A

bntvanc

ela©

K

1

ahh

mdf

0d

Cell Calcium 41 (2007) 221–234

Cellular and subcellular localization of Na+–Ca2+ exchanger proteinisoforms, NCX1, NCX2, and NCX3 in cerebral cortex

and hippocampus of adult rat

Andrea Minelli a,1, Pasqualina Castaldo b,1, Pietro Gobbi c,Sara Salucci c, Simona Magi b, Salvatore Amoroso b,∗

a Institute of Physiological Sciences, University of Urbino “Carlo Bo”, Urbino, Italyb Department of Neuroscience, Section of Pharmacology, Universita Politecnica delle Marche, Ancona, Italy

c Institute of Morphological Sciences, University of Urbino “Carlo Bo”, Urbino, Italy

Received 4 April 2006; received in revised form 24 May 2006; accepted 13 June 2006Available online 17 August 2006

bstract

Na+–Ca2+ exchanger (NCX) controls cytosolic Ca2+ and Na+ concentrations ([Ca2+]i and [Na+]i) in eukaryotic cells. Here we investigatedy immunocytochemistry the cellular and subcellular localization of the three known NCX isoforms, NCX1, NCX2 and NCX3, in adult rateocortex and hippocampus. NCX1–3 were widely expressed in both brain areas: NCX1 immunoreactivity (ir) was exclusively associatedo neuropilar puncta, while NCX2–3 were also detected in neuronal somata and dendrites. NCX1–3 ir was often identified around bloodessels. In both neocortex and hippocampus, all NCX isoforms were prominently expressed in dendrites and dendritic spines contacted bysymmetric axon terminals, whereas they were poorly expressed in presynaptic boutons. In addition, NCX1–3 ir was detected in astrocytes,otably in distal processes ensheathing excitatory synapses. All NCXs were expressed in perivascular astrocytic endfeet and endothelialells.

The robust expression of NCX1–3 in heterogeneous cell types in the brain in situ emphasizes their role in handling Ca2+ and Na+ in both

xcitable and non-excitable cells. Perisynaptic localization of NCX1–3 in dendrites and spines indicates that all isoforms are favourablyocated for buffering [Ca2+]i in excitatory postsynaptic sites. NCX1–3 expressed in perisynaptic glial processes may participate in shapingstrocytic [Ca2+]i transients evoked by ongoing synaptic activity.

2006 Elsevier Ltd. All rights reserved.

eywords: Na+–Ca2+ exchange; Rat brain; Synapses; Dendritic spines; Astrocytic processes; Blood–brain barrier

ai[

. Introduction

Na+–Ca2+ exchanger (NCX) is a plasma membrane

ntiporter mainly involved in maintaining cytosolic Ca2+

omeostasis. NCX couples uphill Ca2+ extrusion to down-ill Na+ influx (forward mode); alternatively, it can operate

∗ Corresponding author at: Dipartimento di Neuroscienze, Sezione di Far-acologia, Universita Politecnica delle Marche, Via Tronto 10/A, Torrette

i Ancona, I-60020 Ancona, Italy. Tel.: +39 071 2206176;ax: +39 071 2206178.

E-mail address: s.amoroso@univpm.it (S. Amoroso).1 Equally contributed to this work.

tpcbctea

143-4160/$ – see front matter © 2006 Elsevier Ltd. All rights reserved.oi:10.1016/j.ceca.2006.06.004

s Na+ efflux–Ca2+ influx pathway (reverse mode), depend-ng on membrane potential and transmembrane ion gradients1,2].

NCX plays relevant role in neural cells, where varia-ions of cytosolic Ca2+ concentration, [Ca2+]i, represent aivotal event in many physiological and pathological pro-esses. The contribution of NCX to shaping [Ca2+]i evokedy cell stimulation was demonstrated in neurons and astro-

ytes in culture [3–8] and in situ [9–12]. It has been reportedhat NCX regulates [Ca2+]i, vesicle recycling and terminalxcitability in presynaptic sites of cultured neurons [13–15]nd participates in neurotransmitter and hormone release

2 Calcium

[m[

iaRmmNaaats[ea

ssnNNt[

2

2

ieoPrmw1mpsNw5e

2

uiow

ip7puwufaltc

2

aBwbaK

2

ts(nrpw40tcwtht

Lblcbb(ttj

22 A. Minelli et al. / Cell

16,17]. It has been also shown that NCX is involved inechanisms leading to neuronal and glial excitotoxicity

18–22].Three genes have been cloned encoding distinct NCX

soforms, NCX1–3, all expressed in adult CNS [23–27]nd sharing similar ionic affinities and pharmacology [28].egional distribution of NCX1–3 transcripts and proteinsapped in adult rat brain by in situ hybridization and lighticroscopy immunohistochemistry, revealed that the threeCX isoforms are abundantly expressed in several brain

reas, often with overlapping distribution (as in neocorticalnd cerebellar cortices, hippocampus, hypothalamic area),lthough with quantitative and topographical differences inheir distribution patterns pointing to a selective expres-ion of NCX isoforms in different circuitry components29]. To date, detailed anatomical survey of NCX isoformsxpression in various elements of CNS in situ is yet to beccomplished.

In the present study, we used light and electron micro-copic immunohistochemistry to investigate the cellular andubcellular localization of NCX1–3 isoforms in adult rateocortex and hippocampus, two brain structures where allCX isoforms are abundantly expressed [29] and whereCX was recently proposed to modulate synaptic plas-

icity [11] and neuroprotection during cerebral ischemia30,31].

. Materials and methods

.1. Cell cultures

Recombinant baby hamster kidney (BHK) cells express-ng either cardiac NCX1 or rat brain NCX2 and NCX3xchanger subtypes, and hybridoma cells producing mon-clonal anti-NCX2 antibody (kindly provided by Dr. H.orzig) were cultured as previously described [32]. Briefly,ecombinant BHK cells were grown in a 1:1 DMEM/F12edium (v/v) (Invitrogen, Carlsbad, CA) supplemetedith 10% fetal bovine serum (FBS), 100 U/ml penicillin,00 �g/ml streptomycin and 0.1 �g/ml fungizone. Hybrido-as were maintained in a 1:1 DMEM/F12 medium sup-

lemented with 20% FBS, 100 U/ml penicillin, 100 �g/mltreptomycin, 1 �g/ml amphotericin B and 1% NutridomaS supplement (Roche Molecular Biochemicals). The cellsere grown in a H2O-saturated atmosphere containing% CO2 and 95% air while supernatant was harvestedvery 24 h.

.2. Rat tissue preparation

Sprague–Dawley male adult albino rats (200–300 g) were

sed in these studies. Care and handling of animals were donen compliance with the regulations of the Ethical Committeef the University of Urbino. Rats were deeply anesthetizedith 12% chloral hydrate and perfused through the ascend-

deAo

41 (2007) 221–234

ng aorta with physiological saline followed by either 4%araformaldehyde (PFA) in phosphate buffer (PB, 0.1 M; pH.4) or 3.5% PFA and 1% glutaraldehyde in PB. Animalserfused with either PFA or PFA plus glutaraldehyde weresed for light microscopic studies, while only rats perfusedith glutaraldehyde-containing fixative were considered forltrastructural analyses. Brains were postfixed in 4% PFAor 2–12 h. Thirty-micrometer thick sections were cut withVibratome in either coronal or parasagittal plane and col-

ected in phosphate buffered-saline (PBS). Sections for elec-ron microscopy were always processed within 24 h fromutting.

.3. Antibodies

NCX1 protein was detected by using a commerciallyvailable mouse monoclonal IgG antibody (R3F1; Swant,ellinzona, Switzerland) [1]. For NCX2 protein detection,e used a hybridoma cell-produced monoclonal IgM anti-ody (W1C3) [32]. Finally, NCX3 protein was detected byrabbit polyclonal IgG antibody generously provided by Dr..D. Philipson [32].

.4. Immunoblotting studies

Rats were perfused with cold 4 mM Tris–HCl, pH 7.4, con-aining 0.32 M sucrose, 1 mM EDTA, 0.5 mM phenylmethyl-ulphonyl fluoride (PMSF), and 0.5 mM N-ethylmaleimideNEM). Membrane proteins derived from BHK recombi-ant cells expressing NCX1, NCX2, and NCX3 and fromat tissues (heart, cortex and hippocampus) were prepared asreviously described [33]. Briefly, BHK cells and rat tissuesere homogenized in 6 vol. of ice-cold homogenating buffer:mM Tris–HCl, pH 7.4; 0.32 M sucrose, 1 mM EDTA;.25 mM dithiothreitol; protease inhibitor cocktail mixtureablets (Roche Diagnostics, Milan, Italy). Homogenates wereentrifuged 1000 × g at 4 ◦C for 15 min, and supernatantsere then centrifuged at 100,000 × g at 4 ◦C for 1 h to obtain

he crude membrane fraction. The pellet was resuspended inomogenating buffer and stored at −70 ◦C for immunoblot-ing.

Membrane proteins (40 �g) were solubilized in 2×aemmli Sample Buffer containing 5% 2-mercaptoethanol,oiled for 5 min, and analyzed by SDS-PAGE (8% polyacry-amide). Separated proteins were transferred electrophoreti-ally to nitrocellulose (Amersham Biosciences, Milan, Italy),locked in 5% non-fat dry milk in PBS and then incu-ated overnight at 4 ◦C with the primary antibody: R3F1dilution, 1:1000), W1C3 (dilution, 1:1000), or NCX3 (dilu-ion, 1:4000). Membranes were then incubated (1 h at roomemperature) with the appropriate secondary antibody con-ugated to horseradish peroxidase (Santa Cruz CA, USA;

ilution, 1:1000). Immunolabelled bands were revealed by annhanced chemiluminescence detection system (ECL + Plus;mersham Biosciences, Milan, Italy). Images were capturedn a ChemiDoc station (Bio-Rad, Milan, Italy), and ana-

Calcium

ls

2

2

ctsipmwi411sldcpudp

so

2p

(bwpgfYfIlcYwPgmi

tacas

2

stidiaiupacso(oci

2

filImio

ftfi(oomapnsrc

3

3

A. Minelli et al. / Cell

yzed with the Quantity one (Bio-Rad, Milan, Italy) analysisoftware.

.5. Immunocytochemical experiments

.5.1. Immunoperoxidase procedureGroups of four serial sections taken at various rostro-

audal or medio-lateral levels were processed in parallel:hree sections were used for NCX detection, and the fourthection for Nissl staining. Free-floating sections were rinsedn PBS containing 1% H2O2 for quenching endogenouseroxidase and in 1% sodium borhydride in PBS to mini-ize aspecific binding due to aldehydic residues. Sectionsere then incubated in 10% non-immune goat serum (NGS)

n PBS plus 0.2% Triton X-100, and then overnight at◦C in NCX1–3 primary antibodies diluted in PBS plus% NGS. NCX1 antibody was used at concentrations of:750, NCX2 at 1:500 and NCX3 at 1:500. The next day,ections were incubated in the appropriate goat biotiny-ated secondary antibodies (Vector Lab, Burlingame, CA)iluted 1:200 in PBS plus 1% NGS. Sections were then pro-essed according to the avidin–biotin peroxidase complexrocedure (Vector; PK-6100). Finally, the reaction prod-ct was demonstrated by 3′,3-diaminobenzidine tetrahy-rochloryde (DAB; 40 mg/50 ml) with 0.03% hydrogeneroxide.

Sections intended for immunoperoxidase electron micro-cope studies were processed as described above, with thenly exception that Triton X-100 was omitted.

.5.2. Preembedding silver-enhanced immunogoldrocedure (SEI)

Sections were incubated with 1% bovine serum albuminBSA) in PBS (PBS–BSA), then overnight in primary anti-odies (concentrated as above) in PBS–BSA. After severalashes in PBS–BSA, sections were incubated in the appro-riate secondary antibodies conjugated to 1.4 nm colloidalold particles: goat anti-mouse IgG and goat anti-rabbit IgGor NCX1 and NCX3 detection, respectively (Nanoprobes,aphank, NY), diluted 1:200 in PBS–BSA with 1% NGS;

or NCX2 detection, due to the lack of gold-conjugated antigM antibody, sections were first incubated with biotiny-ated goat anti-mouse IgM antibody and then with a 1.4 nmolloidal gold-conjugated goat anti-biotin IgG (Nanoprobes,aphank, NY; 1:200 in PBS–BSA with 1% NGS). Sectionsere rinsed in PBS–BSA, post-fixed in 1% glutaraldehyde inBS (10 min) and briefly washed in deionized water; colloidalold labelling was then intensified using a silver enhance-ent kit (HQ silver, Nanoprobes, Yaphank, NY) for 3–5 min

n dark room.For both immunoperoxidase and immunogold ultrastruc-

ural studies, controls were performed by omitting primary

ntisera or by substituting it with NGS 10% in PBS; in theseases, signal was virtually absent and showed no preferentialssociation with plasma membranes or specific subcellulartructures.

Nh(1

41 (2007) 221–234 223

.6. Electron microscopy

After completion of the immunohistochemical procedure,ections were postfixed in 2.5% glutaraldehyde in PB, andhen for 1 h in OsO4 (1% and 0.2% in PB for immunoperox-dase and immunogold material, respectively). After dehy-ration in graded series of ethanol, sections were clearedn propylene oxide, flat-embedded in Epon-Spurr betweencetate sheets (Aclar; Ted Pella, Redding, CA), and polymer-zed at 60 ◦C for 72 h. Embedded sections were examinednder a dissecting microscope, and neocortical and hip-ocampal areas of interest were excised with a razor bladend mounted on cured resin pyramidal blocks utilizing ayanoacrylic glue. From each sample, one semithin (1 �m)ection was cut with a Reichert ultramicrotome and mountedn glass slides for light microscopic inspection. Ultrathin70–100-nm thick) sections were cut either from the surfacer from the edge (i.e. perpendicular to the plane section),ounterstained with uranyl acetate and lead citrate, and exam-ned with a Philips CM 10 electron microscope.

.7. Data analysis

All data were collected from the parietal cortex androm hippocampal CA1 and CA3 subfields. Identification ofmmunolabeled and unlabeled profiles was based on estab-ished morphological and morphometrical criteria [34,35].n sections processed with silver-intensified immunogoldethod, a neuropilar profile was considered labelled when

t had at least two immunoparticles on its plasma membraner cytoplasmic structures.

Counts of NCX1–3 immunostained profiles were per-ormed in single plane ultrathin sections from DAB-reactedissue. Sections from neocortex (areas from deep and super-cial layers were separately examined) and hippocampusfrom stratum radiatum of CA1 and CA3) were analysed inrder to assess, for each NCX isoform, the relative percentagef labelled profiles belonging to dendrites (including proxi-al and distal dendrites as well as dendritic spines), to fibers

nd axon terminals and to astrocytic processes. Neuropilarrofiles whose nature could not be unambiguously recog-ized were considered “unidentified”. In both brain areas, 12quared mesh nickel grids (size: 40 �m × 40 �m) from twoats were scanned at the electron microscope and all DAB-ontaining neuropilar profiles were classified and counted.

. Results

.1. Antibodies characterization by Western blots

Western blots on crude membrane protein fractions from

CX1-expressing BHK cells and from rat neocortical andippocampal homogenates revealed that, in all samples, R3F1NCX1) antibody labelled a predominant band at around60 kDa (Fig. 1A). This band represents the mature glyco-

224 A. Minelli et al. / Cell Calcium

Fig. 1. NCX1–3 immunoblotting. Characterization of the three NCXexchanger isoform-specific antibodies by immunoblots on membrane pro-tein fractions (40 �g proteins/lane) obtained from recombinant BHK cellsthat selectively express NCX1 (ncx1), NCX2 (ncx2), NCX3 (ncx3) isoformsand from rat tissue homogenates (heart; cortex, ctx; and hippocampus, hip):(A) NCX1, (B) NCX2, and (C) NCX3 immunoblots.

smtbsr[hbp

ttfioNa

3

timimcafiacieNsIa(aiKsasstiwIl

owNa

41 (2007) 221–234

ylated form of NCX1 protein [36,37]. Immunoblots withAb W1C3 (NCX2) antibody revealed a major band close

o 60 kDa (Fig. 1B) in NCX2-expressing BHK cells and inoth neocortical and hippocampal homogenates. All sampleshowed an additional, faint band in the high molecular weightange between 131 and 250 kDa, as previously reported32]. In both NCX3-expressing BHK cells and brain tissueomogenates, NCX3 antibody revealed a single immunola-elled band at around 120 kDa (Fig. 1C), in agreement withrevious reports [32].

Since NCX2 and NCX3 isoforms are not expressed inhe cardiac muscle [24], immunoblotting on membrane frac-ions from heart homogenates were used as further controlor antisera specificity: while NCX1 antibody revealed anmmunolabelled band of the same molecular weight as thatbserved in brain homogenates (Fig. 1A), no NCX2- andCX3-labelled bands were found in heart tissue (Fig. 1B

nd C).

.2. Light microscopy

Immunolabelling for all three exchangers was observedhrough the depth of the cerebral cortex (Fig. 2A–C), show-ng no obvious regional variations along the rostro-caudal and

edio-lateral neocortical axes. In the parietal cortex, NCX1r did not display a layer-specific pattern, except for being

ore intense in layer I (Fig. 2A). NCX1 ir was mostly asso-iated to small punctate structures (Fig. 2D), while fibersnd cell bodies were never labelled; positive puncta werenely dispersed in the neuropil (Fig. 2D), and only occasion-lly they were more dense around the profile of unstainedell somata. NCX2 and NCX3 immunolabelling was morentense in layer I and in supragranular and infragranular lay-rs as compared to layer IV (Fig. 2B and C). In all layers,CX2 and NCX3 ir was associated to puncta and heavily

tained pyramidal and non-pyramidal neurons (Fig. 2I–L).n neurons, NCX3 ir was mostly confined to cell somatand occasionally spread to the very proximal arborisationFig. 2L); conversely, NCX2 ir was detected in basal andpical dendrites of pyramidal neurons (Fig. 2G and I) andn proximal dendrites of non-pyramidal neurons (Fig. 2J and). Throughout the neocortex, NCX2 and NCX3 ir was often

een in association with blood vessel profiles (Fig. 2H). Inddition, ir for NCX1–3 was observed in many astrocytes ofubcortical white matter and corpus callosum, showing smalloma and several irregular processes radiating in all direc-ions (Fig. 2E). Few NCX2-labelled astrocytes were foundn the cortical parenchyma, notably in layer I, where theyould often send processes towards the pia mater (Fig. 2F).

ntense labelling for NCX2 and NCX3 was also found in theeptomeninges (Fig. 2B, C and F).

In the hippocampus (Fig. 3A–C), strong NCX1 ir was

bserved in strata oriens and radiatum of CA1 and CA3,hereas labelling for NCX2 and NCX3 was less intense.CX1 ir was virtually absent in pyramidal cell layer of

ll hippocampal fields and in granule cell layer of dentate

A. Minelli et al. / Cell Calcium 41 (2007) 221–234 225

Fig. 2. NCX1–3 ir in the cerebral cortex. Distribution of NCX1 (A), NCX2 (B), and NCX3 (C) ir in the parietal cortex of adult rats. Roman numerals indicatecortical layers. NCX1 ir is associated to densely distributed punctate structures (D; from layer III). Several subcortical astrocyte-like cells are NCX1 positive (E).NCX2 ir is associated to puncta and neurons (G; from layers I to II); several labelled dendrites (open arrows) run towards layer I (G); a labelled astrocyte-likec . IntensN neurond ell bodi(

gehaTpiEi

i(ts(

ell is observed near the pial surface, and arachnoid is intensely stained (F)CX2 ir is detected in both pyramidal (I; from layer V) and non-pyramidalispersed through the neuropil (L; from layers III) as well as in neuronal cD–L).

yrus, where both NCX2 and NCX3 were instead highlyxpressed (Fig. 3A–C). Weak NCX1 ir was observed in theilus and stratum moleculare of dentate gyrus, where NCX2nd NCX3 exhibited higher level of staining (Fig. 3A–C).he morphological features of NCX1–3 ir in the hippocam-

us were similar to those observed in the neocortex. NCX1r was exclusively associated to small puncta (Fig. 3D and). Moderate-to-strong NCX2 and NCX3 ir was detected

n pyramidal cell bodies and processes, in granule cells and

w(ac

e NCX2 ir is visible around blood vessel profile (H; from layer I). Intenses (J and K; from layers III and V, respectively). NCX3 ir is in puncta finelyes (L; black arrow). bv: blood vessel. Bars: 150 �m for (A–C); 30 �m for

n interneurons located in strata oriens/radiatum of CA1–3Fig. 3F, H, and I) and in the hilus/stratum moleculare of den-ate gyrus; immunolabelled dendrites of variable length andize were observed in the stratum radiatum of CA1 and CA3Fig. 3F and I); sparse NCX2- and NCX3-positive puncta

ere diffusely distributed throughout hippocampal neuropil

Fig. 3F–I). Blood vessel profiles appeared rimmed by NCX2nd NCX3 ir (Fig. 3H). Occasionally, NCX2-labelled astro-ytes were observed.

226 A. Minelli et al. / Cell Calcium 41 (2007) 221–234

Fig. 3. NCX1–3 ir in the hippocampus. Distribution of NCX1 (A), NCX2 (B), and NCX3 (C) ir in the hippocampus of adult rats. NCX1 ir is associated tointensely stained, densely distributed punctate structures ((D) and (E); from CA1 stratum radiatum and CA3 stratum oriens, respectively). NCX2 ir is associatedt d vessec X3 ir (I( vessel.

3

siot

nvynhT

dbcaa(eib

wtif(tba

f(mdbibtppc

o puncta scattered in the neuropil ((F) and (G); from CA1) and around blooell layer (F) and interneurons in the stratum radiatum (G) are stained. NCblack arrow) and interneurons (open arrow). DG: dentate gyrus; bv: blood

.3. Electron microscopy

The ultrastructural features of NCX1–3 isoforms expres-ion were remarkably similar in the two brain regions exam-ned; therefore, cellular and subcellular localization patternsf NCX1–3 in neocortex and hippocampus are describedogether.

All three exchanger isoforms were highly expressed ineurons; notably, they were localized in dendritic profiles ofarious size and in dendritic spines. In fact, quantitative anal-ses revealed that the majority of NCX1–3-immunostainedeuropilar profiles counted in the parietal cortex and in theippocampal CA1 belonged to dendrites and spines (seeable 1).

Dense NCX1–3 immunolabelling was observed inistal dendrites that were often contacted by unla-elled axon terminals forming asymmetric synapses;lumps of NCX1–3 reaction product were found inpposition to microtubules, packed around mitochondriand associated to the inner side of plasma membrane

Figs. 4A, 5B, and 6B). Silver-enhancement immunogoldxperiments (SEI) confirmed the presence of NCX1–3mmunoparticles in apposition to dendritic plasma mem-rane (Figs. 4B and 5D) and revealed that some particles

otsa

l (open arrowheads in H; from stratum radiatum); neurons in the pyramidal; from CA3) is observed punctuate structures, as well as in pyramidal cellsBars: 500 �m for (A–C); 30 �m for (D–I).

ere in the vicinity of postsynaptic densities facing axonerminals (Figs. 5D and 6C). All dendritic spines contain-ng DAB reaction product for the various NCX isoformsormed asymmetric synapses with unlabelled axon terminalsFigs. 4C, 5E, and 6D). In SEI-labelled spines, immunopar-icles for NCX1–3 were sometimes found apposed to mem-rane regions close to postsynaptic specializations (Figs. 4Dnd 6E).

While NCX1 ir was mostly confined to distal dendrites, iror NCX2 and NCX3 was often found in proximal dendritesFig. 5B), where clumps of ir were associated to microtubules,itochondria and plasma membrane. Labelled proximal den-

rites were often contacted by axon terminals, all unla-elled, many of them forming asymmetric contacts (Fig. 5C);mmunoparticles for NCX2–3 were found in membrane areasoth in the vicinity of and at distance from synaptic junc-ions. In addition, ir for NCX2–3 was observed in neuronalerikarya, mostly in pyramidal cells (Figs. 5A and 6A): here,atches of labelling were scattered in the cytoplasm and asso-iated to the cytoplasmic side of nuclear envelope, to cisterns

f the rough endoplasmic reticulum and Golgi apparatus ando mitochondria; sometimes, clumps of ir were seen in appo-ition to plasma membrane, usually in areas not contacted byxon terminals.

A. Minelli et al. / Cell Calcium 41 (2007) 221–234 227

Table 1Quantitative analysis of NCX1–3-labelled neuropilar profiles sampled from parietal cortex and hippocampus

NCX1 NCX2 NCX3

Number of profiles Profiles per mesh Number of profiles Profiles per mesh Number of profiles Profiles per mesh

Cerebral cortexDendrites/spines 618 (54.7%) 25.9 ± 7 789 (51.7%) 33 ± 4.4 869 (58.6%) 37 ± 6.3Axon fibers/terminals 20 (1.8%) 0.8 ± 1.2 53 (3.5%) 2.7 ± 2.9 65 (4.4%) 2.9 ± 2.8Glial processes 388 (34.3%) 16.1 ± 9 523 (34.3%) 22 ± 4.9 490 (33%) 20.3 ± 4.3Unidentified 104 (9.2%) 4.3 ± 2.3 161 (10.5%) 6.6 ± 2.5 60 (4%) 2.6 ± 1.4

Total labelled profiles 1130 (100%) 47.1 ± 11.4 1526 (100%) 63.7 ± 14.1 1484 (100%) 61.9 ± 16.1

HippocampusDendrites/spines 868 (40.4%) 36.4 ± 19 889 (51%) 37 ± 8.5 890 (57.1%) 36.6 ± 6Axon fibers/terminals 38 (1.8%) 1.75 ± 2 51 (2.9%) 2 ± 1.4 83 (5.3%) 3.4 ± 2.9Glial processes 1162 (54.1%) 48.7 ± 32 732 (42%) 30.5 ± 9 463 (29.7%) 19.3 ± 9.5Unidentified 80 (3.7%) 3.5 ± 2.6 72 (4.1%) 3 ± 1.1 124 (7.9%) 5.1 ± 2.1

Total labelled profiles 2148 (100%) 89.3 ± 23.5 1744 (100%) 72.5 ± 18.2 1560 (100%) 65 ± 15.6

For each NCX isoform, numbers and relative percentages (in parentheses) of labelled profiles as well as numbers of labelled profiles per mesh (average ± S.D.)are reported.

Fig. 4. Ultrastructural localization of NCX1 ir in the cerebral cortex and hippocampus. (A) Deep neocortical layers: numerous dendritic profiles (den) containclumps of reaction product variably associated to microtubules, mitochondria (open arrowhead), and plasma membrane (black arrow). (B) CA1 stratumradiatum: several dendritic profiles (den) show NCX1 immunoparticles associated to microtubules (open arrows), mitochondria (open arrowhead) and plasmamembrane (black arrows). A labelled myelinated axon (ax) is also evident. (C) Superficial neocortical layers: two labelled dendritic spines (black arrows; sp)receive an unlabelled axon terminal (asterisks) forming asymmetric contacts (triangles). An adjacent astrocytic profile (asp) is also labelled (black arrow). (D)CA1 stratum radiatum: dendritic spine (sp) receiving an unlabelled terminal (asterisk) forming asymmetric contact (triangles) contains NCX1 immunoparticles(black arrows), one of which is near the postsynaptic junction. (E) Deep neocortical layers: a labelled distal astrocytic process (black arrow) is in appositionto unlabelled dendritic spines (asterisk) receiving asymmetric contact. (F and G) CA1 stratum radiatum: labelled fiber (F) shows clumps of reaction productscattered in the axoplasm (open arrow). (G) An axon terminal (axt) contacting an unlabelled dendrite (den) shows a membrane cluster of immunoparticlesopposite to the synaptic specialization (black arrow). Calibration bars: 0.5 �m for (A); 0.25 �m for (B–G).

228 A. Minelli et al. / Cell Calcium 41 (2007) 221–234

Fig. 5. Ultrastructural localization of NCX2 ir in the cerebral cortex and hippocampus. (A) NCX2 immunolabelling in neuronal perikarya (black arrows):example of a pyramidal cell soma in neocortical layer V. (B) CA1 stratum radiatum: several dendritic profiles (den) are heavily stained with raction product (openarrows). An astrocytic process (asp) is also shown containing NCX2 ir in its cytoplasm and plasma membrane (black arrow). (C) In CA1 stratum radiatum, alarge positive dendritic profile (den) is contacted by numerous unlabelled axon terminals (asterisks): ir is associated to microtubules (open arrow), mitochondria(open arrowhead), and plasma membrane regions (black arrow) near the synapses. (D) Superficial neocortical layers: NCX2-positive distal dendrites (den)are shown, one of which is targeted by an unlabelled axon terminals (asterisk). (E) Superficial neocortical layers: a labelled dendritic spine (sp; black arrow)is contacted by an unlabelled axon terminal (asterisk) forming asymmetric junction. (F) Deep neocortical layers: NCX2 labelling in a myelinated fiber: ir isassociated to neurotubules (open arrow) and plasma membrane (black arrow). (G and H) Astrocytic processes (asp) express NCX2 ir. CA3 stratum radiatum:examples of astrocytic processes sheathing synaptic structures (asterisks) in both DAB labeled (G) and immunogold (H) material: in both figures, labellingappears scattered in the cytoplasm (open arrows) or associated to plasma membrane (black arrows) and mitochondria (open arrowheads). nu: nucleus; cyt:cytoplasm. Bars: 1 �m for (A–H); 0.5 �m for (C); 0.25 �m for (D–G).

A. Minelli et al. / Cell Calcium 41 (2007) 221–234 229

Fig. 6. Ultrastructural localization of NCX3 ir in the cerebral cortex and hippocampus. (A) Neocortical layer V: NCX3 ir in several distal dendrites (den).In a neuronal cell body, labelling is scattered within the cytoplasm and associated to plasma membrane (black arrow), perinuclear cisterns (open arrows) andmitochondria (open arrowhead). Nucleus (nu) is unlabelled. In the inset (A1), similar features of perikaryal NCX3 staining are shown at higher magnificationin a different neuron. (B) CA3 stratum radiatum: several dendritic profiles (den) contain clumps of ir associated to microtubules (open arrow), mitochondria(open arrowhead) and to plasmalemma (black arrows). (C) CA1 stratum radiatum: an immunogold-labelled dendrite (black arrow) receives two unlabelledaxon terminals (asterisks) forming asymmetric contacts. (D) CA1 stratum radiatum: labelled dendritic spine (sp; black arrow) forming asymmetric contact withan unlabelled terminal (asterisk). (E) Superficial neocortex: a NCX3 labelled spine (sp; black arrows) forming asymmetric contact with an unlabelled terminal(asterisk) is ensheathed by a NCX3-positive astrocytic process (asp) bearing immunoparticles on its plasma membrane. (F and G) CA1 stratum radiatum: NCX3immunoparticles in a myelinated fiber (F) and an axon terminal (axt; G) contacting an unlabelled spine (asterisk). In G, a perisynaptic astrocytic process is alsolabelled (black arrows). (H) Deep cortical layers: intense NCX3 ir is visible on plasma membrane of two distal astrocytic processes (black arrows; asp), one ofwhich is close to an asymmetric synapse (asterisks indicates the unlabelled axon terminals). (I) Deep cortical layers: a glial process (asp) shows NCX3 labellingscattered in the cytoplasm (open arrows) and associated to mitochondria (open arrowheads). nu: nucleus; cyt: cytoplasm. Bars: 0.5 �m for (A–A1–C–I); 1 �mfor (B); 0.2 �m for (D–G and H); 0.25 �m for (E and F).

230 A. Minelli et al. / Cell Calcium 41 (2007) 221–234

Fig. 7. NCX ir in blood–brain barrier. NCX2 (A) and NCX3 (C) ir in endothelial cells lining the lumen of neocortical (A) and hippocampal CA1 (C) capillaries.Labelling is scattered in the cytoplasm (open arrows) and associated to discrete regions of both luminal (black arrows) and basal membrane (black arrow).( immunoa cytic pre eus. Ba

moafrad(

oinefiwpcwcegtmwctapt(

tta

4

cpacclipeietd

aewnfi

B) Cerebral cortex: a perivascular astrocytic profile (asp) bearing NCX2rrows). (D) Hippocampal CA1: NCX3 ir (black arrow) in perivascular astrondothelial cell membrane is also visible (arrow). bv: blood vessel; nu: nucl

NCX1–3 were poorly expressed in axon fibers and ter-inals (Table 1). NCX1–3-positive fibers contained clumps

f ir variably associated to microtubules and, occasion-lly, to plasma membrane (Figs. 4F, 5F and 6F). Theew NCX1–3-labelled axon terminals formed all asymmet-ic synapses; immunoparticles for NCX1–3 were found inpposition to plasma membrane (although never in regionsirectly facing postsynaptic densities) and to mitochondriaFigs. 4G and 6G).

In both cerebral cortex and hippocampus, NCX1–3 ir wasbserved in astrocytic cell bodies and processes, character-zed by irregular contour adapting to the profile of adjacenteuropilar elements. Quantitative analyses revealed the pres-nce of a conspicuous percentage of NCX1–3-positive pro-les belonging to glial processes (Table 1). Ir for NCX1–3as particularly intense in distal astrocytic processes, wherelasma membrane was intensely stained. Labelled distal pro-esses were often in contiguity of synaptic structures, as theyere found adjacent to axon terminals making asymmetric

ontacts with dendrites and spines (Figs. 4E, 5G and 6H). SEIxperiments consistently showed the presence of NCX1–3old particles in astrocytic membranes directly facing synap-ic structures (Fig. 5H). While NCX1 ir in astrocytes was

ainly restricted to distal processes, ir for NCX2 and NCX3as present also in astrocytic cell bodies and thick pro-

esses, where patches of reaction product were scattered inhe cytoplasm and associated with organelles, mitochondria

nd plasma membrane (Fig. 6I). Several NCX1–3-labellederivascular astrocytic end-feet were observed in appositiono the basal lamina around the capillary endothelial wallFig. 7B and D). In many cases, the endothelial cells con-

bgai

particles on plasma membrane (black arrow) and in the cytoplasm (openocess (asp) apposed to capillary basal lamina. A thin rim of labelling in thers: 1 �m for (A); 0.25 �m for (B); 0.5 �m for (C and D).

ained clumps of NCX1–3 ir in their cytoplasm or associatedo both luminal and basal sides of plasma membrane (Fig. 7And C).

. Discussion

The present work provides the first detailed analysis of theellular and subcellular localization of various NCX isoformroteins in CNS in situ. Major results can be summarizeds follows: (i) NCX1–3 are widely distributed through neo-ortex and hippocampus and are expressed in heterogeneousell populations, including neurons, astrocytes and epithe-ial cell; (ii) neuronal NCX1–3 are preferentially locatedn dendrites and dendritic spines, while glial NCX1–3 arerominently expressed in astrocytic processes ensheathingxcitatory synapses; (iii) NCX1–3 are consistently expressedn capillary endothelial cells and in perivascular astrocyticndfeet. Major sites of NCXs expression in synaptic struc-ures, astrocytic processes and endothelium are schematicallyepicted in Fig. 8.

Previous immunocytochemical studies in hippocampalnd neocortical cell cultures reported that NCX2 protein isxpressed in glial cells and is absent in neurons [32,38],hereas NCX1 and NCX3 were found to be exclusivelyeuronal [32]. Our in situ data, however, did not con-rm these observations. The possibility that NCX1–3 anti-

odies used here cross-react in situ with unrelated anti-ens seems unlikely, since: (i) immunoblots on neocorticalnd hippocampal membranes revealed the same NCX1–3-mmunolabelled bands as in NCX-expressing BHK cells; (ii)

A. Minelli et al. / Cell Calcium

Fiap

vpive

4

paNarii[tlni

seeaabpgouoatt

fiin

iiasrafaisMdtm“irmsiprcNpiPfstit

4

cnNpc

crria(

ig. 8. Schematic drawing illustrating the main sites of NCX1–3 localizationn synaptic structures, astrocytic processes and endothelial cells. Distinctnatomical structures are at different size scales. as: astrocyte; asp: astrocyticrocess; axt: axon terminal; cap: capillary; den: dendrite; end: endothelium.

ariable amounts of mRNAs for all three NCX isoforms werereviously detected in neocortical and hippocampal neuronsn situ [29]. Differences between in vitro and in situ obser-ations may thus possibly reflect changes of NCX proteinsxpression in neural cells maintained in culture.

.1. NCXs in neurons

In both neocortex and hippocampus, all NCX isoforms arerominently expressed in dendrites and spines, whereas theyre poorly present in fibers and axon terminals. Moreover,CX1–3 expression is often detected in plasma membrane

reas close to postsynaptic specializations forming asymmet-ic junctions. Dendritic polarization and perisynaptic local-zation of NCX1–3 expression indicate that in situ all NCXsoforms are situated in favourable position for bufferingCa2+]i in excitatory postsynaptic sites. Evidence showinghat NCX activity can affect postsynaptic Ca2+ transients fol-owing glutamate receptors activation in rat brain primaryeurons [6] and in cerebellar Purkinjie cells in vivo [12] aren support of this view.

Presynaptic NCX expression was proposed by functionaltudies showing that NCX activity can influence terminalxcitability and neurotransmitter release [13–15,18]; how-ver, anatomical evidence documenting the presence of NCXt presynaptic terminals was so far partial, because avail-ble only for NCX1, and rather inconclusive, since eminentlyased on immunofluorescence confocal studies in hippocam-al cell cultures [13] and calyx synapses of chicken ciliaryanglion [39] reporting either high [13] or low [39] levelsf NCX1 colocalization with presynaptic markers. Presentltrastructural investigations conducted in situ unambigu-

usly reveal that NCX1–3 exchangers are expressed only inrestricted population of neocortical and hippocampal axon

erminals (see Table 1), thus pointing to a minor presynap-ic expression of all NCX isoforms in living brain. Thus, our

tiee

41 (2007) 221–234 231

ndings strongly suggest that the physiological role of NCXsn synaptic transmission is prevalently exerted at the postsy-aptic side.

NCX1–3 are highly expressed in dendritic spines form-ng asymmetric contacts, thus suggesting their involvementn modulating Ca2+ transients in axo-spinous synapses. Ingreement, recent studies in cultured hippocampal neuronshowed that reducing the efficacy of NCX-mediated calciumemoval causes a slowing of the decay of Ca2+ transientst individual spine heads and an enhancement of Ca2+ dif-usion from the spine head into the parent dendrite [8]. Welso show that neocortical and hippocampal non-pyramidalnterneurons, the majority of which are smooth or sparselypiny [40], express high levels of NCX2 and NCX3 isoforms.

oreover, dendritic perisynaptic expression of NCX1–3 isetected in membrane regions close to non-spinous synap-ic contacts. Together, these findings suggest that NCXs

ay contribute to shaping postsynaptic Ca2+ responses inspine-free” dendritic regions, otherwise lacking morpholog-cally specialized structures, i.e. spines, that would physicallyestrict calcium diffusion. In support of this view, fast NCX-ediated Ca2+ extrusion has been shown to reduce the spatial

pread of [Ca2+]i increase in aspiny dendrites of corticalnterneurons after individual synapse stimulation [10]. Tem-oral and spatial shaping of postsynaptic [Ca2+]i transientsepresents a critical factor for synapse-specific Ca2+ signalsompartmentalization and synaptic plasticity [41]. Studies inCX2-deficient transgenic mice revealed that mutant hip-ocampal neurons exhibit delayed [Ca2+]i clearance follow-ng depolarization and enhanced long-term potentiation [11].resent ultrastructural findings indicate that all NCX iso-orms are well located in situ to influence postsynaptic Ca2+

ignals and thus, besides reinforcing functional evidence inransgenic mice [11], do not emphasize the notion of ansoform-selective involvement of NCX2 in regulating synap-ic plasticity.

.2. NCXs in glial cells

Modulation of [Ca2+]i plays a central role in many astro-ytic functions, including reciprocal astrocyte-neuron sig-alling and glial toxicity [42–44]. Here we show that allCX isoforms are highly expressed in neocortical and hip-ocampal astrocytes, thus emphasising the role of NCX inontrolling [Ca2+]i homeostasis of brain astrocytes in situ.

It is generally accepted that spilled-out glutamate in vivoan activate its receptors and evoke Ca2+ responses in sur-ounding astrocytes [44–47]. Present ultrastructural findingseveal that, in rat neocortex and hippocampus, all NCXsoforms are consistently expressed in astrocytic processesdjacent to synaptic structures forming asymmetric junctionsi.e. presumably glutamatergic; see Ref. [48]); this observa-

ion indicates that a conspicuous fraction of glial NCX1–3s located in astrocytic domains that, for being in clos-st proximity of release sites, can be reached by glutamatescaped from the cleft. Studies conducted in cultured corti-

2 Calcium

ciaatiscdtpNtipi[d[

iivtsagmaplptwrdsigtpNt

4

wrteaspbr

tdt[

erccm

4

palenitr

A

b“aN

R

32 A. Minelli et al. / Cell

al astrocytes [4,7,49–51] and in cerebellar Bergmann glian situ [52] have shown that glial [Ca2+]i transients associ-ted with the activation of ionotropic glutamatergic receptorsre mainly promoted by reverse-mode, NCX-mediated Ca2+

ransport. Present evidence of a robust expression of NCX1–3n perisynaptic glial processes provides strong anatomicalupport to the view that these exchangers may be espe-ially suited for mediating Ca2+ influx in astrocytes in situuring adjacent synaptic activity in physiological condi-ions. Interestingly, we show that not all distal astrocyticrocesses ensheathing synapses display the same level ofCX1–3 expression, and some of them are devoid of staining,

hus pointing to a functional heterogeneity of NCX activityn affecting Ca2+ responses in different subpopulations oferisynaptic glial processes. In line with recent studies show-ng that astrocytes selectively respond to different synapses43], these findings emphasize the role of NCX activity inetermining synapse-specific Ca2+ responses in astrocytes43,44].

NCX1–3 activity in perisynaptic glial processes may havemportant functional implications. As recently documentedn vitro [53], glutamate-evoked, NCX-mediated [Ca2+]i ele-ation in glial cells could, in turn, trigger important func-ional response, i.e. astrocytic release of neuroactive sub-tances on neighbouring neurons [43]. In addition, NCXctivity may dynamically balance glial Na+ transmembraneradients during ongoing activation of Na+-permeant gluta-atergic receptors. Perisynaptic glial processes in neocortex

nd hippocampus express Na+-dependent glutamate trans-orters [54–57] which are essential for maintaining extracel-ular glutamate homeostasis [58]. Therefore, although directroofs are lacking, NCX1–3 in glial processes are likelyo colocalize and share the same ionic microenvironmentith glutamate transporters. By extruding Na+ operating in

everse, NCX1–3 could be critical for maintenance of Na+-ependent glutamate-clearing ability of glial cells duringustained synaptic activity. Since glial glutamate transports an important component of nerve tissue defense againstlutamate neurotoxicity [59,60], NCX activity in perisynap-ic glial cells may play significant neuroprotective role inathological conditions. Evidence showing that NCX1 andCX3 exert neuroprotective role in rat ischemic cerebral cor-

ex [30,31].

.3. NCXs in blood–brain barrier

Ca2+-induced contractions of endothelial cytoscheletonas proposed to prolong blood–brain barrier (BBB) dis-

uption after osmotic stress [61]. Present findings showinghat in neocortex and hippocampus all NCX isoforms arexpressed in capillary endothelial cells emphasise the role ofll NCX isoforms in controlling endothelial Ca2+ homeosta-

is in situ; in addition, by revealing NCX1–3 expression inerivascular astrocytic endfeet, they point to a major contri-ution of all NCX exchangers in actively regulating BBB bar-ier function, in line with pharmacological studies showing

41 (2007) 221–234

hat NCX blockers have synergistic effect on BBB openinguring hypertonic infusions [62,63] and affect the forma-ion of brain vasogenic oedema after radiofrequency lesions64].

Recent studies showed that [Ca2+]i transients in astrocytendfeet can cause cerebrovascular vasoconstriction [65]; theemarkable expression of NCX1–3 in perivascular glial pro-esses in situ suggests that NCX activity in astrocytic endfeetould play an important role in the glial control of brainicrocirculation [66].

.4. Conclusions

Present anatomical studies in rat cerebral cortex and hip-ocampus show that, in both brain regions, all NCX isoformsre expressed in neurons and astrocytes as well as in endothe-ial cells, thus pointing to a widespread role of the threexchangers in maintaining Ca2+ homeostasis in excitable andon-excitable CNS cell types in situ. The subcellular local-zation patterns of NCX1–3 in neurons and astrocytes supportheir possible involvement in synaptic transmission and neu-oprotection.

cknowledgements

This work was supported by grants from RSA to S.A. andy grants from MIUR-Cofin and from Universita di UrbinoCarlo Bo” to P.G. The authors wish to thank Drs. H. Porzignd K.D. Phillipson for generously providing NCX2 andCX3 antisera, respectively.

eferences

[1] K.D. Philipson, D.A. Nicoll, Sodium-calcium exchange: a molecularperspective, Annu. Rev. Physiol. 62 (2000) 111–133.

[2] L. Annunziato, G. Pignataro, F. Di Renzo, Pharmacology of brainNa+/Ca2+ exchanger: from molecular biology to therapeutic perspec-tives, Pharmacol. Rev. 56 (2004) 633–654.

[3] R.J. White, I.J. Reynolds, Mitochondria and Na+/Ca2+ exchange bufferglutamate-induced calcium loads in cultured cortical neurons, J. Neu-rosci. 15 (1995) 1318–1328.

[4] K. Takuma, T. Matsuda, H. Hashimoto, J. Kitanaka, S. Asano, Y.Kishida, A. Baba, Role of Na+/Ca2+ exchanger in agonist-inducedCa2+ signalling in cultured rat astrocytes, J. Neurochem. 67 (1996)1840–1845.

[5] K.R. Hoyt, S.R. Arden, E. Aizenman, I.J. Reynolds, Reverse Na+/Ca2+

exchange contributes to glutamate-induced intracellular Ca2+ concen-tration increases in cultured rat forebrain neurons, Mol. Pharmacol. 53(1998) 742–749.

[6] N.S. Ranciat-McComb, K.S. Bland, J. Huschenbett, L. Ramonda, M.Bechtel, A. Zaidi, M.L. Michaelis, Antisense oligonucleotide suppres-sion of Na+/Ca2+ exchanger activity in primary neurons from rat brain,Neurosci. Lett. 294 (2000) 13–16.

[7] J.P. Smith, L.A. Cunningham, L.D. Partridge, Coupling of AMPA

receptors with the Na+/Ca2+ exchanger in cultured rat astrocytes, Brain.Res. 887 (2000) 98–109.

[8] E. Korkotian, D. Holcman, M. Segal, Dynamic regulation of spine-dendrite coupling in cultured hippocampal neurons, Eur. J. Neurosci.20 (2004) 2649–2663.

Calcium

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

A. Minelli et al. / Cell

[9] L. Fierro, R. DiPolo, I. Llano, Intracellular calcium clearance in Purk-inje cell somata from rat cerebellar slices, J. Physiol. 510 (1998)499–512.

10] J.H. Goldberg, G. Tamas, D. Aronov, R. Yuste, Calcium microdomainsin aspiny dendrites, Neuron 40 (2003) 807–821.

11] D. Jeon, Y.M. Yang, M.J. Jeong, K.D. Philipson, H. Rhim, H.S. Shin,Enhanced learning and memory in mice lacking Na+/Ca2+ exchanger2, Neuron 38 (2003) 965–976.

12] Y.T. Kim, Y.J. Park, S.Y. Jung, W.S. Seo, C.K. Suh, Effects ofNa+/Ca2+ exchanger activity on the alpha-amino-3-hydroxy-5-methyl-4-isoxazolone-propionate-induced Ca2+ influx in cerebellar Purkinjeneurons, Neuroscience 131 (2005) 589–599.

13] H. Reuter, H. Porzig, Localization and functional significance of theNa+/Ca2+ exchanger in presynaptic boutons of hippocampal cells inculture, Neuron 15 (1995) 1077–1084.

14] A. Bouron, H. Reuter, A role for intracellular Na+ in the regulationof synaptic transmission and turnover of the vesicular pool in culturedhippocampal cells, Neuron 17 (1996) 969–978.

15] A. Doi, Y. Kakazu, N. Akaike, Na+/Ca2+ exchanger in GABAergicpresynaptic boutons of rat central neurons, J. Neurophysiol. 87 (2002)1694–1702.

16] M. Taglialatela, L.M. Canzoniero, E.J. Cragoe Jr., G. Di Renzo, L.Annunziato, Na+/Ca2+ exchange activity in central nerve endings. II.Relationship between pharmacological blockade by amiloride ana-logues and dopamine release from tuberoinfundibular hypothalamicneurons, Mol. Pharmacol. 38 (1990) 393–400.

17] G. Di Renzo, S. Amoroso, A. Bassi, A. Fatatis, M. Cataldi, A.M.Colao, G. Lombardi, L. Annunziato, Role of the Na+/Ca2+ and Na+/H+

antiporters in prolactin release from anterior pituitary cells in primaryculture, Eur. J. Pharmacol. 294 (1995) 11–15.

18] S. Amoroso, S. Sensi, G. Di Renzo, L. Annunziato, Inhibition ofthe Na+/Ca2+ exchanger enhances anoxia and glucopenia-induced[3H]aspartate release in hippocampal slices, J. Pharmacol. Exp. Ther.264 (1993) 515–520.

19] S. Amoroso, M. De Maio, G.M. Russo, A. Catalano, A. Bassi, S. Mon-tagnani, G.F. Di Renzo, L. Annunziato, Pharmacological evidence thatthe activation of the Na+/Ca2+ exchanger protects C6 glioma cells dur-ing chemical hypoxia, Br. J. Pharmacol. 121 (1997) 303–309.

20] S. Amoroso, A. Tortiglione, S. Secondo, A. Catalano, S. Montagnani,G. Di Renzo, L. Annunziato, Sodium nitroprusside prevents chemicalhypoxia-induce cell death through iron ions stimulating the activity ofthe Na+/Ca2+ exchanger in C6 glioma cells, J. Neurochem. 74 (2000)1505–1513.

21] R. Fern, Intracellular calcium and cell death during ischemia in neonatalrat white matter astrocytes in situ, J. Neurosci. 18 (1998) 7232–7243.

22] L. Kiedrowski, A. Czyz, G. Baranauskas, X.F. Li, J. Lytton, Differentialcontribution of plasmalemmal Na/Ca exchange isoforms to sodium-dependent calcium influx and NMDA excitotoxicity in depolarizedneurons, J. Neurochem. 90 (2004) 117–128.

23] D.A. Nicoll, S. Longoni, K.D. Philipson, Molecular cloning and func-tional expression of the cardiac sarcolemmal Na+–Ca2+ exchanger,Science 250 (1990) 562–565.

24] S.L. Lee, A.S. Yu, J. Lytton, Tissue-specific expression ofNa+–Ca2+ exchanger isoforms, J. Biol. Chem. 269 (1994) 14849–14852.

25] Z. Li, S. Matsuoka, L.V. Hryshko, D.A. Nicoll, M.M. Bersohn, E.P.Burke, R.P. Lifton, K.D. Philipson, Cloning of NCX2 isoform of theplasma membrane Na+–Ca2+ exchanger, J. Biol. Chem. 269 (1994)17434–17439.

26] D.A. Nicoll, B.D. Quedneau, Z. Qui, Y.R. Xia, A.J. Lusis, K.D. Philip-son, Cloning of a third mammalian Na+–Ca2+ exchanger, NCX3, J.Biol. Chem. 271 (1996) 24914–24921.

27] L. Yu, R.A. Colvin, Regional differences in expression of transcriptsfor Na+/Ca2+ exchanger isoforms in rat brain, Brain Res. Mol. BrainRes. 50 (1997) 285–292.

28] B. Linck, Z. Qiu, Z. He, Q. Tong, D.W. Higelmann, K.D. Philipson,Functional comparison of the three isoforms of the Na+/Ca2+ exchanger

[

[

41 (2007) 221–234 233

(NCX1, NCX2, NCX3), Am J Physiol. (Cell Physiol. 43) 274 (1998)C415–C423.

29] M. Papa, A. Canitano, F. Boscia, P. Castaldo, S. Sellitti, H. Porzig, M.Taglialatela, L. Annunziato, Differential expression of the Na+–Ca2+

exchanger transcripts and proteins in rat brain regions, J. Comp. Neurol.461 (2003) 31–48.

30] G. Pignataro, R. Gala, O. Cuomo, A. Tortiglione, L. Giaccio, P.Castaldo, R. Sirabella, C. Matrone, A. Canitano, S. Amoroso, G.F.Di Renzo, L. Annunziato, Two sodium/calcium exchanger geneproducts, NCX1 and NCX3, play a major role in the develop-ment of permanent focal cerebral ischemia, Stroke 35 (2004) 2566–2570.

31] G. Pignataro, A. Tortiglione, A. Scorziello, L. Giaccio, A. Sec-ondo, B. Severino, G. Caliendo, V. Santagada, S. Amoroso, G.F.Di Renzo, L. Annunziato, Evidence for a protective role played byNa+/Ca2+ exchanger in cerebral ischemia induced by middle cere-bral artery occlusion in male rats, Neuropharmacology 46 (2004) 439–448.

32] T. Thurneysen, D.A. Nicoll, K.D. Philipson, H. Porzig, Sodium/calciumexchanger subtypes NCX1, NCX2, and NCX3 show cell-specificexpression in rat hippocampal cultures, Mol. Brain Res. 107 (2002)145–156.

33] M. Papa, F. Boscia, A. Canitano, P. Castaldo, S. Sellitti, L. Annunziato,M. Taglialatela, Expression pattern of the ether-A-gogo-related (ERG)K+ channel-encoding genes ERG1, ERG2, and ERG3 in the adult ratcentral nervous system, J. Comp. Neurol. 466 (2003) 119–135.

34] A. Peters, S.L. Palay, H.deF. Webster, The Fine Structure of the Ner-vous System: Neurons and Their Supporting Cells, 3rd ed., OxfordUniversity Press, New York, 1991.

35] A. Privat, M. Gimenez-Ribotta, J.L. Ridet, Morphology of astrocytes,in: H. Kettenmann, B. Ransom (Eds.), Neuroglia, Oxford UniversityPress, New York, 1995, pp. 3–22.

36] K.D. Philipson, S. Longoni, R. Ward, Purification of the cardiacNa+–Ca2+ exchange protein, Biochim. Biophys. Acta 945 (1988)298–306.

37] F. Van Eylen, A. Bollen, A. Herchuelz, NCX1 Na+/Ca2+ exchangersplice variants in pancreatic islet cells, J. Endocrinol. 168 (2001)517–526.

38] S. He, A. Ruknudin, L.L. Bambrick, W. Jon Lederer, D.H. Schulze,Isoform-specific regulation of the Na+/Ca2+ exchanger in rat astrocytesand neurons by PKA, J. Neurosci. 18 (1998) 4833–4841.

39] M. Juhaszova, P. Church, M.P. Blaustein, E.F. Stanley, Location of cal-cium transporters at presynaptic terminals, Eur. J. Neurosci. 12 (2000)839–846.

40] A. Fairen, J. DeFelipe, J. Regidor, Nonpyramidal neurons, in: A. Peters,E.G. Jones (Eds.), Cerebral Cortex, Plenum Press, New York, 1984, pp.201–253.

41] R.S. Zucker, Calcium- and activity-dependent synaptic plasticity, Curr.Opin. Neurobiol. 9 (1999) 305–313.

42] A. Verkhratsky, R.K. Orkand, H. Kettenmann, Glial calcium: home-ostasis and signalling function, Physiol. Rev. 78 (1998) 99–141.

43] G. Perea, A. Araque, Glial calcium signalling and neuron-glia commu-nication, Cell Calcium 38 (2005) 375–382.

44] G. Perea, A. Araque, Properties of synaptically-evoked astrocyte cal-cium signal reveal synaptic information processing by astrocytes, J.Neurosci. 25 (2005) 2192–2203.

45] J. Grosche, V. Matyash, T. Moller, A. Verkhratsky, A. Reichembach, H.Kettenmann, Microdomains for neuron-glia interaction: parallel fibersignalling to Bergmann glial cells, Nat. Neurosci. 2 (1999) 139–143.

46] F. Aguado, J.F. Espinosa-Parrilla, M.A. Carmona, E. Soriano, Neuronalactivity regulates correlated network properties of spontaneous calciumtransients in astrocytes in situ, J. Neurosci. 22 (2002) 9430–9444.

47] O. Peters, C.G. Schipke, Y. Hashimoto, H. Kettenmann, Differentmechanisms promote astrocytic Ca2+ waves and spreading depressionin the mouse neocortex, J. Neurosci. 23 (2003) 9888–9896.

48] I. Dori, M. Petrou, J.G. Parnavelas, Excitatory transmitter aminoacid-containing neurons in the rat visual cortex: a light and electron

2 Calcium

[

[

[

[

[

[

[

[

[

[[

[

[

[

[

[

[

34 A. Minelli et al. / Cell

microscopic immunocytochemical study, J. Comp. Neurol. 290 (1989)169–184.

49] W.T. Kim, M.G. Rioult, A.H. Cornell-Bell, Glutamate-induced calciumsignalling in astrocytes, Glia 11 (1994) 173–184.

50] W.F. Goldman, P.J. Yarowsky, M. Juhaszova, B.K. Krueger, M.P.Blaustein, Sodium/calcium exchange in cortical astrocytes, J. Neurosci.14 (1994) 5834–5843.

51] V.A. Golovina, L.L. Bambrick, P.J. Yarowsky, B.K. Krueger, M.P.Blaustein, Modulation of two functionally distinct Ca2+ stores in astro-cytes: role of the plasmalemmal Na/Ca exchanger, Glia 16 (1996)296–305.

52] S. Kirischuk, H. Kettenmann, A. Verkhratsky, Na+/Ca2+ exchangermodulates kainite-triggered Ca2+ signalling in Bergmann glial cellsin situ, FASEB J. 11 (1997) 566–572.

53] B. Benz, G. Grima, K.Q. Do, Glutamate-induced homocysteic acidrelease from astrocytes: possible implication in glia-neuron signalling,Neuroscience 124 (2004) 377–386.

54] F.A. Chaudry, K.P. Lehre, M. van Lookeren Campagne, O.P. Ottersen,N.C. Danbolt, J. Storm-Mathisen, Glutamate transporters in glialplasma membranes: highly differentiated localizations revealed byquantitative ultrastructural immunocytochemistry, Neuron 15 (1995)711–720.

55] F. Conti, S. De Biasi, A. Minelli, J.D. Rothstein, M. Melone, EAAC1,a high-affinity glutamate transporter, is localized to astrocytes andGABAergic neurons besides pyramidal cells in the rat cerebral cortex,Cerebral Cortex 8 (1998) 108–116.

56] K.P. Lehre, N.C. Danbolt, The number of glutamate transporter sub-

type molecules at glutamatergic synapses: chemical and stereologicalquantification in young adult rat brain, J. Neurosci. 18 (1998) 8751–8757.

57] A. Minelli, P. Barbaresi, R.J. Reimer, R.H. Edwards, F. Conti, The glialglutamate transporter GLT-1 is localized both in the vicinity of and at

[

41 (2007) 221–234

distance from axon terminals in the rat cerebral cortex, Neuroscience108 (2001) 51–59.

58] N.C. Danbolt, Glutamate uptake, Prog. Neurobiol. 65 (2001) 1–105.59] J.D. Rothstein, M. Dykes-Hoberg, C.A. Pardo, L.A. Bristol, L. Jin,

R.W. Kuncl, Y. Kanai, M.A. Hediger, Y. Wang, J.P. Schielke, D.F.Welty, Knockout of glutamate transporters reveals a major role forastroglial transport in excitotoxicity and clearance of glutamate, Neuron16 (1996) 675–686.

60] K. Tanaka, K. Watase, T. Manabe, K. Yamada, K. Watanabe, K. Taka-hashi, H. Iwana, T. Nishikawa, N. Ichihara, T. Kikuchi, S. Okuyama,N. Kawashima, S. Hori, M. Takimoto, K. Wada, Epilepsy and exacer-bation of brain injury in mice lacking the glutamate transporter GLT-1,Science 276 (1997) 1699–1702.

61] S.I. Rapoport, Osmotic opening of the blood–brain barrier: principles,mechanisms, and therapeutic applications, Cell Mol. Neurobiol. 20(2000) 217–230.

62] A.K. Bhattacharjee, T. Nagashima, T. Kondoh, N. Tamaki, The effectsof Na+/Ca2+ exchange blocker on osmotic blood–brain barrier disrup-tion, Brain Res. 900 (2001) 157–162.

63] M. Ikeda, A.K. Bhattacharjee, T. Kondoh, T. Nagashima, N. Tamaki,Synergistic effect of cold mannitol and Na+/Ca2+ exchange blocker onblood brain barrier opening, Biochem. Biophys. Res. Commun. 291(2002) 669–674.

64] Y. Koyama, S. Matsui, S. Itoh, M. Osakada, A. Baba, T. Matsuda,The selective Na+/Ca2+ exchange inhibitor attenuates brain edema afterradiofrequency lesion in rats, Eur. J. Pharmacol. 489 (2004) 193–196.

65] S.J. Mulligan, B.A. MacVicar, Calcium transients in astrocyte endfeet

cause cerebrovascular constrictions, Nature 431 (2004) 195–199.

66] M. Zonta, M.C. Angulo, S. Gobbo, B. Rosengarten, K.A. Hossmann,T. Pozzan, G. Carmignoto, Neuron-to-astrocyte signalling is central tothe dynamic control of brain microcirculation, Nat. Neurosci. 6 (2003)43–50.