Background
Amyotrophic lateral sclerosis (ALS) is the most common adult-onset motor neuron disorder. It is characterized by progressive degeneration of upper and lower motor neurons leading to paralysis and, unfortunately, to patient’s death within 2 to 5 years. Nearly 10 % of ALS cases are familial and 90 % are sporadic. Expanded hexanucleotide repeats in C9orf72 account for approximately 30 % of familial cases, mutations in superoxide dismutase 1 (SOD1) for 20 % whereas other genes like TAR DNA-binding protein (TDP-43), fused in sarcoma (FUS), p62/SQSTM1 and Ubiquilin-2 (UBQLN2) account for less than 10 % [
1]. The main pathogenic mechanisms of ALS are still a mystery. Numerous cellular dysfunctions have been linked to ALS physiopathology including oxidative stress, protein inclusions, inflammatory processes, RNA processing and endoplasmic reticulum stress (ER-stress) [
2].
Ubiquilin-2 acts as an important player in the ubiquitin proteasome system (UPS) by connecting the UPS and ubiquitinated proteins. It is also implicated in autophagy, cell cycle progression and cell signaling. UBQLN2 possesses an N-terminal ubiquitin-like domain, a C-terminal ubiquitin-associated domain and a PXX domain essential for protein-protein interaction [
3]. Originally, five X-linked mutations in UBQLN2 gene have been discovered in ALS/FTD familial cases [
4]. All these mutations are located in the PXX domain and one of the most frequent is P497H. Patients with mutant UBQLN2
P497H develop cytoplasmic inclusions positive for major proteins implicated in ALS such as TDP-43, ubiquitin, FUS and p62. Furthermore, ALS/FTD patients without UBQLN2 mutation also express UBQLN2 positive inclusions, supporting an important role of this protein in ALS physiopathology [
4]. More than ten UBQLN2 mutations have been currently described in ALS, not only in the PXX domain [
5‐
8]. UBQLN2 is also implicated in other neurological disorders such as FTD [
4], Alzheimer’s disease [
9] and Huntington’s disease [
10].
Nuclear Factor kappa-B (NF-κB) is a transcription factor implicated in inflammation. NF-κB is formed by members of Rel/NF-κB family such as p50, p52, p65 (RelA), RelB or c-Rel in homo or heterodimeric complexes. The complex composed of p65 and p50 has been the most characterized. A wide variety of extracellular signals lead to NF-κB activation, including cytokines, infectious agents or oxidants. Almost all signals that trigger the NF-κB signaling pathway converge on activation of a molecular complex that contains a serine residue-specific IκB kinase (IKK). In the classical (canonical) NF-κB pathway, activation of the IKK complex leads to phosphorylation mediated by IKKβ of IκB-α, which is subsequently targeted for intracellular ubiquitination and degradation by the proteasome complex. This releases p65 NF-κB from IκB-α inhibitor and the phosphorylated p65 form is then transported to nucleus where it binds to specific response elements (RE) affecting transcription of various genes involved in a diversity of biological processes such as immunity, inflammatory, stress response and development [
11]. NF-κB has an emerging role in ALS or other neurological disorders. NF-κB activity is increased in human neuroblastoma cells expressing mutant SOD1
G93A [
12] and it is up-regulated in motor neurons of sporadic ALS cases [
13]. Our group reported previously that TDP-43 interacts with NF-κB and that NF-κB mRNA levels are abnormally up-regulated in the spinal cord of ALS patients [
14]. Furthermore, NF-κB inhibition by administration of Withaferin A, a known NF-κB inhibitor, reduced ALS disease symptoms in a TDP-43 transgenic mouse model [
14] and extended lifespan of mutant SOD1 ALS mice [
15]. Longevity of mutant SOD1 mice was also increased by microglia-specific inhibition of NF-κB pathway [
16]. These data suggest a central role for the NF-κB pathway in ALS pathogenesis.
Here, we used a NF-κB-luciferase reporter assay to examine the effect of UBQLN2 overexpression on NF-κB activity. We have determined that up-regulation of UBQLN2 enhances NF-κB activation in Neuro2A cells. We also used small interference RNA (siRNA) against UBQLN2 to prevent NF-κB activation by UBQLN2. Treatment of transfected cells with different MAPK inhibitors suggested that NF-κB activation by UBQLN2 species resulted from p38 MAPK activation. Moreover, we found that up-regulation of UBQLN2 protein species causes aggregation and cytoplasmic accumulation of TDP-43. Evidence is presented that UBQLN2 inclusions are dynamic structures which can increase in size over time. Finally, we report that UBQLN2 upregulation enhances ER-stress response and NF-κB-mediated neuronal vulnerability to death caused by exposure to toxic mediator TNF-α.
Discussion
Here, we report that an upregulation of Ubiquilin-2 species can enhance NF-κB activity in neuronal cells. Many lines of evidence support this conclusion: (a) nuclear levels of NF-κB phospho-P65 were increased in Neuro2A cells overexpressing hUBQLN2
WT and hUBQLN2
P497H, and after TNF-α treatment (Fig.
1a, b); (b) luciferase reporter activity of NF-κB was also increased in cells overexpressing hUBQLN2
WT or hUBQLN2
P497H (Fig.
1c, d); and (c) down-regulation of hUBQLN2 by the use of siRNA targeting hUBQLN2 succeeded in reducing NF-κB activity to its basal level (Fig.
1e). So, we detected significant increase of NF-κB activity by UBQLN2 species up-regulation using three different approaches. It should be noted that an up-regulation of UBQLN2 alone was not sufficient to activate NF-κB and that Neuro2A cells required to be primed with TNF-α.
These results are consistent with the view of a converging role for NF-κB in ALS pathogenesis. Several ALS-linked proteins were found to modulate the NF-κB pathway: TDP-43 upregulation can enhance NF-κB activation [
14], mutation in valosin-containing protein (VCP) in mice resulted in NF-κB hyperactivation [
35,
36] and suppression of Optineurin (OPTN) led to neuronal death via NF-κB pathway [
37]. Fuse in Sarcoma (FUS) was found to enhance NF-κB activation induced by physiological stress [
38] and NF-κB-related inflammation was increased in mouse deficient in Progranulin (PGRN) [
39]. Furthermore, there are reports that NF-κB inhibition by Withaferin A conferred neuroprotection in transgenic mouse models [
14,
15].
Our findings that an up-regulation of UBQLN2 species enhances NF-κB activity in a TNF-α treatment dependent manner (Fig.
1) led us to investigate the cellular mechanisms underlying this activation. Evidence for MAPK implication came from the observation that: (a) phospho-p38 MAPK levels were increased in cells transfected with UBQLN2 species mainly with TNF-α treatment (Fig.
2a and
d) and (b) there was inhibition of NF-κB activation due to UBQLN2 species by treatment with p38 inhibitor but not with p42/44 or JNK inhibitors (Fig.
2g,
h). Thus, our results suggest an involvement of p38 MAPK in the enhancement of NF-κB activation by UBQLN2. Figure
2b is showing a hypothetical model of NF-κB activation by UBQLN2 inclusions. Because phospho-p38 MAPK was robustly increased in UBQLN2 overexpressing cells and because its inhibitor SB203580 blocked a significant increase of NF-κB activity in UBQLN2 species-transfected Neuro2A cells, we propose that p38 is the main activator of NF-κB p65 due to UBQLN2 overexpression. This is the first report of NF-κB activation via p38 MAPK pathway by an up-regulation of ALS-linked UBQLN2 mutant. This finding is in line with previous reports of up-regulation of p38 MAPK levels in motor neurons of SOD1
G93A mice [
40,
41].
A common pathological feature of ALS/FTD is the redistribution of TDP-43 from nucleus to cytoplasm in neurons of spinal cord and brain [
42]. Here, we report that physiological (Fig.
3i) expression of hUBQLN2
WT or hUBQLN2
P497H in Neuro2A can promote formation of cytoplasmic inclusions that progressively sequester TDP-43. The evidence is based on the immunolocalization of UBQLN2 and TDP-43 (Fig.
3a, c), the increased TDP-43 levels in insoluble fraction of Neuro2A transfected with UBQLN2
WT and UBQLN2
P497H (Fig.
3d, e, f) and the increased levels of TDP-43 in the cytosol of Neuro2A cells overexpressing UBQLN2 species (Fig.
3h). These results are consistent with previous reports that ALS patients with UBQLN2 mutation are exhibiting UBQLN2 aggregates positives for TDP-43 [
4,
5] and that UBQLN2 binds TDP-43 with high affinity [
43]. A co-localization of either UBQLN2
WT or UBQLN2
P497H with c-TDP-43 has also been shown in Neuro2A cells co-transfected with c-TDP-43 and UBQLN2 [
4] but not in UBQLN2
P497H transgenic mouse [
27] or in UBQLN2
P497H transgenic rats [
28]. This inconsistency can be explained by the low to moderate level of UBQLN2 protein in these in vivo models. It has been proposed that interaction and aggregation between UBQLN2 and TDP-43 is concentration dependent [
4,
43]. Indeed, we noted that TDP-43 was easier to detect in UBQLN2 aggregates at 48 h than 24 h after transfection which correlate to UBQLN2 aggregates sizes (Fig.
3g-
4a).
So, factors in ALS which may contribute to cytosolic accumulation of UBQLN2 such as proteasome deficiency [
4,
44] might also contribute to cytosolic TDP-43 accumulations. The early pathogenic mechanisms underlying TDP-43 recruitment to UBQLN2 inclusions has to be clarified. Recent report suggested that the binding of UBQLN2 to the C-terminal tail of TDP-43 reduces TDP-43 affinity for nucleic acids and may inhibit its physiologic function and increase its aggregation [
43]. Another study has suggested that the oxidation of RRM1 domain of TDP-43 may cause protein aggregation [
45]. Although, there is substantial evidence for oxidative stress in ALS physiopathology [
46], it is still unknown whether UBQLN2 aggregation can cause oxidative stress which lead to TDP-43 RRM1 domain oxidation. However, we report that UBQLN2 up-regulation cause a cellular stress which leads to MAPK activation. Interestingly, kinases have been recently shown to have a critical role in TDP-43 accumulation in stress granules following a chronic stress [
47]. The phosphorylation by kinases may modulate the association of TDP-43 with stress granules and, through a chronic process, could lead to the pathogenic aggregates find in ALS. Consequently, a time dependent recruitment of TDP-43 in UBQLN2 inclusions (Fig.
3g) could be explained in part by chronic kinases activation caused by UBQLN2 dysregulation.
We propose that UBQLN2 overexpression increased vulnerability to neuronal death. Many lines of evidence support this conclusion: (a) protein levels of cleaved caspase-3 and caspase-12, an ER-stress related caspase, were increased in cells overexpressing UBQLN2 species when treated with TNF-α (Fig.
5a); (b) most of cleaved caspase-3 positive cells were positive for UBQLN2 (Fig.
5f, h); and (c) the number of living cells counted by MTS assay were decreased in Neuro2A transfected with UBQLN2
WT or UBQLN2
P497H (Fig.
5i). Moreover, we observed that UBQLN2-induced neuronal death can be decreased with WA treatment (Fig.
5i), which could suggest a role for NF-κB activation in neuronal death. We have previously shown that TDP-43 overexpression can enhance microglial toxicity toward neighboring neurons via NF-κB pathway and that NF-κB inhibition by Withaferin A treatment of TDP-43 mouse model reduces ALS disease symptoms [
14]. Withaferin A also reduced levels of misfolded SOD1 and extended lifespan of mutant SOD1 ALS mice [
15]. NF-κB is also known to modulates apoptosis in neurons treated with glutamate [
48] or with chemicals inducing DNA-damage [
49]. Our results are similarly consistent with previous studies with UBQLN2
P497H rats reporting neuronal loss in the cortex and dentate gyrus and a glial activation surrounding neuronal damages [
28]. These results, taken together, suggest that the up-regulation of UBQLN2 species can act like TDP-43 toxicity in neurons and that neurons bearing UBQLN2 inclusions become more vulnerable to toxic mediator TNF-α secreted by activated microglia in context of inflammation.
There is evidence that mutations in UBQLN2 can slow down degradation of this protein [
4,
44]. Thus, such impairment of protein turnover can lead to an increase in the steady state levels of mutant UBQLN2. A similar post-transcriptional dysregulation in levels of WT UBQLN2 in ALS cases without genetic mutation cannot be excluded. Indeed, UBQLN2 accumulations have been detected in ALS patients without a mutation in the ubiquilin-2 gene [
4]. Our observation that both WT and mutant UBQLN2 species can form inclusions when overexpressed in transfected Neuro2A cells (Fig.
3) are in line with data previously reported by Deng et al. [
4].
Methods
Cell culture, transfection and cell treatment
Almost all experiments were done with Neuro2A cells, which are mouse neuroblastoma cells. Neuro2A cells were growth in Dulbecco’s Modified Eagle Medium (DMEM) with 10 % fetal bovine serum (FBS), 1 % penicillin-streptomycin and 1 % glutamine. Cells were transfected at 80-90 % confluence with lipofectamine 2000 according to manufacturer protocol. Opti-MEM media was replaced by normal growth media 24 h after transfection. PCMV-hUBQLN2
WT and pCMV-hUBQLN2
P497H were used for transfection (see section
plasmids construction). We used pCDNA3 as control plasmid. Cells were collected at 24 h or 48 h after transfection consistent with the experiment. Recombinant mTNF-α (R&D systems, Minneapolis) treatment (20 ng/ml) was done 4 h before collecting the cells. HEK293 cells, which are human embryonic kidney cells, for Fig.
3i were cultured with same protocol as Neuro2A.
For kinase pathways investigation, MAPK specific inhibitors U0126-EtOH, SP600125 and SB203580 (ApexBio, Houston) dissolved at 1 mg/ml in DMSO were used at 10 μM in a 1 h pre-treatment and then TNF-α (20 ng/ml) was added to the cells for 4 h. Human UBQLN2 siRNAs were purchased from Origene (catalog No: SR309321, Rockville). SiRNA SR309321A: rGrGrCrArGrCrUrCrArUrUrArUrGrG rCrUrArArUrCrCrACA was used for the experiments. Cells were co-transfected with siRNA (10 nM) and control plasmid, pCMV-hUBQLN2WT or pCMV-hUBQLN2P497H. NF-κB activation (luciferase assay) was measure at 48 h after transfection and after 4 h TNF-α treatment.
Protein extraction and Western blot analysis
After collecting the cells, cytoplasmic and nuclear extraction or total proteins extraction was realized. Cytoplasmic buffer contained 10 mM HEPES pH 7.5, 10 mM KCl, 1.5 mM MgCl2, 0.34 M sucrose, 10 % glycerol, 1 mM PMSF, 10 mM NaF and 1 mM Na2VO3. After ice lysis and low speed centrifugation, nuclear fraction was obtained with sonication in buffer containing 3 mM EDTA, 0.2 mM EGTA, 1 mM PMSF, 10 mM NaF, 1 mM Na2VO3. 30 μg of proteins was loaded into 10 % SDS-page gels. Total protein was obtained with buffer containing 0.15 M NaCl, 0.05 M tris pH 7.4, 10 % glycerol and 1 % Triton X-100. Cells were incubated on ice for 30 min and then centrifuged to collect the supernatant.
For aggregates assay, we first collected the cells at 48 h after transfection and dissolved them into a re-suspending buffer containing 10 mM tris, 1 mM EDTA and 100 mM NaCl. Then, we added same volume of extraction buffer 1 containing 10 mM Tris, 1 mM EDTA, 100 mM NaCl, 1 % NP-40, protease and phosphatase inhibitor 1X. After sonication and high speed centrifugation (>100 000 g) for 5 min, we kept the supernatant (soluble fraction). We added extraction buffer 2 containing 10 mM Tris, 1 mM EDTA, 100 mM NaCl, 0,5 % NP-40 and protease and phosphatase inhibitor 1X to the pellet 1, sonicated and centrifuged again (>100 000 g for 5 min). After removing the supernatant, we extracted the pellet 2 with 10 mM Tris, 1 mM EDTA, 100 mM NaCl, 0,5 % NP-40, 0,25 % SDS, 0,5 % Deoxycholic acid and phosphatase and protease inhibitor 1X. Finally, we sonicated and kept supernatant 3 (insoluble fraction) [
50].
Antibodies used were phospho-NF-κB p65 (Cell signaling, Whitby, 1:1000), NF-κB p65(Santa-cruz, Dallas, 1:1000), phospho-IκB-α (Cell signaling, 1:1000), IκB-α (Santa-Cruz, 1:1000), actin (Milipore, Etobicoke, 1:20,000), p84 nuclear matrix (Abcam, Cambridge, 1:1000), GAPDH (Abcam, 1:1000), TARDBP (Proteintech, Chicago, 1:2500), FLAG M2 (Sigma-Aldrich, Saint-Louis, 1:1000), UBQLN2 (Abcam, 1:1000), phospho-SAPK/JNK (Cell signaling, 1:1000), SAPK/JNK (Cell signaling, 1:1000), phospho-p42/44 (Cell signaling, 1:1000), p42/44 (Cell signaling, 1:1000), phospho-p38 (Cell signaling, 1:1000) p38 (Cell signaling, 1:1000), cleaved-caspase-3 (Cell signaling, 1:1000), caspase-12 (Cell signaling, 1:1000), phospho-PERK (Cell signaling, 1:1000), PERK (Sigma-Aldrich, 1:1000), ATF6 (Imgenex, 1:1000) and BiP (Cell signaling, 1:1000).
Luciferase assay
Neuro2a cells were stably transfected with pGL4.32 [luc2p/NF-κB-RE/Hygro] (Promega, Madison) and then transfected with previous plasmids. After 24 or 48 h, cells were lysed with glo lysis buffer (Promega, Madison) and 1 volume of Bright Glo luciferase assay system (Promega, Madison) was added after 5 min. Each sample was duplicated and the experiments were repeated more than 5 times. Luciferase activity was measured with Enspire reading machine in a 96 wells plate.
Immunofluorescence
48 h after transfection with pCMV-UBQLN2WT and pCMV-UBQLN2P497H, cells were fixed with 4 % PFA and methanol on 10 mm coverslip. Goat serum 10 % was used for blocking. First antibody was incubated overnight at 4 °C and 1 h at room temperature for secondary antibody. Primary antibodies were monoclonal Flag M2 (Sigma-Aldrich, Saint-Louis, 1:100), TARDBP (Proteintech, Chicago, 1:600), NF-κB p65 (Santa-Cruz, 1:200), IκB-α (Santa-Cruz, 1:200) and SQSTM1/P62 (Cell signaling, 1:200). Secondary antibodies were alexa-fluor 488 goat anti-rabbit (1:500) and alexa-fluor 594 goat anti-mouse (1:500).
Plasmids construction
Human UBQLN2 gene was obtained by PCR from RP11 human BAC: 43 N15. Gene was amplified using the primer 5′GGGGAATT CATGGACTACAAGGACGACGATGACAA GGCTGAGAATGGCGAGAGCAGCGGC-3′ (forward) and 5′-GGGGCGGCCGC TGGGGT GGGATAATCCTCCTAAAC-3′ (reverse). The forward primer introduced an EcoRI restriction site and a flag tag in C-terminal. The reverse primer introduced a NotI restriction site. The P497H mutation was introduced by mutagenesis. The PCR products were ligated into pCDNA3 plasmids with same restriction sites. The plasmids finally drove human UBQLN2 wild-type or P497H mutant c-terminally fused to Flag under the control of a CMV promoter. We named these two plasmids pCMV-hUBQLN2WT and pCMV-hUBQLN2P497H.
To create a FeGFP-hUBQLN2 plasmid, we amplified UBQLN2 sequence from previous pCMV-UBQLN2 plasmids by PCR using the primer 5′-dGGGACGACGGATCCG CTGAGAATGGCGAGAGCAGCGGCCC-3′ (forward) and 5′-dGGGACGACGCGG CCGCTTACGATGGCTGGGAGCCCAG -3′ (reverse). The forward and reverse primers introduced BamHI and NotI restriction sites without the flag tag in c-terminal. The PCR product was ligated into pCDNA3-Flag-eGFP plasmid with the same restriction sites. The plasmids finally drove human UBQLN2 c-terminally fused to Flag-eGFP under the control of a CMV promoter. We named these two plasmids pCMV-FeGFP-hUBQLN2
WT and pCMV-FeGFP-hUBQLN2
P497H. Human wild type TDP-43 gene was used for DsRed-TDP-43 plasmid construction like previously described [
51].
MTS assay
The number of live cells in proliferation was measured by MTS assay 48 h after transfection. Celltiter 96 AQueous One Solution Cell Proliferation Assay (Promega, Madison) was a colorimetric assay for determining the cell viability. MTS tetrazolium compound was reduced in a colored formazan by live cells. Cells were growth in a 24 wells plate. Live cells were washed with PBS X2 and then collected. We took 100 μl of media and putted them in a 96 wells reading plate. 20 μl of One solution Reagent was added and cells were incubated at 37 °C for 1 h. Absorbance was measured at 490 nm. Viable cell was calculated by dividing absorbance in transfected cells by absorbance in non-transfected cells and then reported in percentage of control transfected cells. Cells were either treated or not with Withaferin A (Enzo Life Sciences) 0.5 μM for 2 h and TNF-α 20 ng/ml was added to media for 4 h.
Live cell imaging
Neuro2A cells were transfected with pCMV-FeGFP-hUBQLN2WT, pCMV-FeGFP-hUBQLN2P497H or DsRed-TDP-43. After 24 h, opti-MEM was removed and replaced by DMEM 10 % FBS. Cells were then putted in NIKON ECLIPSE TE2000-E live imaging system. Pictures were taken every 2 min over a 2 h period using Metamorph software.
Statistical analysis
Statistical significance was assessed using GraphPad Prism software. We used Student’s unpaired t-test for generating the p-values. We considered p < 0.05 as statistically significant.
Competing interests
The authors declare no conflict of interest.
Authors’ contributions
J-PJ conceived and supervised this project. GS provided the DsRed-TDP-43, DP did construct the pCMV-UBQLN2WT and pCMV-UBQLN2P497H plasmids, KD performed some western blot analysis and all other experiments were realized by VP-M. VP-M and J-PJ analysed the data and wrote the paper. All authors read and approved the final manuscript.