|Home | About | Journals | Submit | Contact Us | Français|
DHCR24/seladin-1, a crucial enzyme in sterol synthesis, is of lower abundance in brain areas affected by Alzheimer's disease. While high levels of DHCR24/seladin-1 exert antiapoptotic function by conferring resistance against oxidative stress, the molecular mechanism for this protective effect is not fully understood. Here we show that DHCR24/seladin-1 expression is up-regulated in an acute response and down-regulated in a chronic response to oxidative stress. High levels of DHCR24/seladin-1 were associated with elevated cholesterol concentrations and a general increase in cholesterol biosynthesis upon oxidative stress exposure in neuroblastoma SH-SY5Y cells. DHCR24/seladin-1 overexpression conferred resistance to oxidative stress in a cholesterol-dependent manner. Mutating the reductase activity within DHCR24/seladin-1 abolished this protective effect. Conversely, DHCR24/seladin-1 levels diminished upon chronic exposure to oxidative stress. Low levels of DHCR24/seladin-1 were associated with reduced p53 levels, independent of DHCR24 activity and cholesterol concentrations. Additionally, ablation of DHCR24/seladin-1 prevented apoptosis of primary neurons in a p53-dependent manner and reduced the response of critical p53 targets due to deficient stabilization of p53 and therefore elevated p53 ubiquitination and degradation. Our findings reveal a dual capacity of DHCR24/seladin-1, which appears to be involved in two mechanistically independent prosurvival effects, exerting an acute response and a chronic response to oxidative stress.
The 3β-hydroxysterol-Δ24 reductase (DHCR24) is a broadly expressed oxidoreductase, sharing homologies with a family of flavin adenine dinucleotide-dependent reductases (26). The dhcr24 gene is the human orthologue of the diminuto/dwarf1, initially identified in plants, where it is required for the synthesis of brassinosteroids, a group of plant sterols that are essential for normal growth and development (4, 18, 23). In mammals, DHCR24 plays an indispensable role in cholesterol biosynthesis, catalyzing the conversion of desmosterol to cholesterol (6, 26, 27). However, DHCR24 was also described in a different context: dhcr24 expression was shown to be down-regulated in brain areas affected by Alzheimer's disease (11) and was therefore named seladin-1, the selective Alzheimer's disease indicator 1, suggesting an association of DHCR24/seladin-1 with the selective vulnerability of neurons in the affected brain areas. Conversely, high levels of DHCR24/seladin-1 exert protective functions, conferring resistance against oxidative stress and protecting cells from apoptotic cell death (2, 9, 11, 16). Endogenous DHCR24/seladin-1 levels are highly up-regulated upon acute oxidative stress (3, 28), while expression declines to very low levels upon chronic exposure (3), suggesting that DHCR24/seladin-1 plays a role in integrating cellular responses to oxidative stress. However, the precise molecular mechanism for this protective effect is not known. Intriguingly, recent findings identified an interaction between DHCR24/seladin-1 and the tumor suppressor protein p53. Following oncogenic and oxidative stress, seladin-1 binds to p53 in fibroblasts, thus displacing the E3 ubiquitin ligase Mdm2 from p53, which results in p53 accumulation (28). These data argue in favor of a potential tumor suppressor role of DHCR24/seladin-1, indicating that low protein levels enhance p53 degradation and thus prevent senescence in cellular response to Ras/p53-mediated oncogenic signaling (28). However, it was shown that DHCR24 activity of seladin-1, i.e., the oxidoreductase activity of the protein in cholesterol biosynthesis, is not required for p53 binding and the p53-dependent oxidative stress response (28).
Naturally high transcription levels of DHCR24/seladin-1 and gene transfer of DHCR24/seladin-1 cDNA into various cell lines were associated with elevated cholesterol concentrations (3, 9), suggesting a possible role for cholesterol in the protection process. We recently showed that overexpression of DHCR24/seladin-1 in human neuroblastoma SH-SY5Y cells leads to elevated levels of cellular and membrane cholesterol, thereby enhancing the formation of lipid rafts in these cells (6).
Oxidative stress leads to the production of reactive oxygen species which attack lipid membrane constituents such as unsaturated phospholipids, glycolipids, and cholesterol, resulting in cellular dysfunction and cell death (10). During apoptosis, sterol regulatory element-binding proteins (SREBPs), which regulate expression of genes involved in lipid and cholesterol homeostasis (30), were shown to be activated by caspase cleavage (21). These observations have led to the hypothesis that cholesterol may be required in the early stages of apoptosis to maintain plasma membrane integrity (25). Moreover, plasma membrane compartments rich in cholesterol may participate in signal transduction pathways activated upon oxidative stress, and thus enhance prosurvival pathways, while cholesterol depletion appears to increase apoptotic events triggered by hydrogen peroxide treatment (29). In addition, methyl-β-cyclodextrin-mediated cholesterol depletion causes apoptosis, which is associated with caspase-3 activation and Akt inactivation, while cholesterol reload replenishes rafts on the cell surface and restores Akt activation and cell viability (13, 29).
In the present work, we analyzed the role of DHCR24/seladin-1 in integrating cellular responses to oxidative stress by employing primary neurons and neuroblastoma SH-SY5Y cells. We found that DHCR24/seladin-1 expression is up-regulated in an acute response and down-regulated in a long-term response to oxidative stress. In this context, protective effects of high DHCR24/seladin-1 levels were mediated by increased cellular cholesterol concentrations, which resulted from a generally enhanced cholesterol biosynthesis after exposure to oxidative stress. In contrast, protection against oxidative stress mediated by low levels of DHCR24/seladin-1 was associated with reduced levels of p53 and elevated p53 ubiquitination, while overexpression of DHCR24/seladin-1 stabilized p53, independent of DHCR24 activity and cholesterol concentrations. These findings reveal a dual capacity of DHCR24/seladin-1, which appears to be involved in two mechanistically different prosurvival effects, exerting an acute response as well as a late response to oxidative stress.
A heterozygous breeding pair with target depletion of one DHCR24 allele was received from E. Feinstein (Quark Biotech, Inc.). Mice were bred and genotyped as described previously (27). All animal experiments and husbandry were performed in compliance with national guidelines and were approved by the veterinary authorities of the Canton of Zurich.
SH-SY5Y cells were cultured in Dulbecco modified Eagle medium-F12 medium (Invitrogen) containing 10% fetal calf serum (FCS), 5% horse serum, 5,000 U/ml penicillin, and 5,000 mg/ml streptomycin.
DHCR24/seladin-1-overexpressing SH-SY5Y cells and enhanced green fluorescent protein (EGFP)-overexpressing control cells were generated as described previously (6). The mutated N294T/K306N DHCR24/seladin-1 construct was made using the QuikChange II XL site-directed mutagenesis kit (Stratagene) according to the manufacturer's instructions. The primer sequences are listed in Table Table11.
For cholesterol upload, cells were grown in culture medium containing 5, 10, and 20 μg/ml cholesterol (Sigma) for 12 h. Amphotericin B treatment was carried out according to the manufacturer's protocol (Sigma). Cell viability after amphotericin B treatment was determined as described below.
Primary cultures of cortical neurons from wild-type, DHCR24+/−, and DHCR24−/− mice were prepared from embryos as described previously (1). Neurons (glial contamination, <5%) were cultured in Dulbecco modified Eagle medium (Invitrogen) supplemented with 10% FCS and 5 μg/ml cholesterol.
The viability of the individual cultures was determined using the in vitro toxicology assays based on lactate dehydrogenase (LDH) or on Tox-1 (thiazolyl blue tetrazolium bromide [MTT]) (both from Sigma) according to the manufacturer's instructions.
Two oligonucleotides recognizing the DHCR24/seladin-1 sequence were purchased from Dharmacon (sense, GAAGUACGUCAAGCUGCGUUU; antisense, ACGCAGCUUGACGUACUUCUUUU). SH-SY5Y cells were transfected for 4 h with a final concentration of 100 nM using Lipofectamine (Invitrogen) according to the manufacturer's instructions. Experiments were carried out 72 h after transfection.
Human bone marrow cells were obtained from the iliac crest of normal donors marrow aspirates as previously described (3). Melanoma cell lines used were obtained from clinical specimens and cultured under standard conditions. The melanoma cell lines 20842P and 20842M2, respectively, named S/P and S/M2, were established from the same melanoma patient (9).
Membrane fractions from SH-SY5Y cells were prepared as described before (6). Lipids were extracted from cells and membrane pellets according to the Bligh-Dyer method using chloroform-methanol (5). Cholesterol was measured using the Amplex red cholesterol kit (Invitrogen) according to the manufacturer's instructions. The amount of cholesterol was normalized to the corresponding protein amounts in cultured cells.
Tosylactivated M-280 beads (Dynal) were coated with anti-p53 antibody in 0.1 M borate buffer (pH 9.5) for 24 h at 37°C while shaking at 1,400 rpm (approximately 3 μg antibody per 107 beads). Beads were washed in phosphate-buffered saline and blocked with 0.2 mM Tris buffer (pH 8.5) for 4 h at 37°C while shaking. Unbound antibody was removed by washing with 1% Triton in phosphate-buffered saline for 10 min at 4°C. Cells were lysed in a 20 mM Tris buffer (pH 7.5) containing 137 mM NaCl, complete proteinase inhibitor, and 1% sodium dodecyl sulfate (SDS) to also access the nuclear fraction. Equal amounts of protein per sample were diluted to a final SDS concentration of 0.05% in 20 mM Tris buffer (pH 7.5) to allow antibody binding and incubated with the antibody-coupled beads for 4 h at 4°C. Beads were washed in a 50 mM Tris buffer (pH 7.4) containing 150 mM NaCl, 1 mM dithiothreitol, 0.1% SDS, 1% NP-40, and 0.25% deoxycholate and boiled in 1× sample loading buffer for SDS-polyacrylamide gel electrophoresis for 10 min at 95°C.
SDS-polyacrylamide gel electrophoresis and Western blotting was performed as described previously (6). Primary neurons and SH-SY5Y cells were homogenized in 20 mM Tris buffer (pH 7.5) containing 137 mM NaCl, 1% SDS, and complete proteinase inhibitor mix (Roche). Polyclonal anti-DHCR24/seladin-1 antibody was generated in rabbit against the N terminus of the human protein (nanoTools, Teningen, Germany). Rat antihemagglutinin (Roche Molecular Biochemicals), monoclonal anti-p53 (Santa Cruz), monoclonal anti-p-ERK (Santa Cruz), polyclonal anti-pAkt1/2/3 (Santa Cruz), polyclonal anti-cleaved caspase-3 (Cell Signaling), polyclonal anti-p21 (Santa Cruz), monoclonal anti-MDM2 (Santa Cruz), polyclonal anti-Puma (Cell Signaling), polyclonal anti-Bax, monoclonal antiubiquitin (Chemicon), monoclonal anti-β-actin (Abcam), and anti-GAPDH (glyceraldehyde-3-phosphate dehydrogenase) (Biodesign) antibodies were used for Western blot analysis. Quantification was carried out by densitometry of the scanned films under conditions of nonsaturated signal using the ImageJ software.
Total RNA was prepared from cell cultures using the RNeasy kit (Qiagen), treated with RNase-free DNase (Qiagen), and quality checked on a 2% agarose gel. One microgram was reverse transcribed with the QuantiTect reverse transcription kit (Qiagen). Real-time PCR was performed in triplicates using Full Velocity SYBR green QPCR master mix (Stratagene) reagent according to the manufacturer's protocol. The analysis was repeated if the standard deviation of the triplicate measurements was higher than 2.5% of the mean. Primers were used at a final concentration of 500 nM. The sequences of the primers used for real-time PCR are listed in Table Table1.1. Relative mRNA levels of the genes of interest were normalized to GAPDH expression using the simplified comparative threshold cycle ΔCT method [2(CT GAPDH − CT gene of interest)].
Data were statistically analyzed by a two-tailed t test, with results deemed significant when the P value was <0.05. In all graphs, means ± standard deviations are shown.
To determine how oxidative stress affects endogenous DHCR24/seladin-1 levels, we exposed naïve neuroblastoma SH-SY5Y cells to 100 μM H2O2 for up to 12 h. DHCR24/seladin-1 mRNA levels increased in response to oxidative stress in a time-dependent manner, reaching the maximum after 4 h. However, chronic exposure to H2O2 resulted in decreased DHCR24/seladin-1 expression, with mRNA levels dropping significantly below baseline after 12 h (Fig. (Fig.1A).1A). Intriguingly, we found high levels of DHCR24/seladin-1 associated with high levels of cholesterol. In this context, cholesterol levels increased moderately after 1 hour of H2O2 treatment, reached maximum levels after 4 h, and decreased again to the baseline after 8 to 10 h (Fig. (Fig.1B1B).
Since DHCR24/seladin-1 catalyzes the very last steps in cholesterol biosynthesis (26), we aimed to assess whether oxidative stress influences other cholesterol biosynthesis-related enzymes. Therefore, we analyzed gene expression of 3-hydroxy-3-methyl-glutaryl-coenzyme A (HMG-CoA) reductase, the enzyme catalyzing the rate-limiting step in isoprenoid/cholesterol synthesis, and of SREBP1, which regulates expression of genes involved in lipid and cholesterol homeostasis. Expression of HMG-CoA reductase and SREBP1 was moderately increased upon H2O2 exposure for 1 h (143% ± 9.9% and 251% ± 24%, respectively) and for 4 h (207% ± 16% and 265% ± 103%, respectively), while expression of both genes returned to normal after 8 and 12 h (Fig. (Fig.1C1C).
To characterize the effect of DHCR24/seladin-1 overexpression on cell viability upon exposure to oxidative stress, we analyzed human neuroblastoma SH-SY5Y cells stably transfected with human DHCR24/seladin-1 under control of the cytomegalovirus promoter. SH-SY5Y cells expressing EGFP under control of the same promoter were used as controls (Fig. (Fig.2A).2A). Cellular and membrane cholesterol levels were significantly higher upon DHCR24/seladin-1 overexpression (168% ± 9.1% and 190% ± 11.4%, respectively) (Fig. (Fig.2B),2B), while endogenous desmosterol levels were reduced by more than 70% (data not shown). In a recent study, we showed that cholesterol increment due to DHCR24/seladin-1 overexpression influences membrane composition, leading to enhanced formation of lipid rafts (6). Exposure of SH-SY5Y cells to H2O2 revealed significantly higher cell viability in DHCR24/seladin-1-overexpressing cultures (82% ± 6.9% of untreated control values) compared to EGFP-expressing control cultures (45% ± 7.2% of untreated control values) (Fig. (Fig.2C2C).
To evaluate whether, independent of DHCR24/seladin-1 levels, increased cholesterol concentrations are implicated in the observed protective mechanism, we added cholesterol to membranes of naïve SH-SY5Y cells and exposed these cells to oxidative stress. To this end, we first determined the ideal dose of cholesterol with respect to membrane cholesterol levels and cell viability (Fig. (Fig.2D).2D). Cholesterol measurements of membrane sterols revealed that the presence of 5 μg/ml did not affect membrane cholesterol concentrations, while 10 μg/ml cholesterol significantly increased membrane cholesterol levels (140% ± 15%). The presence of 20 μg/ml cholesterol resulted in highly elevated levels (169% ± 8.3%) but was toxic and induced cell death. Therefore, the dosage of 10 μg/ml cholesterol was chosen for the following experiments. To confirm that cholesterol taken up from the medium was correctly integrated in membranes of SH-SY5Y cells, we compared the amphotericin B susceptibilities of control, DHCR24/seladin-1-overexpressing, and cholesterol-uploaded SH-SY5Y cells. Amphotericin B complexes with sterol-rich membranes forming aqueous pores, resulting in cell lysis (8, 12), and therefore is an appropriate indicator for altered membrane lipid composition. Amphotericin B treatment followed by a cell viability assay confirmed increased membrane cholesterol levels in DHCR24/seladin-1-overexpressing and cholesterol-treated cells, indicating a proper integration of the uploaded cholesterol in the membranes of naïve SH-SY5Y cells at 10 μg/ml (Fig. (Fig.2E).2E). Cholesterol-treated cells exhibited a significantly higher resistance to oxidative stress than control cells (75% ± 6.3% and 52% ± 8.4% of untreated control values after 0.5 and 2 h, respectively) (Fig. (Fig.2F),2F), similar to what was observed for DHCR24/seladin-1-overexpressing cells.
Both proapoptotic and antiapoptotic signaling pathways become activated in response to H2O2 (17). Consistent with earlier reports showing that protein kinase B/Akt and ERK play an important prosurvival role depending on membrane cholesterol concentrations (13, 29), we found ERK and Akt phosphorylation to be significantly increased in DHCR24/seladin-1-overexpressing cultures (140% ± 10% and 232% ± 50%, respectively) and high-cholesterol cultures (154% ± 16% and 174% ± 26%, respectively) (Fig. 2G and H) after 2 h of treatment with 200 μM H2O2. These data provide evidence of a cholesterol-related prosurvival effect that is per se independent of DHCR24/seladin-1 protein levels, indicating that 2 h after H2O2 treatment, high cholesterol levels are the key determinant of the observed protective function. Exposure of cells to H2O2 led to increased p53 levels compared to control conditions, but p53 protein levels were not altered between the three groups (Fig. (Fig.2G2G).
In order to analyze whether the intrinsic DHCR24 activity is crucial for the protective effect observed in DHCR24/seladin-1-overexpressing cultures, we inserted two mutations (N294T/K306N) in the DHCR24/seladin-1 cDNA, which were shown to abrogate the DHCR24 activity (26). Overexpression of DHCR24/seladin-1 carrying the N294T/K306N double mutation (Fig. (Fig.3A)3A) neither resulted in increased cholesterol concentrations (Fig. (Fig.3B)3B) nor protected cells from oxidative stress. Compared to cells overexpressing wild-type DHCR24/seladin-1, N294T/K306N-DHCR24/seladin-1-overexpressing and naïve SH-SY5Y cells released significantly higher levels of LDH into the culture medium following H2O2 treatment (Fig. (Fig.3C),3C), further speaking in favor of a protective effect of high DHCR24/seladin-1 levels which is related to its DHCR24 activity, i.e., the concomitant increase in cholesterol concentrations.
High endogenous levels of DHCR24/seladin-1 were found in various peripheral human tissues and different cell types (3, 9, 11). To evaluate whether DHCR24/seladin-1 expression is essentially associated with other genes implicated in cholesterol synthesis or may be regulated independently, we analyzed expression profiles of HMG-CoA reductase and SREBP1 in cells with different endogenous DHCR24/seladin-1 expression levels. In a first set of experiments, we analyzed human mesenchymal stem cells (hMSC), which were recently shown to express high levels of DHCR24/seladin-1 in association with elevated cholesterol levels (3). In contrast, neuronal differentiation of these cells (hMCS-n) resulted in markedly decreased DHCR24/seladin-1 gene expression associated with lower concentrations of cellular cholesterol (3). Expression analyses with five independent hMSC/hMSC-n cultures confirmed these previous data, revealing an average reduction of 68% ± 22% in DHCR24/seladin-1 expression after neuronal differentiation (Fig. (Fig.4A).4A). Moreover, HMG-CoA reductase (Fig. (Fig.4B)4B) and SREBP1 (Fig. (Fig.4C)4C) expression in differentiated hMSC-n cultures was notably reduced, by 48% ± 30% and 68% ± 28%, respectively, compared to that in their undifferentiated counterparts. Direct comparison of DHCR24/seladin-1 and HMG-CoA reductase mRNA abundances in hMSC-n and hMSC revealed a correlation (r2) of 0.9332 (P < 0.001) (Fig. (Fig.4D),4D), demonstrating a linear relationship between the expression levels of the two genes. A strong correlation was also found between DHCR24/seladin-1 and SREBP1 expression (R2 = 0.6919; P < 0.05) (not shown).
To confirm these findings with yet another cell type, we analyzed cholesterol-related gene expression in melanoma cells derived from a cutaneous metastasis (S/M2) and from the autologous primary tumor (S/P). In S/M2 cells, DHCR24/seladin-1 expression was significantly higher than in S/P cells, which was shown to be associated with elevated levels of cellular cholesterol and a higher resistance against oxidative stress (9). In three independent S/P and S/M2 cultures, direct comparison of the expression levels of HMG-CoA reductase, SREBP1, and DHCR24/seladin-1 revealed a significant correlation (r2 = 0.8906 and P < 0.05 for HMG-CoA reductase versus DHCR24/seladin-1 [Fig. [Fig.4E]4E] and r2 = 0.8618 and P < 0.05 for SREBP1 versus DHCR24/sealdin-1 [Fig. [Fig.4F]),4F]), confirming a linear relationship between DHCR24/seladin-1 and the investigated cholesterol synthesis-related genes. Additionally, expression levels of all three genes correlated significantly with the cellular cholesterol concentrations of S/P and S/M2 cultures (r2 = 0.9771 and P < 0.001 for DHCR24/seladin-1 [Fig. [Fig.4G],4G], r2 = 0.8463 and P < 0.05 for HMG-CoA reductase [Fig. [Fig.4H],4H], and r2 = 0.7997 and P < 0.05 for SREBP1 [not shown]).
While several findings demonstrate that high DHCR24/seladin-1 levels exert a protective effect against oxidative stress, low levels of DHCR24/seladin-1 were shown to prevent senescence, resulting in increased p53 degradation and subsequent inhibition of the cell cycle arrest following oncogenic stress (28). To determine how long-term exposure to oxidative stress affects endogenous p53 levels, we treated naïve SH-SY5Y cells with 100 μM H2O2 for up to 12 h. The resulting protein pattern of p53 in response to oxidative stress was similar to that of DHCR24/seladin-1: increasing in a time-dependent manner, reaching the maximum level between 3 and 6 h, and decreasing to baseline levels after 10 to 12 h (Fig. (Fig.5A).5A). Remarkably, both p53 and DHCR24/seladin-1 levels increased up to sixfold, followed by a marked decline of approximately 50% below baseline levels after 12 h of H2O2 treatment (Fig. 5B and C). While p53 protein levels were significantly reduced after long-term H2O2 treatment of naïve SH-SY5Y cells (Fig. 5A and C), treatment of SH-SY5Y cells overexpressing DHCR24/seladin-1 did not result in decreased p53 levels (Fig. 5D and E). Additionally, overexpression of DHCR24/seladin-1 containing the N294T/K306N double mutation abolished p53 reduction (Fig. 5D and E), giving further evidence for a possible DHCR24/seladin-1-mediated stabilization of p53, independent of DHCR24 activity.
To evaluate the effect of DHCR24/seladin-1 ablation on neuronal survival, we took advantage of mice deficient in both dhcr24 alleles (DHCR24−/−) (27). As expected, DHCR24/seladin-1 expression in primary neurons derived from DHCR24−/− mice was not detectable (Fig. (Fig.6A).6A). DHCR24−/− mice entirely lack brain cholesterol, while the direct precursor desmosterol accounts for over 99% of all sterols in brains of these mice (6, 20, 27). However, neurons grown in cholesterol-containing medium revealed no difference in cellular sterol content (Fig. (Fig.6B),6B), indicating that endogenous desmosterol and absorbed cholesterol make up membrane sterol concentrations that are comparable to those in wild-type neurons. To confirm that sterols were correctly integrated in neuronal membranes, we compared amphotericin B susceptibilities of wild-type (DHCR24+/+) and DHCR24−/− neurons. Cell viability after amphotericin B treatment was not changed between the two cultures, which is indicative of a similar amount of properly integrated membrane sterols in these neurons (Fig. (Fig.6C).6C). Moreover, there was no difference between the cultures in morphology or cell viability prior to H2O2 treatment (data not shown).
To evaluate the effect of DHCR24/seladin-1 ablation on neuronal survival after oxidative stress induction, we exposed primary neurons of wild-type and DHCR24−/− mice to H2O2. Cell viability was not altered between the two cultures after 10 min of treatment, while significantly less cell death was detected in DHCR24−/− cultures treated with H2O2 for 100 min (84% ± 1.7% of untreated control values) than in treated wild-type controls (61% ± 2.5% of untreated control values) (Fig. (Fig.6D).6D). In addition, densitometric measurements of Map-2-stained neurons revealed markedly higher numbers of surviving neurons in DHCR24−/− cultures than in wild-type controls after 3 h of exposure to H2O2 (data not shown). To monitor possible changes in cell signaling responsible for the difference in the susceptibility to oxidative stress, we analyzed the phosphorylation states of ERK and Akt. Cell signaling patterns did not reveal obvious differences between wild-type and DHCR24−/− neurons at any time point: phosphorylation of ERK and Akt was elevated after 10 min of H2O2 exposure but markedly declined to similar extents after 100 min in neurons of either genotype (Fig. (Fig.6E).6E). The basal levels of p53 in untreated cultures were similar for the two genotypes (Fig. (Fig.6F,6F, upper panel). However, p53 levels were considerably lower in neurons derived from DHCR24−/− mice (53% ± 10%) (Fig. 6F and G) than in wild-type cultures after 100 min of H2O2 exposure. Moreover, significantly lower levels of cleaved caspase-3 were observed in DHCR24−/− cultures after oxidative stress exposure (25% ± 3.5%) (Fig. 6F and H), confirming that DHCR24/seladin-1 ablation protected against oxidative stress-induced apoptosis.
To evaluate the effect of DHCR24/seladin-1 depletion on p53 function, we analyzed the transcription levels of critical p53 targets, using an siRNA approach to down-regulate DHCR24/seladin-1 in SH-SY5Y cells. Nontargeting siRNAs were used to transfect control cultures. Three days after transfection, DHCR24/seladin-1 transcripts were down-regulated by more than 80% in seladin-1 siRNA-treated cultures compared to controls (Fig. (Fig.7A).7A). However, DHCR24/seladin-1 down-regulation did not alter basic expression of p53 and p53 target genes such as those for p21, puma, and noxa (Fig. (Fig.7A).7A). Intriguingly, the expression patterns of p53, p21, puma, and noxa were altered between seladin-1 siRNA and control cultures after long-term H2O2 exposure (8 h). While gene expression of p53 was higher, expression of p21, puma, and noxa was lower in DHCR24/seladin-1-depleted cultures than in controls (Fig. (Fig.7B).7B). To clarify how DHCR24/seladin-1 status affects the p53 protein level, we determined total and ubiquitinated p53 levels at different time points after H2O2 exposure. DHCR24/seladin-1 siRNA-mediated depletion did not influence the basal p53 level but reduced the p53 protein increment upon H2O2 exposure for 1, 3, and 4 h. In addition, p53 protein levels were considerably lower in seladin-1-depleted cultures than in controls after long-term exposure (10 h) (Fig. (Fig.7C),7C), confirming the notion that DHCR24/seladin-1 stabilizes p53 and protects it from ubiquitin-mediated degradation. To corroborate this assumption, we immunoprecipitated p53 from DHCR24/seladin-1-depleted and control cell homogenates and evaluated p53 ubiquitination during H2O2 exposure. While in control cultures only faint levels were detectable, DHCR24/seladin-1 depletion resulted in increased levels of ubiquitinated p53, both before and after 1 to 3 h of H2O2 exposure (Fig. (Fig.7D).7D). These results show that p53 turnover is substantially influenced by DHCR24/seladin-1. Western blot analyses showed that MDM2, p53, Bax, p21, and Puma-β protein levels after 3 and after 8 h of H2O2 exposure were considerably lower in DHCR24/seladin-1-depleted than in control cultures (Fig. 7E and F), while no apparent differences in basal protein levels were detected before H2O2 exposure (not shown). However, Puma-α showed an inverted pattern, with higher levels in DHCR24/seladin-1-depleted cells and lower levels in control cultures at both time points.
These data show that DHCR24/seladin-1 depletion substantially alters p53-mediated long-term responses to H2O2-induced oxidative stress, presumably due to a reduced stabilization of p53 by DHCR24/seladin-1 and therefore subsequent ubiquitination.
The data presented here provide evidence for a dual role of DHCR24/seladin-1 in protecting against oxidative stress. We found that up-regulation of DHCR24/seladin-1 upon oxidative stress is part of a general increase in cholesterol biosynthesis, leading to elevated membrane cholesterol levels. In contrast, DHCR24/seladin-1 levels decline below baseline upon chronic exposure to oxidative stress. This down-regulation seems to be part of another prosurvival strategy, which is independent of membrane cholesterol but depends on an interaction of DHCR24/seladin-1 with the tumor suppressor protein p53. The seemingly paradoxical factor that both too much and too little DHCR24/seladin-1 appear to provide protection against oxidative stress might be explained by the fact that the two protective mechanisms are completely different and independent of each other: high DHCR24/seladin-1 levels protect the cell from oxidative stress through a cholesterol-dependent mechanism, possibly involving prosurvival factors such as Akt, while low DHCR24/seladin-1 levels enhance the resistance against oxidative stress by altering p53 status and function.
Except for the brain, DHCR24/seladin-1 is expressed in most peripheral organs (11). Differential expression of DHCR24/seladin-1 is also found in several types of cancer (9, 15, 16), and the protective effect of high DHCR24/seladin-1 levels conferring resistance from apoptosis was described for various cell types (2, 9, 11, 16), while the mechanism for this protective effect is not fully understood. Therefore, we aimed to investigate the molecular mechanism of the prosurvival effect of high DHCR24/seladin-1 levels and whether it is related to the DHCR24 activity of the protein. Moreover, we aimed to determine whether DHCR24/seladin-1 expression is up-regulated in an independent fashion or if high levels of endogenous DHCR24/seladin-1 are essentially associated with a generally high cholesterol biosynthesis.
Cholesterol biosynthesis is tightly regulated by a feedback regulatory system. This includes, among other factors, sterol intermediates (22) that can function as direct effectors acting on the degradation process of the enzyme HMG-CoA reductase, catalyzing a rate-limiting step in isoprenoid biosynthesis. Moreover, all genes regulated by sterols contain a sterol-responsive element within the 5′-flanking region. Through the binding of SREBPs, sterols are able to control the transcription of genes (30). In a positive feedback scenario, increased cholesterol production is dependent on an altered expression level of many genes encoding enzymes in the isoprenoid biosynthesis pathway, which begins with the condensation of acetoacetyl-CoA with acetyl-CoA to produce mevalonate in a set of reactions, including the rate-limiting step catalyzed by HMG-CoA reductase. Several consequent reactions are required to finally produce cholesterol via either the Kandutsch-Russell pathway with 7-dehydrocholesterol as the direct precursor or the Bloch synthesis pathway with desmosterol as the ultimate precursor of cholesterol (26).
Increased cholesterol production through mere overexpression of DHCR24/seladin-1 is limited by the amount of sterol precursors entering the Bloch or Kandutsch pathway and therefore is likely to be dependent on the basic activity of cholesterol biosynthesis and thus the precursor molecules present in the respective cell line. This might explain why DHCR24/seladin-1 overexpression in certain cell lines leads to elevated levels of cellular and membrane cholesterol (6), while it fails to increase cholesterol concentrations in others (9).
We show that in addition to DHCR24/seladin-1, HMG-CoA reductase and SREBP1 expression also significantly increased upon acute H2O2 treatment, resulting in increased concentrations of membrane cholesterol. These data suggest that elevation of DHCR24/seladin-1 is part of a general up-regulation of the isoprenoid/cholesterol pathway and consecutively increased cholesterol synthesis, potentially counteracting lipid peroxidation associated with oxidative stress and thus maintaining plasma membrane integrity. Moreover, cholesterol seems to mediate the observed protective effect, as determined by cholesterol upload to naïve SH-SY5Y cells and by overexpressing N294T/K306N-DHCR24/seladin-1 lacking the DHCR24 activity, which did not protect cells from apoptotic insult.
It was shown that lipid rafts mediate inhibition of apoptotic cell death (29). In our study, elevated membrane cholesterol levels induced by overexpression of DHCR24/seladin-1 or by cholesterol upload experiments were associated with increased phosphorylation of Akt and ERK, two prosurvival signaling factors that are known to be phosphorylated in several cell types upon exposure to a variety of proapoptotic stimuli (17, 24). Our findings are consistent with previous studies demonstrating that cholesterol depletion attenuated Akt and ERK phosphorylation and enhanced H2O2-induced apoptosis in cells with reduced membrane cholesterol concentrations, suggesting that plasma membrane compartments rich in cholesterol participate in signal transduction pathways activated by oxidative stress (13, 29). In our study, enhanced phosphorylation of ERK and Akt paralleled the elevation of cholesterol concentrations in response to oxidative stress, possibly indicating a link between these prosurvival signals and de novo cholesterol biosynthesis. However, the investigation of the molecular mechanisms and the signaling consequences of these findings go beyond the scope of this study and need further examination in future experiments.
A possible mechanism of how DHCR24/seladin-1 exerts its protective effect was proposed by Greeve et al., who suggested that seladin-1 is a death substrate for caspases, since distinct cleavage products of DHCR24/seladin-1 were obtained after growth factor deprivation (11). However, we did not detect any cleavage products of endogenous or overexpressed DHCR24/seladin-1 after H2O2 treatment using N- and C-terminal anti-DHCR24/seladin-1 antibodies. In addition, overexpression of DHCR24/seladin-1 carrying a mutation at a predicted caspase cleavage site (D122E) prevented H2O2-induced apoptosis of SH-SY5Y cells to a similar extent as wild-type DHCR24/seladin-1 (data not shown). This discrepancy might be due to the different cytotoxic insult used in our study and the ensuing apoptosis regulatory pathway.
DHCR24/seladin-1 levels were shown to naturally vary also between different cell types or clones thereof. For example, DHCR24/seladin-1 expression levels differ considerably between differentiated and nondifferentiated hMSC (3) and between melanoma cells derived from a cutaneous metastasis and the autologous primary tumor (9). To determine whether the expression of DHCR24/seladin-1 correlates with the expression of other cholesterol-related genes and the respective cellular cholesterol concentration, we utilized the above-mentioned cell lines and found significant positive correlations between expression levels of DHCR24/seladin-1 and HMG-CoA reductase, as well as DHCR24/seladin-1 and SREBP1, independent of their absolute expression levels. We also found a positive correlation between high DHCR24/seladin-1 levels and increased cholesterol concentrations in the investigated cell lines, a finding that is in accordance with recently published results (3, 9), suggesting that the expression level of DHCR24/seladin-1 is closely associated with the general activity of cholesterol biosynthesis.
In contrast to the substantial up-regulation of DHCR24/seladin-1 as an acute response to H2O2-induced oxidative stress, chronic H2O2 treatment significantly decreased DHCR24/seladin-1 protein levels. Intriguingly, low levels of endogenous DHCR24/seladin-1 after long-term treatment with H2O2 were associated with markedly reduced p53 levels in these cells. Our data are in line with previous results showing that p53 production is rapidly increased in neurons in response to oxidative stress (7) and is reduced again after long-term exposure to other stressors, including H2O2 (19). The observation that overexpression of wild-type DHCR24/seladin-1 resulted in stabilized p53 levels confirmed the hypothesis of a dynamic interaction between these two proteins. However, abrogation of DHCR24 activity did not alter preserved p53 levels in DHCR24/seladin-1-overexpressing cultures, indicating that the relation between p53 and DHCR24/seladin-1 is independent of its DHCR24 activity, as proposed earlier (28), and thus unrelated to membrane cholesterol concentrations.
Apparently, the protection mechanism mediated by high DHCR24/seladin-1 levels, i.e., high cholesterol levels, has to be of overriding importance. Although a high abundance of DHCR24/seladin-1 resulted in increased p53 stabilization, cells with high DHCR24/seladin-1 levels were more resistant to oxidative stress. This might be explained by the nature of the subsequent prosurvival machinery that is initiated by elevated membrane cholesterol concentrations. Detailed analyses on the molecular mechanism of the two counteracting strategies are ongoing.
To determine the effect of DHCR24/seladin-1 ablation on neuronal survival upon exposure to oxidative stress, we analyzed primary neuronal cultures derived from mice deficient in both dhcr24 alleles (27). DHCR24/seladin-1 deficiency leads to a complete loss of cholesterol in brains of these mice, while desmosterol, the direct precursor of cholesterol, accumulates and accounts for over 99% of total sterols in brains of DHCR24 knockout mice (6, 20). In addition to the increased concentrations of endogenous desmosterol, primary neurons grown in serum-containing medium take up sufficient amounts of cholesterol from the medium and therefore exhibit no alterations in membrane organization and integrity (our unpublished observations). These observations are in line with recently published data showing that cholesterol supplementation rescues membrane-related deficits in DHCR24−/− mouse embryonic fibroblasts (14). In our neuronal cultures, ablation of DHCR24/seladin-1 by genetic means was clearly associated with reduced p53 levels, higher resistance to oxidative stress, and diminished cleaved caspase-3 protein abundance after apoptotic stimuli. These data indicate that diminished DHCR24/seladin-1 levels provide deficient stabilization of p53 and therefore protect neurons from p53-mediated apoptotic cell death. This assumption was substantiated in SH-SY5Y cell experiments, where DHCR24/seladin-1 depletion by means of siRNA gene targeting clearly affected p53 status and function upon H2O2 exposure. Although the p53 pathway was activated in DHCR24/seladin-1-depleted cells upon H2O2 exposure, gene expression of p53 target genes such as those for p21, Puma, and Noxa was lower in DHCR24/seladin-1-depleted cells than in controls. In addition, protein level of p53 targets such as MDM2, Bax, p21, and Puma-β were reduced after H2O2 exposure in DHCR24/seladin-1-depleted cells. Furthermore, p53 turnover was evidently influenced by DHCR24/seladin-1 levels, as revealed by a higher degree of p53 ubiquitination upon DHCR24/seladin-1 depletion. Interestingly, p53 protein levels were reduced, while gene expression in DHCR24/seladin-1-depleted cells was increased, after long-term exposure to oxidative stress, which is likely to be part of a compensating mechanism due to increased p53 ubiquitin-mediated degradation.
However, it is still a matter of debate whether DHCR24 and seladin-1 are two truly independent functions of the same protein or whether they share certain regulatory mechanisms; attempts to define the factors responsible for regulating DHCR24/seladin-1 expression in response to oxidative stress are ongoing.
Taken together, the results of the present study demonstrate that DHCR24/seladin-1 plays a dual role in neuroprotection. While the acute protective effect of high DHCR24/seladin-1 levels seems to be related to cholesterol production, the prosurvival impact of low protein levels appears to be mediated by the tumor suppressor protein p53.
We thank E. Feinstein (Quark Biotech Inc.) for providing the DHCR24-deficient mice.
This work was supported by grants from the Swiss National Science Foundation (3200BO-112616 and 451NF40-111381) (NCCR Neuro), the University of Zurich, the Swiss Academy of Medical Sciences, and the Hermann Klaus, Hartmann Müller, and Novartis Foundations to M.H.M.; by grants from the National Institutes of Health (NINDS R01 NS046006) to F.L.H.; and by EU grants LSHM-CT-2003-503330 (APOPIS) and DFGSFB6027 to R.M.N.
Published ahead of print on 5 November 2007.