecular mechanism of ANT1 transcription, we cloned an 1132-bp (− 1000 to + 132 bp) fragment located in the 5′ -flanking region of the human ANT1 gene ( Fig. 2A and B) into promoterless vector pGL3-Basic. + 1 was denoted by the first nucleotide of exon 1 in ensemble transcript ENST00000281456.10. The promoter construct showed a high luciferase activity, indicating that the region − 1000 + 132 bp contained the functional promoter region of the human ANT1 gene. As expected, the luciferase activity of ANT1 promoter (pANT1-LucA) reflected by dual-luciferase assay was significantly lower in NF-κ B/p65 transfected cells than in controls (6.814 ± 0.6215 RLU compared with 14.28 ± 1.919 RLU), indicating that the 1.1-kb fragment contained NF-κ B responsive elements. To identify the location of NREs, five truncated pANT1-Luc (B-F) were constructed, containing different regions of ANT1 gene promoter. Dual-luciferase activity showed that NF-κ B/p65 decreased the activity of ANT1 promoters, with the exception of pANT1-LucF (Fig. 2C), indicating that the putative NREs may be located from + 1 to + 61 bp. Chromatin immunoprecipitation (ChIP) showed that NF-κ B/p65 antibody specifically pulls down a genomic DNA region of − 74 bp to + 108 bp containing the putative NREs ( Fig. 2D and E). The deletion plasmids were confirmed by sequencing and restriction enzyme digestion on a 1.5% agarose gel. Vector size is 4.7 kb, and the ANT1 gene 5′ -flanking fragment inserts ranged from 0.1 kb to 1.1 kb. M, marker. (C) HEK293 cells were co-transfected with NF-κ B/p65 expression vector and various ANT1 promoter deletion constructs. Plasmid pRL-TK was used to normalise transfection efficiency, and dual luciferase activities were measured at 24 hours by a luminometer (GloMax 20/20 Luminometer, Promega, Wisconsin, USA). The histogram depicts the mean ratios of relative luciferase units of deletions to pGL3-Basic ± SEM (n = 4); *p < 0.05, Student's t test. RLU, relative luciferase unit. (D) Anti-NF-κ B/p65 (#8242; CST) was used to immunoprecipitate the cross-linked NF-κ B/p65-DNA complex in ChIP assay in HEK293 cells. The immunoprecipitates were analysed with Anti-NF-κ B/p65 antibody. IgG was used as negative control. H.C., heavy chain. Cropped gels were displayed. (E) A pair of primers was used to amplify ANT1 promoter region in ChIP. Signals amplified from input were used as size markers for PCR. IgG and H 2 O were used as negative controls. Lane 1 is input. Lane 2 is immunoprecipitate by anti-NF-κ B/p65 antibody. Lane 3 is immunoprecipitate by IgG antibody and Lane 4 uses H 2 O as a blank template control. M, marker. Cropped gels were displayed.
Scientific RepoRts | 7:44708 | DOI: 10.1038/srep44708 Bioinformatic analysis of ANT1 promoter region of − 74 bp to + 108 bp using JASPAR software (JASPAR 2016) 34 revealed three putative NREs located at + 2 bp to + 14 bp, + 20 bp to + 32 bp and + 49 bp to + 61 bp. To further identify the putative NREs, three putative NREs (NRE1, NRE2 and NRE3) were synthesised, spanning the region from + 1 to + 61 bp of ANT1 promoter (Fig. 3A). EMSA was performed using the consensus NRE as a probe and 50x excess of the three putative ANT1 NREs as competitors. The NRE/NF-κ B band (Lane 2 of Fig. 3B) could be out-competed using the cold consensus NRE (Lane 3 of Fig. 3B), as well as NRE1 and NRE3 spanning from + 1 to + 20 bp and + 41 to + 61 bp (Lanes 5 and 7 of Fig. 3B). Competitors using the mutant consensus NRE or the NRE2 spanning from + 21 to + 40 bp could not out-compete the shifted band (Lanes 4 and 6 of Fig. 3B). In addition, the single mutation of either NRE1 or NRE3 in the pANT1-lucA construct did not abolish the effect of NF-κ B in the ANT1 promoter, while the dual mutations abolished the repressing effects of NF-κ B on the ANT1 The nucleotide sequence was from − 70 bp to + 121 bp region of ANT1 promoter containing the three putative NRE sites. Those bold bases depict the mutant sites in our NRE1 and NRE3. NRE1 site (5′ -AAGGGGGAGCTGCGGGCCAG-3′ ) was mutated to NRE1 mutant (5′ -AAGaaacAaCTGttGGCCAG-3′ ) and NRE3 (5′ -CGCAGGGTCGGGGA CTGCGCG-3′ ) was mutated to NRE3 mutant (5′ -CGCAGGGTCaacaAaTGttCa-3′ ). (B) EMSA was performed with IRDye 800-labelled consensus NRE oligo. Competition assays were performed by unlabelled consensus NRE, mutant consensus NRE and three putative NREs from ANT1 promoter (50x). NRE, NF-κ B responsive elements. (C) HEK293 cells were co-transfected with NF-κ B/p65 expression vector and pANT1-NRE1mut, pANT1-NRE3mut or pANT1-NRE1&3mut. The Renilla luciferase vector pRL-TK was used to normalize transfection efficiency. And dual luciferase activities were measured 24 hours after transfection by a luminometer (GloMax 20/20 Luminometer, Promega, Wisconsin, USA). The histogram depicts the mean ratios of relative luciferase units of mutants to pGL3-Basic ± SEM (n = 4); *p < 0.05, Student's t test. RLU, relative luciferase unit. (D) IRDye 800-labelled NRE1 or NRE3 oligonucleotides from ANT1 promoter were used as probes in EMSA. The addition of anti-NF-κ B/p65 antibody further shifted the NREs-NF-κ B complex band to a higher molecular weight (super-shift). The results of longer time running of EMSA (cropped) were shown below to make the super-shift much clear.
Scientific RepoRts | 7:44708 | DOI: 10.1038/srep44708 T98G cells were transfected with pSiANT1 and primary neurons were infected with lentivirus LV-SiANT1 or negative control (pSiCON or LV-SiCON) for 72 hours. Cell lysates were separated by SDS-PAGE and analysed with anti-ANT1 antibody (ab110322; Abcam). Cropped blots were displayed. The histogram depicts the mean ratios of ANT1 protein to β -actin ± SEM (n = 3); **p < 0.01, Student's t test. (C) ATP/ADP exchange rates were determined in T98G cells with TNFα treatment, ANT1 knockdown (pSiANT1 transfected) and ANT1 overexpression (pANT1 transfected). ANT1 knockdown and TNFα treatment decreased ATP/ADP exchange rate while ANT1 overexpression increased the rate. The histogram depicts the means of the ATP/ADP exchange rates in treatment groups ± SEM (n = 3); *p < 0.05, **p < 0.01, Student's t test, as compared to their respective control groups (CON). (D,E) ATP levels in T98G (D) and neurons (E) were measured by a luminometer using an ATP determination kit (A22066; Invitrogen). The histogram depicts the means of the cellular ATP level in treatment groups ± SEM (n = 3); ** p < 0.01, Student's t test, as compared to their respective control groups (CON). (F) Total and cleaved caspase-3 were determined by caspase-3 (#9665, CST) and cleaved caspase-3 (#9664, CST) antibodies in T98G cells treated by TNFα (10 ng/ml) for 0, 90, 180 and 360 minutes. Cropped blots were displayed. The histogram depicts the mean ratios of cleaved caspase-3 protein level to total caspase-3 protein level ± SEM (n = 3); NS, no significant difference, One-way ANOVA test, as compared with initial time point.
promoter, indicating that the ANT1 promoter contains two NREs at + 1 bp to + 20 bp and + 40 bp to + 61 bp (Fig. 3C). Furthermore, oligonucleotide probes using the ANT1 gene promoter NRE1 and NRE3 sites were synthesised and end-labelled with IRDye 800 infrared dye. A shifted NRE/NF-κ B band was observed after addition of nuclear extract (Lane 3 of Fig. 3D), and the shifted band was further shifted to a greater molecular weight after addition of anti-NF-κ B antibody, suggesting the specificity of NRE and NF-κ B binding complexes (Fig. 3D). The results of longer time running of EMSA were shown below to make the super-shift much clear. Taken together, these results demonstrate that the NF-κ B transcription factor binds to the two NREs in ANT1 promoter at + 1 to + 20 bp and + 41 to + 61 bp.
NF-κB lowers the ATP/ADP exchange rate and ATP level through ANT1. Our data show that the gene transcription of ANT1 may be negatively regulated by NF-κ B signalling. Because the primary function of ANT1 is the exchange of cytosolic ADP and intramitochondrial ATP, we then examined whether NF-κ B modulates the ANT1-dependent ATP/ADP exchange rate and subsequently affects ATP production. The ANT1 knockdown effects of plasmid pSiANT1 (SiANT1) and lentivirus SiANT1 (LV-SiANT1) were validated in T98G and neurons ( Fig. 4A and B). T98G cells were treated with TNFα (10 ng/ml) for three hours to activate NF-κ B signaling. The data show that the ATP/ADP exchange rate in TNFα treated T98G cells was reduced to about 75% of the control cells (p < 0.01, Bar 1 vs. Bar 2 of Fig. 4C). The exchange rate was reduced to 45.28 ± 3.937% in ANT1 knockdown cells (p < 0.01, Bar 3 vs. Bar 4 of Fig. 4C). In addition, the exchange rate was increased to 167.4 ± 20.21% in ANT1 over-expressed cells (p = 0.029, Bar 5 vs. Bar 6 of Fig. 4C). Consistent with the reduction in ATP/ADP exchange rate, ATP levels were reduced to 45.46 ± 4.092% in T98G cells transfected with pSiANT1 (p < 0.01, Bar 1 vs. Bar of Fig. 4D). And ATP levels were reduced to 48.11 ± 4.552% in primary rat neurons infected with LV-SiANT1 (p < 0.01, Bar 1 vs. Bar 2 of Fig. 4E). Similarly, TNFα treatment decreased ATP production in T98G cells to 56.39 ± 3.269% (p = 0.004, Bar3 vs. Bar 4 of Fig. 4D). And TNFα treatment decreased ATP production in primary neurons to 60.43 ± 6.021% (p = 0.001, Bar 3 vs. Bar 4 of Fig. 4E). There was no change in caspase-3 activation in T98G cells after TNFα treatment, indicating that the reduction of ATP had nothing to do with apoptosis (Fig. 4F). All these results indicate that NF-κ B activation by TNFα treatment may diminish the ATP/ADP exchange rate and reduce the ATP level through regulating the ANT1 gene expression.
NF-κB decreases the Ca 2+ -induced mPTP opening level via ANT1. ANT1 had been identified as being responsible for Ca 2+ -induced mPTP opening, and it has been reported that more Ca 2+ than usual is required to activate the mPTP of mitochondria lacking ANT1 35,36 . To verify whether the Ca 2+ -induced mPTP opening level is altered by NF-κ B signalling, T98G cells were treated with the TNFα (10 ng/ml) for three hours, or transfected by pSiANT1. Primary neurons were treated with TNFα (10 ng/ml) for three hours or transfected by lentivirus LV-SiANT1. Ca 2+ ionophore ionomycin (5 μ M) was used to trigger mPTP opening. mPTP opening is reflected by the decreased percentage of initial calcein fluorescence. The basal level of calcein fluorescence showed no difference in either T98G or neurons (Bars 1-4 of Fig. 5A, and Bars 1-4 of Fig. 5B). With the treatment of ionomycin, calcein fluorescence was less reduced by TNFα treatment or ANT1 knockdown than the controls (Bars 5-8 of Fig. 5A, and Bars 5-8 of Fig. 5B). There was no change in the levels of calcein fluorescence in T98G or neurons after treatment with bongkrekate (BKA), an inhibitor of Ca 2+ -induced mPTP opening (Bars 9-12 of Fig. 5A, and Bars 9-12 of Fig. 5B). In addition, with treatment of carboxyatractylate (CATR), an activator of Ca 2+ -induced mPTP opening, TNFα and ANT1 knockdown also showed less decrease of calcein fluorescence compared with the control (Bars 13-16 of Fig. 5A, and Bars 13-16 of Fig. 5B). To confirm this result, a simple measurement of mitochondrial swelling upon Ca 2+ overload were used in T98G. Transfected or TNFα treated T98G cells were added with CATR or BKA, and Ca 2+ -induced mitochondrial swelling was assayed by the decrease in absorbance at 540 nm. TNFα and ANT1 knockdown also showed less level of mitochondrial swelling than their respective controls ( Fig. 5C and D). Taken together, these data imply that NF-κ B activation by TNFα treatment may decrease the Ca 2+ -induced mPTP opening level via ANT1 expression.
NF-κB increases mitochondrial membrane potential (Δψ m ) and ROS production. The mitochondria in ANT1-deficient neurons increases mitochondrial membrane potential (Δ ψ m ), indicating an inherent regulation between ANT1 and Δ ψ m 37 . To elucidate whether NF-κ B signalling may affect the Δ ψ m via ANT1, mitochondrial Δ ψ m were examined in T98G cells and primary neurons treated with TNFα or transfected by pSiANT1 and lentivirus LV-SiANT1. ANT1 knockdown and TNFα (10 ng/ml) treatment increased Δ ψ m level in T98G cells by ~1.5 fold or ~1.3 fold (p < 0.05, Bar 1 vs. Bar 2, and Bar 3 vs. Bar 4 of Fig. 5E). ANT1 knockdown and TNFα (10 ng/ml) treatment elevated the Δ ψ m by ~2.6 fold or ~1.5 fold in primary neurons (p < 0.05, Bar 1 vs Bar 2, and Bar 3 vs. Bar 4 of Fig. 5F). We next determined the cellular ROS, since alterations in mitochondrial membrane potential usually induce ROS generation. In T98G cells, the ROS was increased by ~1.6 fold or ~2.3 fold by TNFα (10 ng/ml) treatment and ANT1 knockdown (p < 0.05, Bar 1 vs. Bar 2, and Bar3 vs. Bar 4 of Fig. 5G). And ROS in primary neurons were increased by ~4.0 fold and ~4.4 fold after TNFα (10 ng/ml) treatment and ANT1 knockdown by lentivirus LV-SiANT1 (p < 0.05, Bar 1 vs. Bar 2, and Bar 3 vs. Bar 4 of Fig. 5H). To confirm this, the production of superoxide radicals was examined by dihydroethidium (DHE) staining in T98G. ANT1 knockdown and TNFα also showed ~1.5 fold and ~2 fold increase in the mean intensity of DHE fluorescence (p < 0.05, Bar 1 vs. Bar 2, in 5I and Bar 1 vs. Bar 2 of Fig. 5J). These results indicate that ROS are increased by NF-κ B signalling through reduction of ANT1.

Discussion
NF-κ B is a transcriptional regulator involved in many pivotal roles in cellular functions including inflammatory responses, tumorigenesis and apoptosis. Our study here identified ANT1 as a novel target of NF-κ B signalling. We showed that NF-κ B/p65 decreased ANT1 mRNA and protein expression by binding to the two NREs located at + 1 bp to + 20 bp and + 41 bp to + 61 bp of the ANT1 promoter. Furthermore, NF-κ B signalling affected the normal mitochondrial function relating to ANT1, including mitochondrial ATP/ADP exchange, calcium-induced mPTP opening and mitochondrial membrane potential, and subsequently led to impairment of ATP production and over-generation of reactive oxygen species. Our study of interactions of NF-κ B and ANT1 therefore reveals a molecular linkage between inflammation and mitochondrial dysfunction. Previous study showed that ANT1 was decreased in inflammatory heart in mice and ANT1 knockdown increased swollen mitochondria and mitochondrial ROS 31 . Our study here provided a molecular mechanism in inflammation induced mitochondrial dysfunctions.
Upon activation of NF-κ B signalling by such as TNFα treatment, the Iκ Bα serine residues 32/36 will be phosphorylated by IKK and subjected to ubiquitination and degradation by proteasome, NF-κ B is translocated into nucleus and downregulates ANT1 gene transcription. Iκ Bα , the inhibitor and cytoplasmic partner of NF-κ B, has been shown to physically interact with ANT1 in mitochondrial intermembrane space, and Iκ Bα ·NF-κ B complex may exert a transcriptionally independent regulatory function on apoptosis 38 . And it may be possible that NF-κ B could affect ANT1 function through a protein-protein interaction with ANT1. Whether downregulation of ANT1 by NF-κ B affects Iκ Bα ·NF-κ B complex localization and function in mitochondria remains unknown though Iκ Bα localization in mitochondria is not affected by its phosphorylation or degradation 38 . Overexpression of ANT1 recruits Iκ Bα ·NF-κ B complex into mitochondria and decreases NF-κ B transcriptional activity in nucleus 39 , which may further affect ANT1 transcription and leads to increased ANT1 expression. There are several levels of interactions between ANT1 and NF-κ B in nucleus and in mitochondria. Over a certain level of ANT1 expression, a vicious cycle between ANT1 and NF-κ B may effect to facilitate cell death. Consistent with this consumption, ANT1 overexpression in cells can dominantly induce cell apoptosis 19 .
Our results showed ANT1 reduction by NF-κ B or knockdown reduced ATP/ADP exchange rate and ATP production. The reduction of ATP in cells with ANT1 knocked down or reduced by NF-κ B activation was probably due to the decreased ADP levels in mitochondria. The reduction of ATP may not be a result of mitochondrial apoptosis signalling, since the mitochondrial inner membrane potential was increased concomitantly. In addition, caspase-3 activation was not changed by TNFα , indicating no apoptosis was induced though there was a significant reduction of ATP. TNFα has also been shown to reduce ATP production by inhibiting oxidative phosphorylation through tyrosine phosphorylation of cytochrome c oxidase 27 . Our studies here suggested ATP can also be reduced due to reduction of ANT1 transcription by TNFα .
Mitochondria are required for cellular bioenergetics, and their dysfunction has been considered as a central cytopathology of neurodegenerative diseases including Alzheimer's disease and Parkinson's disease 40 . Our study here implies that neuroinflammation, a key feature in neurodegenerative diseases, may lead to dysregulation of ANT1 protein in mitochondria and dysfunctions of mitochondria that in turn leads to excessive ROS production. ROS is normally produced in the mitochondrial electron-transport chain (ETC) during respiration, and excess ROS may damage cellular lipids, proteins and DNA, inhibiting their normal functioning 41 . Our finding that NF-κ B/p65 may elevate ROS levels, as down-regulated ANT1 does, is also supported by reports of increased ROS production and oxidative stress in hearts of ANT1 knockout mouse 42 . Recent studies have shown mitochondrial ROS can drive proinflammatory cytokine productions including IL-1β , IL-6 and TNF, resulting in activation of caspase-1-activating complexes known as inflammasomes 43,44 . Human disease with chronic inflammation such as Crohn' disease, diabetes and atherosclerosis are also characterized with excessive ROS 43 . The harmful positive feedback loop between inflammation and ROS induces excessive cell and tissue damage and ultimately leads to destruction of normal tissue and chronic inflammation. Our study here implicated the ANT1 gene is regulated by inflammation and ANT1 reduction successively induces ROS production, thus providing a molecular linkage between inflammation and ROS. Cell cultures. HEK293, HEK293T and T98G human glioblastoma cells were cultured in high glucose Dulbecco's modified Eagle's medium supplemented with 10% FBS. Neuronal cells for primary cultures were from Wistar rat embryos at 17-18 days as described previously 45 . All cells were maintained at 37 °C in an incubator containing 5% CO 2 . The experimental protocols were approved by the Animal Care and Protection Committee of Shandong University and institutional Ethics Committees of Qilu Hospital, and in compliance with ARRIVE guideline.

Chromatin immunoprecipitation (ChIP) and electrophoretic mobility shift assay (EMSA). ChIP
and EMSA were performed as previously described 48 . The association of exogenous NF-κ B with ANT1 promoter in HEK293 cells was confirmed using a chromatin immunoprecipitation assay kit # 17-295, Millipore) following the manufacturer's protocol. Transfected by NF-κ B/p65 plasmid, cells each about 5 × 10 6 cells) were cross-linked by formaldehyde (final concentration of 1%) for 10 min at 37 °C, and then washed by cold PBS twice. The cells were centrifuged and pellets were lysed by 100 μ l 1% SDS lysis buffer and sheared by sonication. The sonicated cell supernatant was then diluted by 9 fold (900 μ l) ChIP dilution buffer and 1/50 (20 μ l) was accepted as input. After the cross-linked proteins and DNA pulled down with p65 antibody (1:100, #8242, CST) or normal IgG as a negative control overnight at 4 degree, the cross links (both input and immunoprecipitated group) were reversed and DNA was recovered by phenol/chloroform extraction. The extract ed DNA was re-dissolved in 20 μ l PCR-grade water. 1 μ l of input or immunoprecipitated DNA were used as templates and confirmed by PCR (35 cycles) using following promoter primers (ChIP-F:5′ -CACCTGCCCAGCCAATGC-3′ and ChIP-R: 5′ -CGCAGGCAGCCCGTTCGT-3′ ). Products of ChIP-PCR were separated on a 1% agarose gel with ethidium bromide. Immunoprecipitation of proteins, after ChIP with antibodies against NF-κ B/p65, was confirmed by Western blot analysis before the ChIP-PCR analysis. For EMSA assay, infrared dye-labeled probe (50 nM) were used in respective incubation and the three sense sequences of ANT1 NREs were 5′ -AAGGGGGAGCTGCGGGCCAG (NRE1), 5′ -GCGGCGGCCCCCTAGCGTCG (NRE2) and 5′ -CGCAGGGTCGGGGACTGCGCG (NRE3). Consensus NRE and mutant NRE were 5′ -AGTTGAGGGGACTTTCCCAGGC, 5′ -CAAAAATGTTCAAAAATGTT.
Scientific RepoRts | 7:44708 | DOI: 10.1038/srep44708 ATP level measurement. ATP level was determined according to a method developed by Yang et al. 49 and measured by a bioluminescence assay using ATP determination kit (A22066; Invitrogen, Waltham, USA) as indicated by the manufacturer. Briefly, cultured cells were counted. 5 × 10 5 cells were collected with a centrifugation at 800 g for 5 min and suspended in 500 μ l boiling water. The cell pellets were then centrifuged at 12,000 g for 10 min and the supernatants were collected. We used the 10 μ l of supernatants per assay in a final reaction volume of 100 μ l. Bioluminescence of ATP was acquired by a luminometer (20/20,Promega Glomax). Luciferase values were converted to nanomoles of the amount of ATP by plotting against a standard curve with certain concentrations of ATP (1 nM to 1 μ M).
Analysis of ATP-ADP exchange rate. The ATP-ADP exchange rate was measured as previously described 50,51 . ADP-ATP exchange rate mediated by the ANT1 was determined with the addition of BeF 3− and Na 3 VO 4 to media of digitonin-permeabilized T98G cells. Cells were cultured in 12-wells plates, suspended in the buffer (8 mM KCl, 110 mM K-gluconate, 10 mM NaCl, 10 mM Hepes, 10 mM KH 2 PO 4 , 5 μ M EGTA, 10 mM Mannitol, 25 μ M AP 5 Petronilli 52 . And the decreased percentage of initial calcein fluorescence could be accepted as mPTP opening level. For determination of mitochondrial swelling, the isolated mitochondria in T98G cells were used to measure the mitochondrial swelling on Ca 2+ overload. Briefly, cells (about 10 7 cells) were washed by PBS at room temperature for twice and collected after the centrifugation at 1000 g for 5 min. Then the cell pellets were suspended in ice-cold buffer (150 mM MgCl 2 , 10 mM KCl, 25 mM Tris HCl, 1 mM EDTA, 0.25 M Sucrose, PH 7.4) containing protease and phosphatase inhibitor cocktail, homogenized and centrifuged at 1000 g for 10 min at 4 °C. The supernatants were centrifuged at 8000 g for 15 min at 4 °C and the resulted pellets contained the mitochondrial fractions. Then the pellets were suspended in mitochondrial swelling buffer (125 mM KCl, 2 mM K 2 HPO 4 , 1 mM MgCl 2 , 20 mM Hepes, 5 mM glutamate, 5 mM malate and 2 μ M rotenone, pH7.4) and 100 μ M CaCl 2 to trigger the mitochondrial swelling. ANT1 ligand CATR (1 μ M) or BKA (5 μ M) was used to trigger or inhibit mPTP opening. Ca 2+ -induced mitochondrial swelling was assayed by the decrease in absorbance at 540 nm and the curves represent typical recordings from experiments of at least three different mitochondrial preparations.

Measurement of intracellular ROS.
For the determination of ROS, 5 μ M dichlorofluorescein diacetate (DCFH-DA) staining was performed using reactive oxygen species assay kit following the manufacturer's instructions. DCFH-DA was deacetylated intracellularly by nonspecific esterase, which was further oxidized by ROS to the fluorescent compound dichlorofluorescein (DCF). DCF fluorescence was detected by FACScan flow cytometer (FACS, AriaIII). For each sample 30 000 events were collected. The mean DCFH-DA fluorescence intensity was determined using FlowJo software (FlowJo 10.0.7). Transfected or TNFα treated T98G cells cultured in 6-wells plate were harvested and then incubated with 5 μ M Dihydroethidium (DHE, S0063, Beyotime) for 45 min at 37 °C in the dark. Subsequently, cells were washed three times with PBS (700 g × 5 min), and the fluorescence intensity of DHE was assayed by FACScan flow cytometer (FACS, AriaIII). For each sample 30 000 events were collected. The mean DHE fluorescence intensity was determined using FlowJo software (FlowJo 10.0.7) Fluorescent imaging of the mitochondrial membrane potential (Δψ m ). Cells cultured in 96-wells plate were washed once by PBS and incubated with 50 nM tetramethylrhodamine methylester perchlorate (TMRM) at 37 °C for 90 min before washing and mounting in Hanks' buffered salt solution for visualization. Depolarized or inactive mitochondria have decreased membrane potential and fail to sequester TMRM. The quantification of the fluorescent intensity was done using the Image J software (Image J 1.46r, NIH, Baltimore, MD).

Statistical Analysis.
All the experiments were repeated at least three times. For immunoblotting, one representative picture was shown, while quantifications were calculated from at least three independent experiments. All data are shown as means ± SEM. Comparisons between 2 groups were performed using Student's t tests and