Proc. Natl. Acad. Sci. USAVol. 92, pp. 3948-3952, April 1995Neurobiology

D-Serine, an endogenous synaptic modulator: Localization toastrocytes and glutamate-stimulated release

(N-methyl-D-aspartate/glycine site/kainate/D-amino acid oxidase/colloidal gold)

MICHAEL J. SCHELL, MARK E. MOLLIVER, AND SOLOMON H. SNYDER*Departments of Neuroscience, Pharmacology and Molecular Sciences, and Psychiatry, Johns Hopkins University School of Medicine, 725 North Wolfe Street,Baltimore, MD 21205

Contributed by Solomon H. Snyder, January 6, 1995

ABSTRACT Using an antibody highly specific for D-serine conjugated to glutaraldehyde, we have localized endog-enous D-serine in rat brain. Highest levels of D-serine immu-noreactivity occur in the gray matter of the cerebral cortex,hippocampus, anterior olfactory nucleus, olfactory tubercle,and amygdala. Localizations of D-serine immunoreactivitycorrelate closely with those of D-serine binding to the glycinemodulatory site of the N-methyl-D-aspartate (NMDA) recep-tor as visualized by autoradiography and are inversely corre-lated to the presence of D-amino acid oxidase. D-Serine isenriched in process-bearing glial cells in neuropil with themorphology of protoplasmic astrocytes. In glial cultures of ratcerebral cortex, D-serine is enriched in type 2 astrocytes. Therelease of D-serine from these cultures is stimulated byagonists of non-NMDA glutamate receptors, suggesting amechanism by which astrocyte-derived D-serine could mod-ulate neurotransmission. D-Serine appears to be the endoge-nous ligand for the glycine site of NMDA receptors.

D amino acids have long been known to play a prominent role inbacterial physiology. There have been sporadic reports of endog-enous D amino acids in animals, but mostly in invertebrates (1).Recently several groups have reported the existence of D-serinein serum (2), urine (3), and brain (4, 5) of mammals, includinghumans (6). Hashimoto et at (7) showed with biochemicaltechniques that the distribution of D-serine in rat brain regionsparallels the distribution of N-methyl-D-aspartate (NMDA) sub-types of glutamate receptors. This is of interest because stimu-lation ofNMDA receptors by glutamate requires the binding ofglycine (8, 9), and D-serine mimics glycine at these sites (10).To ascertain whether endogenous D-serine plays a physio-

logic role in neurotransmission, we have developed polyclonalantibodies that are selective for D-serine. We now report alocalization of D-serine throughout rat brain that closelyparallels NMDA receptors and is concentrated in gray-matterastrocytes of the telencephalon. We also demonstrate thatD-serine is released from type 2 astrocyte cultures stimulatedwith agonists of non-NMDA glutamate receptors.

MATERIALS AND METHODSMaterials. D-[3H]Serine (22 Ci/mmol; 1 Ci = 37 GBq) was

provided by Steven Hurt (New England Nuclear). Antibody toglial fibrillary acidic protein (GFAP) was from Dako, andantibody to glutamine synthetase was from Chemicon. Glu-taraldehyde was from EM sciences. Gold chloride was fromAldrich. Alkaline phosphatase-coupled anti-rabbit antibodywas from The Jackson Laboratories. The peroxidase Elitestaining kit was from Vector Laboratories. Glutamate receptordrugs were from Research Biochemicals (Natick, MA). Cell

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement" inaccordance with 18 U.S.C. §1734 solely to indicate this fact.

culture reagents were from GIBCO/BRL. All other reagentswere from Sigma.

Preparation of Polyclonal Antiserum to D-Serine. D-Serinewas coupled to bovine serum albumin (BSA) with glutaralde-hyde and then reduced with NaBH4 (11). After extensivedialysis against water, the conjugate was adsorbed to freshlyprepared 45-nm colloidal gold particles (12). A rabbit wasimmunized intradermally every 3 weeks with the BSA conju-gate alone and intravenously with the gold particles. Beforeuse, all D-serine antiserum used in this study was incubated for2 hr at room temperature with Sepharose beads coupled toglutaraldehyde-treated BSA, to eliminate antibodies not se-lective for D-serine (13). Liquid-phase conjugates of variousamino acids to glutaraldehyde were prepared identically to theoriginal immunogen, except that BSA was omitted and freealdehyde groups were blocked with excess Tris. For dot blotscreens, various amino acids were coupled to dialyzed rat braincytosol with glutaraldehyde as described above for D-serine/BSA conjugates and then spotted on nitrocellulose. Afterovernight incubation with the primary antiserum, blots werevisualized with an alkaline phosphatase-coupled anti-rabbitsecondary antibody. High-affinity stereoselective antibodiesappeared after 5 months of immunization.

Immunohistochemistry. Anesthetized rats (age, 45-50 days)were perfused through the aorta for 30 sec with 37°C oxygen-ated Krebs/Henseleit buffer and then at 60 ml/min with 500ml of 37°C 5% glutaraldehyde/0.5% paraformaldehyde con-taining 0.2% Na2S2O5 in 0.1 M sodium phosphate (pH 7.4).Brains were postfixed in the same buffer for 2 hr at roomtemperature. After cryoprotection for 2 days at 4°C in 50 mMsodium phosphate, pH 7.4/0.1 M NaCl/20% (vol/vol) glyc-erol, brain sections (20-40 ,um) were cut on a sliding micro-tome. Free-floating brain sections were reduced for 20 minwith 0.5% NaBH4 and 0.2% Na2S2O5 in Tris-buffered saline(TBS: 50 mM Tris HCl, pH 7.4/0.15 M NaCl), washed for 45min at room temperature in TBS containing 0.2% Na2S2O5,blocked with 4% normal goat serum for 1 hr in the presenceof 0.2% Triton X-100, and incubated overnight at 4°C with theD-serine antiserum diluted 1:5000 to 1:1000 in TBS containing2% goat serum and 0.1% Triton X-100. Immunoreactivity (IR)was visualized with the Vectastain ABC Elite kit. To testimmunohistochemical specificity, liquid-phase conjugates ofglutaraldehyde and L-serine or D-serine (0.2 mM amino acid)were incubated overnight with diluted antibody (1:1000) be-fore incubation with brain sections. Immunohistochemistrywas routinely done in the presence of the L-serine conjugate.

D- [3H] Serine Autoradiography. Fresh-frozen brain sections(20 ,Lm) on slides were incubated with 100 nM D-[3H]serine for45 min on ice in 50 mM Tris citrate (pH 7.4) as described forstrychnine-insensitive glycine autoradiography (14). The pat-

Abbreviations: IR, immunoreactivity; NMDA, N-methyl-D-aspartate;GFAP, glial fibrillary acidic protein; BSA, bovine serum albumin;AMPA, a-amino-3-hydroxy-5-methyl-4-isoxazolepropionate.*To whom reprint requests should be addressed.

3948

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

20, 2

020

Proc. Natl. Acad. ScL USA 92 (1995) 3949

tern produced with D-[3H]serine strongly resembles the patternobtained for [3H]glycine binding. Specificity of binding waschecked in tissue wipe tests; the glycine-site antagonist 5,7-dichlorokynurenic acid (0.01 mM) displaced 75% of boundlabel.D-Amino Oxidase Enzyme Histochemistry. Anesthetized

rats were perfused at 4°C with 500 ml of2% paraformaldehydein 0.1 M phosphate buffer (pH 7.4) and postfixed for 2 hr at4°C. Free-floating tissue sections were incubated in the histo-chemistry medium described by Horiike et al (15).Primary Astrocyte Cultures. Mixed glial cultures were

prepared from rat cerebral cortex at postnatal day 1-2 (16).After 10-12 days, the cultures were shaken for 1 hr at 37°C todeplete microglia and then shaken overnight to dislodge apopulation of type 2 astrocytes from the underlying layer oftype 1 astrocytes. The dislodged cells were replated and grownfor 2-5 days before use. For immunocytochemistry, cells werewashed twice in phosphate-buffered saline and then fixed with5% glutaraldehyde in 0.1 M phosphate buffer (pH 7.4) for 1 hrat room temperature. In some experiments, cells were exposedto 50 AM kainate for 30 min before fixation.

D- [3H] Serine Uptake and Release. Type 2 astrocyte-enriched cultures were plated on poly(D-lysine)-coated 1.5-cm-diameter multiwell dishes and grown for 3-5 days. Releaseexperiments were done as described by Levi and Patrizio (17).In brief, each well was washed twice in 1 ml of oxygenatedKrebs-Ringer medium and then incubated for 15 min at 37°Cwith 0.5 ,uCi of D-[3H]serine in 0.5 ml. Cells were then exposedcontinuously to glutamate-related drugs, and release of radio-activity into the medium was measured. Released radioactivitycould be oxidzed by D-amino acid oxidase, confirming that ithad not been metabolized by the cells.

RESULTSTo develop an antiserum to D-serine as it would occur in fixedbrain tissue, we immunized rabbits with a glutaraldehydeconjugate of D-serine and BSA. To ascertain the sensitivity andspecificity, we coupled various amino acids to dialyzed ratbrain cytosol with glutaraldehyde and spotted the preparationsonto nitrocellulose. Dot blots were probed with a 1:5000dilution of the D-serine antibody in the presence or absence of0.5 mM D- or L-serine glutaraldehyde conjugate (Fig. 1). Theantibody detects 0.01 nmol of D-serine but not 0.001 nmol. Theantibody is about 100-fold less sensitive to L-serine than toD-serine and shows even less IR to glycine, L-glutamate,L-alanine, and D-aspartate, with no reactivity to y-aminobutyrate.

Ab Alone Ab + L-ser-GA Ab + D-ser-GAC-GAD-ser-C-GA * *0L-ser-C-GA v0gly-C-GAL-glu-C-GAGABA-C-GAL-ala-C-GAD-asp-C-GA

-12 -11-10 -9 -8 -12 -11-10 -9 -8 -12 -11-10 -9 -8

log Amino Acid (moles)

FIG. 1. D-Serine antibody (Ab) specificity. Dialyzed rat braincytosol (C) was coupled to various amino acids with glutaraldehyde(GA) and then reduced with NaBH4. Serial 1:10 dilutions of eachconjugate were spotted onto three pieces of nitrocellulose and thenprobed with a 1:5000 dilution of antibody. (Left) IR of the antibodyalone. (Center) IR when 500 ,uM liquid-phase L-serine glutaraldehydeconjugate is included in the incubation. (Right) IR when the sameconcentration of D-serine glutaraldehyde conjugate is included in-stead.

Preabsorption of the antibody with 0.5 mM L-serine glutaralde-hyde conjugate (Fig. 1 Center) abolishes the small amounts ofcrossreactivitywith other amino acids but diminishes sensitivity toD-serine minimally; preabsorption with the same concentration ofD-serine glutaraldehyde conjugate eliminates IR (Fig. 1 Left).Likewise, in tissue sections, the D-serine conjugate eliminates IR(data not shown).Immunohistochemical staining of rat brain reveals highly

selective localizations of D-serine (Fig. 2 Top). IR is concen-trated in the telencephalon, with very low levels in the hind-brain. The most intense IR occurs in the anterior olfactorynuclei, olfactory tubercle, hippocampus, rostral cerebral cortex(especially superficial layers), and corpus striatum. Intense IRalso occurs in the amygdala (data not shown). IR occursthroughout all neuropil in the hippocampus and in the hilus ofthe dentate gyrus, but is much less in the pyramidal cell layerand is absent from the granule cell layer of the dentate gyrus.Substantial IR occurs in the thalamus and hypothalamus. Themidbrain, pons, and medulla have almost negligible IR. In thecerebellar molecular layer, IR is lower than in the forebrain butis clearly present relative to other parts of the cerebellum,which are unstained.The distribution of endogenous D-serine strongly resembles

that ofNMDA receptors labeled by the binding of D-[3H]serineto the glycine recognition site of the NMDA receptor (Fig. 2Middle). D-Serine binding is highest in the CAl region of thehippocampus in a layered pattern, with the most intenselabeling occurring in the strata radiatum and oriens and verylittle binding in the pyramidal cell layer, in agreement with

FIG. 2. Comparison of D-serine IR (Top), D-[3H]serine binding(Middle), and D-amino (D-AA) oxidase enzymatic activity (Bottom) insagittal sections of adult rat brain. White areas represent positivestaining. Preabsorption of D-serine antiserum with D-serine glutaral-dehyde conjugate but not L-serine glutaraldehyde conjugate abolishesIR (data not shown). AON, anterior olfactory nuclei; C, cerebralcortex; GR, granule cell layer of cerebellum; H, hippocampus; MOL,molecular layer of cerebellum; OT, olfactory tubercule; S, striatum.

Neurobiology: Schell et aL

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

20, 2

020

Proc. Natl. Acad. Sci. USA 92 (1995)

previous reports for glycine binding (14). Binding is also highin the lower blade of the dentate gyrus but is low in the granulecell layer. Substantial binding occurs in anterior olfactorynuclei, olfactory tubercule, rostral cerebral cortex, and corpusstriatum. High D-serine binding occurs in the thalamus, wherelevels of endogenous D-serine are only moderate. The mid-brain, pons, and medulla show negligible levels of binding. Thepattern of D-serine binding differs from endogenous D-serinein the cerebellum, where D-serine binding occurs in the granulecell layer with little or none in the molecular layer.

Horiike et at (15, 18, 19) have observed that D-amino acidoxidase is concentrated in astrocytes of the hindbrain andcerebellum. We confirm these findings. The localization ofD-amino acid oxidase is almost exactly the inverse of D-serine,and no D-amino acid oxidase is detected in the telencephalon.The molecular layer of the cerebellum is a special case; notonly does it possess the highest levels of D-amino acid oxidaseactivity in brain, but it is also the only region where D-serineand D-amino acid oxidase appear colocalized.The presence of endogenous D-serine in telencephalic gray

matter enriched in NMDA receptors suggested that D-serinemight exist in neurons. We were surprised to observe underhigh magnification that D-serine IR was associated exclusivelywith glia. We observe no D-serine-immunopositive neurons inany brain region (Fig. 3). D-Serine-positive glia have small cellbodies and are bipolar or multipolar. IR is concentrated in cellbodies and proximal processes and appears punctate and intra-

cellular. In neuropil, IR is present not only in individual cells butalso appears to fill most of the extraneuronal space. In the lowerblade of the dentate gyrus and in layer 1 of the cerebral cortex,D-serine-positive cells orient their processes toward the pialsurface. In the frontal cortex, immunostained cells are oftenoriented radially along pyramidal cell processes, while in the deepcortex and striatum, D-serine cells have no obvious pattern. Infiber tracts of the telencephalon, numerous intensely labeled cellsstand out against an unstained background. Immunoreactive cellsare numerous around blood vessels, suggesting that they areastrocytes. In double-labeling experiments (data not shown), weobserve D-serine-immunopositive cells which are also positive forthe astrocytic markers GFAP or glutamine synthetase. However,a substantial portion of D-serine-immunopositive glial cells areGFAP negative.To determine which glial types make D-serine, we prepared

glial cultures from cerebral cortex (16). In primary mixed glialcultures, type 1 astrocytes are polygonal and form a confluentlayer on the culture dish; process-bearing type 2 astrocytes,oligodendrocytes, 02A progenitors, and microglia grow ontop. Many cells in this top layer display strong D-serine IR,while the underlying type 1 cells display a weaker but stilldetectable IR. Enriched fractions of type 2 astrocytes areobtained by overnight shaking of the culture flask. Almost allof these cells stain positively for D-serine (Fig. 4). About halfof these cells are positive for the astrocytic marker GFAP, andmost are positive for glutamine synthetase (data not shown).

CONTROL

FIG. 3. D-Serine IR in the telencephalon of 50-day-old rats. Darkareas indicate positive IR. (a) In the hippocampus high levels of IR arepresent in all the neuropil and in the hilus of the dentate gyrus. (b) Inthe striatum IR is present in neuropil; strongly stained astrocytes(arrows) appear frequently around blood vessels (diamond). (c) Detailof the lower blade of the dentate gyrus. Arrows indicate stronglystained astrocytes, especially near blood vessels (diamonds), whereasgranule neurons are unstained. (d) In the cerebral cortex, staining ishighest in superficial layers (I and II), and neurons appear as negativeimages. (e) Higher magnification of the dentate granule cells empha-sizes D-serine-immunopositive glial process (arrows) surrounding un-labeled neuronal cell bodies. (f) Higher magnification of cerebralcortex emphasizes IR that is often radially oriented and surroundingpyramidal neuron cell bodies and processes (arrowheads) and bloodvessels (diamond). G, granule cell layer; H, hilus of dentate gyrus; L,stratum lacunosum; MOL, molecular layer of dentate gyrus; 0,stratum oriens; R, stratum radiatum. (Bars = 100 ,tm in a and c; 50mum in b, d, andf; and 10 ,um in e).

+ KAINATE

FIG. 4. D-Serine IR in type 2 astrocyte cultures from rat cerebralcortex. (Upper) IR is concentrated in cell bodies and in proximalprocesses. (Lower). IR is substantially reduced after treatment ofidentically prepared cultures with 50 ,uM kainate for 30 min. (Bar =20 gm.)

3950 Neurobiology: Schell et at

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

20, 2

020

Proc. NatL Acad Sci. USA 92 (1995) 3951

Levi et at (20) have reported that various glutamate derivativescan stimulate the release of 'y-aminobutyrate and other endog-enous L amino acids, including serine, from cortical type 2astrocyte cultures, with kainate being the most potent releaser.We labeled type 2 astrocytic cultures with D-[3H]serine andmonitored release of radioactivity following stimulation of thecultures (Fig. 5). Depolarizing with 40 mM KCl fails to releaseD-[3H]serine, similar to the results reported for L-serine (17).NMDA (100 ,uM) also fails to release D-serine. In contrast,kainate (100 ,uM) markedly augments release of D-[3H]serinefrom type 2 astrocytes. The non-NMDA agonists AMPA andquisqualate also augment D-[3H]serine release but are less effi-cacious than kainate. 6-Cyano-7-nitroquinoxaline-2,3-dione, an

antagonist at non-NMDA receptors, completely prevents thestimulation produced by kainate and AMPA. To determinewhether the endogenous pool of D-serine can be released by thesame stimuli that release preloaded D-[3H]serine, we exposedtype 2 astrocytes to 50 ,uM kainate for 30 min and then fixed thecells and probed with the D-serine antibody (Fig. 4). In compar-ison with identical cultures of unstimulated cells, we observe a

substantial reduction of IR in the kainate-treated cultures. Thedifference between stimulated and unstimulated cultures is ap-parent upon gross examination of the immunolabeled slides, evenwithout microscopy.

DISCUSSIONThe main findings of this study are that D-serine is localized tomultipolar glia in areas of the brain enriched in NMDAreceptors and that D-serine release from type 2 astrocytes isselectively enhanced by non-NMDA glutamate receptor ago-nists. Glycine is often called a "coagonist" with glutamate atthe NMDA receptor, since occupation of the glycine site isabsolutely required to gate the channel (10, 21). In theiroriginal description of glycine's effect on NMDA receptors,Johnson and Ascher (8) suggested that endogenous levels ofglycine suffice to saturate the glycine site, so that receptorstimulation with a nerve impulse involves only glutamate. Onthe other hand, several investigators have shown that admin-istration of glycine-site agonists, including D-serine, stimulate

Control

KCI, 40 mM

Glu, 100 FM

Glu, 300 FM

KA, 10lM

KA, 100 FM

KA, 100 FM

AMPA, 100 jM

AMPA, 100 jiM

QA, 100 iLM

NMDA, 100 jM

0 5 10 15 20 25 30

Percent 1H]D-Ser Released

FIG. 5. D-[3H]Serine efflux from type 2 astrocyte cultures isstimulated by non-NMDA receptor agonists. Cells were preloadedwith D-[3H]serine for 20 min and then exposed continuously to variousstimuli. Data are the means of two to four experiments, each intriplicate, whose values varied <10%. Basal (control) efflux (mean ±

SD) was 8.3 ± 1% of loaded counts, and efflux after 100 ,uM kainateisoxazole was 25.6 ± 5.4%. AMPA, a-amino-3-hydroxy-5-methyl-4-propionate; CNQX, 6-cyano-7-nitroquinoxaline-2,3-dione; KA, kain-ate; QA, quisqualate.

NMDA receptors in vivo, indicating that the sites are notnormally saturated (22-24).We suggest that the predominant endogenous ligand for the

glycine site is D-serine and not glycine. D-Serine levels in cerebralcortex (5, 7) are about a third of glycine levels (7, 25). D-Serineis concentrated in gray-matter astrocytes near NMDA receptors.In contrast, the regional distribution of glycine in the telenceph-alon does not match NMDA receptors. Immunohistochemicalstudies using antibodies specific for glycine (11,26, 27) report verylow IR in hippocampus and cortex, regions with the highestdensities of NMDA receptors. These localizations contrast withglycine IR in the brainstem and spinal cord, which occurs in aneuronal pattern consistent with glycine's role as an inhibitoryneurotransmitter (28-30). In terms of its overall localizationthroughout the brain, the distribution of glycine is virtually theopposite of NMDA receptors (25).

In the adult cerebellum, D-serine and NMDA receptors donot colocalize; D-serine is only in the molecular layer, while theD-serine binding site is only in the granule cell layer. Underhigh magnification, D-serine in the cerebellum occurs selec-tively in Bergmann glia, where it is colocalized with D-aminoacid oxidase (M.J.S. and S.H.S., unpublished data). Duringpostnatal week 2, when D-serine develops, Bergmann gliaprovide a scaffolding for the migration of granule cells fromthe external granule layer to the internal granule layer. Thisprocess involves NMDA receptors, as migration is enhanced byapplication of glycine and inhibited by NMDA receptor an-

tagonists (31). Perhaps D-serine released by Bergmann gliamodulates granule cell migration.

Astrocytes in vivo are traditionally categorized as fibrous orprotoplasmic (32). Fibrous astrocytes are so designated be-cause they are found mainly in white matter and contain manyfilaments which stain positively for GFAP. Protoplasmic as-

trocytes have multipolar processes, are found mainly in graymatter, and express lower levels of GFAP. D-Serine astrocytesin the adult rat in vivo more closely resemble protoplasmicastrocytes, although numerous strongly stained cells are

present in white matter as well. D-Serine astrocytes resembletype 2 astrocytes in culture, as they share a multipolar shapeand postnatal expression (33). Indeed, the D-serine antibodylabels type 2 astrocytes intensely. Some D-serine cells areGFAP negative, similar to some type 2 astrocytes and all oftheir progenitors (02A cells). Type 2 astrocytes express highlevels of ionotropic non-NMDA glutamate receptors, of boththe kainate- and AMPA-preferring varieties (34-37). Ourresults show that type 2 astrocytes release D-serine whenstimulated with non-NMDA glutamate receptor agonists.These data are consistent with D-serine release occurringthrough a carrier-mediated mechanism (20). In this model thedepolarization of nerve terminals releases glutamate, which inturn stimulates ionotropic glutamate receptors on type 2astrocytes. The influx of Na+ may allow Na+-dependent aminoacid transporters to function in reverse (Fig. 6).At any synapse where the glycine site is not fully occupied,

activation of NMDA receptors may be a coordinated processwhereby glutamate released by the presynaptic neuron relievesthe Mg2+ block and binds to the NMDA site, and then D-serinereleased from astrocytes binds to the glycine site to gate thechannel. Since D-serine uptake in brain occurs poorly relativeto L-serine or glycine uptake (M.J.S. and S.H.S., unpublisheddata), and no known enzymes can metabolize D-serine in thetelencephalon, D-serine may remain in the synaptic spacelonger than conventional neurotransmitters.By analogy with bacterial enzymes, most mechanisms pro-

posed for the biosynthesis of D-serine in worms and insectsinvolve a racemase (1). A particularly tempting possibility forD-serine biosynthesis in brain involves the glycine cleavagepathway (38), a multienzyme complex localized to the mito-chondria and capable of transforming glycine to serine. Theregional distribution of glycine cleavage activity in the brain

+ CNQX, 30 pLM

+ CNQX, 30 tiM

I I I I

Neurobiology: Schell et at

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

20, 2

020

Proc. Natl. Acad. Sci. USA 92 (1995)

NMDA-R

FIG. 6. Model depicting the role of astrocyte-derived D-serine inneurotransmission at glutamatergic synapses. NMDA-R, NMDA re-

ceptor.

resembles that of D-serine (38), and immunohistochemicalmapping of the P-protein component of the glycine cleavagecomplex in brain reveals localizations to the mitochondria ofprocess-bearing astrocytes in the telencephalon (39). Furtherstudy may reveal whether the glycine cleavage pathway isstrictly chiral.

We thank Elizabeth O'Hearn for helpful discussions. This work was

supported by Public Health Service Grants MH18501 and DA00266,Research Scientist Award DA00074 to S.H.S., and a grant of the W.M. Keck Foundation. M.E.M. was supported by Public Health ServiceResearch Grant DA044331 from the National Institute on DrugAbuse.

1. Corrigan, J. J. (1969) Science 164, 142-149.2. Nagata, Y., Konno, R., Yasumura, Y. & Akino, T. (1989)

Biochem. J. 257, 291-292.3. Bruckner, H., Haasmann, S. & Friedrich, A. (1994)Amino Acids

6, 205-211.4. Hashimoto, A., Nishikawa, T., Hayashi, T., Fujii, N., Harada, K.,

Oka, T. & Takahashi, K. (1992) FEBS Lett. 296, 33-36.5. Nagata, Y., Horiike, K. & Maeda, T. (1994) Brain Res. 634,

291-295.6. Hashimoto, A., Kumashiro, S., Nishikawa, T., Oka, T., Taka-

hashi, K., Mito, T., Takashima, S., Doi, N., Mizutani, Y.,

Yamazaki, T., Kaneko, T. & Ootomo, E. (1993)J. Neurochem. 61,783-786.

7. Hashimoto, A., Nishikawa, T., Oka, T. & Takahashi, K. (1993) J.Neurochem. 60, 783-786.

8. Johnson, J. W. & Ascher, P. (1987) Nature (London) 325, 529-531.

9. Reynolds, I. J., Murphy, S. N. & Miller, R. J. (1987) Proc. Natl.Acad. Sci. USA 84, 7744-7748.

10. Kemp, J. A. & Leeson, P. D. (1993) Trends Pharmacol. Sci. 14,20-25.

11. Campistron, G., Buijs, R. M. & Geffard, M. (1986) Brain Res.376, 400-405.

12. Pow, D. V. & Crook, D. K. (1993)J. Neurosci. Methods 48, 51-63.13. Ottersen, 0. P., Storm-Mathisen, J., Madsen, S., Skumlien, S. &

Stromhaug, J. (1986) Med. Biol. 64, 147-158.14. McDonald, J. W., Penney, J. B., Johnston, M. V. & Young, A. B.

(1990) Neuroscience 35, 653-668.15. Horiike, K., Arai, R., Tojo, H., Yamano, T., Nozaki, M. & Maeda,

T. (1985) Acta Histochem. Cytochem. 18, 539-550.16. Levison, S. W. & McCarthy, K. D. (1991) in Culturing Nerve Cell,

eds. Banner, G. & Goslin, K. (MIT Press, Cambridge, MA), pp.309-336.

17. Levi, G. & Patrizio, M. (1992) J. Neurochem. 58, 1943-1952.18. Horiike, K., Tojo, H., Arai, R., Yamano, T., Nozaki, M. & Maeda,

T. (1987) Brain Res. Bull. 19, 587-596.19. Horiike, K., Tojo, H., Arai, R., Nozaki, M. & Maeda, T. (1994)

Brain Res. 652, 297-303.20. Levi, G., Gallo, V. & Patrizio, M. (1992) in Progress in Brain

Research, eds. Yu, A. C. H., Hertz, L., Norenberg, M. D., Sykova,E. & Waxman, S. G. (Elsevier, Amsterdam), pp. 243-250.

21. Thomson, A. M. (1990) Prog. Neurobiol. 35, 53-74.22. Singh, L., Oles, R. J. & Tricklebank, M. D. (1990) Br. J. Phar-

macol. 99, 285-288.23. Wood, P. L., Emmett, M. R., Tadimeti, S. R., Mick, S., Cler, J. &

Iyengar, S. (1989) J. Neurochem. 53, 979-981.24. Thiels, E., Weisz, D. J. & Berger, T. W. (1992) Neuroscience 46,

501-509.25. Aprison, M. H., Shank, R. P. & Davidoff, R. A. (1969) Comp.

Biochem. Physiol. 28, 1345-1355.26. van den Pol, A. N. & Gorcs, T. (1988) J. Neurosci. 8, 472-492.27. Pourcho, R. G., Goebell, D. J., Jojich, L. & Hazlett, J. C. (1992)

Neuroscience 46, 643-656.28. Snyder, S. H., Young, A. B., Bennett, J. P. & Mulder, A. H.

(1973) Fed. Proc. Fed. Am. Soc. Exp. Biol. 32, 2039-2047.29. Curtis, D. R. & Johnston, G. A. (1970) Nature (London) 225,

1258-1259.30. Werman, R., Davidoff, A. & Aprison, M. H. (1967) Nature

(London) 214, 681-683.31. Komuro, H. & Rakic, P. (1993) Science 260, 95-97.32. Privat, A. & Rataboul, P. (1986) in Astrocytes, ed. Fedorff, S.

(Academic, New York), pp. 105-129.33. Raff, M. C. (1989) Science 243, 1450-1455.34. Bowman, C. L. & Kimelberg, H. K. (1984) Nature (London) 311,

656-659.35. Usowicz, M. M., Galla, V. & Cull-Candy, S. G. (1989) Nature

(London) 339, 380-383.36. Wyllie, D. J. A., Mathie, A., Symonds, C. J. & Cull-Candy, S. G.

(1991) J. Physiol. (London) 432, 235-258.37. Patneau, D. K., Wright, P. W., Winters, C., Mayer, M. L. & Gallo,

V. (1994) Neuron 12, 357-371.38. Daly, E. C., Nadi, N. S. & Aprison, M. H. (1976) J. Neurochem.

26, 179-185.39. Sato, K., Yoshida, S., Fujiwara, K., Tada, K. & Tohyama, M.

(1991) Brain Res. 567, 64-70.

3952 Neurobiology: Schell et al

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

20, 2

020