Functional Melatonin Receptors andMetabolic Coupling in Cultured Chick

AstrocytesAKIHITO ADACHI,* ARJUN K. NATESAN,* MELISSA G. WHITFIELD-RUCKER,*

SHANNON E. WEIGUM, AND VINCENT M. CASSONE**Department of Biology, Biological Clocks Program, Texas A&M University,

College Station, Texas

KEY WORDS astrocytes; circadian; melatonin; metabolism

ABSTRACT Melatonin is an important hormone regulating circadian clocks in birds,but the specific cellular sites of action are not completely known. The present study wasdesigned to determine whether astrocytes derived from chick brain contained functionalmelatonin receptors. Primary cell cultures of diencephalon astrocytes that express glialfibrillary acidic protein (GFAP), but not neuron-specific enolase (NSE) immunoreactiv-ity, were employed to determine the cellular distribution and physiological role for thethree known receptor subtypes. Saturation and Scatchard analysis of 2-[125I]iodomela-tonin binding demonstrated melatonin receptor binding with a high affinity and apharmacological profile similar to that obtained from brain. In situ hybridization forreceptor subtypes revealed Mel1A and Mel1C receptor mRNA, but not Mel1B. Adminis-tration of pharmacological levels of melatonin acutely inhibited forskolin-stimulated2-deoxyglucose (2DG) uptake, while rhythmic administration of physiological levels ofmelatonin gradually imposed a rhythm in 2DG uptake and of the release of both lactateand pyruvate into the medium. These results indicate that (1) there are functional Mel1Aand Mel1C melatonin receptors in astrocyte-rich cultures, and (2) rhythmic administra-tion of melatonin plays an important role in the regulation of astrocytic metabolicactivity. Together, the data suggest that the circadian secretion of melatonin probablyplays a role in the global metabolic economy of the avian brain through rhythmicregulation of metabolism in astrocytes. GLIA 39:268–278, 2002. © 2002 Wiley-Liss, Inc.

INTRODUCTION

The pineal gland and the retinae are important com-ponents of the avian circadian system, influencingrhythmic aspects of behavior and physiology via thecircadian secretion of the hormone melatonin (Cassone,1998). Melatonin is produced rhythmically by photore-ceptive pinealocytes within the pineal gland and by theretinal photoreceptor layer during the night, when an-imals are placed in light:dark (LD) cycles; the rhyth-micity persists when animals are maintained underconstant environmental conditions such as continuousdarkness (DD) (Klein et al., 1997; Bernard et al., 1999).In chickens, this rhythm is due primarily to the dailyand circadian regulation of transcription of genes en-coding key enzymes in the melatonin biosyntheticpathway in vivo and in vitro (Bernard et al., 1997).

Thus, the rhythmic secretion of this hormone is a directoutput of the circadian clock of birds (Cassone, 1998;Klein et al., 1997).

Grant sponsor: National Institutes of Health; Grant number: RO1 NS35822;Grant number: PO1 NS39546.

Dr. Adachi is currently at the Department of Anatomy and Neurobiology,Kinki University, Osaka, Japan.

Dr. Natesan is currently at the Laboratory of Cellular and Molecular Regula-tion, Bldg. 36, National Institute of Mental Health, Bethesda, MD.

Dr. Weigum is currently at the Department of Chemistry, University of Texas,Austin, TX.

*These authors contributed equally and are listed alphabetically.

**Correspondence to: Vincent M. Cassone, Department of Biology, Texas A&MUniversity, College Station, TX 77843. E-mail: vmc@mail.bio.tamu

Received 9 January 2002; Accepted 8 May 2002

DOI 10.1002/glia.10109

Published online 15 July 2002 in Wiley InterScience (www.interscience.wiley.com).

GLIA 39:268–278 (2002)

© 2002 Wiley-Liss, Inc.

The sites of melatonin action have been examinedextensively in several avian species by in vitro autora-diography, using a melatonin agonist, 2-[125I]iodomela-tonin (IMEL) (Rivkees et al., 1989; Yuan and Pang,1992; Cassone et al., 1995). Characterization of IMELbinding in chick brain indicates a high-affinity mem-brane bound receptor linked to a Gi GTP-binding pro-tein second messenger system (Rivkees et al., 1989;Cassone and Natesan, 1997; Reppert, 1997; Masanaand Dubocovich, 2001). IMEL binding is widespreadthroughout the avian brain but is particularly dense inretinorecipient and integrative structures involved invision (Rivkees et al., 1989; Cassone and Brooks, 1991;Siuciak and Dubocovich, 1993; Cassone et al., 1995).

Three melatonin receptors, the Mel1A, Mel1B, andMel1C melatonin receptor subtypes, have been isolated,cloned, sequenced, and at least partially characterizedin domestic chickens, the subject of the present study,as well as other vertebrate species (Liu et al., 1995;Reppert et al., 1995; Reppert, 1997). Receptors similarto Mel1A and Mel1B are found in all vertebrate groups,whereas Mel1C has only been found in nonmammalianvertebrates (Reppert, 1997). These receptors havestructural features placing them in a 7-trans-mem-brane domain G-protein-coupled receptor family andexpress similar pharmacological and biochemical char-acteristics. However, these receptor subtypes are dif-ferentially expressed in chick brain (Reppert et al.,1995). Mel1A is expressed by neurons in the chickbrain, particularly in structures important to visualsystem and circadian function (Reppert et al., 1997;Cassone and Natesan, 1997; Natesan and Cassone,2002). Mel1B is expressed primarily in the retina, aswell as in optic tectum and cerebellum, at much lowerlevels of expression (Liu et al., 1995; Natesan andCassone, 2002). Mel1C is widely expressed in chickbrain, especially in areas that contain few neuronal cellbodies, such as the ependyma and optic chiasm (Rep-pert et al., 1995). This distribution has suggested to usthat the Mel1C receptor subtype might be expressed innon-neuronal elements of the central nervous system(CNS), including glial cells. The present study demon-strates that cultured astrocytes isolated from chickdiencephalons express high-affinity melatonin binding,both Mel1A and Mel1C receptor mRNA, but not Mel1BmRNA, and a physiological effect of the hormone. Thesignificance of this expression is discussed in terms ofthe possible role of melatonin in the circadian control ofbrain metabolism.

METHODS AND MATERIALSAstrocyte-Rich Culture Preparation

Glial cell cultures were prepared from embryonic17-day-old (E17) chick diencephalic tissues obtainedfrom white Leghorn eggs purchased from Hy-Line In-ternational (Bryan, TX). Embryos were removed fromthe egg and the brain was dissected out from the skull.Meninges and superficial blood vessels were carefullyremoved, and the diencephalon was placed in Dulbec-

co’s modified Eagle’s medium (DMEM, Sigma; 180 mMglucose). Tissues were manually dissociated and tritu-rated gently in DMEM containing 0.1% trypsin. Cellswere then filtered through nylon mesh (180-�m porediameter). This suspension was centrifuged for 5 minat 1,000g. The supernatant was decanted, and the pel-let was resuspended in DMEM containing 10% fetalbovine serum (FBS; HyClone, Logan, UT) and 1% of anantibiotic cocktail containing 5 mg penicillin, 5 mgstreptomycin, and 5 mg neomycin per ml of buffer. Theresultant cell suspension was filtered again throughnylon mesh (77-�m pore diameter) and seeded into75-cm2 tissue culture flasks.

Cultures were maintained at 37°C with 95% air and5% CO2 in a humidified Napco CO2 incubator. Themedium was changed after 3 days to DMEM containing10% FBS without the antibiotic culture and waschanged everyday thereafter until the cultures hadreached confluence (�10 days).

Astrocyte Culture Immunohistochemistry

A subset of the confluent cultures was replated ontoLab-Tek chamber slides and allowed to grow to near-confluence (4–5 days). Cells were fixed for 10 min in 4%paraformaldehyde in 0.1 M phosphate buffer (pH 7.4)and were then rinsed three times with phosphatebuffer for 15 min each. Then slides were incubated in a1:500 dilution of rabbit anti-glial fibrillary acidic pro-tein (GFAP; Incstar, Stillwater, MN) or rabbit anti-neuron-specific enolase (NSE; Chemicon International,Temecula, CA) in phosphate-buffered saline (PBS) con-taining 1% Triton-X and 5% goat serum (GT) for 48 h at4°C. Slides were then rinsed in phosphate buffer asabove and were incubated in a 1:200 dilution of goatanti-rabbit IgG in PBS-GT for 90 min at room temper-ature. Slides were rinsed three more times in phos-phate buffer and then incubated in avidin-biotin com-plexes conjugated to horseradish peroxidase (HRP) for1 h (Vector Pharmaceuticals). Finally, slides were in-cubated in a solution of diaminobenzamine and hydro-gen peroxide until desired contrast was obtained.

2-[125I]Iodomelatonin Binding

Once cells became confluent after the first passage,DMEM with 10% FBS was decanted from the flasks,and the cells in each were scraped into 50 mM Tris, 1mM EDTA, and 5 mM MgCl2 (homogenate buffer; pH �7.4) with a cell scraper (Falcon). Cells were homoge-nized and centrifuged at 30,000g for 20 min at 4°C.After the supernatant was decanted, pellets werewashed with 500 �l homogenate buffer, centrifuged,and resuspended in 1 ml homogenate buffer. The ho-mogenates were vortexed, and a 20-�l aliquot fromeach sample was reserved for Total Protein Assay(Sigma). After protein determination, homogenateswere aliquoted into 100-�g protein samples. Each sam-

269MELATONIN AND ASTROCYTES

ple was then brought up to a 100-�l final volume by theaddition of homogenate buffer.

To determine specific IMEL binding in cultured cells,a modified version of the radioreceptor assay used byPiketty and Pelletier (1993) was employed. Briefly,100�l of homogenate buffer containing 1 of 9 concen-trations (20, 50, 150, 200, 400, 600, 800, and 1,000 pM)of IMEL (DuPont; 2,000 Ci/mmol) was added to theduplicate samples described above to form 200 �l sam-ples with final IMEL concentration of 10, 25, 75, 100,200, 300, 400, and 500 pM. The same concentrationsand volumes of IMEL were added to a parallel set ofduplicate samples containing 2 nM nonradioactive mel-atonin (for a final concentration of 1 nM), to determinenonspecific binding. These samples were incubated at37°C for 1 h in a shaking water bath. The reaction wasstopped by the addition of 2 ml 30 �g/ml protaminesulfate in 50 mM Tris buffer with 4 mM CaCl2 at 4°C.Samples were vortexed and then centrifuged at30,000g for 20 min at 4°C. The supernatant was de-canted; the samples were dried and counted on a Beck-man Gamma 4000 gamma counter. Saturation curveswere produced from these data, and Scatchard analy-ses were performed. Binding specificity was deter-mined by comparing the relative efficacies of relatedindoles in inhibiting IMEL binding at 100 pM. Samplesfrom culture dishes (100 �g protein) were incubated inthe presence of 10 concentrations of melatonin, 2-io-domelatonin, N-acetylserotonin, 6-hydroxymelatonin,and 5-hydroxytryptamine ranging from 10 mM to 10pM. Incubations were performed, stopped, and ana-lyzed as above.

In Situ Hybridization

cDNA was generated for Mel1A, Mel1B, and Mel1Creceptors using polymerase chain reaction (PCR) toamplify a region of full-length melatonin receptorcDNAs. The primers used were A1-GCCACCATCCT-CATCTTC, A2-GCCGTTCGCTCGCATCCC; B1-TAT-ATGTCATAGCTTTGCCTA, B2-CTTGGCTTTGTTT-CTGACTTG; C1-CCCACTGTCCATCGTGAC, and C2-AGTCCGAAGCATGAGCAG. A 120-bp fragment of theMel1A gene, a 300-bp fragment of the Mel1B gene, anda 350-bp fragment of the Mel1C gene were amplifiedand gel-purified. cDNA was generated from chick optictectum using a gene-specific reverse transcriptase pro-tocol (enzyme and protocol from Stratagene). Thesegenes were ligated into pCR-Script (Invitrogen) andwere transformed into Escherichia coli. T3 and T7 poly-merase reactions were performed using a nonisotopiclabeling mix containing DIG-labeled UTP (Roche Bio-chemical). Cultures were replated onto slides as forimmunohistochemistry. For control chick brain in situhybridizations, chicks at ZT6 (6 h after lights areturned on) were anesthetized with 90 mg/kg ketamineand 10 mg/kg xylazine. They were then transcardiallyperfused with 0.75% saline, followed by 0.1 M sodiumphosphate buffer containing 10% sucrose and were fro-zen in �40°C isopentane. Brains were transversely

sectioned on a cryostat at 20 �m and were mountedonto histological slides.

Slides containing both astrocyte cultures and controlbrain tissues were fixed for 10 min in 4% paraformal-dehyde. The slides were washed in PBS, followed bytwo washes in PBS containing 100 mM glycine, 0.3%Triton-X in phosphate buffer for 15 min, rinsed in PBS,and then by a 10-min acetylation. Slides were prehy-bridized for 2 h in 4� SSC containing 50% formamide.DIG probe was diluted in hybridization buffer (40%formamide, Denhardt’s reagent, 4� SSC, 10 mM DTT,10% dextran sulfate, 1 mg/ml yeast tRNA and 1 mg/mlsalmon testes DNA) and added to slides. Coverslipswere put on the slides, and they were incubated at 50°Cin a moisture chamber overnight. After hybridization,slides were washed for 30 min at 50°C in 5� SSC with10 mM DTT, followed by a 20-min wash at 65°C in asolution containing 50% formamide, 2� SSC, and 10mM DTT. Slides were rinsed twice in Tris buffer fol-lowed by treatment with RNase A at 37°C. Slides werethen washed 4 times at 2� SSC, followed 0.1� SSC at37°C.

Slides were then incubated in blocking reagent (5%goat serum, 0.3% Triton X-100, 1% bovine serum albu-min (BSA) in Tris-Cl buffer) for 30 min at room tem-perature. The slides were then incubated in anti-DIGantibody (Roche Biochemical) diluted 1:1,000 in aTris-Cl buffer containing 2.5% goat serum, 0.3% TritonX-100 for 2 h at 37°C. The slides were then washed andincubated in a color–substrate solution containingNBT, BCIP, and MgCl2 as defined by the manufacturer(Roche Biochemical). After desired contrast was ob-tained, the reaction was stopped by incubating theslides in 0.2 M EDTA.

Acute Effects of Melatonin on2-Deoxy[14C]Glucose Uptake in

Astrocyte-Rich Cultures

First-passage cells were split, replated in 25-cm2 Fal-con tissue culture flasks (4 ml DMEM with 10% FBS),and allowed to grow to confluence (7 days). Flasks (N �5 per treatment) were then exposed to one of six treat-ments: (1) DMSO vehicle; (2) 10 �M melatonin inDMSO; (3) 10 �M forskolin in DMSO; (4) 10 �M fors-kolin plus 10 �M melatonin in DMSO; (5) 10 ng/mlpertussis; and (6) 100 ng/ml pertussis, 10 �M forskolin,plus 10 �M melatonin in DMSO. Cultures were ex-posed to treatments for 15 min before each received 0.2�Ci/ml 2DG for 1 h. After 1 h in 2DG, the DMEM wasdecanted off and reserved for later counting. The cellswere scraped in 2 ml homogenate buffer and centri-fuged at 2,000g for 5 min. The supernatant was de-canted, and the cells were resuspended in 1 ml bufferand sonicated for 30 s. Aliquots (200 �l) of sonicatewere placed in scintillant and counted; 100-�l aliquotswere reserved for total protein assay.

To determine the rates of 2DG uptake and the bio-chemical compartments into which the label was incor-porated in cultured astrocytes, cells were replated in

270 ADACHI ET AL.

25-cm2 Falcon tissue culture flasks (4 ml DMEM withFBS) and allowed to grow confluence as above. Dishes(N � 2 per timepoint) were then rinsed and processedafter 2DG exposure for 15, 30, 45, 60, 90, 120, 150, 180,210, and 240 min by a modification of the methods byNewman et al. (1990) and Banhegy et al. (1997).

The medium was decanted and counted later. Cellswere then scraped as above into 1 ml homogenatebuffer, 500 �l 0.6 M perchloric acid was added to eachsample, and then each sample was sonicated on ice for30 s. Samples were centrifuged at 15,000g for 15 min at4°C, and the protein pellet was reserved for countinglater. The supernatant was then mixed with 5 �lmethyl orange and 100 �l 3 M KOH, bringing the pH to9–10, allowed to precipitate on ice for 15 min and thencentrifuged at 15,000g for 15 min. The perchloratepellet was reserved for counting as the glycogen com-partment. 2-deoxyglucose-6-phosphate (2DG6P) wasseparated from free 2DG by incubating the remainingsupernatant in 500 �l 0.2 M ZnSO4 and then 500 �lsaturated Ba(OH)2, which resulted in the immediateprecipitation of 2DG6P from solution. Samples werecentrifuged as above, and the pellet containing 2DG6Pwas reserved for counting, while 500 �l of the super-natant containing 2DG was placed in scintillation vialsin duplicate. All pellets were eluted in 1 ml homogenatebuffer, and 500 �l of the suspension were placed inscintillation vials in duplicate as above. All sampleswere allowed to dry under a hood before scintillant wasadded, and the samples were counted Beckman scintil-lation counter.

Effects of Rhythmic Administration ofMelatonin on Glucose Metabolism

Once cells became confluent after the first passage,they were exposed to rhythmic administration of mel-atonin by changing feeding medium from a mediumcontaining melatonin (500 pg/ml; 2.15 nM) to a mediumwithout melatonin every 12 h. Zeitgeber time (ZT) 0was defined as the time the medium was changed frommedium containing melatonin to medium without mel-atonin, while ZT 12 was defined as the time mediumwas changed from medium without melatonin to me-dium containing the hormone. The feeding mediumwas the same as described above with the exceptionthat it didn’t contain pyruvate or phenol red, as thesesubstances interfere with the assays. Cells were incu-bated with 0.2 �Ci/ml 2DG for 1 h and collected every4 h at ZT2, ZT6, ZT10, ZT14, ZT18, and ZT22 on the4th and 8th days after the rhythmic administrationhad started. The methods to prepare the pellet and tocount the radioactivity were the same as above. As acontrol, we also examined 2DG uptake of the cells thatwere not exposed to rhythmic administration of mela-tonin.

To determine whether pyruvate and lactate are re-leased rhythmically in response to melatonin, cellswere maintained as above. Medium (N � 6 per time-

point) was collected every 6 h at ZT 6, ZT12, ZT 18, andZT 24 for 2 days, starting on the 3rd and 7th day afterthe beginning of the rhythmic melatonin administra-tion. Measurement of pyruvate and lactate content wasperformed employing an enzymatic colorimetric assaykit (Sigma).

Changes in 2-deoxyglucose uptake and in lactate andpyruvate levels were analyzed by analysis of variance(ANOVA). Newman-Keuls post-hoc analyses were per-formed when statistically significant changes were de-tected.

RESULTSGFAP Expression in

Diencephalon Astrocyte Cultures

Immunocytochemistry was performed on astrocytecultures for GFAP and NSE as shown in Figure 1. Invivo, GFAP is present only in astrocytes throughoutthe brain. In these cultures, approximately 98% of thecells expressed GFAP. NSE is a marker for neuronalcell types and staining was not found in these cultures.

Fig. 1. Immunohistochemistry on astrocyte cultures. Antibodies toeither glial fibrillary acidic protein (GFAP) (top) or nonspecific enolase(NSE) (bottom) were used on astrocyte cultures. GFAP labels morethan 95% of cells in the culture, whereas NSE does not show anysignal. Scale bar � 10 �m.

271MELATONIN AND ASTROCYTES

IMEL Binding in Astrocyte Cultures

Specific high-affinity IMEL binding was observedin these astrocyte-rich cultures. Binding increased

in a concentration-dependent manner; it saturatedat higher concentrations and was reversible with 1nM melatonin (Fig. 2A; nonspecific binding). Scat-chard analysis of this binding showed a high affinity

Fig. 2. Iodomelatonin binding in as-trocyte-rich cultures. A: Astrocyte-rich cultures contain specific, high af-finity of the melatonin agonist2-[125I]iodomelatonin. The saturationcurve shows that binding saturatesand is reversible by incubation withnonradioactive melatonin. Inset: Scat-chard analysis of the slope and x-in-tercept indicated the KD was 86.4 pMand the BMAX was 10.3 fmol/mg pro-tein. B: Competition for iodomelatoninbinding. Competition of 100 pM2-[125I]iodomelatonin by several in-dolyl compounds in astrocyte-rich cul-tures indicates that 2-iodomelatoninand melatonin were equivalent incompeting for binding. 6-hydroxy-melatonin and N-acetylserotonin fol-lowed these compounds in competingfor binding. Serotonin (5-hydroxytryp-tamine) was not effective at �10�4 M.

272 ADACHI ET AL.

with a KD of 86.4 pM, and a BMAX of 10.3 fmol/mgprotein (Fig. 2A inset). These values are similar tovalues obtained from chick brain (Rivkees et al.,1989).

The pharmacological characteristics for inhibition ofspecific IMEL binding closely paralleled similar studiesin chick brain and in receptors expressed in acutelytransfected COS-7 cells (Rivkees et al., 1989; Reppertet al., 1995). The rank order of inhibition of IMELbinding for one site binding models was 2-iodomelato-nin (Ki � 1.6 � 10�10 M) � melatonin (Ki � 4.8 � 10�10

M) � 6-hydroxymelatonin (Ki � 3.8 � 10�7 M) � N-acetylserotonin (Ki � 1.1 � 10�5 M) � 5-hydroxytryp-tamine (Ki � 10�4 M) (Fig. 2B). The Hill coefficient forIMEL binding was calculated to be 0.76, similar to thecoefficient determined for IMEL binding in chick brainof 0.82 (Rivkees et al., 1989). A Hill coefficient of �1suggests more than one binding site, and a two-sitemodel fit, with the data better than a one-site model foriodomelatonin and melatonin in the competition stud-ies. For iodomelatonin, the Ki at the first site was 8.9 �10�11 M, while a second lower affinity site was found at5.8 � 10�7 M. For melatonin, the Ki at the first site was3.3 � 10�10 M, while a second lower affinity site wasfound at 2.6 � 10�7 M, using a two-site model.

In Situ Hybridization

In situ hybridization with DIG-labeled riboprobes forMel1A, Mel1B and Mel1C indicated all astrocytes ex-pressed Mel1C mRNA and a significant subset alsoexpressed Mel1A. The mRNA for the Mel1C receptorsubtype was seen in virtually all of the cultured astro-cytes (Fig. 3E). Mel1A receptor mRNA expression wasalso observed, but in only about 25% of the cells (Fig.3A). Mel1B antisense and sense probe reactions for eachsubtype were negative (Fig. 3C,D). Northern blot anal-ysis and RNase protection assay on astrocyte culturesconfirmed the presence of Mel1A and Mel1C but notMel1B receptor mRNA (Natesan, 2001; data notshown). As a positive control, in situ hybridization forMel1B was performed on chick brain, nucleus rotundus,which contains neurons expressing Mel1B receptormRNA (A.K. Natesan and V.M. Cassone, unpublishedobservations), while hybridization using the senseprobe for Mel1B receptor in brain was negative (Fig.3H,I). Both Mel1A and Mel1C receptor mRNA stainingwithin the astrocyte was primarily in the perinucleararea, demonstrating noticeably clear profiles of the nu-cleus and no hybridization within the astrocytic pro-cesses.

Acute Effects of Melatonin on2-Deoxy[14C]Glucose Uptake in

Astrocyte-Rich Cultures

Administration of 10 �M melatonin inhibited forsko-lin-stimulated 2DG uptake by cultured astrocytes (P �0.01) (Fig. 4A). This effect was reversed by pertussis

toxin. No effect of melatonin was observed if cells werenot previously stimulated with forskolin. Kinetic anal-ysis of melatonin inhibition of forskolin-stimulated2DG uptake by these cells revealed a surprisingly dy-namic effect of the hormone (Fig. 4B). First, forskolininduced a rapid increase in 2DG uptake. This uptakeprofile is paralleled by an increase in labeled 2-deoxy-glucose-6-phosphate (2DG6P). In response to forskolin,much lower incorporation of 14C label was detected inthe glycogen fraction, followed shortly afterward bylabel in the protein fraction, attributed to incorporationinto glycoproteins. Melatonin administration de-creased the rate at which forskolin stimulated 2DG and2DG6P in the cells. Conversely, melatonin increasedforskolin-induced incorporation of 14C into glycogen(Fig. 4B). No effect of melatonin was detected on incor-poration of 14C into the protein fraction.

Effects of Rhythmic Administration ofMelatonin on Glucose Metabolism

Rhythmic administration of physiological, blood lev-els concentrations of melatonin induced 2-DG uptakerhythms, which were in antiphase to melatonin admin-istration; high uptake was observed when melatoninwas not present and low uptake was observed whenmelatonin was present in the medium. By the 4th dayof melatonin administration these rhythms of 2DG up-take were statistically significant (P � 0.01) and by day8 the amplitude of the rhythm increased 2-fold (Fig.5A). The cells that were not exposed to rhythmic ad-ministration of melatonin did not show significant vari-ability indicative of circadian rhythmicity (P � 0.14,data not shown).

Rhythmic administration of melatonin inducedrhythmic release of both lactate and pyruvate in phasewith the rhythm of 2DG uptake (Fig. 5B). Melatoninadministration for 3 to 4 days induced a low-amplituderhythm of both lactate (ANOVA; F (7,40) � 0.955, P �0.47) and pyruvate (F � (7,48) � 5.14, P � 0.01) con-tent such that levels were high when melatonin waswithdrawn and levels were low when melatonin waspresent. The difference between the peak and thetrough (amplitude) of the lactate rhythm was 158.3 �22.1 �M, and the amplitude of the pyruvate rhythmwas 26.3 � 4.3 �M (mean �SEM; n � 6–7 per time-point). The amplitude of the rhythm increased after 8to 9 days of melatonin administration (P � 0.01);378.3� 50.2 �M for the lactate rhythm and 133.4 � 6.8�M for pyruvate (Fig. 5C) (n � 6 per timepoint).

DISCUSSION

The present study demonstrates that astrocytes cul-tured from embryonic chick diencephalon express func-tional Mel1A and Mel1C melatonin receptors with io-domelatonin binding properties biochemically andpharmacologically indistinguishable from those ofchick brain (Rivkees et al., 1989; Reppert et al., 1995)

273MELATONIN AND ASTROCYTES

Fig. 3. In situ hybridization. In situ hybridization was performedon astrocyte cultures, using digoxygenin-labeled riboprobes. Mel1Aantisense probes hybridized to 25% of cells in culture (A) and Mel1Cantisense was observed in virtually every cell in culture (E,G). Mel1Bantisense (C) and sense controls (B,D,F) for all three receptor sub-

types were negative. As Mel1B hybridizations were negative, a posi-tive control for Mel1B was analyzed. The Mel1B probe stains neuronsin the nucleus rotundus of the chick (H), whereas the sense probe isnegative (I). Scale bar � 10 �m.

274 ADACHI ET AL.

(Figs. 1–3). Further, the data show that melatoninprofoundly influences astrocytic metabolism in a waythat is consistent with the in vivo physiological effectsof the hormone (Cassone and Brooks, 1991; Lu andCassone, 1993; Cantwell and Cassone, 2002). That is,in vitro, melatonin decreases glucose uptake, decreasesrelease of both lactate and pyruvate, and marginallyincreases glycogen and glycoproteins content (Figs. 4and 5). In vivo, 2DG uptake is high during the day, lowduring the night, and is inhibited by exogenous mela-tonin (Cassone, 1988; Cassone and Brooks, 1991; Cant-well and Cassone, 2002), which is normally presentduring the night (Klein et al., 1997).

The data are also consistent with the notion that,although melatonin receptor subtypes are very similarbiochemically and pharmacologically, they are ex-pressed by different cell-types and therefore are likelyto serve disparate functions based on the cell-type(s)expressing them. In vivo, areas devoid of neurons ex-press Mel1C receptor mRNA while expression of Mel1Areceptor mRNA is very high in regions containing ahigh density of neuronal cell bodies (Reppert et al.,1995). However, this pattern of expression does notappear to be mutually exclusive, at least in vitro.

Immunohistochemistry for GFAP, which is gliaspecific, indicates that more than 95% of the cells in

Fig. 4. A: Melatonin inhibits forsko-lin-stimulated 2-deoxy-[14C]glucoseuptake by astrocytes. Melatonin(10�M) had no effect on these cells byitself but inhibited forskolin-stimu-lated uptake. This effect of melatoninwas blocked by pertussis toxin, sug-gesting a Gi GTP-binding protein me-diated the effect. Pertussis itself hadno effect. Asterisks indicate signifi-cance at P � 0.01 relative to forskolintreatment. B: Timecourse of the ef-fects of melatonin on forskolin-stimu-lated 2-deoxy[14C]glucose (2DG) up-take. Forskolin (2DG forsk) increased2DG uptake. The rate of uptake wasslowed by melatonin (2DG forsk-mel).There was a similar effect of [14C] in-corporation into 2-deoxyglucose-6-phosphate (2DG6P forsk vs 2DG6Pforsk-mel), but a slight increase in in-corporation of 14C into glycogen storesand/or increase in glycogen turnover(glycogen forsk vs glycogen forsk-mel).There was no effect of melatonin onincorporation of 14C into the proteinfraction.

275MELATONIN AND ASTROCYTES

the culture contain GFAP-expressing astrocytes.Similarly, virtually all of the cultured cells expressMel1C receptor mRNA, so that it is likely all theastrocytes express this subtype. Yet, some 25% of thecells also express Mel1A receptor mRNA. Because allthese cells express GFAP, and none expresses NSE,the most parsimonious view is that these culturedastrocytes primarily express Mel1C, but some expressboth Mel1A and Mel1C. The functional significance of

the coexpression in the same cells is at this point notclear. However, several lines of evidence suggestthese receptors, like many 7-trans-membrane do-main receptors (Maggio et al., 1999; Salahpour et al.,2000), form oligomeric complexes. Preliminary datafrom our laboratory indicate that melatonin receptorprotein may be expressed in monomers, dimers, ortetramers (Karaganis, Bailey, Adachi and Cassone,unpublished observations). This differential expres-

Fig. 5. A: Diurnal changes of the 2-deoxy-[14C]glu-cose uptake by cultured astrocytes under rhythmic ad-ministration of melatonin. A statistically significantrhythm was observed after 4 days of melatonin in vitro(F). The amplitude of the rhythm was increased after 8days of melatonin (E). The medium was changed every12 h from medium with melatonin (black bar) to thatwithout melatonin (white bar). N � 6 per timepoint�SEM. Diurnal changes in the release of pyruvate (B)and lactate (C) from cultured astrocytes under rhythmicadministration of melatonin. A low-amplitude rhythmof both substances was observed after 4 days of melato-nin treatment (F). The amplitude of this rhythm wasdramatically increased on days 7–8. N � 6 per time-point �SEM. *, significance at P � 0.05 relative totimepoints in which melatonin was present; **, P � 0.01significance.

276 ADACHI ET AL.

sion and/or oligomerization may influence second-messenger signaling (Brydon et al., 1999; Conway etal., 2000; Masana and Dubocovich, 2001).

It is now well established that central nervous astro-cytes are metabolically coupled with neurons and thatglucose uptake, phosphorylation and polymerizationinto glycogen stores are dependent on levels of neuro-nal activity (Tsacopoulos and Magistretti, 1996;Wiesinger et al., 1997). Glucose is transported fromcapillaries into astrocytes and is phosphorylated byhexokinase to glucose-6-phosphate (G6P), from whichglycogen synthesis, glycolysis, or the pentose phos-phate pathway may commence. Neurons also possessglucose transporters, but the major metabolic pathwayis oxidative (Sokoloff, 1984). During periods of rela-tively low metabolic demand, G6P is converted to gly-cogen via a three-step process (Wiesinger et al., 1997).When metabolic demands increase, or when astrocytesare stimulated, or both, substances that increase intra-cellular cAMP can increase glycolytic activity, and gly-cogen stores are reduced by glycogenolysis. This pro-cess fuels astrocytic activities, such as the Na/K

ATPase pump involved in regulation of extracellularK concentrations, and enhances conversion of pyru-vate to lactate, which is then exported for neuronalmetabolism (Tsacopoulos and Magistretti, 1996).

Many neuroactive substances affect glial metabolism(Quach et al., 1978; Hara et al., 1989; Tsacopoulos andMagistretti, 1996; Pellerin and Magistretti, 1994). Sub-stances that increase intracellular cAMP concentrationcan increase 2DG uptake and glucose catabolism eitherdirectly or indirectly and can decrease glycogen contentvia increased glycogenolysis. This process involves in-duction of a CCAAT/enhancer binding protein (C/EBP),which may act as a cAMP-inducible early-immediategene in astrocytes, hepatocytes, and adipocytes (Cardi-naux and Magistretti, 1996). Among substances in thiscategory are norepinephrine, acting via -adrenergicreceptors, pituitary adenylate cyclase-activating pep-tide (PACAP), and vasoactive intestinal polypeptide(VIP) (Tsacopolis and Magistretti, 1996).

The present data are consistent with the notion that,in birds, at least, melatonin secreted by the pinealgland and retinae during the night regulates astrocytesand central nervous metabolism via a Gi GTP-bindingprotein in a fashion which is opposite those discussedabove for substances that increase cAMP levels. First,pharmacological concentrations of melatonin preventforskolin-stimulated 2DG uptake by these cells (Fig. 4).Second, physiologically relevant concentrations of thehormone reveal a more subtle effect. That is, rhythmicmelatonin administration can impose a rhythm ofpyruvate and lactate release from these cultures and arhythm of 2DG uptake (Fig. 5). Why this effect takes5–7 days of melatonin administration is not clear. Per-haps, astrocytes contain a damped circadian oscillatorthemselves which are gradually synchronized over thisperiod by the periodic melatonin treatment. Alterna-tively, but not exclusively, receptor-ligand kinetics, re-ceptor-G-protein kinetics and/or receptor-receptor oli-gomerization, which may affect receptor kinetics, may

develop in response to the periodic melatonin adminis-tration.

Diurnal nonmammalian vertebrates exhibit highneuronal activity during the day and low activity dur-ing the night (Cassone, 1988; Lu and Cassone, 1993;Cantwell and Cassone, 2002), while the release of mel-atonin is high during the night and low during the day(Klein et al., 1997). Melatonin may synchronize therhythm of neuronal activity by inhibiting the release oflactate and pyruvate during the night, playing a role inthe regulation of basal metabolism. Rather than releas-ing energy on demand as with excitatory neurotrans-mitters, melatonin may regulate a shift from daytimeglycolytic metabolism, when these diurnal birds aremost active, to a phase of energy storage at night, whenthe animals are usually asleep. In fact, melatonin caninhibit both 2DG uptake and 2DG6P accumulation,whereas 2DG incorporation into glycogen stores orturnover, or both, is increased (Fig. 4).

In conclusion, we have shown that astrocyte-rich cul-tures derived from chick diencephalon have functional,high-affinity melatonin receptors. These are biochemi-cally and pharmacologically similar to those expressedby chick brain and regulate astrocyte metabolism byinhibiting glycolytic activity and shifting energy me-tabolism to a phase of glucose storage. This regulationmay be an important feature of the circadian and di-urnal regulation of brain metabolism.

ACKNOWLEDGMENTS

This research was supported by NIH grants RO1NS35822 and PO1 NS39546 (to V.M.C). Many thanksto Gregg Allen, Elizabeth Cantwell, Barbara Earnest,David Earnest, Jennifer Peters, and Ellen Natesan fortechnical assistance and helpful discussions.

REFERENCES

Banhegyi G, Marcolongo P, Fulceri R, Hinds C, Burchell A, BenedettiA. 1997. Demonstration of a metabolically active glucose-6-phos-phate pool in the lumen of liver microsomal vesicles. J Biol Chem272:13584–13590.

Bernard M, Iuvone PM, Cassone VM, Roseboom PH, Coon SL, KleinDC. 1997. Avian melatonin synthesis: photic and circadian regula-tion of serotonin N-acetyltransferase mRNA in the chicken pinealgland and retina. J Neurochem 68:213–224.

Bernard M, Guerlotte J, Greve P, Grechez-Cassiau A, Iuvone MP,Zatz M, Chong NW, Klein DC, Voisin P. 1999. Melatonin synthesispathway: circadian regulation of the genes encoding the key en-zymes in the chicken pineal gland and retina. Reprod Nutr Dev39:325–334.

Brydon L, Petit L, de Coppet P, Barrett P, Morgan PJ, Strosberg AD,Jockers R. 1999. Polymorphism and signalling of melatonin recep-tors. Reprod Nutr Dev 39:315–324.

Cantwell EL, Cassone VM. 2002. Daily and circadian fluctuation in2-deoxy[14C]glucose uptake in circadian and visual system struc-tures of the chick brain: effects of melatonin. Brain Res Bull 57:603–611.

Cardinaux JR, Magistretti PJ. 1996. Vasoactive intestinal peptide,pituitary adenylate cyclase-activating peptide, and noradrenalineinduce the transcription factors CCAAT/enhancer binding protein(C/EBP)-beta and C/EBP delta in mouse cortical astrocytes: involve-ment in cAMP-regulated glycogen metabolism. J Neurosci 16:919–929.

277MELATONIN AND ASTROCYTES

Cassone VM. 1988. Circadian variation in 2-deoxyglucose uptakewithin the hypothalamic suprachiasmatic region of the house spar-row, Passer domesticus. Brain Res 459:178–182.

Cassone VM. 1998. Melatonin’s role in vertebrate circadian rhythms.Chronobiol Int 15:457–473.

Cassone VM, Brooks DS. 1991. The sites of melatonin action in thehouse sparrow brain. J Exp Zool 260:302–309.

Cassone VM, Natesan AK. 1997. Time and time again: the phylogenyof melatonin as a transducer of biological time. J Biol Rhythms12:489–497.

Cassone VM, Brooks DS, Kelm TA. 1995. Comparative distribution of2[125I]iodomelatonin binding in the brains of diurnal birds: out-group analysis with turtles. Brain Behav Evol 45:241–256.

Conway S, Canning SJ, Howell HE, Mowat ES, Barrett P, Drew JE,Delagrange P, Lesieur D, Morgan PJ. 2000. Characterisation ofhuman melatonin mt(1) and MT(2) receptors by CRE-luciferasereporter assay. Eur J Pharmacol 390:15–24.

Hara M, Matsuda Y, Hirai K, Okumura N, Nakagawa H. 1989. Char-acteristics of glucose transport in neuronal cells and astrocytes fromrat brain in primary culture. J Neurochem 52:902–908.

Klein DC, Coon SL, Roseboom PH, Weller JL, Bernard M, Gastel JA,Zatz M, Iuvone PM, Rodriguez IR, Begay V, Falcon J, Cahill GM,Cassone VM, Baler R. 1997. The melatonin rhythm-generatingenzyme: molecular regulation of serotonin N-acetyltransferase inthe pineal gland. Recent Prog Horm Res 52:307–357.

Liu F, Yuan H, Sugamori KS, Hamadanizadeh A, Lee FJ, Pang SF,Brown GM, Pristupa ZB, Niznik HB. 1995. Molecular and func-tional characterization of a partial cDNA encoding a novel chickenbrain melatonin receptor. FEBS Lett 374:273–278.

Lu J, Cassone VM. 1993. Daily melatonin administration synchro-nizes circadian patterns of brain metabolism and behavior in pinea-lectomized house sparrows, Passer domesticus. J Comp Physiol A173:775–782.

Maggio R, Barbier P, Colelli A, Salvadori F, Demontis G, Corsini G.1999. G protein-linked receptors: pharmacological evidence for theformation of heterodimers. J Pharmacol Exp Ther 291:251–257.

Masana MI, Dubocovich ML. 2001. Melatonin receptor signaling:finding the path through the dark. Science http://stke.scencemag.org/cgi/content/full/sigtrans;2001/107/pe39.

Natesan AK. 2001. Molecular analyses of melatonin receptors in thechick central nervous system: role for multiple receptor sub-types.PhD dissertation. Texas A&M University.

Natesan AK, Cassone VM. 2002. Melatonin receptor sub-type mRNAlocalization and rhythmicity in chick retina. Vis Neurosci (in press).

Newman GC, Hospod FE, Patlak CS. 1990. Kinetic model of 2-deoxy-glucose metabolism using brain slices. J Cereb Blood Flow Metab10:510–526.

Pellerin L, Magistretti PJ. 1994. Glutamate uptake into astrocytesstimulates aerobic glycolysis: a mechanism coupling neuronal ac-tivity to glucose utilization. Proc Natl Acad Sci U S A 91:10625–10629.

Piketty V, Pelletier J. 1993. Melatonin receptors in the lamb parstuberalis/median eminence throughout the day. Neuroendocrinol-ogy 58:359–365.

Quach TT, Rose C, Schwartz JC. 1978. [3H]Glycogen hydrolysis inbrain slices: responses to neurotransmitters and modulation of nor-adrenaline receptors. J Neurochem 30:1335–1341.

Reppert SM. 1997. Melatonin receptors: molecular biology of a newfamily of G protein-coupled receptors. J Biol Rhythms 12:528–531.

Reppert SM, Weaver DR, Cassone VM, Godson C, Kolakowski LF Jr.1995. Melatonin receptors are for the birds: molecular analysis oftwo receptor subtypes differentially expressed in chick brain. Neu-ron 15:1003–1015.

Rivkees SA, Cassone VM, Weaver DR, Reppert SM. 1989. Melatoninreceptors in chick brain: characterization and localization. Endocri-nology 125:363–368.

Salahpour A, Angers S, Bouvier M. 2000. Functional significance ofG-protein-coupled receptors. Trends Endocrinol Metab 11:163–168.

Siuciak JA, Dubocovich ML. 1993. Effect of pinealectomy and thelight/dark cycle on 2-[125I]iodomelatonin binding in the chick optictectum. Cell Mol Neurobiol 13:193–202.

Sokoloff L. 1984. Metabolic probes of central nervous system activityin experimental animals and man. Sunderland, MA: Sinauer.

Tsacopoulos M, Magistretti PJ. 1996. Metabolic coupling between gliaand neurons. J Neurosci 16:877–885.

Wiesinger H, Hamprecht B, Dringen R. 1997. Metabolic pathways forglucose in astrocytes. Glia 21:22–34.

Yuan H, Pang SF. 1992. [125I]iodomelatonin binding sites in thechicken brain: diurnal variation and effect of melatonin injection orpinealectomy. Biol Signals 1:208–18.

278 ADACHI ET AL.