Astroglial regulation of apolipoprotein E expression in neuronal cells. Implications for Alzheimer's disease.

Although apolipoprotein (apo) E is synthesized in the brain primarily by astrocytes, neurons in the central nervous system express apoE, albeit at lower levels than astrocytes, in response to various physiological and pathological conditions, including excitotoxic stress. To investigate how apoE expression is regulated in neurons, we transfected Neuro-2a cells with a 17-kilobase human apoE genomic DNA construct encoding apoE3 or apoE4 along with upstream and downstream regulatory elements. The baseline expression of apoE was low. However, conditioned medium from an astrocytic cell line (C6) or from apoE-null mouse primary astrocytes increased the expression of both isoforms by 3-4-fold at the mRNA level and by 4-10-fold at the protein level. These findings suggest that astrocytes secrete a factor or factors that regulate apoE expression in neuronal cells. The increased expression of apoE was almost completely abolished by incubating neurons with U0126, an inhibitor of extracellular signal-regulated kinase (Erk), suggesting that the Erk pathway controls astroglial regulation of apoE expression in neuronal cells. Human neuronal precursor NT2/D1 cells expressed apoE constitutively; however, after treatment of these cells with retinoic acid to induce differentiation, apoE expression diminished. Cultured mouse primary cortical and hippocampal neurons also expressed low levels of apoE. Astrocyte-conditioned medium rapidly up-regulated apoE expression in fully differentiated NT2 neurons and in cultured mouse primary cortical and hippocampal neurons. Thus, neuronal expression of apoE is regulated by a diffusible factor or factors released from astrocytes, and this regulation depends on the activity of the Erk kinase pathway in neurons.

The ⑀4 allele of the gene encoding apolipoprotein (apo) 1 E has been genetically linked to late-onset familial and sporadic Alz-heimer's disease (AD) and has a gene-dose effect on the risk and age of onset of the disease (1)(2)(3)(4)(5). Individuals with two copies of the ⑀4 allele have a 50 -90% chance of developing AD by the age of 85, and those with one copy have about a 45% chance (1,6). Only about 20% of the general population develops AD by the age of 85 (1).
ApoE is found in amyloid plaques and neurofibrillary tangles, two neuropathological hallmarks of AD (7)(8)(9)(10)(11)(12)(13), but its role in their pathogenesis is unclear. ApoE4 has several adverse effects that might explain the association between AD and the ⑀4 allele. It modulates the deposition and clearance of amyloid ␤ peptides and plaque formation (14 -21), impairs the antioxidative defense system (22), dysregulates neuronal signaling pathways (23), disrupts cytoskeletal structure and function (24,25), and alters the phosphorylation of tau and the formation of neurofibrillary tangles (26 -30). However, the mechanisms of these effects are still largely unknown, and it is not known which are the primary effects and which are subsequent or downstream effects.
Initially, apoE was thought to be synthesized in the brain only by astrocytes, oligodendrocytes, and ependymal layer cells (31,32). However, under diverse physiological and pathological conditions, central nervous system (CNS) neurons also express apoE, albeit at lower levels than astrocytes (33)(34)(35)(36)(37)(38)(39)(40). ApoE protein and mRNA are found in cortical and hippocampal neurons in humans (39) and in transgenic mice expressing human apoE under the control of the human apoE promoter (41). In rats treated with kainic acid to induce excitotoxic stress, apoE expression is induced in surviving hippocampal neurons, as determined by in situ hybridization and anti-apoE immunohistochemistry (42). Neuronal expression of apoE can be induced in human brains after cerebral infarction (43). ApoE is also expressed in primary cultured human and rat CNS neurons (44) and in many human neuronal cell lines, including SY-5Y, Kelly, and NT2 cells (32,(45)(46)(47).
In this study we used Neuro-2a cells transfected with human apoE genomic DNA, human NT2/D1 neuronal precursor cells, and mouse primary cortical or hippocampal neurons to investigate how apoE expression is regulated in neurons. Here we demonstrate that neuronal expression of apoE can be induced by astrocyte-conditioned medium and that the astroglial regulation of apoE expression in neurons is controlled by the extracellular signal-regulated kinase (Erk) pathway.
Cell Cultures-Neuro-2a cells (American Type Culture Collection, Manassas, VA) were maintained at 37°C in a humidified 5% CO 2 incubator in MEM containing 10% FBS supplemented with nonessential amino acids, penicillin, and streptomycin.
Human neuronal precursor NT2/D1 cells were kindly provided by Dr. Virginia M.-Y. Lee (University of Pennsylvania School of Medicine, Philadelphia, PA) and maintained in Opti-MEM-I (Invitrogen) containing 5% FBS and penicillin/streptomycin (59). The cells were treated with retinoic acid to induce differentiation into NT2 neurons (59,60).
Primary cultures of cortical or hippocampal neurons were prepared from embryonic day 17 wild-type or apoE-null mice (61,62). As determined by immunostaining with cell-specific antibodies, Ͼ95% of cells in 6-day-old cortical cultures in vitro are positive for neuron-specific enolase (63).
Primary astrocytes were prepared from embryonic day 17 wild-type or apoE-null mice (64). As determined by immunostaining with cellspecific antibodies, Ͼ95% of cells in 6-day-old astrocyte cultures in vitro are positive for glial fibrillary acidic protein (63).
Preparation of Astrocyte-conditioned Medium-Rat astrocytic C6 cells were grown to 80% confluence in DMEM containing 20% FBS in T175 flasks, washed 3 times with serum-free DMEM, and incubated with 10 ml of serum-free DMEM for 24 h. The medium was collected, centrifuged to remove cellular debris, and stored at Ϫ80°C.
The apoE-null mouse primary astrocytes were grown in DMEM containing 20% FBS in 20-cm dishes, washed 3 times with serum-free DMEM, and incubated with 10 ml of serum-free DMEM for 24 h. The conditioned medium was collected, centrifuged, and stored at Ϫ80°C.
Generation of Neuro-2a Cells Stably Expressing Human ApoE3 or ApoE4 -Neuro-2a cells were cotransfected with a 17-kilobase human apoE genomic DNA construct encoding apoE3 or apoE4 and a plasmid containing the neomycin resistance gene by using LipofectAMINE (Invitrogen) (65). The construct consisted of 5 kilobases of 5Ј-flanking region, four exons, three introns, and 8 kilobases of 3Ј-flanking region of the apoE gene. Stably transfected cells were selected by growing them in MEM containing 10% FBS and 400 g/ml G418 for more than 2 weeks and cloned by a cell sorter (65). Cell lines stably expressing apoE3 or apoE4 at similar levels, as determined by reverse transcription (RT)-PCR and anti-apoE Western blotting, were used in this study.
Preparation of Cell Lysates and Western Blotting-Neuro-2a cells transfected with apoE, NT2/D1 neuronal precursor cells, fully differentiated NT2 neurons, and mouse primary cortical and hippocampal neurons were grown to 80% confluence in six-well plates and then incubated with medium conditioned by C6 or primary astrocytes (1:1 dilution in serum-free MEM or Opti-MEM) in the presence or absence of an inhibitor of the Erk (U0126, 5 M), JNK inhibitor 1 (5 M), or p38 (SB203580, 2.5 M) pathway at noncytotoxic doses for 24 h. The cells were harvested, lysed in ice-cold lysis buffer (50 mM Tris/HCl, pH 8.0, 150 mM NaCl, 0.1% SDS, 1% Nonidet p-40, 0.5% sodium deoxycholate, and a mixture of protease inhibitors) for 30 min, and centrifuged at 13,000 rpm for 15 min. Proteins in the supernatant were subjected to SDS-PAGE and detected by anti-apoE Western blotting (28). The fulllength apoE bands were scanned, normalized to cellular proteins, and reported as arbitrary units.
In some experiments 95% primary cortical neurons from apoE-null mice were mix-cultured with 5% primary astrocytes from wild-type mice in serum-free medium for 24 h. ApoE in the cell lysate and medium was then determined as described above.
Quantification of ApoE mRNA by Real-time RT-PCR-Total RNA from apoE3-or apoE4-transfected Neuro-2a cells treated with medium conditioned by C6 or primary astrocytes was isolated with the RNeasy mini kit (Qiagen). RT-PCR was performed with 10 ng of total RNA and the RT-PCR kit from Invitrogen. The RT-PCR product of apoE was 75 bp. The RT-PCR product of the glyceraldehyde-3-phosphate dehydrogenase, which was used as an internal standard, was 112 bp. Different amounts of input RNA, cycling temperatures, and cycle numbers were evaluated to assure a linear response of the apoE and the glyceraldehyde-3-phosphate dehydrogenase signals. The RT-PCR products were measured in real time with an ABI-PRISM-7700 sequence detector from Applied Biosystems (Foster City, CA). The apoE mRNA levels were then calculated and normalized to the internal glyceraldehyde-3phosphate dehydrogenase mRNA standards.
Immunocytochemistry-Mouse primary cortical neurons were grown for 6 days on glass cover slips (Fisher) coated with polylysine in 24-well plates, washed 3 times with serum-free medium, and incubated for 24 h with serum-free medium conditioned by C6 astrocytes (1:1 dilution in serum-free Opti-MEM). After incubation the neurons were fixed, permeabilized, and stained with polyclonal anti-apoE and a fluoresceincoupled secondary antibody (Vector Laboratories) (28). The immunofluorescent-labeled slides were mounted in VectaShield (Vector Laboratories) and viewed with an MRC-1024 laser scanning confocal system (Bio-Rad) mounted on an Optiphot-2 microscope (Nikon).
Statistical Analysis-Results are reported as the mean Ϯ S.D. Differences were evaluated by t test or analysis of variance.

Astroglial Regulation of ApoE Expression in Neuro-2a
Cells-To investigate how apoE expression is regulated in neurons, we transfected Neuro-2a cells with a large fragment of human apoE genomic DNA (Fig. 1A) (66) and selected stable cell lines expressing matched levels of apoE3 and apoE4 (Fig.  1B). All of the transfectants expressed low levels of apoE (Ͻ100 ng of apoE/mg of cell protein/24 h), suggesting that the apoE gene promoter is not very active in Neuro-2a cells. However, medium conditioned by an astrocytic C6 cell line or apoE-null mouse primary astrocytes increased apoE secretion 4 -10-fold and apoE mRNA expression 3-4-fold, with no difference between the isoforms (Fig. 2, A-C). These results suggest that astrocytes secrete a factor or factors that regulate apoE expression in Neuro-2a cells. In preliminary studies, the putative astrocyte-derived factor was sensitive to heat or protease treatment, suggesting a protein (data not shown).
Astroglial Regulation of ApoE Expression in Neuro-2a Cells Is Dependent on the Erk Pathway-Next, we sought to identify the pathway involved in the astroglial regulation of apoE expression in Neuro-2a cells. At noncytotoxic levels, the Erk pathway inhibitor U0126 almost completely abolished the increased expression of apoE3 (Fig. 3, A and B) or apoE4 (data not shown) induced by astrocyte-conditioned medium. Inhibitors of other mitogen-activated protein kinase pathways, JNK inhibitor-1 and SB203580 (for the p38 pathway), had no significant effects on apoE expression (Fig. 3A), although they did inhibit the JNK and p38 pathways, respectively, in control assays (data not shown). U0126 also inhibited the baseline expression of apoE3 (Fig. 3C) and apoE4 (data not shown). Thus, astroglial regulation of apoE expression in Neuro-2a cells is dependent on the Erk pathway.
ApoE Expression in Human NT2 Neurons Is Regulated by Astroglia-To avoid a potential artificial effect of transgene integration on apoE expression in transfected Neuro-2a cells, we also studied human neuronal precursor NT2/D1 cells. These cells expressed apoE constitutively (Fig. 4, three far left lanes), as shown previously (47). When the cells were treated with retinoic acid to induce differentiation, apoE expression increased and then diminished (Fig. 4, six middle lanes), suggesting that it is regulated by signaling pathways related to cell proliferation or differentiation. Incubation of fully differentiated human NT2 neurons with medium conditioned by C6 astrocytes rapidly up-regulated apoE expression (Fig. 4, three far right lanes), suggesting that an astroglia-derived factor or factors can also induce apoE expression in fully differentiated human NT2 neurons. Similar results were also obtained from fully differentiated human NT2 neurons treated with medium conditioned by apoE-null mouse primary astrocytes (data not shown).

Astroglial Regulation of ApoE Expression in Mouse Primary
Neurons Is Also Dependent on the Erk Pathway-To study neuronal apoE expression in a more physiological scenario, we next assessed apoE expression in mouse primary neurons. In both cortical neurons (Fig. 5A) and hippocampal neurons (Fig.  5B), apoE was expressed at low levels; the levels increased in response to incubation with astrocyte-conditioned medium, and the increase was almost completely abolished by U0126 (Fig. 5,  A and B), consistent with involvement of the Erk pathway. Similar results were obtained from mouse primary neurons treated with medium conditioned by apoE-null mouse primary astrocytes (data not shown). As shown by anti-glial fibrillary acidic protein Western blotting, the neuronal cultures contained undetectable levels of astrocytes (Fig. 5C). The absence of detectable apoE in the cell lysate and medium of a mixed culture containing 95% primary cortical neurons from apoEnull mice and 5% primary astrocytes from wild-type mice (our

FIG. 2. Astroglial regulation of apoE expression in apoE-transfected Neuro-2a cells incubated in the presence or absence of astrocyte-conditioned medium.
ApoE-transfected Neuro-2a cells were incubated in the presence or absence of medium conditioned by C6 astrocytes (C6-CM) (A and C) or by apoE-null mouse primary astrocytes (1°-CM) (B and C) in serum-free MEM at 37°C for 24 h. After incubation, the medium (80 l) was subjected to anti-apoE Western blotting, and the apoE bands were scanned and normalized to cellular protein levels (n ϭ 6 per treatment). C, total mRNA was isolated from the cells, and apoE mRNA levels were determined by real-time RT-PCR (n ϭ 4 per treatment).

FIG. 3. Regulation of apoE expression in Neuro-2a cells by astroglia-derived factor(s) is dependent on the Erk pathway. A,
Neuro-2a cells stably expressing human apoE3 were incubated alone (Control) or with C6-conditioned medium (C6-CM) in the presence of various mitogen-activated protein kinase pathway inhibitors in serumfree MEM at 37°C for 24 h: U0126 (5 M) for the Erk pathway, JNK inhibitor 1 (JNKI-1; 5 M) for the JNK pathway, and SB203580 (2.5 M) for the p38 pathway. After incubation, the medium (80 l) was subjected to anti-apoE Western blotting, and the apoE3 bands were scanned and normalized to cellular protein levels (n ϭ 5 per treatment). B, Neuro-2a cells stably expressing human apoE3 were incubated alone (Control) or with medium conditioned by apoE-null mouse primary astrocytes (1°-CM) in the presence or absence of U0126 (5 M) in serum-free MEM at 37°C for 24 h. After incubation, apoE3 levels in the medium (80 l) were determined as described above (n ϭ 6 per treatment). C, Neuro-2a cells stably expressing human apoE3 were incubated alone (Control) or with U0126 (5 M) in serum-free MEM at 37°C for 24 h. After incubation, apoE3 levels in the medium (80 l) were determined as described above (n ϭ 6 per treatment). primary neuronal cultures contained Ͼ95% neurons) further proved that the apoE detected in the primary neuronal cultures came from neurons but not the potentially contaminated astrocytes (data not shown). These results indicate that mouse primary cortical and hippocampal neurons also express low levels of apoE, which can be further stimulated by an astrogliaderived factor or factors through the Erk pathway.
Interestingly, only ϳ10% of apoE generated in mouse primary cortical neurons was secreted into the culture medium during a 24-h incubation (Fig. 6A), whereas more than 60% of apoE generated in mouse primary astrocytes was secreted (Fig.  6B). Similar results were obtained during a 12-h incubation (data not shown). These results suggest that neuron-generated apoE tends to accumulate intracellularly, whereas astrocytegenerated apoE is more likely to be secreted. Anti-apoE immunofluorescent staining revealed that apoE generated in mouse primary neurons in response to treatment with medium conditioned by C6 astrocytes was packed into vesicles distributed along the neurites (Fig. 6C). This finding suggests that the neuron-generated apoE is transported from the soma to growth corns to support neurite outgrowth (24,25,67). DISCUSSION This study demonstrates that neuronal expression of apoE can be induced by a factor or factors secreted by astrocytes and that the astroglial regulation of apoE expression in neurons is controlled by the Erk pathway. The astroglial regulation of neuronal apoE expression may be relevant to neurodegeneration related to aging and AD. Glial fibrillary acidic protein, a marker of astroglial cells, increases with age (68), especially in the hippocampus and entorhinal cortex, which are susceptible to neurodegeneration in AD (69 -72). Astrocytosis, the proliferation and activation of astrocytes (73), is more pronounced in AD brains than in normal brains and usually occurs in areas surrounding neuritic plaques (74,75) and neurons bearing neurofibrillary tangles (75)(76)(77)(78). Importantly, activated astrocytes in AD brains secrete many cytokines, including interleukins 1 and 6, tumor necrosis factor-␣, and acute-phase inflammatory proteins, such as ␣ 2 -macroglobulin and C-reactive protein (79 -82). We speculate that in AD brains the astrocytederived factor or factors, whose effects were observed in the current study, induce neuronal expression of apoE, as observed in previous studies (35,39). None of the astrocyte-derived cytokines mentioned above altered apoE expression in neuronal cells (data not shown). Additional studies are needed to identify the factor or factors that induce neurons to express apoE.
The involvement of the Erk pathway in astroglial regulation of apoE expression in neurons is intriguing. The Erk pathway plays a crucial role in cell proliferation and differentiation (83). In AD and stroke patients, significant numbers of neurons in the neocortex and hippocampus appear to reenter the cell cycle (84 -92); in many of them, the Erk pathway was activated (93)(94)(95)(96)(97), perhaps by amyloid ␤ peptides, oxidative stress, or astroglia-derived cytokines (94,97,98). Thus, it is reasonable to speculate that neuronal activation of the Erk pathway in brains of AD or stroke patients participates in the regulation of apoE expression in neurons. ApoE is expressed in some neurons in AD or stroke brains (39,43). Interestingly, apoE can also regulate cell proliferation (99,100) and activation of the Erk pathway in smooth muscle cells (101) and rat primary hippocampal neurons (102). Thus, there could be a feedback loop between Erk pathway activation and apoE expression in neurons.
Increased apoE levels have been demonstrated in the plasma FIG. 4. Astroglial regulation of apoE expression in human NT2 neurons. Undifferentiated human neuronal precursor NT2/D1 cells (Control), NT2/D1 cells treated with retinoic acid for 1 or 4 weeks, and retinoic acid-differentiated NT2 neurons treated with C6-conditioned medium (C6-CM) were incubated in serum-free medium for 24 h. The medium (80 l) was subjected to anti-apoE Western blotting, and the apoE bands were scanned and normalized to cellular protein levels (n ϭ 6 per treatment).
FIG. 5. Erk pathway-dependent regulation of apoE expression in mouse primary cortical and hippocampal neurons by astroglia-derived factor(s). Mouse primary cortical (A) or hippocampal (B) neurons were incubated alone (Control) or with C6-conditioned medium (C6-CM) in the presence or absence of the Erk pathway inhibitor U0126 in serum-free MEM at 37°C for 24 h. After incubation, the cells were lysed. The lysates (100 g of total protein) were subjected to anti-apoE Western blotting, and the apoE bands were scanned and normalized to cellular protein levels (n ϭ 5 per treatment). C, lysates of mouse primary cortical neurons (200 g of total protein) or mouse primary astrocytes (40 g of total protein) were analyzed by antiglial fibrillary acidic protein Western blotting. and brains of AD patients independently of apoE genotype in some (103)(104)(105) but not all studies (106 -108). Recent epidemiological studies suggest that some APOE promoter polymorphisms that alter the transcription of APOE (105,109,110) are associated with increased risk for AD (111)(112)(113)(114)(115)(116)(117); however, this observation has not been confirmed in some populations (118 -121). Thus, polymorphisms within the proximal promoter of APOE may cause increased apoE levels in the brain, thereby increasing the risk for AD, especially for those carrying the ⑀4 allele (122). Transcription of APOE is regulated by an array of tissue-specific cis-acting regulatory elements distributed over a 20-kilobase region spanning the APOE locus (110,(123)(124)(125)(126)(127)(128)(129)(130)(131). Although our current study does not directly address the mechanisms by which the astrocyte-derived factor or factors regulate the transcription of APOE, we speculate that the factor or factors may trigger a signal pathway, such as the Erk pathway, that leads to activation of certain tissue-specific cis-acting regulatory elements of APOE that might be modified by APOE promoter polymorphisms.
Based on the current and previous studies (32)(33)(34)(35)(36)(37)(38)(39)(41)(42)(43)(44)(45)(46)(47), we hypothesize that acute injury, such as head trauma, or chronic injury, such as damage related to amyloid ␤ deposition or oxidative stress, of the CNS causes astroglial proliferation and activation (astrocytosis), leading to secretion of a factor or factors that activate the Erk pathway and induce neuronal expression of apoE (Fig. 7). Neuronal expression of apoE in response to brain injury probably promotes neuronal repair and remodeling and protects neurons from injury (42,(132)(133)(134). However, in apoE4 carriers, intraneuronal proteolytic processing of apoE4 and generation of neurotoxic apoE4 fragments (28,30) may turn a neuroprotective response into a pathogenic process. In this scenario it may be advantageous to inhibit neuronal production of apoE. Our results suggest that this might be accomplished by inhibiting astrocytosis, blocking specific diffusible astroglial factors, or modulating the Erk pathway in neurons.
FIG. 6. Neuron-generated apoE tends to accumulate intracellularly, whereas astrocyte-generated apoE tends to be secreted. Mouse primary cortical neurons (A) or primary astrocytes (B) were incubated with serum-free medium for 24 h at 37°C. After incubation, apoE in cell lysates (100 g of total protein) or medium (100 l) was determined by anti-apoE Western blotting and semiquantified by scanning the films. C, anti-apoE immunofluorescent staining of a mouse primary cortical neuron treated with C6-CM for 24 h. The image was taken by a confocal scanning microscope (ϫ400).

FIG. 7.
A working model of astroglial regulation of apoE expression in neuronal cells. We hypothesize that acute or chronic CNS injury causes astroglial proliferation and activation (astrocytosis), leading to secretion of a factor or factors that induce neuronal expression of apoE. CNS insults might also affect neuronal apoE expression directly.