Effects of Oxidative Stress on the Solubility of HRD1, a Ubiquitin Ligase Implicated in Alzheimer’s Disease

The E3 ubiquitin ligase HRD1 is found in the endoplasmic reticulum membrane of brain neurons and is involved in endoplasmic reticulum-associated degradation. We previously demonstrated that suppression of HRD1 expression in neurons causes accumulation of amyloid precursor protein, resulting in amyloid β production associated with endoplasmic reticulum stress and apoptosis. Furthermore, HRD1 levels are significantly decreased in the cerebral cortex of Alzheimer’s disease patients because of its insolubility. The mechanisms that affect HRD1 solubility are not well understood. We here show that HRD1 protein was insolubilized by oxidative stress but not by other Alzheimer’s disease-related molecules and stressors, such as amyloid β, tau, and endoplasmic reticulum stress. Furthermore, we raise the possibility that modifications of HRD1 by 4-hydroxy-2-nonenal, an oxidative stress marker, decrease HRD1 protein solubility and the oxidative stress led to the accumulation of HRD1 into the aggresome. Thus, oxidative stress-induced HRD1 insolubilization might be involved in a vicious cycle of increased amyloid β production and amyloid β-induced oxidative stress in Alzheimer’s disease pathogenesis.


Introduction
Alzheimer's disease (AD), a progressive neurodegenerative disorder, is the most common cause of dementia in the elderly. The well-known neuropathological hallmarks of AD are extracellular senile plaques, which are aggregates of toxic amyloid b (Ab) peptides, such as Ab40 and Ab42. Abs, Ab40 and Ab42, are peptides generated from the amyloid precursor protein (APP) through sequential proteolytic cleavages by b-secretase (BACE1) and c-secretase complexes [1,2]. APP, a type I transmembrane glycoprotein with a short intracellular carboxyl terminus, is folded and N-glycosylated in the endoplasmic reticulum (ER). The ''Swedish'' APP double mutation results in Ab overproduction, causing early onset familial AD [3]. Aggregated Ab peptides form characteristic senile plaques in the brain tissues of AD patients, whereas intracellular neurofibrillary tangles (NFTs) are composed of paired helical filaments of hyperphosphorylated microtubuleassociated protein tau [4] in affected cortical and subcortical neurons. A number of tau mutants, such as P301L, V337M, and R406W, accelerate the aggregation of tau into filaments [5]. Tau pathology is a later event in AD progression, probably triggered by Ab-dependent hyperphosphorylation of tau [6]. Toxic Ab peptides and hyperphosphorylated tau both interfere with numerous neuronal functions, such as ER function and intracellular trafficking of proteins [7].
ER plays a key role in protein synthesis; newly synthesized membrane and secretory proteins mature in ER, through protein processing, glycosylation, and disulfide bond formation. Various stresses, including hypoxia, glucose starvation, and viral infection, affect ER function and lead to ER stress, which is characterized by the accumulation of unfolded proteins in the ER. Under such conditions termed ER stress, a series of signaling pathways, the unfolded protein response (UPR), including ER chaperone induction and ER-associated degradation (ERAD) are activated in response to unfolded proteins accumulated in the ER [8,9]. In the ERAD pathway, which is an ER protein quality control system and a defense mechanism against ER stress, ERAD target proteins are removed from the ER by retrograde transport to the cytosol, where they are degraded by the ubiquitin-proteasome system [10,11].
Our recent studies have implicated ER stress and ERAD dysfunction in AD pathogenesis [2]. The E3 ubiquitin ligase HRD1, which is a human homolog of yeast Hrd1p/Der3p, forms a complex with its stabilizing factor SEL1L, a human homolog of yeast Hrd3p, in the ER membrane [12]. Furthermore, in the brain, HRD1 is only expressed in neurons, and not glia [13]. In addition, HRD1 is also localized to neural stem/progenitor cells in the subventricular zone of the adult mouse [14]. HRD1 and SEL1L expression induced by ER stress play major roles in ERAD and protect against ER stress-induced apoptosis [15,16]. HRD1 is involved in the degradation of 3-hydroxy-3-methylglutaryl coenzyme A reductase, CD-3d, TCR-a, p53, and several neurodegenerative disease-related proteins, such as Parkin-associated endothelin receptor-like receptor, prion protein, and huntingtin protein [17][18][19][20][21]. Misfolded MHC class I heavy chain, Nrf1, and Z variant a1-antitrypsin are also identified as a substrate for HRD1 [22,23].
We recently demonstrated that HRD1 expression promotes the ubiquitination and degradation of unfolded APP in the ERAD pathway, resulting in decreased Ab production. Conversely, suppression of HRD1 expression leads to APP accumulation and Ab production in neurons. In addition, we found that AD-affected neurons are under ER stress and, a significant decrease in HRD1 levels in the NP-40-soluble fraction was observed in the cerebral cortex of AD patients [2], which negatively correlated with Ab accumulation levels in the human cerebral cortex [24]. Furthermore, we found an increase in the HRD1 levels in the NP-40insoluble fraction, suggesting protein insolubilization [25]. Because the mechanism(s) underlying the change in HRD1 solubility are unclear, we have investigated the possible roles of AD-related molecules and stresses, such as Ab, tau, ER stress, and oxidative stress.

Transgenic Mouse Brain Samples
Transgenic mice expressing hemizygous human Swedish double mutated APP (APP SWE : Tg2576) or the four-repeat isoform of human tau (Tau P301L : JNPL3) were purchased from Taconic (Germantown, NY, USA) and maintained in a room at 22-24uC under a constant 12 h light/12 h dark cycle. All animal experiments were performed in accordance with NIH Guidelines for Care and Use of Laboratory Animals and approved by the Committee for Animal Research at Kyoto Pharmaceutical University. Each mouse was euthanized by cervical spine dislocation without anesthesia, and subsequently the brain tissue was carefully isolated.

Plasmids
Expression vectors for wild-type (wt) human APP 695 (tagged with FLAG and 66His epitopes at the C terminus), human PS2, and the human tau isoforms four-repeat tau (0N4R tau) and P301L were gifts of Dr. Toshiharu Suzuki (Hokkaido University, Japan), Dr. Takeshi Iwatsubo (University of Tokyo, Japan), and Dr. Tomohiro Miyasaka (Doshishya University, Japan), respectively.

Cycloheximide Assay
Normal SH-SY5Y cells were pretreated with 25 mg/ml cycloheximide (CHX) for 5 min before treated with or without 100 mM H 2 O 2 . Cells were harvested at 6 hours after a treatment of H 2 O 2 .
For RNA analysis, cells were homogenized in TRI reagent (Sigma-Aldrich) and reverse transcribed into cDNA using Super-Script VILO cDNA synthesis kit (Invitrogen). Gene expression was analyzed in duplicate with the TaqMan-based real-time PCR assay using a 7500 Real-Time PCR System (Applied Biosystems, Foster City, CA, USA), as described previously [12].

ELISA for Ab Peptides
Secreted Ab peptides were measured by a standard sandwich ELISA using a human Ab  and (1-42) assay kit from IBL (Takasaki, Japan). For APP overexpression, N2a cells stably expressing PS2 were transfected with APP-FLAG using Lipofectamine LTX reagent (Invitrogen).

Statistical Methods
Results were compared using two-tailed Student's t-test or twotailed multiple t-test with Bonferroni correction following analysis of variance. All data are expressed as mean 6 standard error of mean. A p value less than 0.05 was considered statistically significant.

HRD1 Solubility in Neurons was Independent of Ab Peptide or Tau Levels
We previously showed that HRD1 levels in the NP-40-soluble fraction and Ab peptide accumulation are correlated in the human cerebral cortex. However, it is unclear whether Ab generation is dependent on HRD1, or whether HRD1 is decreased by Abinduced neurotoxicity. In addition, although Ab peptide accumulation in the cerebral cortex is a pathological hallmark of AD, the mechanism of Ab toxicity remains unclear. To determine whether HRD1 solubility was affected by Ab, we transiently overexpressed APP in N2a cells stably expressing PS2 to promote Ab production and found increased Ab peptide secretion; approximately 50-fold for Ab40 24 h after transfection, and approximately 20-fold for Ab42 (Fig. 1A). However, no change in the HRD1 levels was observed in the NP-40-insoluble fraction (Fig. 1B). These results indicate that HRD1 solubility was unrelated to the Ab level. On the other hand, the HRD1 levels in the NP-40-soluble fraction were significantly increased in APP-overexpressing cells (Fig. 1B,  C). Additionally, because the GRP78 levels also tend to increase in the NP-40-soluble fraction, it seems that the APP-overexpressing cells were under mild ER-stress conditions (Fig. 1B, C).
The accumulation of hyperphosphorylated tau is believed to play a crucial but undefined role in AD pathogenesis. We assessed the possible effects of tau on HRD1 solubility in N2a cells using transient overexpression of tau 0N4R (wt tau containing 4 Cterminal repeat tau but with no N-terminal insert) or the tau mutant P301L. The level of these tau proteins was increased in the NP-40-soluble and -insoluble fractions compared with that in N2a cells transfected with an empty vector (mock). However, HRD1 and SEL1L levels in the NP-40-soluble and -insoluble fractions were unchanged (Fig. 1D, E). These results show that the accumulation of soluble or insoluble tau had no effect on HRD1 and SEL1L solubility.

HRD1 Solubility was Unaffected by Ab and Tau Overexpression in Transgenic Mice
Short-term overexpression of Ab or tau had no effect on HRD1 or SEL1L solubility in vitro. However, HRD1 solubility might be affected over a longer period. We investigated this possibility by assaying HRD1 levels in the cerebral cortex of hemizygous APP SWE transgenic mice, which express the Swedish form of human APP (Tg2576), and in Tau P301L transgenic mice, which express the four-repeat isoform of human tau (JNPL3). Tg2576 mice are an AD model that develops amyloid plaques but without neurodegenerative changes, whereas JNPL3 is a model of tauopathy that develops NFTs with neuronal loss [26]. Consistent with our in vitro data (Fig. 1), HRD1 and SEL1L levels in the NP-40-insoluble fraction were not decreased in these transgenic mice compared with the wt controls ( Fig. 2A, B). However, the HRD1 levels in the NP-40-soluble fraction was significantly increased in Tg2576 mice ( Fig. 2A), consistent with the experimental result from APP and PS2-overexpressing N2a cells (Fig. 1B, C). These results indicate that HRD1 might act as a defensive factor against APP-and/or Ab-induced neurotoxicity.

ER Stress did not Affect HRD1 Solubility in Neurons
We previously demonstrated that HRD1 levels in the cerebral cortex of AD patients was significantly decreased by insolubility and that AD brains were under ER stress [2,25]. Therefore, we next investigated the effect of ER stress on HRD1 solubility in N2a cells using the ER stress inducers tunicamycin (an inhibitor of protein N-linked glycosylation) and thapsigargin (an inhibitor of the ER Ca 2+ -ATPase). Expression of HRD1, SEL1L, and GRP78 mRNAs, as well as HRD1, SEL1L, and PDI levels in the NP-40soluble fraction, were all significantly increased by tunicamycin or thapsigargin (Fig. 3A, B). However, no significant changes in the HRD1, SEL1L, and PDI levels were observed in the NP-40insoluble fraction from tunicamycin-or thapsigargin-treated cells (Fig. 3C). These results suggest that although HRD1 protein was induced by ER stress, this did not affect HRD1 solubility.

Oxidative Stress Induced HRD1 Insolubility
Ab deposition occurs in blood vessels in the vicinity of cortical micro-infarcts. However, it remains unclear whether and how oxidative stress from chronic cerebral ischemia or cerebrovascular accident is involved in AD pathogenesis [27]. To investigate the influence of oxidative stress on HRD1 and SEL1L solubility, we exposed human neuroblastoma SH-SY5Y cells to exogenous oxidative stress using H 2 O 2 . We found that this stress increased HRD1 and SEL1L levels in the NP-40-insoluble fraction in an H 2 O 2 concentration-dependent manner, whereas their levels in the NP-40-soluble fraction were not significantly affected (Fig. 4A). Interestingly, there was no effect of H 2 O 2 on PDI. We next exposed SH-SY5Y cells to endogenous oxidative stress using rotenone, an inhibitor of the mitochondrial electron transport chain and observed similar effects. The HRD1 and SEL1L levels in the NP-40-insoluble fraction were increased in a concentrationdependent manner, but those in the NP-40-soluble fraction remained unchanged (Fig. 4B). These results indicate that oxidative stress increases the HRD1 and SEL1L levels in the insoluble fraction.

Soluble HRD1 Protein Levels were Maintained by de novo Protein Synthesis
In this study, we did not detect any decrease in HRD1 levels in the NP-40-soluble fraction due to any of the various stressors examined. Turnover of intracellular protein in most neuroblastoma cell lines is faster than in differentiated neurons; therefore, de novo protein synthesis in neuroblastoma cells might mask any decrease in soluble HRD1 protein due to insolubility. To test this hypothesis, we performed a cycloheximide chase assay. In the presence of cycloheximide, the soluble HRD1 protein levels were significantly decreased by H 2 O 2 treatment (Fig. 4C). This result indicates that although oxidative stress actually reduced soluble protein levels, soluble HRD1 protein level was maintained by de novo protein synthesis in neuroblastoma cells.

4-HNE Modification of HRD1
To further elucidate the influences of oxidative stress on HRD1 and SEL1L solubilization, SH-SY5Y cells were exposed to 4-HNE, an unsaturated aldehydic product of membrane lipid peroxidation that modifies cellular DNA and proteins and induces apoptosis [28][29][30]. Interestingly, the HRD1 and SEL1L levels in the NP-40-insoluble fraction were increased by exposure to 4-HNE (Fig. 5A), whereas those in the NP-40-soluble fraction remained unchanged (Fig. 5B). To investigate whether HRD1 protein were modified by 4-HNE, we exposed purified HRD1 to 4-HNE in vitro under non-reducing conditions and observed an increase in the molecular weight of HRD1 (Fig. 5C). This raises the possibility that 4-HNE modifications are involved in HRD1 solubility.

Oxidative Stress Induced HRD1 Aggresome
It is generally accepted that protein aggregates, such as amyloid senile plaques, NFTs, and Lewy bodies, are insoluble. The aggresome is an aggregate of several denatured ER proteins exported to the microtubule-organizing center (MTOC), which is localized around the nucleus and mainly composed of c-tubulin.
To determine if insoluble HRD1 was a component of the aggresome, we examined HRD1 localization in cells under oxidative stress. Double immunofluorescence staining using anti-HRD1 and anti-c-tubulin antibodies showed that HRD1 proteins were present in aggregates adjacent to ER following exposure to H 2 O 2 , rotenone, or 4-HNE (Fig. 6). In addition, the aggregated HRD1 protein was present near MTOC. Therefore, these results suggest that the insoluble HRD1 induced by the oxidative stressors was localized to the aggresome.

Discussion
Our previous studies showed that HRD1 levels in the cerebral cortex inversely correlates with Ab accumulation levels [24]. However, it is unclear whether the decrease in HRD1 protein causes Ab generation or whether Ab neurotoxicity decreases HRD1 levels. Recent reports showed that Ab neurotoxicityinduced apoptosis is due to activation of ER stress-specific initiator caspases, including mouse caspase-12 and human caspase-4 [31,32]. Furthermore, accumulation of Ab in neurons could interfere with the ubiquitin-proteasome system by inhibiting the proteasome and deubiqitinases [33,34]. Because HRD1 is a component of the ubiquitin-proteasome system, HRD1 may also be a target for interference by Ab. In other words, HRD1 insolubility is likely to be caused by Ab.
We here found that the Ab level had no effect on HRD1 solubility in N2a cells stably expressing PS2. However, this experimental system is controversial, in part because the effects of Ab on protein solubility may only become apparent over longer periods. Furthermore, we did not confirm whether the oligomeric and/or fibrillar Ab were formed by endogenous Ab. To solve these problems, we analyzed the cerebral cortex of the APP SWE transgenic mouse line Tg2576. Consistent with the in vitro results, insoluble HRD1 was not increased in the cerebral cortex of Tg2576 mice. In this study, we employed Tg2576 mice to evaluate the effect of accumulated Ab without hyperphosphorylated tau and neurodegenerative change on HRD1 protein solubility. We found that other factors (besides accumulated Ab) were necessary for the insolubilization of HRD1 protein in AD. These in vivo and in vitro results suggest the possibility that a decrease in HRD1 protein levels precedes Ab accumulation in the cerebral cortex of AD patients.
Increased accumulation of hyperphosphorylated tau is also closely involved in AD pathogenesis. However, it is unclear whether phosphorylation and/or accumulation of tau affects ER protein stability. Our results here show that the ubiquitin ligase HRD1 and its stabilizing factor SEL1L, which are components of ERAD, were unaffected by the level of insoluble tau in neurons. Consistent with this, the HRD1 and SEL1 levels were normal in the cerebral cortex of Tau P301L transgenic mice.
Ubiquitination of tau and phosphorylated tau (p-tau) by HRD1 targets these proteins for degradation by the proteasome [35]. Tau phosphorylation is mainly due to glycogen synthase kinase-3b (GSK-3b) [36], which is activated in relation to ER stress and the ER-associated chaperone Bip, also known as GRP78 [37]. Our results and these data suggest that dysfunction of the protein quality control in the ER leads to tau phosphorylation and accumulation. In addition, ER stress may occur in advance of tau hyperphosphorylation. Moreover, our analysis of the effects of Ab and tau on HRD1 solubility supports the hypothesis that the HRD1 protein decreases before the increase in Ab generation. However, it remains to be fully determined whether tau hyperphosphorylation leads to HRD1 insolubility because we did not assay hyperphosphorylated tau in our experiments. In addition, hyperphosphorylation of tau accumulated in the cerebral cortex of JNPL3 mice is at a low level. Regarding the interaction between HRD1 and p-tau [35], we believe that hyperphosphorylated tau is likely to lead to HRD1 accumulation and/or insolubility.
In a previous study, we found that the cerebral cortex of AD patients is under ER stress [2]. However, it remains unclear whether AD-related molecules, such as Ab and hyperphosphorylated tau, cause ER stress, or whether they accumulate in response to ER stress. Our previous results showed that suppression of HRD1 expression causes APP accumulation and ER stressinduced apoptosis [2]. Based on this, we suggest that the decrease and insolubility of HRD1 precede Ab accumulation and ER stress. In this study, we found that tunicamycin-and thapsigargininduced ER stress had no effects on HRD1 solubility, although it did increase HRD1 protein expression, suggesting that HRD1 upregulated may protect against ER stress. Moreover, HRD1 upregulation did not affect its solubility, suggesting that increases in ER stress response-induced proteins do not provoke further ER stress but act to protect against it.
We found that insoluble HRD1 in neuronal cell lines was increased by oxidative stressors, such as H 2 O 2 , rotenone, and 4-HNE, resulting in the presence of HRD1 in aggresomes. In fact, increased oxidative stress is linked to AD progression. Furthermore, Ab peptides are deposited in the vicinity of cortical microinfarcts [27]. Although ischemia is known to cause oxidative stress, it is unclear how oxidative stress affects Ab generation and thus AD onset. Our findings raise the possibility that the complex of the ubiquitin ligase HRD1 with SEL1L is a target of oxidative stressinducing agents, such as reactive oxygen species and 4-HNE. Furthermore, these findings could provide important insights into the link between oxidative stress and Ab generation. Thus, oxidative stress-induced inhibition of HRD1 may promote Ab generation and accumulation.
Although we did not assess whether HRD1 enzyme activity was impaired by oxidative stress, we identified possible oxidative modification of the HRD1 protein. LC-MS/MS analysis revealed cysteine residues in the RING-finger domain of HRD1 as candidates for oxidative modification. In addition, 3D protein structure prediction of the RING-finger domain of HRD1 using the SWISS-MODEL (http://swissmodel.expasy.org) showed that the cysteine residues in RING are localized at the protein surface to coordinate zinc. Therefore, these cysteine residues in the RING of HRD1 may be subjected to oxidative modification. Furthermore, recent reports indicate that the cysteine residues in RINGfinger domains of ubiquitin ligase are sensitive to nitrosative and oxidative modifications [38]. Furthermore, the enzyme activities of Parkin and XIAP (RING-type ubiquitin ligases, which are the same type of HRD1) are regulated by S-nitrosylation of the cysteine residues in the RING-finger domain [39][40][41]. Therefore, the enzyme activity of HRD1 may also be regulated by oxidative stressors though the oxidative modification of RING. In the present study, we found that the 4-HNE induced HRD1 insolubilization. It is well known that 4-HNE specifically modifies cysteine, histidine, and lysine residues [42]. Therefore, the 4-HNE-induced HRD1 insolubilization might be caused by 4-HNE modifications. In this study, we found the possibility that HRD1 protein was subject to 4-HNE modifications in vitro. Additionally, our recent study found that this 4-HNE-induced molecular shift in the HRD1 protein was almost completely prevented by Nethylmaleimide, which causes irreversible alkylation of cysteine (data not shown). This raises the possibility that 4-HNE modifications are involved in HRD1 solubility. However, we have not identified 4-HNE modifications to the cysteine residues of HRD1 protein by LC-MS/MS analysis. This result of LC-MS/ MS suggests the possibility that 4-HNE might be indirectly involved in oxidative modification of cysteine residues of HRD1 protein and affect HRD1 protein solubility. Because recent reports showed that 4-HNE induced oxidative stress though mitochondrial dysfunction [43,44], 4-HNE-induced HRD1 insolubilization may be due to indirect oxidation as a result of mitochondrial dysfunction. Therefore, more detailed studies are needed to determine whether 4-HNE directly or indirectly modifies HRD1 and affects its solubility. On the other hand, the localization of HRD1 to the aggresome may indicate irreversible damage to HRD1 because the aggresome is mostly comprised of denatured proteins [45]. In general, it is believed that HRD1 localizes and functions in the ER, whereas denatured HRD1 is selectively removed from the ER via the ERAD pathway. Therefore, incorporation of HRD1 into aggresomes implies the loss of HRD1 function.
In this study, we demonstrated that de novo protein synthesis masks decrease in soluble HRD1 protein due to oxidative stressinduced insolubilization. Since AD patient's brains appears to be exposed to various stress conditions, other stresses also implicated in soluble HRD1 depletion mechanism in addition to oxidative stress. For instance, ER stress response induces translational arrest. In fact, ER stress and protein synthesis inhibition were observed as AD pathological events [15,40]. To elucidate more precise HRD1 insolubilization mechanism, it is necessary to investigate the effects of multiple stresses on HRD1 protein solubility.
We investigated age-related changes in soluble and insoluble HRD1 in the cerebral cortex of wt mice. We found that the level of soluble HRD1 protein remained constant with age, whereas insoluble HRD1 did not increase with age ( Figure S1). These results suggest that HRD1 might be insolubilized by abrupt increases in oxidative stress, such as cerebral infarction, but not age-associated biological change.
Recent reports showed that Ab enhances the ER stress response through mitochondrial dysfunction [41] and that ER-mitochondrial interactions are involved in Ab-induced apoptosis [42]. Furthermore, oxidative stress triggered by mitochondrial dysfunction may be a key pathogenic trigger in AD pathogenesis [45]. Although these reports indicate a close relationship between Ab and oxidative stress, HRD1 insolubility was only caused by oxidative stress. In this study, endogenous Ab was insufficient to induce oxidative stress. Furthermore, because our experiments raised the possibility that HRD1 is a defensive factor against Abinduced neurotoxicity,oxidative stress-induced HRD1 insolubilization might interfere with ERAD systems, leading to a vicious cycle of increased Ab production and Ab-induced oxidative stress. Subsequently, accumulation of Ab and ER stress promotes neurodegeneration and AD pathology. Our findings support a role for loss of HRD1-mediated ERAD in AD pathogenesis. Figure S1 Age-related changes in HRD1 protein solubility. HRD1 in the cerebral cortex of 1.0, 1.5, and 2.0-year-old C57BL mice. The total lysates of NP-40-soluble (A) and -insoluble (B) fractions were analyzed by western blotting, quantified, and expressed as a dot plot. (PDF)