Noradrenaline induces expression of peroxisome proliferator activated receptor gamma (PPARγ) in murine primary astrocytes and neurons

Authors


Address correspondence and reprint requests to Michael T. Heneka, Department of Neurology, Neuroinflammation Research Group, Sigmund Freud Str. 25, 53127 Bonn, Germany. Tel. +49 228287 6255, Fax: +49 228287 5024, E-mail: m.heneka@uni-bonn.de

Abstract

Cerebral inflammatory events play an important part in the pathogenesis of Alzheimer's disease (AD). Agonists of the peroxisome proliferator-activated receptor gamma (PPARγ), a nuclear hormone receptor that mediates anti-inflammatory actions of non-steroidal anti-inflammatory drugs (NSAIDs) and thiazolidinediones, have been therefore proposed as a potential treatment of AD. Experimental evidence suggests that cortical noradrenaline (NA) depletion due to degeneration of the locus ceruleus (LC) – a pathological hallmark of AD – plays a permissive role in the development of inflammation in AD. To study a possible relationship between NA depletion and PPARγ-mediated suppression of inflammation we investigated the influence of NA on PPARγ expression in murine primary cortical astrocytes and neurons. Incubation of astrocytes and neurons with 100 µm NA resulted in an increase of PPARγ mRNA as well as PPARγ protein levels in both cell types. These effects were blocked by the β-adrenergic antagonist propranolol but not by the α-adrenergic antagonist phentolamine, suggesting that they might be mediated by β-adrenergic receptors. Our results indicate for the first time that PPARγ expression can be modulated by the cAMP signalling pathway, and suggest that the anti-inflammatory effects of NA on brain cells may be partly mediated by increasing PPARγ levels. Conversely, decreased NA due to LC cell death in AD may reduce endogenous PPARγ expression and therefore potentiate neuroinflammatory processes.

Abbreviations used
AD

Alzheimer's disease

cAMP

cyclic adenosine monophosphate

COX-2

cyclo-oxygenase-2

dbcAMP

dibutyryl cyclic adenosine monophosphate

DBD

DNA-binding domain

DIV

days in vitro

DMEM

Dulbecco's modified Eagle medium

15d-PGJ2

15-deoxy-Δ12,14-prostaglandin J2

FCS

fetal calf serum

GDH

glyceraldehyde 3-phosphate dehydrogenase

GM-CSF

granulocyte macrophage colony-stimulating factor

iNOS

inducible form of nitric oxide

LC

locus ceruleus

M-CSF

macrophage colony-stimulating factor

NA

noradrenaline

NSAIDs

non-steroidal anti-inflammatory drugs

PBS

phosphate-buffered saline

PKA

protein kinase A

PMSF

phenylmethylsulfonyl fluoride

PPARγ

peroxisome proliferator activated receptor gamma

PPREs

peroxisome proliferator response elements

PVDF

polyvinylidene difluoride

SDS–PAGE

sodium dodcyl sulfate polyacrylamide gel electrophoresis

TZDs

thiazolidinediones

There is compelling evidence for a role of inflammatory events in the development and progression of Alzheimer's disease (AD; Akiyama et al. 2000). The importance of inflammation in AD is demonstrated by a number of studies showing a protective effect of non-steroidal anti-inflammatory drugs (NSAIDs) against AD (McGeer et al. 1990; Stewart et al. 1997; Wyss-Coray and Mucke 2000). The action of NSAIDs seems not to involve the classical target cyclo-oxygenase-2 (COX-2), as selective COX-2 inhibitors such as nimesulide had no effect on the progression of the disease (Aisen et al. 2002). Because it has been shown that several NSAIDs (Lehmann et al. 1997) as well as the antidiabetic thiazolidinediones (TZDs; Thieringer et al. 2000) and the naturally occurring agonist 15-deoxy-Δ12,14-prostaglandin J2 (15d-PGJ2; Kliewer et al. 1994; Forman et al. 1995) can activate peroxisome proliferator activated receptor gamma (PPARγ), we proposed that the actions of NSAIDs in AD are mediated through PPARγ activation (Heneka et al. 2001; Landreth and Heneka 2001).

PPARγ belongs to a group of nuclear hormone receptors that are closely related to the thyroid hormone and retinoid receptors (Wahli et al. 1995; Desvergne and Wahli 1999). PPARγ mRNA is found in hippocampus, retina, brain stem, cerebellum, and cortex (Braissant et al. 1996), although the expression levels are low compared to adipose tissue. Already known endogenous ligands include several polyunsaturated fatty acids and the prostaglandin metabolite 15d-PGJ2. The activation of PPARγ reduces the expression of proinflammatory cytokines and the inducible form of nitric oxide synthase (iNOS) in macrophages (Lemberger et al. 1996; Ricote et al. 1998b), monocytes (Jiang et al. 1998), microglial cells (Bernardo et al. 2000), and neurons (Heneka et al. 2000), suggesting a role of PPARγ in inflammatory events.

The factors and signals that regulate PPARγ expression are not well characterized. In cells of the adipocyte lineage, expression of PPARγ is induced by dexamethasone and 3-isobutyl-1-methylxanthine via induction of the CCAAT enhancer-binding proteins C/EBPβ and C/EBPδ (Wu et al. 1999; Hamm et al. 2001). Moreover, expression of PPARγ is induced upon activation of PPARβ/δ in 3T3C2 fibroblasts (Bastie et al. 1999). In macrophages, it has been shown that macrophage colony-stimulating factor (M-CSF) and granulocyte/macrophage colony-stimulating factor (GM-CSF) can induce expression of PPARγ (Ricote et al. 1998a). In pneumocytes, PPARγ expression can be induced by cyclic adenosine monophospate (cAMP) analogs (Michael et al. 1997). These observations suggest that intracellular cAMP levels could modulate PPARγ expression.

A pathological hallmark of AD is the degeneration of the locus ceruleus (LC), with consecutive loss of noradrenergic projections and reduced cortical noradrenaline (NA) levels (Mann et al. 1982; Kalaria et al. 1989). Interestingly, reduced cortical NA levels significantly correlate with the numbers of Aβ plaques, neurofibrillary tangles, and the severity of dementia in AD (Mann et al. 1982; Bondareff et al. 1987). Recently, it has been shown that experimentally induced loss of LC neurons increases the response to Aβ-evoked inflammation in cortical projection areas, including neuronal iNOS expression (Heneka et al. 2002). This represents the first model proposing a link between loss of a neurotransmitter and occurrence of inflammatory events in AD. NA has well documented anti-inflammatory effects in vitro: it blocks TNF-α (Hu et al. 1991), IL-1β (Willis and Nisen 1995), and iNOS expression in astrocytes (Feinstein et al. 1993; Feinstein 1998; Galea and Feinstein 1999), as well as inflammatory activation of microglial cells (Lee et al. 1997; Loughlin et al. 1993). In part, these in vitro anti-inflammatory effects seem to be mediated by activation of β-adrenergic receptors and intracellular elevation of cAMP (Hu et al. 1991; Willis and Nisen 1995; Pahan et al. 1997; Galea and Feinstein 1999; Qi et al. 2000), leading to an increase in inhibitory IkB proteins and therefore reduced inflammatory gene transcription (Gavrilyuk et al. 2002). However, the complete signal transduction cascade mediating noradrenergic anti-inflammatory effects is not yet well defined.

Because at least some of the anti-inflammatory actions of NA are mediated via β-adrenergic receptors and intracellular elevation of cAMP, and the expression of PPARγ itself can be induced by cAMP, we investigated the possible influence of NA on the expression of PPARγ as a possible mediator of noradrenergic anti-inflammatory action in brain. In this study, we demonstrate that NA regulates transcription and expression of PPARγ in murine primary astrocytes and neurons via activation of the β-adrenergic receptor. These results suggest that decreased cortical NA content in AD may cause a reduction of PPARγ-mediated anti-inflammatory defences, and thereby augment pro-inflammatory events in LC projection areas.

Experimental procedures

Materials

DMEM, neurobasal A medium, phosphate-buffered saline (PBS), penicillin, streptomycin, B-27 supplement, and trizol were purchased from Gibco Life Technologies (Karlsruhe, Germany). Fetal calf serum (FCS) was from PAN Biotech (Aidenbach, Germany). Bovine serum albumin, ARA-C, trypsin, trypsin-inhibitor, deoxyribonuclease 1, +/– arterenol, isoproterenol, phentolamine, propranolol, dibutyryl cyclic adenosine monophosphate (dbcAMP), mouse monoclonal anti β-actin antibody (Clone AC-15), horseradish peroxidase-conjugated goat anti-rabbit IgG for detection of PPARγ, and actinomycin D were from Sigma (Deisenhofen, Germany). Synthetic oligonucleotides were from Thermo Hybaid Interactiva (Ulm, Germany). Taq polymerase and cDNA reagents were purchased from Invitrogen Life Technologies (Carlsbad, CA, USA). Rabbit polyclonal anti PPARγ antibody was from Alexis Biochemicals (Lausen, Switzerland); horseradish peroxidase-conjugated rabbit anti-mouse IgG for detection of β-actin was obtained from Dianova (Hamburg, Germany).

Cell culture

Primary cultures of cerebral cortical astrocytes were prepared from 1-day-old CD1 mice (Harlan Winkelmann, Borchen, Germany). Brains were removed aseptically from the skulls, the meninges excised carefully, and the cortex dissected free. Brains were first mechanically dissected and afterwards cells dissociated by passage through needles of decreasing gauges (1.2 × 40, 0.8 × 40, and 0.5 × 16 mm) with a 10-mL syringe. No trypsin was used for dissociation. The cells were seeded on poly-l-lysine-coated 100-mm dishes (Nunc, Wiesbaden, Germany) in DMEM containing 10% FCS and 1% penicillin/streptomycin and subsequently incubated at 37°C in an atmosphere containing 5% CO2/95% air. The culture medium was renewed on day in vitro (DIV) 5 and cells used for experiments on DIV 10.

Primary cultures of cerebral cortical neurons were prepared from 1-day-old CD1 mice (Harlan Winkelmann, Borchen, Germany) essentially as described. Briefly, cells were dissociated from freshly dissected forebrains (as described above) by mechanical disruption in the presence of trypsin and DNAse 1 and then plated on poly-l-lysine pre-coated 6-well plates in neurobasal A medium containing B27 supplement and 1% penicillin/streptomycin. Cells were seeded at a density of 2.1 × 100 000 cells/cm2. ARA-C (10 µm) was added to the culture medium 24 h after plating of the cells to arrest the growth of non-neuronal cells. Cultures were fed 5 mm d-glucose on DIV 7 and used for experiments at DIV 8. Cultures generated by this method have been shown to contain more than 95% neurons (Nicoletti et al. 1986; Heneka et al. 1999).

Treatment of cells

During all incubations, primary cultures were maintained at 37°C in an atmosphere containing 5% CO2. For each experiment, NA was freshly diluted in PBS from a stock solution of 10 mm, and the respective amount was then added to the cells. Treatment with isoproterenol was essentially the same. The adrenergic antagonists propranolol and phentolamine were dissolved in PBS and added to the cells 15 min before exposure to NA. Similarly, cells were exposed to cycloheximide and actinomycin D 15 min before treatment with NA. At the end of the various periods of incubation, cells were washed three times with ice-cold PBS, and total RNA or protein was extracted according to standard protocols.

Western blot analysis

At the respective time points, cells were scraped of in PBS containing 0,1 mg/mL of phenylmethylsulfonyl fluoride (PMSF). After centrifugation (3000 g, 10 min, 4°C), supernatant was removed and cells resuspended in RIPA buffer [150 mm NaCl, 10 mm Tris–HCl pH 8.0, 1% NP-40, 0.5% deoxycholic acid, 0.1% sodium dodecyl sulfate (SDS), 5 mm EDTA] containing 0.7% PMSF, 0.2% aprotinin and 0.2% leupeptin and kept on ice for 15 min Cell lysates were obtained by centrifugation (13 000 g, 15 min, 4°C). Forty micrograms of protein (protein concentration was determined by Bradford protein assay) were used for each sample. Samples were then mixed with an equal amount of 2 × SDS sample buffer, boiled at 98°C for 5 min, centrifuged and separated by 10% SDS polyacrylamide gel electrophoresis (SDS–PAGE). Afterwards, proteins were transferred to polyvinylidene difluoride (PVDF) membranes. Protein loading was controlled by ponceau-S staining of membranes. After incubation for 1 h in blocking solution (0.05% Tween-20 and 5% skim milk in PBS), membranes were incubated overnight with primary antibody to PPARγ (1 : 1000), diluted in PBS/Tween-20 containing 1% bovine serum albumin and 0.1% sodium azide. Thereafter, blots were washed extensively in PBS/-20, and incubated for 2 h in blocking solution with horseradish peroxidase-conjugated goat anti-rabbit IgG (1 : 3000) for detection of PPARγ. Bands were then visualized by enhanced chemiluminescence reagent. PPARγ protein could be detected at approximately 45 kDa, the specificity of the antibody used for all western blot experiments shown has been confirmed by western blotting of mouse neuroblastoma (N2a) cells transiently transfected with PPARγ cDNA as well as of embryonic mouse PPARγ knockout and wild type fibroblasts (Fig. 1; Rosen et al. 1999). For detection of β-actin, membranes were afterwards incubated with stripping solution (62.5 mm Tris–HCl pH 6.8; 2% SDS, 100 mm 2-mercaptoethanol) for 30 min at 60°C, and after extensive washing with water and PBS again incubated with blocking solution for 1 h. Membranes were then incubated overnight with primary antibody to β-actin (1 : 5000; diluted in PBS containing 0.05% Tween-20 and 5% skim milk), again washed extensively with PBS/Tween-20, and incubated for 2 h with blocking solution containing horseradish peroxidase-conjugated rabbit anti-mouse IgG (1 : 10000). Representative blots of PPARγ and β-actin from the same membrane are shown.

Figure 1.

Specificity of antibodies directed against PPARγ. Western blot analysis of lysates prepared from N2a cells that were transiently transfected with a PPARγ cDNA, showing the presence of an expected band of approximately 45 kDa. The specificity of the anti-PPARγ antibody was further confirmed by the absence of bands using lysates prepared from PPARγ knockout (–/–), but not from wild-type (+/+) fibroblasts.

RNA preparation and RT–PCR

Total RNA was prepared using Trizol according to manufacturer's procedures. Thereafter, aliquots of total RNA (1 µg) were converted into cDNA using random hexamer primers, and amplified by PCR using primers specific for PPARg or glyceraldehyde 3-phosphate dehydrogenase (GDH). The primers used for PPARγ mRNA were 5′-ATGCTGGCCTCCCTGATGAATAAA-3′ for forward primer and 5′-CTGGAGCACCTTGGCGAACA-3′ for reversed primer, yielding a 228-bp product. The primers used for murine GDH mRNA was 5′-TCACCAGGGCTGCCATTTGC-3′ for forward primer, and 5′-GACTCCACGACATACTCAGC-3′ for reversed primer, which yield a 304-bp product. The mRNA levels were determined in parallel aliquots of cDNA. The amount of GDH mRNA served as a control for the evaluation of PPARγ mRNA levels. PCR conditions were for GDH 23 cycles of denaturation at 93°C for 25 s; annealing at 55°C for 25 s; and extension at 72°C for 25 s; followed by 5 min at 72°C. PCR conditions of PPARγ differed only in the number of cycles (35 cycles). The number of cycles yielding best results in order to distinguish between different amounts of cDNA was assessed for each primer (data not shown). PCR products were separated by electrophoresis through 2% agarose gels containing 0.1 µg/mL ethidium bromide.

Densitometry and statistical analysis

For all PCR and western blots, bands were scanned and intensities measured using the freeware NIH Image J program. Each experiment has been performed at least three times and was statistically analyzed. Data were calculated as the percentage of control (from each gel/blot) which served as basis for further statistical analysis. Data are given as mean ± SEM. Differences between controls and the different groups were assessed by student's t-test (two tailed, unpaired); a p-value less than 0.05 was considered to be significant.

Results

Effect of NA on PPARγ expression in murine primary astrocytes

In order to study the effect of NA on PPARγ transcription, astrocytes were incubated with 100 µm NA for several periods of time. This concentration of NA was initially selected based upon several previous studies (Frohman et al. 1988; Hasko et al. 1998; Feinstein et al. 1993; Gavrilyuk et al. 2002) in which the effects of NA on glial inflammatory gene expression showed the most robust effects at 100 µm. The levels of PPARγ mRNA, assessed by semiquantitative RT–PCR, were found increased after exposure to NA, with the maximum effect (approximately twofold) at 4 h (Fig. 2a). A significant induction of PPARγ mRNA levels was already detected 2 h after incubation with NA, persisting up to 8 h, indicating a rapid and lasting effect. The levels of GDH remained unchanged by 100 µm NA (Fig. 2b).

Figure 2.

NA induces PPARγ transcription and expression in murine primary astrocytes. Analysis of PPARγ mRNA (a) and protein levels (c and d) in astrocytes exposed to 100 μm NA at different time points. One gel/blot representative of three separate experiments is displayed. GDH and β-actin served as controls (b and d). The intensity of each band was measured by densitometry; graphs demonstrate mean and SEM of each time point. Asterisks indicate significant differences between control and respective time points (*p < 0.05, **p < 0.01; two-tailed, unpaired Student's t-test).

In parallel, a significant increase in PPARγ protein was already observed at 12 h after stimulation with 100 µm NA (Figs 2c and d), with a maximal increase at 24 h after incubation with this neurotransmitter. β-actin levels did not differ between treatments at all time points analyzed (Fig. 2d).

Incubation with NA increased both PPARγ mRNA and protein levels (Fig. 3). A significant increase in mRNA levels was observed between 10 and 100 µm NA, with a maximal increase of approximately threefold versus control values occurring between 10 and 30 µm (p < 0.05 vs. control); and a statistically more significant increase (approximately twofold) present at 100 µm NA (p < 0.01 vs. control). PPARγ protein levels were also increased by incubation with NA, with an almost significant increase due to incubation at 10 µm NA, and with significant and similar increases (roughly 2.3-fold control values) observed at 30 and 100 µm NA (p < 0.01 vs. control). For the following studies the concentration of 100 µm was used because maximal changes in PPARγ protein levels were observed at this concentration. Co-incubation with actinomycin D (5 µg/mL), an inhibitor of RNA transcription, and 100 µm NA abolished the effect of NA on PPARγ mRNA levels, indicating that NA leads to de novo synthesis of PPARγ mRNA (Fig. 3a). Correspondingly, co-application of 100 µm NA and cycloheximide (30 µg/mL), an inhibitor of protein translation, reversed the effect of NA on PPARγ protein levels (Figs 3c and d).

Figure 3.

NA induces PPARγ expression in a dose-dependent manner. Analysis of PPARγ mRNA and protein levels in astrocytes exposed to different concentrations of NA at 4 h (for mRNA) and 20 h (for protein; a, c and d). Additionally, astrocytes were treated with 100 μm NA plus actinomycin D (a) or cycloheximide (c and d) for inhibition of de novo transcription and translation, respectively. One gel/blot representative of three separate experiments is displayed. GDH and β-actin served as control (b and d). The intensity of each band was measured by densitometry; graphs demonstrate mean and SEM of each time point. Asterisks indicate significant differences between control and different NA concentrations (*p < 0.05, **p < 0.01; two-tailed, upaired Student's t-test). Crosses indicate significant differences between treatment with 100 μm NA plus actinomycin D/cycloheximide and treatment with different concentrations of NA alone (†p < 0.05, ††p < 0.01; two-tailed, unpaired Student's t-test).

Effect of NA on PPARγ expression in murine primary neurons

To investigate the effect of NA on neuronal PPARγ mRNA and protein levels, cells were incubated with 100 µm NA for several periods of time. The maximal increase in neuronal PPARγ mRNA levels was already observed after 1 h exposure (Fig. 4a). The maximal increase was estimated to be fivefold compared to controls, suggesting that the magnitude of NA-induction was more pronounced in neurons than in astrocytes. As it has been shown for astrocytes, treatment of neurons with actinomycin D (5 µg/mL) together with 100 µm NA for 1 h completely abolished the effect of NA on PPARγ mRNA levels, suggesting that the observed fivefold increase in PPARγ mRNA levels by NA is indeed due to induction of de novo transcription (Fig. 4a). Within 8 h after initial NA stimulation, PPARγ mRNA levels were not significantly elevated when compared to control, demonstrating that the stimulatory effect on neurons is not as long-lasting as in astrocytes. The levels of GDH remained unchanged by 100 µm NA (Fig. 4b).

Figure 4.

NA induces PPARγ transcription and expression in murine primary neurons. Analysis of PPARγ mRNA and protein levels in neurons exposed to 100 μm NA at different time points (a, c and d). Additionally, treatment with actinomycin D together with 100 μm NA at 1 h is displayed (a). One gel/blot representative of three separate experiments is displayed. GDH and β-actin served as control (b and d). The intensity of each band was measured by densitometry; graphs demonstrate mean and SEM of each time point. Asterisks indicate significant differences between control and respective time points (*p < 0.05, **p < 0.01; two-tailed, unpaired Student's t-test). Crosses indicate significant differences between treatment with 100 μm NA plus actinomycin D and treatment with NA alone at 1 h after beginning of treatment (††p < 0.01; two-tailed, unpaired Student's t-test).

Corresponding to the time-course observed for neuronal PPARγ mRNA levels, a highly significant increase in PPARγ protein levels was detectable at 5 h after exposure to NA, and already at 10 h after exposure to NA PPARγ protein levels have returned to baseline levels (Figs 4c and d). These results suggest differences in the regulation – and maybe the function – of PPARγ in neurons versus astrocytes, as the rapid increase and decrease of PPARγ mRNA and protein levels detected in neurons does not correspond to the slower, but persisting increase in PPARγ mRNA levels and protein levels observed in astrocytes.

Effect of adrenergic agonists and antagonists on PPARγ expression in astrocytes and neurons

In order to study which adrenergic receptor is involved in the regulation of PPARγ expression, neurons and astrocytes were incubated with different adrenergic receptor agonists and antagonists. In astrocytes, 100 µm isoproterenol, a β-adrenergic agonist, induced a similar increase in PPARγ mRNA levels (approximately 2.1-fold induction) and PPARγ protein expression (roughly 3.6-fold induction) as NA (Figs 5a, c and d).

Figure 5.

Effect of adrenergic agonists and antagonists on PPARγ mRNA and protein levels in astrocytes. Incubation of astrocytes with NA (100 μm) plus propranolol (Pro, 10 μm), NA (100 μm) plus phentolamine (Phe, 10 μm), as well as with isoproterenol (Iso, 100 μm), and with dbcAMP (cAMP, 1 μm; a, c and d). Several concentrations of isoproterenol (10 μm, 30 μm, 100 μm) and dbcAMP (1 μm, 10 μm, 100 μm) were tested (e and f). Analysis of PPARγ mRNA levels was performed after 4 h and 8 h, analysis of protein levels after 20 h incubation with the respective substances (a, c–f). One gel/blot representative of three separate experiments is displayed. GDH and β-actin served as control (b, d and f). Asterisks indicate significant differences between control and different agonists/antagonists (*p < 0.05, **p < 0.01; two-tailed, unpaired Student's t-test); crosses indicate significant differences between cells treated with NA plus propranolol and cells treated with NA alone or with isoproterenol or dbcAMP (†p < 0.05, two-tailed, unpaired Student's t-test).

For further investigation of the signal transduction pathway downstream of adrenergic receptors, we studied the influence of intracellular cAMP elevation on PPARγ levels, as activation of β-adrenergic receptors can induce an increase of the intracellular second messenger cAMP. Incubation of astrocytes with the membrane-permeable analog of cAMP, dbcAMP, at a concentration of 1 µm resulted in a significant increase in PPARγ protein levels with an approximately twofold induction (Figs 5c and d).

Although isoproterenol was already effective in induction of PPARγ protein levels at concentrations of 10 µm and 30 µm, the most significant increase was observed at a concentration of 100 µm, thereby following a similar dose–response pattern as does NA itself (Figs 5e and f). On the contrary, dbcAMP seemed to be most effective at the lowest concentration (1 µm), but it still significantly elevated PPARγ protein levels at concentrations of 10 µm and 100 µm (Figs 5e and f). However, statistical analysis did not detect any significant difference among the three concentrations tested.

In neurons, stimulation with 100 µm isoproterenol led to a 1.8-fold increase in PPARγ mRNA and a 1.9-fold increase in PPARγ protein levels, resembling the effect of NA (Figs 6a, c and d). Incubation of neurons with 1 µm dbcAMP resulted in an approximately 1.9-fold increase in PPARγ protein levels, thereby proving to be as effective as NA or the β-adrenergic agonist isoproterenol (Figs 6c and d).

Figure 6.

Effect of adrenergic agonists and antagonists on PPARγ mRNA and protein levels in neurons. Incubation of neurons with NA (100 μm) plus propranolol (Pro, 10 μm), NA (100 μm) plus phentolamine (Phe, 10 μm), as well as with isoproterenol (Iso, 100 μm), and with dbcAMP (cAMP, 1 μm). Analysis of PPARγ mRNA levels was performed after 2 h and 6 h, analysis of protein levels after 5 h incubation with the respective substances (a, c and d). One gel/blot representative of three separate experiments is displayed. GDH and β-actin served as control (b and d). Asterisks indicate significant differences between control and different agonists/antagonists at the respective time points (*p < 0.05 two-tailed, unpaired Student's t-test). Crosses indicate significant differences between cells treated with NA plus propranolol and cells treated with NA alone or with isoproterenol or dbcAMP at the respective time points (†p < 0.05, two-tailed, unpaired Student's t-test).

In astrocytes (Figs 5a, c and d) as well as in neurons (Figs 6a, c and d), the β-adrenergic antagonist propranolol (10 µm) completely antagonized the effects of 100 µm NA, whereas the α-adrenergic antagonist phentolamine did not affect NA-induced PPARγ transcription and protein expression. These results suggest that the NA-induced effects on PPARγ mRNA and protein are mediated by β-adrenergic and not by α-adrenergic receptors, leading to an increase in intracellular cAMP levels.

Discussion

The present study shows that NA increases PPARγ transcription and expression in murine primary astrocytes and neurons. In astrocytes, the maximum effect on PPARγ mRNA levels was found at 4 h after exposure to 100 µm NA, with a twofold increase compared to the baseline amount. Neurons revealed a more rapid increase, noticed already at 1 h after noradrenergic stimulation accompanied by a higher degree of induction (fivefold increase in PPARγ mRNA levels). Whereas the highest levels of astrocytic PPARγ protein were detected at 24 h, neurons revealed a faster increase in PPARγ protein that was maximal already at 5 h after initial exposure to NA. Induction of PPARγ mRNA and protein by NA resulted from de novo synthesis, as these effects were blocked by coincubation with actinomycin D and cycloheximide.

In dose–response studies (Fig. 3) we observed, as others have (Frohman et al. 1988; Hasko et al. 1998; Feinstein et al. 1993; Gavrilyuk et al. 2002), that relatively high concentrations of NA were necessary to observe maximal stimulatory effects. Because we only added NA once at the beginning of the experiment, this requirement may reflect a need for continuous exposure to a certain concentration of NA, or to the presence of NA at a later time during the incubation period; in both cases an initial high dose may be needed to maintain sufficiently high levels at later times.

Apart from its role as a classical neurotransmitter, NA has a number of additional functions, such as induction of astrocytic synthesis and release of neurotrophic factors (Schwartz and Mishler 1990), as well as regulation of glucose uptake and glycogen synthesis in astrocytes (Sorg and Magistretti 1992; Allaman et al. 2000). In addition, anti-inflammatory effects of NA on glial cells have been described. These effects are likely to be mediated by β-adrenergic receptor activation, with consecutive release of G proteins, activation of adenylate cyclase and elevation of intracellular cAMP levels (Galea and Feinstein 1999), as the β-adrenergic agonist isoproterenol (Chambers et al. 1993; Pahan et al. 1997; Szabo et al. 1997; Farmer et al. 2000), cAMP analogs or elevators (Feinstein 1998; Pahan et al. 1997) resulted in an equal reduction of glial inflammation as NA itself, whereas noradrenergic actions were inhibited by simultaneous application of the β-adrenergic antagonist propranolol (Feinstein et al. 1993). In line with these findings, we demonstrate that NA positively regulates PPARγ expression by activation of the β-adrenergic receptor, as the β-adrenergic agonist isoproterenol is as effective in enhancing PPARγ expression as NA, whereas the effect of NA could be significantly reduced by the β-adrenergic antagonist propranolol but not by the α-adrenergic antagonist phentolamine. Moreover, application of the membrane-permeable cAMP analog dbcAMP resulted in similar increase in PPARγ protein levels in astrocytes as well as in neurons, suggesting that NA acts via activation of β-adrenergic receptors, leading to an intracellular increase of cAMP.

However, the signal transduction cascade downstream is not yet well defined; a few studies demonstrate an activation of protein kinase A (PKA) resulting in reduced inflammatory gene transcription, but the nuclear transcription factors involved in this process have not yet been identified (Galea and Feinstein 1999; Gavrilyuk et al. 2002).

The main, if not exclusive, source of noradrenergic cortical afferents is the locus ceruleus (LC) which undergoes degeneration in AD resulting in cortical NA depletion (Mann et al. 1982; Bondareff et al. 1987; Cohen et al. 1997). In addition, an upregulation of cortical β2-adrenergic receptors has been observed in AD brains (Bondareff et al. 1987; Kalaria and Andorn 1991). Evidence that LC loss is not just an unimportant feature of brain degeneration but rather a potentiating factor in the course of the disease comes from in vivo experiments demonstrating that the inflammatory reaction upon local injection of β-amyloid is increased in animals after cortical NA depletion (Heneka et al. 2002), with a reversal of this effect upon coinjection of NA or isoproterenol. Notably, there is a positive correlation between the extent of noradrenergic depletion and plaque count, inflammatory processes and clinical status pre-mortem in AD (Bondareff et al. 1987). The results presented in this paper suggest that the noradrenergic system may be involved in suppression of inflammation in vivo, and diseases featuring loss of noradrenergic projections like AD may facilitate therefore inflammatory responses upon diverse stimuli.

PPARγ belongs to a group of nuclear receptors that are closely related to the thyroid hormone and retinoid receptors (Wahli et al. 1995; Desvergne and Wahli 1999). Upon ligand-mediated activation, it undergoes conformational changes, heterodimerizes with RXR and binds to peroxisome proliferator response elements (PPREs) in the promotor region of target genes. Besides ligand-mediated activation, PPARγ can also become activated by means of phosphorylation (Lazennec et al. 2000). This activation is independent from ligand-activation and seems to be mediated by PKA via phosphorylation of PPARγ in the DNA-binding domain (DBD; Lazennec et al. 2000). Moreover, activation of PKA results in a stabilization of the ligand-activated PPARγ to DNA. These findings are especially interesting as PKA represents a potential key position in the signal transduction cascade initiated by NA that eventually leads to activation of PPARγ.

Activation of PPARγ results in a reduction of cytokine generation in β-amyloid-stimulated microglial cells and immunostimulated macrophages (Jiang et al. 1998; Combs et al. 2000), thereby reducing both glial activation and neuronal toxicity (Combs et al. 2000). Therefore, a positive regulation of PPARγ protein levels by NA could be a key mechanism of NA-mediated anti-inflammation. In line with a neuroprotective role of PPARγ, decreased neuronal iNOS expression and prevention of neuronal apoptosis have been observed after treatment of immunostimulated neurons with PPARγ agonists (Heneka et al. 2000). The fact that NA enhances the expression of PPARγ, which itself seems to be an important factor in the defence mechanisms against neuroinflammation, suggests that noradrenergic depletion in AD may result in a continuous decrease of PPARγ expression leaving the AD brain without an important endogenous anti-inflammatory defence system.

Recently, anti-inflammatory drugs activating PPARγ have been proposed for treatment of AD (Heneka et al. 2000; Landreth and Heneka 2001). The findings of the present study suggest that these substances may be only effective in early stages of the disease with sufficient NA-driven PPARγ-levels in neuronal and glial cells. Once PPARγ levels are reduced under a certain, still unknown level, these substances may become less effective. However, it has been shown that some PPARγ agonists induce the expression of their own receptor in vitro and may therefore prove to be useful in restoring brain PPARγ levels (Gimble et al. 1996; Bernardo et al. 2000). Taken together, our results demonstrated that noradrenaline increases PPARγ expression in neurons and astrocytes, which may contribute to the anti-inflammatory effects of noradrenaline in vivo.

Acknowledgements

We kindly thank E.D. Rosen and B. Spiegelman for the generous gift of mouse embryonic PPARγ knockout fibroblasts. This study was supported by a grant from the Sonderforschungsbereich 400 (Teilprojekt A8) to TK and MTH.

Ancillary