Mitochondria Profoundly Influence Apolipoprotein E Biology

Abstract

Background:

Mitochondria can trigger Alzheimer’s disease (AD)-associated molecular phenomena, but how mitochondria impact apolipoprotein E (APOE; apoE) is not well known.

Objective:

Consider whether and how mitochondrial biology influences APOE and apoE biology.

Methods:

We measured APOE expression in human SH-SY5Y neuronal cells with different forms of mitochondrial dysfunction including total, chronic mitochondrial DNA (mtDNA) depletion (ρ0 cells); acute, partial mtDNA depletion; and toxin-induced mitochondrial dysfunction. We further assessed intracellular and secreted apoE protein levels in the ρ0 cells and interrogated the impact of transcription factors and stress signaling pathways known to influence APOE expression.

Results:

SH-SY5Y ρ0 cells exhibited a 65-fold increase in APOE mRNA, an 8-fold increase in secreted apoE protein, and increased intracellular apoE protein. Other models of primary mitochondrial dysfunction including partial mtDNA-depletion, toxin-induced respiratory chain inhibition, and chemical-induced manipulations of the mitochondrial membrane potential similarly increased SH-SY5Y cell APOE mRNA. We explored potential mediators and found in the ρ0 cells knock-down of the C/EBPα and NFE2L2 (Nrf2) transcription factors reduced APOE mRNA. The activity of two mitogen-activated protein kinases, JNK and ERK, also strongly influenced ρ0 cell APOE mRNA levels.

Conclusion:

Primary mitochondrial dysfunction either directly or indirectly activates APOE expression in a neuronal cell model by altering transcription factors and stress signaling pathways. These studies demonstrate mitochondrial biology can influence the biology of the APOE gene and apoE protein, which are implicated in AD.

Keywords

Alzheimer’s disease apolipoprotein E mitochondria mitochondrial DNA

INTRODUCTION

APOE single nucleotide variants define the common APOE2, 3, and 4 isoforms [1]. APOE4 associates with increased, and APOE2 decreased, Alzheimer’s disease (AD) lifetime risk [2]. The underlying mechanisms are unclear. In experimental models the APOE4-encoded peptide, apolipoprotein (apoE) 4, promotes amyloid plaque deposition and neurofibrillary tangle formation [3 –5]. This is consistent with, at least at face value, the amyloid cascade hypothesis premise that amyloid-β (Aβ) accumulation initiates the disease and drives progression through processes that may require tau protein aggregation [6]. ApoE protein, though, affects cells in ways not immediately relevant to Aβ including lipid metabolism, glycolysis, cell cycling, and synapse maintenance [1 , 7–10]. Also, cleavage of the peptide’s C-terminal 27 amino acids, which preferentially happens when folded in an apoE4-associated pattern, creates a mitochondrial targeting sequence within the remaining 272 amino acid peptide [11 –13]. This product physically associates with mitochondria and impairs mitochondrial function.

Regarding the question of APOE-mitochondria connections in AD, we and others report differences between mitochondria from APOE4 carrier and non-carrier subjects [14 –18]. Altered features are structural and functional, and they occur within and outside the brain. It seems reasonable to consider that apoE protein in general, and especially apoE4, may at least partly account for AD brain and systemic mitochondrial defects and that mitochondria might mediate apoE toxicity.

While apoE appears to affect mitochondria, few have considered the reverse possibility, that mitochondria might affect APOE and apoE biology. We believe this is important to consider given data from some models, including cytoplasmic hybrid (cybrid) cell lines, argue factors beyond apoE contribute to AD mitochondrial dysfunction [19, 20]. Here, we considered whether states of primary mitochondrial dysfunction affect APOE and apoE biology. We found mitochondria profoundly influence APOE transcription and identified mechanisms that potentially contribute to this.

MATERIALS AND METHODS

ρ+, ρ0, and cybrid cell culture

SH-SY5Y cells with (ρ+) or without (ρ0) mitochondrial DNA (mtDNA) were maintained in high-glucose, L-glutamine Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 100μg/mL sodium pyruvate, 50μg/mL uridine, and 1% penicillin-streptomycin. The ρ0 cells were generated through chronic ethidium bromide exposure [21]. Upon achieving complete mtDNA depletion ethidium bromide was no longer added to the medium and reversion to ρ+ status did not occur.

Cytoplasmic hybrid (cybrid) cells were generated by fusing SH-SY5Y ρ0 cells with platelets obtained from human subjects enrolled in the University of Kansas Alzheimer’s Disease Research Center (KU ADRC) as previously described [22]. Human subject participation was approved by the Kansas University Medical Center’s Institutional Review Board, and all human subjects provided informed consent. After removing untransformed cells during a selection period in which the resultant cultures were kept in medium containing heat-inactivated FBS, no pyruvate, and no uridine the mtDNA-restored cybrid lines were placed back into the same 10% fetal bovine serum (FBS), 100μg/mL sodium pyruvate, 50μg/mL uridine-supplemented medium used to grow the ρ+ and ρ0 cells.

Our SH-SY5Y ρ+, ρ0, and cybrid cell lines were seeded at 25–50% density and maintained side-by-side in a humidified incubator at 37°C and 5% CO₂. Cells were harvested at ∼80% density for subsequent analyses.

Identification and interrogation of candidate transcription factors and signaling pathways

Transcription factors and cell signaling pathways interrogated in this study were identified through a literature search. Our selections were reinforced by predictive data supplied by bioinformatics tools and databases, including Ensembl, NCBI Gene, GeneCards, UCSC Genome Browser, AliBaba 2.1, and TRANSFAC.

We used small interfering RNA (siRNA) transfection to assess the role of candidate transcription factors. All siRNAs including non-targeting siRNA controls were purchased from Dharmacon (ON-TARGETplus, SMARTPool). Lipofectamine RNAiMAX for use as a transfection reagent was purchased from Thermofisher. Transfection components were incubated in Opti-MEM I Reduced Serum Media before mixing into penicillin/streptomycin-free DMEM, supplemented with pyruvate, uridine, and 10% FBS. Cells were incubated in transfection medium for two days before harvesting. Total siRNA concentration in the transfection mixtures ranged from 25–50 nM.

To manipulate the JNK and ERK candidate signaling pathways we used anisomycin, a JNK pathway activator and U0126, an ERK pathway inhibitor (Cell Signaling Technologies). Anisomycin was used at a final concentration of 50 ng/mL. U0126 was used at a final concentration of 20μM. The treatment duration for each compound was 24 h. The amount of dimethyl sulfoxide (DMSO) did not exceed 1μL/mL for either the experimental or control conditions.

Western blot immunochemistry, fluorescence immunocytochemistry, and ELISA

To generate whole cell protein lysates, cells were washed with ice-cold PBS and placed in RIPA buffer supplemented with Halt Protease and Phosphatase Inhibitor Cocktail (ThermoFisher). The resulting suspensions were vigorously vortexed, incubated on ice for 15-min, re-vortexed, centrifuged at 14,000 rcf for 15 min at 4°C, and the supernatants collected. For cytosolic and nuclear-specific protein lysates, samples were prepared with NE-PER Nuclear and Cytoplasmic Extraction Reagents (ThermoFisher, 78833) according to the kit protocol.

For western blot immunochemistry, protein concentrations were measured using a bicinchoninic acid (BCA) assay. Samples were diluted to allow for the addition of 30–50μg of protein per well and boiled with 1X Laemmli-SDS buffer. Samples were run alongside a prestained protein ladder (10 to 180 kDa, ThermoFisher, 26616). Following gel electrophoresis, we performed wet transfers to polyvinylidene difluoride (PVDF) membranes, which were blocked for 1 h with bovine serum albumin (BSA) and maintained overnight in primary antibody solution at 4°C. Antibodies against apoE were obtained from Abcam. Antibodies against phospho-c-Jun (Ser73), phospho-SAPK/JNK (Thr183/Tyr185), total SAPK/JNK, phospho-p44/42 MAPK (ERK1/2) (Thr202/Tyr204), and total p44/42 MAPK (ERK1/2) were obtained from Cell Signaling Technologies. Table 1 lists the antibody catalogue numbers, dilutions, and the secondary antibodies that were used. Uncropped western blots shown in our data figures are included in the Supplementary Material.

Table 1

Antibodies used for western blot immunochemistry

Antibody Name/Target	Dilution	Source	Catalog #
ApoE (C-terminus)	1:1000	Abcam	ab52607
Goat anti-Rabbit IgG (Heavy Chain), Superclonal Recombinant Secondary Antibody, HRP	1:3500	Invitrogen	A27036
Phospho-p44/42 MAPK (ERK1/2) (Thr202/Tyr204)	1:1000	Cell Signaling Technology	9101
p44/42 MAPK (ERK1/2)	1:1000	Cell Signaling Technology	9102
Phospho-SAPK/JNK (Thr183/Tyr185) (81E11)	1:1000	Cell Signaling Technology	4669
SAPK/JNK	1:1000	Cell Signaling Technology	9252
Phospho-c-Jun (Ser73) (D47G9) XP^®	1:1000	Cell Signaling Technology	3270
β-Tubulin	1:2000	Cell Signaling Technology	2146
Lamin B1	1:1000	Cell Signaling Technology	12586

For fluorescence immunocytochemistry cells were seeded in 8-well chamber slides and grown to 70–80% confluency in a humidified 37°C incubator. Cells were fixed with 4% paraformaldehyde and permeabilized with 0.04% Triton X-100. Cells were blocked in 10% normal donkey serum/10% BSA/0.3 M glycine/0.03% Triton X-100 in PBS for 1 h at room temperature. Slides were incubated with an N-terminal apoE antibody (Abcam, ab51015, 1:50) overnight at 4°C in a humidified chamber. After washing, slides were incubated in Alexa Fluor 488 Donkey anti-rabbit IgG (H+L) secondary antibody (Invitrogen, A21206, 1:300) at room temperature for 1.5 h. Nuclei were stained with DAPI for 5 min. Staining procedures with fluorescent reagents (Alexa Fluor and DAPI) were shielded from light. After washing, slides were dried and mounted in ProLong Gold Antifade reagent. All slides were viewed and quantified with identical laser settings. Fluorescence intensity was measured in each cell by sampling the cytosol in two places using ImageJ software. Intensity of cytosolic fluorescence was quantified in at least 60 cells each for both ρ+ and ρ0 cells, across eight total images. A human apoE ELISA kit was obtained from Abcam (ab108813). Conditioned medium from ρ+ and ρ0 cell cultures was collected and analyzed using this kit according to the manufacturer’s instructions.

RNA extraction, cDNA preparation, and real-time PCR

Cells were washed with ice-cold PBS and RNA from cell cultures was isolated using TRIzol reagent, chloroform, and isopropanol extraction. cDNA was prepared from 1μg input RNA with BIO-RAD iScript Reverse Transcription Supermix (1708841) for reverse transcription according to their protocol. All quantitative PCR reactions were prepared with ThermoFisher TaqMan Universal Master Mix II with UNG (4440038) and gene specific ThermoFisher Taqman primers (Table 2). Real-time PCR was performed using a StepOne Real-Time PCR System (96 well) or QuantStudio 5 Real-Time PCR System (384 well) and their respective software. Relative gene expression was determined using the 2^–ΔΔCt method. Each sample was tested in duplicate or greater, and the reported data reflects the average of the replicates.

Table 2

Taqman primers used for quantitative PCR

Primer Gene Target	Assay ID #
APOE	Hs00171168_m1
ACTB	Hs99999903_m1
CEBPA	Hs00269972_s1
CEBPB	Hs00270923_s1
NR1H2 (LXRB)	Hs01027215_g1
NFE2L2	Hs00975961_g1
NFKB1	Hs00765730_m1
PPARG	Hs01115513_m1
RXRA	Hs01067640_m1
18S	Hs03003631_g1

Toxin-induced models of acute mitochondrial dysfunction

All mitochondrial toxins were diluted from stock solutions prepared in DMSO. SH-SY5Y ρ+ cells were incubated in 0.5μM rotenone, 2.5μM antimycin A, or 5μM oligomycin for 24 h in 12-well or 6-well plates. SH-SY5Y ρ+ cells incubated in 20μM FCCP were incubated for 4 h. Control cells were incubated for the same duration with a commensurate volume of DMSO. The amount of DMSO in the rotenone, antimycin A, and oligomycin treatment condition was kept to no more than 0.5μL/mL, and to 6.5μL/mL for the FCCP condition. Cells were harvested in TRIzol followed by RNA extraction.

To determine cell viability in our toxin-treated cell cultures, cells were seeded in 96 well plates and treated with the toxins noted above. Each treatment condition included eight replicates. Following toxin exposure, cells were washed thoroughly in Hanks Balanced Salt Solution (HBSS) and placed in HBSS containing 5μg/mL propidium iodide (PI) (Sigma-Aldrich, P4170). Cells were incubated for 30 min at 37°C in a humidified incubator. Cells were again thoroughly washed in HBSS, followed by a 10 min incubation in HBSS containing Hoechst dye (ThermoFisher, 62249). Plates were analyzed using a Cytation 1 cell-imaging multi-mode microplate reader (BioTek) with Gen5 software. Total cell counts were determined from Hoechst staining and estimated by the software. Apoptotic, propidium iodide-positive cells were counted manually with ImageJ software. Wells with fewer than 1000 detected cells were excluded, for a minimum of 5 replicates in each treatment condition.

To induce acute, partial mtDNA-depletion we placed ρ+ cells in ethidium bromide as previously described [23]. We analyzed the ethidium bromide-treated cells following 1, 3, and 7 days of treatment.

Statistics

Data analysis and graph design were performed with Microsoft Excel and GraphPad Prism statistical software. Unless otherwise noted, all comparisons were performed by unpaired, two-way Student’s t-test. Continuous values are portrayed as means±SEM.

RESULTS

Neurons do not typically express APOE mRNA at appreciable levels [24]. Consistent with their classification as a human neuronal cell line SH-SY5Y ρ+ cells contained very little APOE mRNA. SH-SY5Y ρ0 cells, though, showed a relative 65-fold increase in APOE mRNA (Fig. 1A).

Fig. 1

Chronic mtDNA depletion increases APOE mRNA and protein. A) Chronic mtDNA depletion in SH-SY5Y cells (ρ0) increased APOE mRNA by ∼65-fold compared to controls (ρ+). B) ELISA of the cell culture media revealed extracellular apoE secretion increased ∼8-fold in the ρ0 group compared to controls (pg apoE/μg total protein). C) Representative whole-cell lysate western blots. Densitometry revealed increased intracellular apoE in the ρ0 group. D) Immunocytochemistry (ICC) reveals significantly increased apoE fluorescence in ρ0 cells compared to controls. Scale bar is set to 20μM. E) Restoration of mtDNA returns APOE mRNA to the baseline level. ns = not significant, ^*p≤0.05, ^****p≤0.0001. Error bars represent SEM.

ApoE protein is normally secreted by the cells that produce it [25]. Consistent with the mRNA data, conditioned medium from the ρ0 cells had approximately 8-fold more apoE protein than conditioned medium from the ρ+ cells (Fig. 1B). Thus, although ρ0 cells secreted more apoE protein this increase was not proportional to their increase in APOE mRNA. We accordingly used immunochemistry and immunocytochemistry to assess the amount of ρ0 cell intracellular apoE protein and found it was increased (Fig. 1C, D). To ensure differences in ρ0 cell APOE expression were specifically due to mtDNA depletion and not to an unanticipated defect acquired during their exposure to ethidium bromide, we transferred mtDNA back into the ρ0 cells to generate cybrid cell lines. Restoring mtDNA and respiratory chain competence lowered the APOE mRNA copy number to the ρ+ level (Fig. 1E).

We previously reported exposing SH-SY5Y ρ+ cells to 2μg/ml ethidium bromide for 1, 3, and 7 days reduced their mtDNA content by approximately 50, 75, and 95% [23]. We measured APOE mRNA and intracellular protein levels in cells treated with 2μg/ml ethidium bromide for those durations and found the 3 and 7-day exposures induced an approximate 4-fold increase in APOE mRNA (Fig. 2A). Partial, acute mtDNA depletion, therefore, boosts APOE gene expression, although not to the same extent as complete, chronic mtDNA depletion. The 3 and 7-day treatments also increased intracellular apoE protein levels (Fig. 2B).

Fig. 2

Acute mtDNA depletion and mitochondrial toxins increases APOE expression. A) Ethidium bromide-treated SH-SY5Y cells were incubated alongside controls for 1, 3, or 7-days. APOE mRNA levels increase within three days of the mtDNA depletion process. B) Representative whole cell-lysate western blot depicting apoE protein in the setting of acute mtDNA depletion. C) Effect of focal respiratory chain inhibition and mitochondrial membrane potential perturbation on APOE mRNA in SH-SY5Y ρ+ cells. Values are expressed as a percent relative to untreated controls. Significant increases were observed in each condition. D) Cell death following acute mitochondrial challenge in ρ+ cells, as measured using propidium iodide and Hoechst co-staining. A significant increase in cell death was observed only in the oligomycin condition. One-way ANOVA with multiple comparisons test. ns, not significant; ^*p≤0.05, ^**p≤0.01, ^***p≤0.001, ^****p≤0.0001. Error bars depict SEM.

We additionally assessed the impact of several mitochondrial toxins on SH-SY5Y ρ+ cell APOE mRNA. The complex I inhibitor rotenone, complex III inhibitor antimycin A, complex V inhibitor oligomycin, and mitochondrial membrane protonophore FCCP each increased APOE mRNA levels (Fig. 2C). To consider the possibility that an induction of cell death, as opposed to mitochondrial dysfunction, was primarily responsible for this we assessed cell viability and found that only the oligomycin condition significantly increased the proportion of dead cells, but even in that case the proportion of dead cells remained below 7% of the total cell number. This suggests different types of primary mitochondrial dysfunction trigger APOE gene expression.

We evaluated the status of transcription factors or components of transcription factors that reportedly facilitate APOE expression (Fig. 3). mRNA levels of CEBPA and the NFKB1 component NF-κB1 were comparable between the ρ+ and ρ0 cells. mRNA levels for CEBPB, LXRB, NFE2L2, and PPARG were higher in the ρ0 cells, while RXRA mRNA was lower. To further interrogate a potential role for these transcription factors in mediating increased SH-SY5Y ρ0 cell apoE, we used siRNA to knock down their expression and determined the effect of each knock-down on the APOE mRNA level. Figure 4A shows the extent of knock-down achieved for each transcription factor. ρ0 cell APOE mRNA levels decreased with C/EBPα and NFE2L2 knock-down and increased with NF-κB1 and PPARγ knock-down (Fig. 4B).

Fig. 3

mRNA levels of candidate genes linked to APOE gene expression. A) No change was observed in the CEBPA level. B-D) CEBPB, LXRB, and NFE2L2 mRNA levels were increased in the ρ0 cells. E) NFKB1 (NF-κB subunit 1) mRNA levels were unchanged. F) PPARG levels were increased in the ρ0 cells. G) RXRA mRNA levels were significantly decreased in the ρ0 cells. ns, not significant; ^*p≤0.05, ^**p≤0.01, ^****p≤0.0001. Error bars depict SEM.

Fig. 4

Impact of transcription factor knock-down on APOE gene expression. A) Efficiency of candidate transcription factor knockdown through siRNA transfections expressed as a percent relative to the corresponding non-targeting siRNA control. The dashed line indicates the baseline established from the non-targeting controls. B) APOE mRNA levels in the setting of candidate gene knockdown. Values are expressed as a percent relative to the corresponding non-targeting siRNA control. Knockdown of NFE2L2 and C/EBPα significantly decreased APOE mRNA while knockdown of NF-κB1 and PPARγ significantly increased APOE mRNA. ^*p≤0.05, ^**p≤0.01, ^****p≤0.0001. Error bars depict SEM.

Gafencu et al. previously reported phosphorylated c-Jun levels inversely correlate with macrophage APOE expression [26]. As mitogen-activated protein kinases (MAPKs) drive c-Jun phosphorylation [27], we assessed the ERK MAPK pathway. ERK pathway activity in ρ0 cells was markedly reduced. Although the ρ0 cells contained more total ERK1/2 than the ρ+ cells, very little of it was phosphorylated (Fig. 5A). The ρ+ cells, on the other hand, showed extensive ERK1/2 phosphorylation. Relative to the ρ+ cells, the ρ0 cells exhibited a 99.7% reduction in their nuclear lysate p-ERK/total ERK ratio and 97.9% reduction in their cytosol p-ERK/total ERK ratio (Fig. 5B). To directly interrogate a potential relationship between ERK signaling and APOE expression, we treated both ρ+ and ρ0 cells with U0126, which inhibits the MEK1/2 enzymes that phosphorylate ERK. U0126 reduced ERK1/2 phosphorylation in both cell types (Fig. 5C). U0126 reduced c-Jun phosphorylation in the ρ+ cells, and there was an apparent trend towards less phosphorylated c-Jun in the ρ0 cells (p = 0.066) (Fig. 5C). U0126 increased APOE mRNA levels in both cell types (Fig. 5D).

Fig. 5

mtDNA depletion alters APOE gene expression through an ERK/c-Jun sensitive mechanism. A) Representative cytosolic and nuclear lysate western blots demonstrating phosphorylated ERK (pERK) and total ERK 1/2 protein levels in SH-SY5Y ρ+ and ρ0 cells. mtDNA-depleted cells showed decreased pERK in both cytosolic and nuclear compartments, while total ERK levels increased. B) The pERK 1/2 to total ERK 1/2 ratio is markedly lower in the ρ0 cells. C) Representative whole-cell lysate western blots demonstrating the effect of the selective MEK 1 and 2 inhibitor, U0126, on levels of phosphorylated c-Jun, pERK, and total ERK 1/2 in ρ+ and ρ0 cells. U0126 incubation decreased ρ+ cell phospho-c-Jun and pERK1/2, while total ERK1/2 increased. The ρ0 cell phospho-c-Jun level trended lower. D) U0126 increased APOE mRNA levels in ρ+ and ρ0 cells, expressed as a fold change relative to the respective ρ+ and ρ0 controls. ^*p≤0.05, ^**p≤0.01, ^****p≤0.0001. Error bars depict SEM.

We also evaluated the SAPK/JNK MAPK pathway. As was the case for the ERK pathway, JNK pathway signaling was reduced. JNK phosphorylation was lower in ρ0 cells (Fig. 6A), the cytosolic p-JNK/total JNK ratio was reduced by 88.2%, and the nuclear p-JNK/total JNK ratio was reduced by 88.7% (Fig. 6B). We treated ρ0 and ρ+ cells with anisomycin, which enhances JNK signaling [28], and this increased c-Jun phosphorylation in both cell types and JNK phosphorylation in the ρ0 cells (Fig. 6C). APOE mRNA levels were markedly lower in the anisomycin-treated cells than they were in the anisomycin-untreated cells. In the ρ+ cells the amount of APOE mRNA decreased by 45.2%, and in the ρ0 cells the amount of APOE mRNA decreased by 73.8% (Fig. 6D).

Fig. 6

mtDNA depletion alters APOE gene expression through a SAPK/JNK sensitive mechanism. A) Representative western blots from cytosolic and nuclear-enriched lysates show phosphorylated JNK (pJNK) and total JNK protein levels in SH-SY5Y ρ+ and ρ0 cells. mtDNA-depleted cells contained less cytosolic and nuclear pJNK. Total JNK levels were unchanged. B) The pJNK to total JNK ratio is markedly lower in the ρ0 cells. C) Representative whole-cell lysate western blots demonstrating the effect of 50 ng/mL anisomycin on phosphorylated c-Jun, pJNK, and total JNK in ρ+ and ρ0 cells. Anisomycin increased ρ0 cell phospho-c-Jun, pJNK, and total JNK. D) Effect of anisomycin-mediated SAPK/JNK pathway activation on APOE mRNA levels in ρ0 and ρ+ SH-SY5Y cells. Anisomcyin decreased APOE mRNA levels in ρ+ and ρ0 cells, expressed as a fold change relative to the respective ρ+ and ρ0 controls. ^*p≤0.05, ^**p≤0.01, ^****p≤0.0001. Error bars depict SEM.

DISCUSSION

In human neuronal SH-SY5Y cells mtDNA depletion, respiratory chain inhibition, and mitochondrial membrane potential perturbation induce robust APOE gene expression. In the case of chronic mtDNA depletion this induction is mediated, at least in part, by mitochondrial dysfunction-driven changes in transcription factor and stress pathway activity. The APOE mRNA that is expressed secondary to a state of primary mitochondrial dysfunction is translated, and the resultant protein accumulates intracellularly (Fig. 7).

Fig. 7

Mitochondria influence APOE and apoE biology. Mitochondrial dysfunction caused by chronic mtDNA depletion or through other causes (acute mtDNA depletion, focal respiratory chain failure, altered mitochondrial membrane potential) increase APOE expression. Factors that mediate increased APOE expression in the case of chronic mtDNA depletion are shown; these factors may or may not act directly at the APOE promoter. Other mechanisms and mediators may also contribute, and factors that mediate increased APOE expression when other types or causes of mitochondrial dysfunction exist are not established. Increased APOE expression is accompanied by increased intracellular and secreted apoE protein, and as the literature reports apoE (or an apoE cleavage product) can impair mitochondrial function, it is worth emphasizing apoE generated because of mitochondrial dysfunction may in turn exacerbate mitochondrial dysfunction.

It is known that unstressed neurons exhibit limited APOE gene expression, while stressed neurons show substantially more expression [24, 25]. Our study finds a particular type of stress, mitochondrial stress, can initiate SH-SY5Y human neuronal cell APOE expression. We cannot say whether this phenomenon extends to other cell types, or the extent to which mechanisms that mediate mitochondrial stress induced APOE expression generalize to non-mitochondrial stress dependent APOE expression.

In some tissues the LXRβ and PPARγ transcription factors bind enhancer elements downstream of the APOE gene to induce its synthesis, although their role in neuron APOE expression is unclear [29, 30]. In this study neither appeared to drive ρ0 cell APOE expression. LXRβ and PPARγ heterodimerize with RXRα to activate gene expression when bound, respectively, to oxysterols and fatty acids [29]. We cannot explain why PPARγ but not RXRα knock-down unexpectedly increased APOE mRNA levels, but note this dissociation suggests PPARγ indirectly, and not directly, affects SH-SY5Y cell APOE expression.

Mitochondrial distress can trigger inflammatory responses [31, 32], but for unclear reasons knock-down of the NF-κB component NF-κB1 also unexpectedly increased APOE expression. In neurons C/EBPβ reportedly induces APOE expression under physiological and pathological conditions [33], but this transcription factor did not appear critical to the observed increase in ρ0 cell APOE mRNA.

The NFE2L2-encoded Nrf2 transcription factor regulates antioxidant responses [34], and previous studies indicate links between oxidative stress and APOE regulation in some cell types [35]. Increased NFE2L2 mRNA in SH-SY5Y ρ0 cells suggests these cells experience some degree or type of oxidative stress, and the observation that NFE2L2 knock-down reduced APOE mRNA argues mitochondrially-generated oxidative stress can enhance or initiate neuronal cell APOE expression. Our data do not resolve whether the Nrf2 transcription factor acts directly at the APOE promoter, or elsewhere in the APOE locus.

C/EBPα, which is recognized to play a role in cell cycle progression in some cell types and especially myeloid cells, and to interact with AP1 proteins [36, 37], also appears to contribute to mitochondrial dysfunction-induced APOE expression in SH-SY5Y ρ0 cells. This raises the possibility that in addition to oxidative stress neuron differentiation status, which is altered in AD [38 –40], may affect neuron APOE expression. Our data do not resolve whether the C/EBPα transcription factor acts directly at the APOE promoter, or elsewhere in the APOE locus.

The ERK pathway was previously implicated in APOE expression [41]. Our study extends this fundamental observation to the SH-SY5Y human neuronal cell line, indicates SH-SY5Y cell mtDNA depletion reduces ERK signaling, and demonstrates ERK pathway suppression subsequently increases APOE expression. The JNK pathway also reportedly attenuates macrophage APOE expression [26]. Our data further establish chronic mtDNA depletion downregulates SH-SY5Y neuronal cell SAPK/JNK pathway signaling, and that this subsequently increases APOE expression.

A terminal consequence of the mitogen-activated ERK and JNK stress pathways is c-Jun phosphorylation [42 –44]. c-Jun phosphorylation enables its heterodimerization to Fos, which generates the AP1 transcription factor [27]. It is possible c-Jun in its unphosphorylated state may act directly at the APOE gene to inhibit its transcription, although we are unaware of any such precedent. Since AP1 typically activates gene transcription, it is difficult to see how a loss of AP1 function might act directly at the APOE gene to activate its expression. A more likely scenario, we believe, is that AP1 initiates the transcription of another gene or genes which in turn go on to inhibit APOE expression. According to this paradigm, when AP1 formation declines because less ERK and JNK MAPK-induced c-Jun phosphorylation occurs, the repressor gene or genes inactivate and no longer repress APOE expression.

Data from others provide an increasingly sophisticated understanding of APOE expression regulation. This includes evidence of interactions between the APOE promoter and regional enhancers [45, 46], and regulation through RNA. Interestingly, Watts et al. recently reported a long non-coding enhancer RNA referred to as the APOE-associated noncoding RNA (AANCR) anneals to the intergenic region between the APOE promoter and the TOMM40 gene [47]. AANCR is partially transcribed in cells that do not normally express APOE and fully transcribed in cells that do express APOE. Future studies should address whether mtDNA depletion or mitochondrial dysfunction upregulate APOE expression through AANCR.

Oxidative stress activates the Nrf2 transcription factor and a variety of stress responses [34]. We are unsure why JNK and ERK pathway activity are reduced in ρ0 cells, especially given the observed increase in NFE2L2 expression. By way of speculation, we note functional relationships between mitochondria and the ERK and JNK pathways include physical interactions [48, 49]. It is not known whether the mitochondrial defects present in ρ0 cells negate or limit those interactions. Also, the chronic nature of the ρ0 defect is perhaps relevant, as we have previously reported substantial cell remodeling occurs as cells transition from states of acute to chronic mtDNA depletion [23]. Finally, it is possible chronic NFE2L2 overexpression creates an environment in which the JNK and ERK pathways are not perpetually maintained in an activated state.

The p38 pathway, like the JNK and ERK pathways, is a MAPK pathway [50]. We did not interrogate the p38 pathway as it did not appear to influence APOE expression in a previous study, while the JNK and ERK pathways did [26]. That study, though, was performed in macrophages and its findings may not extrapolate to neuronal cells. The relevance of the p38 pathway in our models, therefore, is unknown.

It is necessary to emphasize that, as is the case with most human disease-directed, mechanistic-oriented work, our study utilized a model. In this case we used the human SH-SY5Y neuronal cell line, which specifically included the standard ρ+ cell line and a ρ0 line originating from it. We chose this model because it is a widely used neuronal cell model [51]. That these cells are human-derived, we feel, is also relevant because in human cells APOE gene regulation plays out within the context of its broader chromosomal locus through potentially species-specific enhancer-promoter relationships [45 , 52]. In addition, SH-SY5Y cells are easily manipulated, which facilitates mechanistic studies. Perhaps most critically, the SH-SY5Y ρ0 cells constitute an unequivocal model of primary, chronic mitochondrial dysfunction; to ensure the increase in SH-SY5Y ρ0 cell APOE expression was not an artifact of the mtDNA-depletion process we generated cybrid cell lines and showed restoration of mtDNA returned APOE mRNA to the ρ+ cell level.

The SH-SY5Y line is of course a tumor cell line, which begs the question of how rigorously these neuronal cells recapitulate actual neurons. It is possible to differentiate dividing SH-SY5Y ρ+ neuronal cells into non-dividing neuronal cells [51], but SH-SY5Y ρ0 cells are difficult to differentiate. Regardless of questions over the fidelity of the model, it is hard to argue we did not show that a state of primary mitochondrial dysfunction can initiate APOE gene expression in a cell type that, like neurons, typically expresses very little APOE.

It is worth considering the relevance of modelling mitochondrial dysfunction to query AD biology-pertinent questions. To this end we note the etiology of AD is not settled and mitochondrial dysfunction is a well-established feature of this disease [53]. In AD patient-derived brain and non-brain tissues mitochondria show respiratory chain defects [19], and mitochondria from AD patients or in AD patient-derived living cells are relatively depolarized [54 –56]. Importantly, neurons from AD brains contain less mtDNA than neurons from non-AD brains [57 –61]. In AD patient-derived brains, reduced mtDNA copy number associates with increased amounts of the classic AD histopathologies [60, 61], and in the SH-SY5Y neuronal cell model reducing mtDNA copy number initiates tau oligomerization [23]. Some might counter mitochondrial defects in AD, including reduced mtDNA copy number, are simply a consequence of plaque and tangle pathology, or neurodegeneration, but it is not settled that these entities primarily cause AD mitochondrial dysfunction in AD patient brains, and it is especially challenging to attribute mitochondrial defects present outside the brain to plaques, tangles, and neurodegeneration.

Regardless, in our models relative to neuronal cells with normal mitochondria cells with primary mitochondrial defects contained increased amounts of APOE mRNA. Interestingly, a recent report found neurons in human AD autopsy brains, at least in the early stages of AD, also contain elevated APOE mRNA levels [62]. In our study SH-SY5Y ρ0 and partially mtDNA-depleted cells that increase their APOE mRNA also produce more apoE protein, which at least in the ρ0 cells accumulates intracellularly perhaps because they inefficiently secrete it. Intracellular ApoE protein, especially when encoded from an APOE4 allele, is reportedly toxic and can interfere with mitochondrial function itself [11 –13]. Primary mitochondrial dysfunction, therefore, could conceivably initiate a feedback loop with the APOE gene in which apoE protein harms neurons and perhaps even exacerbates mitochondrial dysfunction.

Footnotes

ACKNOWLEDGMENTS

The authors have no acknowledgments to report.

FUNDING

This study was supported by the University of Kansas Alzheimer’s Disease Center (P30 AG072973), R01 AG061194, the Thompson Foundation, the Dow Family Foundation, the Clune Family Foundation, and the Snyder Family Foundation.

CONFLICT OF INTEREST

The authors have no conflict of interest to report.

Russell H. Swerdlow is an Editorial Board Member of this journal but was not involved in the peer-review process nor had access to any information regarding its peer-review.

DATA AVAILABILITY

This manuscript does not contain shared data.

The supplementary material is available in the electronic version of this article: .

References

Mahley

, Rall

Jr (2000) Apolipoprotein E: Far more than a lipid transport protein. Annu Rev Genomics Hum Genet 1, 507–537.

Yamazaki

, Zhao

, Caulfield

, Liu

, Bu

(2019) Apolipoprotein E and Alzheimer disease: Pathobiology and targeting strategies. Nat Rev Neurol 15, 501–518.

Castellano

, Kim

, Stewart

, Jiang

, Demattos

, Patterson

, Fagan

, Morris

, Mawuenyega

, Cruchaga

, Goate

, Bales

, Paul

, Bateman

, Holtzman

(2011) Human apoE isoforms differentially regulate brain amyloid-β peptide clearance. Sci Transl Med 3, 89ra57.

Shi

, Yamada

, Liddelow

, Smith

, Zhao

, Luo

, Tsai

, Spina

, Grinberg

, Rojas

, Gallardo

, Wang

, Roh

, Robinson

, Finn

, Jiang

, Sullivan

, Baufeld

, Wood

, Sutphen

, McCue

, Xiong

, Del-Aguila

, Morris

, Cruchaga

, Fagan

, Miller

, Boxer

, Seeley

, Butovsky

, Barres

, Paul

, Holtzman

(2017) ApoE4 markedly exacerbates tau-mediated neurodegeneration in a mouse model of tauopathy. Nature 549, 523–527.

Lahiri

(2004) Apolipoprotein E as a target for developing new therapeutics for Alzheimer’s disease based on studies from protein, RNA, and regulatory region of the gene. J Mol Neurosci 23, 225–233.

Hardy

, Selkoe

(2002) The amyloid hypothesis of Alzheimer’s disease: Progress and problems on the road to therapeutics. Science 297, 353–356.

, Zhang

, Zhao

(2018) Human ApoE isoforms differentially modulate brain glucose and ketone body metabolism: Implications for Alzheimer’s disease risk reduction and early intervention. J Neurosci 38, 6665–6681.

, Deckelbaum

, Chen

, Vogel

, Talmage

(2001) Apolipoprotein E inhibits serum-stimulated cell proliferation and enhances serum-independent cell proliferation. J Biol Chem 276, 43455–43462.

Lanz

, Carter

, Merchant

(2003) Dendritic spine loss in the hippocampus of young PDAPP and Tg2576 mice and its prevention by the ApoE2 genotype. Neurobiol Dis 13, 246–253.

10.

Rueter

, Rimbach

, Huebbe

(2022) Functional diversity of apolipoprotein E: From subcellular localization to mitochondrial function. Cell Mol Life Sci 79, 499.

11.

Chang

, ran Ma

, Miranda

, Balestra

, Mahley

, Huang

(2005) Lipid- and receptor-binding regions of apolipoprotein E4 fragments act in concert to cause mitochondrial dysfunction and neurotoxicity. Proc Natl Acad Sci U S A 102, 18694–18699.

12.

Chen

, Ji

, Dodson

, Miranda

, Rosenblum

, Reynolds

, Freedman

, Weisgraber

, Huang

, Mahley

(2011) Apolipoprotein E4 domain interaction mediates detrimental effects on mitochondria and is a potential therapeutic target for Alzheimer disease. J Biol Chem 286, 5215–5221.

13.

Mahley

, Weisgraber

, Huang

(2006) Apolipoprotein E4: A causative factor and therapeutic target in neuropathology, including Alzheimer’s disease. Proc Natl Acad Sci U S A 103, 5644–5651.

14.

Valla

, Yaari

, Wolf

, Kusne

, Beach

, Roher

, Corneveaux

, Huentelman

, Caselli

, Reiman

(2010) Reduced posterior cingulate mitochondrial activity in expired young adult carriers of the APOE epsilon4 allele, the major late-onset Alzheimer’s susceptibility gene. J Alzheimers Dis 22, 307–313.

15.

Wilkins

, Koppel

, Bothwell

, Mahnken

, Burns

, Swerdlow

(2017) Platelet cytochrome oxidase and citrate synthase activities in APOE epsilon4 carrier and non-carrier Alzheimer’s disease patients. Redox Biol 12, 828–832.

16.

Wilkins

, Wang

, Menta

, Koppel

, Bothwell

, Becker

, Anderson

, Schwartz

, Pei

, Yellapu

, Chalise

, Gouvion

, Haeri

, Burns

, Swerdlow

(2021) Bioenergetic and inflammatory systemic phenotypes in Alzheimer’s disease APOE ɛ4-carriers. Aging Cell 20, e13356.

17.

Schmukler

, Solomon

, Simonovitch

, Goldshmit

, Wolfson

, Michaelson

, Pinkas-Kramarski

(2020) Altered mitochondrial dynamics and function in APOE4-expressing astrocytes. Cell Death Dis 11, 578.

18.

Yin

, Reiman

, Beach

, Serrano

, Sabbagh

, Nielsen

, Caselli

, Shi

(2020) Effect of ApoE isoforms on mitochondria in Alzheimer disease. Neurology 94, e2404–e2411.

19.

Swerdlow

(2012) Mitochondria and cell bioenergetics: Increasingly recognized components and a possible etiologic cause of Alzheimer’s disease. Antioxid Redox Signal 16, 1434–1455.

20.

Swerdlow

, Parks

, Cassarino

, Maguire

, Bennett

Jr , Davis

, Parker

Jr (1997) Cybrids in Alzheimer’s disease: A cellular model of the disease? Neurology 49, 918–925.

21.

Miller

, Trimmer

, Parker

Jr , Davis

(1996) Creation and characterization of mitochondrial DNA-depleted cell lines with “neuronal-like” properties. J Neurochem 67, 1897–1907.

22.

Swerdlow

, Parks

, Miller

, Tuttle

, Trimmer

, Sheehan

, Bennett

Jr , Davis

, Parker

Jr (1996) Origin and functional consequences of the complex I defect in Parkinson’s disease. Ann Neurol 40, 663–671.

23.

Weidling

, Wilkins

, Koppel

, Hutfles

, Wang

, Kalani

, Menta

, Ryan

, Perez-Ortiz

, Gamblin

, Swerdlow

(2020) Mitochondrial DNA manipulations affect tau oligomerization. J Alzheimers Dis 77, 149–163.

24.

Mahley

, Huang

(2012) Apolipoprotein e sets the stage: Response to injury triggers neuropathology. Neuron 76, 871–885.

25.

Kockx

, Traini

, Kritharides

(2018) Cell-specific production, secretion, and function of apolipoprotein E. J Mol Med (Berl) 96, 361–371.

26.

Gafencu

, Robciuc

, Fuior

, Zannis

, Kardassis

, Simionescu

(2007) Inflammatory signaling pathways regulating ApoE gene expression in macrophages. J Biol Chem 282, 21776–21785.

27.

Karin

, Liu

, Zandi

(1997) AP-1 function and regulation. Curr Opin Cell Biol 9, 240–246.

28.

Cano

, Hazzalin

, Mahadevan

(1994) Anisomycin-activated protein kinases p45 and p55 but not mitogen-activated protein kinases ERK-1 and -2 are implicated in the induction of c-fos and c-jun. Mol Cell Biol 14, 7352–7362.

29.

Courtney

, Landreth

(2016) LXR regulation of brain cholesterol: From development to disease. Trends Endocrinol Metab 27, 404–414.

30.

Yue

, Mazzone

(2009) Peroxisome proliferator-activated receptor gamma stimulation of adipocyte ApoE gene transcription mediated by the liver receptor X pathway. J Biol Chem 284, 10453–10461.

31.

Missiroli

, Genovese

, Perrone

, Vezzani

, Vitto

VAM

, Giorgi

(2020) The role of mitochondria in inflammation: From cancer to neurodegenerative disorders. J Clin Med 9, 740.

32.

Picca

, Calvani

, Coelho-Junior

, Marzetti

(2021) Cell death and inflammation: The role of mitochondria in health and disease. Cells 10, 537.

33.

Xia

, Wang

, Zhang

, Liu

, Yu

, Ye

, Wang

, Ye

, Wang

(2021) C/EBPβ is a key transcription factor for APOE and preferentially mediates ApoE4 expression in Alzheimer’s disease. Mol Psychiatry 26, 6002–6022.

34.

Greco

, Fiskum

(2010) Neuroprotection through stimulation of mitochondrial antioxidant protein expression. J Alzheimers Dis 20(Suppl 2), S427–S437.

35.

Espiritu

, Mazzone

(2008) Oxidative stress regulates adipocyte apolipoprotein e and suppresses its expression in obesity. Diabetes 57, 2992–2998.

36.

Friedman

(2015) C/EBPα in normal and malignant myelopoiesis. Int J Hematol 101, 330–341.

37.

Ramji

, Foka

(2002) CCAAT/enhancer-binding proteins: Structure, function and regulation. Biochem J 365, 561–575.

38.

Zhu

, Castellani

, Takeda

, Nunomura

, Atwood

, Perry

, Smith

(2001) Differential activation of neuronal ERK, JNK/SAPK and p38 in Alzheimer disease: The ‘two hit’ hypothesis. Mech Ageing Dev 123, 39–46.

39.

Zhu

, Lee

, Perry

, Smith

(2007) Alzheimer disease, the two-hit hypothesis: An update. Biochim Biophys Acta 1772, 494–502.

40.

Zhu

, Raina

, Perry

, Smith

(2004) Alzheimer’s disease: The two-hit hypothesis. Lancet Neurol 3, 219–226.

41.

Harris

, Tesseur

, Brecht

, Xu

, Mullendorff

, Chang

, Wyss-Coray

, Mahley

, Huang

(2004) Astroglial regulation of apolipoprotein E expression in neuronal cells. Implications for Alzheimer’s disease. J Biol Chem 279, 3862–3868.

42.

Meng

, Xia

(2011) c-Jun, at the crossroad of the signaling network. Protein Cell 2, 889–898.

43.

Kim

, Choi

(2010) Pathological roles of MAPK signaling pathways in human diseases. Biochim Biophys Acta 1802, 396–405.

44.

Hepp Rehfeldt

, Majolo

, Goettert

, Laufer

(2020) c-Jun N-terminal kinase inhibitors as potential leads for new therapeutics for Alzheimer’s diseases. Int J Mol Sci 21, 9677.

45.

Bekris

, Lutz

, Yu

(2012) Functional analysis of APOE locus genetic variation implicates regional enhancers in the regulation of both TOMM40 and APOE. J Hum Genet 57, 18–25.

46.

Nuytemans

, Lipkin Vasquez

, Wang

, Van Booven

, Griswold

, Rajabli

, Celis

, Oron

, Hofmann

, Rolati

, Garcia-Serje

, Zhang

, Jin

, Argenziano

, Grant

SFA

, Chesi

, Brown

, Young

, Dykxhoorn

, Pericak-Vance

, Vance

(2022) Identifying differential regulatory control of APOE ɛ4 on African versus European haplotypes as potential therapeutic targets. Alzheimers Dement 18, 1930–1942.

47.

Watts

, Grunseich

, Rodriguez

, Liu

, Li

, Burdick

, Bruzel

, Crouch

, Mahley

, Wilson

, Cheung

(2022) A common transcriptional mechanism involving R-loop and RNA abasic site regulates an enhancer RNA of APOE. Nucleic Acids Res 50, 12497–12514.

48.

Galli

, Jahn

, Hitt

, Hesse

, Opitz

, Plessmann

, Urlaub

, Poderoso

, Jares-Erijman

, Jovin

(2009) A new paradigm for MAPK: Structural interactions of hERK1 with mitochondria in HeLa cells. PLoS One 4, e7541.

49.

Dagda

, Zhu

, Chu

(2009) Mitochondrial kinases in Parkinson’s disease: Converging insights from neurotoxin and genetic models. Mitochondrion 9, 289–298.

50.

Ono

, Han

(2000) The p38 signal transduction pathway: Activation and function. Cell Signal 12, 1–13.

51.

Xicoy

, Wieringa

, Martens

(2017) The SH-SY5Y cell line in Parkinson’s disease research: A systematic review. Mol Neurodegener 12, 10.

52.

Bekris

, Millard

, Galloway

, Vuletic

, Albers

, Li

, Galasko

, DeCarli

, Farlow

, Clark

, Quinn

, Kaye

, Schellenberg

, Tsuang

, Peskind

, Yu

(2008) Multiple SNPs within and surrounding the apolipoprotein E gene influence cerebrospinal fluid apolipoprotein E protein levels. J Alzheimers Dis 13, 255–266.

53.

Swerdlow

(2018) Mitochondria and mitochondrial cascades in Alzheimer’s disease. J Alzheimers Dis 62, 1403–1416.

54.

Khan

, Cassarino

, Abramova

, Keeney

, Borland

, Trimmer

, Krebs

, Bennett

, Parks

, Swerdlow

, Parker

Jr , Bennett

Jr (2000) Alzheimer’s disease cybrids replicate beta-amyloid abnormalities through cell death pathways. Ann Neurol 48, 148–155.

55.

Cassarino

, Swerdlow

, Parks

, Parker

Jr , Bennett

Jr (1998) Cyclosporin A increases resting mitochondrial membrane potential in SY5Y cells and reverses the depressed mitochondrial membrane potential of Alzheimer’s disease cybrids. Biochem Biophys Res Commun 248, 168–173.

56.

Bell

, Barnes

, Clemmens

, Al-Rafiah

, Al-Ofi

, Leech

, Bandmann

, Shaw

, Blackburn

, Ferraiuolo

, Mortiboys

(2018) Ursodeoxycholic acid improves mitochondrial function and redistributes Drp1 in fibroblasts from patients with either sporadic or familial Alzheimer’s disease. J Mol Biol 430, 3942–3953.

57.

Coskun

, Beal

, Wallace

(2004) Alzheimer’s brains harbor somatic mtDNA control-region mutations that suppress mitochondrial transcription and replication. Proc Natl Acad Sci U S A 101, 10726–10731.

58.

de la Monte

, Luong

, Neely

, Robinson

, Wands

(2000) Mitochondrial DNA damage as a mechanism of cell loss in Alzheimer’s disease. Lab Invest 80, 1323–1335.

59.

Brown

, Sheu

, Mohs

, Haroutunian

, Blass

(2001) Correlation of the clinical severity of Alzheimer’s disease with an aberration in mitochondrial DNA (mtDNA). J Mol Neurosci 16, 41–48.

60.

Klein

, Trumpff

, Yang

, Lee

, Picard

, Bennett

, De Jager

(2021) Characterization of mitochondrial DNA quantity and quality in the human aged and Alzheimer’s disease brain. Mol Neurodegener 16, 75.

61.

Harerimana

, Paliwali

, Romero-Molina

, Bennett

, Pa

, Goate

, Swerdlow

, Andrews

(2022) The role of mitochondrial genome abundance in Alzheimer’s disease. Alzheimers Dement. doi: 10.1002/alz.12812.

62.

Zalocusky

, Najm

, Taubes

, Hao

, Yoon

, Koutsodendris

, Nelson

, Rao

, Bennett

, Bant

, Amornkul

, Xu

, An

, Cisne-Thomson

, Huang

(2021) Neuronal ApoE upregulates MHC-I expression to drive selective neurodegeneration in Alzheimer’s disease. Nat Neurosci 24, 786–798.