Molecular and Cellular Neuroscience 17, 151–166 (2001)
doi:10.1006/mcne.2000.0937, available online at http://www.idealibrary.com on MCN
A
Calsyntenin-1, a Proteolytically ProcessedPostsynaptic Membrane Protein with aCytoplasmic Calcium-Binding Domain1
Lorenz Vogt,* ,2 Sabine P. Schrimpf,* ,2 Virginia Meskenaite,*Renato Frischknecht,* Jochen Kinter,* Dino P. Leone,*Urs Ziegler,† and Peter Sonderegger* ,3
*Institute of Biochemistry and †Institute of Anatomy, University of Zurich,Winterthurerstrasse 190, CH-8057 Zurich, Switzerland
1bita(
smnwssocpafrrc1aprppiatt
In a screen for proteins released from synapse-formingspinal cord neurons, we found the proteolytically cleavedN-terminal fragment of a transmembrane protein local-ized in the postsynaptic membrane of both excitatory andinhibitory synapses. We termed this protein calsyntenin-1,because it binds synaptic Ca21 with its cytoplasmic do-
ain. By binding Ca21, calsyntenin-1 may modulate Ca21-mediated postsynaptic signals. Proteolytic cleavage ofcalsyntenin-1 in its extracellular moiety generates a trans-membrane stump that is internalized and accumulated inthe spine apparatus of spine synapses. Therefore, thesynaptic Ca21 modulation by calsyntenin-1 may be subjectto regulation by extracellular proteolysis in the synapticcleft. Thus, calsyntenin-1 may link extracellular proteoly-sis in the synaptic cleft and postsynaptic Ca21 signaling.
INTRODUCTION
Extracellular serine proteases play a role in the reg-ulation of learning and memory. For example, neuraltissue plasminogen activator (tPA) is elevated in hip-pocampal neurons upon induction of long-term poten-tiation (LTP; Qian et al., 1993) and in cerebellar Purkinje
eurons of rats after training for a complex motor learn-ng task (Seeds et al., 1995). In line with these observa-ions, a perturbance in late-phase LTP was found inippocampal slices of mice deficient in tPA (Frey et al.,
1 The mouse, chicken, and Drosophila melanogaster calsyntenin-1equences described in this paper are entered in GenBank underccession Nos. AJ289016, AJ289017, and AJ289018, respectively.2 Contributed equally to this work.
tc
3 To whom correspondence should be addressed. Fax: 141 1 635 6831. E-mail: [email protected].
996; Huang et al., 1996) and after pharmacologicallockage of tPA activity (Baranes et al., 1998). Accord-
ngly, enhanced LTP was found in mice overexpressingPA in CNS neurons (Madani et al., 1999) and after bathpplication of recombinant tPA to hippocampal slicesBaranes et al., 1998).
Although these observations implicate extracellularerine proteases in neural plasticity, the molecularechanisms of proteolytic processes at synapses have
ot been worked out. To identify synaptic proteinshich are proteolytically cleaved in the extracellular
pace, we screened for released protein fragments inynapse-forming neuronal cultures. In a pilot analysisf the supernatant of synapse-forming primary CNSultures, we found more than 100 released proteins,resumably representing a mixture of secreted proteinsnd proteolytically cleaved protein fragments derivedrom both neurons and nonneuronal cells. In order toeduce complexity and to restrict the screen to proteinseleased from neurons, we used a compartmental cellulture system (Campenot, 1979; Osterwalder et al.,996; Stoeckli et al., 1989) that allows the selective met-bolic labeling of proteins released from the neuronalrocesses. In cultures of dissociated spinal cord neu-ons we found four proteins that are actively trans-orted into the axodendritic compartment. Here, weresent the purification, the partial amino acid sequenc-
ng, and the cDNA cloning of one of them. The deducedmino acid sequence revealed a novel, unclassified pro-ein containing a large extracellular N-terminal part, aransmembrane segment, and a cytoplasmic tail. Thus,
he 115-kDa protein isolated from the axodendriticompartment represents a proteolytically released frag-
151
tmwetsptmaw
w
152 Vogt et al.
ment. By immunocytochemical and protein analyses,the protein was characterized as a postsynaptic mem-brane protein with a Ca21-binding cytoplasmic domain.Based on these characteristics, we termed it calsynte-nin-1. In consideration of its structural and functionalfeatures, as well as based on its localization, we suggestcalsyntenin-1 to be a dynamic modulator of postsynap-tic Ca21 signals. Related sequences in genomic andcDNA databases indicate that calsyntenin-1 may be arepresentative of a novel gene family.
RESULTS
A 115-kDa Protein Released from the Neurites ofEmbryonic Chicken Spinal Cord Neurons Is aProteolytic Fragment of Calsyntenin-1, aTransmembrane Protein with a Highly AcidicCytoplasmic Domain
In a search for proteins released from neurites wecultivated dissociated spinal cord neurons in a com-partmental cell culture system that provides separateaccess to neuronal cell bodies and neurites (Fig. 1C). Sixdays after plating, when the side compartments hadbecome densely populated by neurites (Fig. 1B), thenewly synthesized proteins were metabolically labeledby adding fresh medium containing [35S]methionine tohe center compartment. After 40 h, the conditioned
edia of both the center and the side compartmentsere harvested and subjected to two-dimensional gel
lectrophoresis followed by fluorographic detection ofhe newly synthesized proteins (Figs. 1D and E). Ashown in Fig. 1D, the supernatant of the center com-artment contained a relatively large number of pro-
eins. In contrast, the supernatant of the side compart-ent contained only four strongly labeled protein spots
nd a few very weak spots. In accordance with previousork with the same culture system (Stoeckli et al., 1989),
the abundance of the weak spots in the side compart-ment was below 10% of the abundance of the corre-sponding spots of the center compartment. Therefore,the weak spots with clearly identifiable counterparts inthe medium of the center compartments were taken asreference proteins for proteins diffused into the sidecompartment across the connecting film of medium. Inthree independent experiments, the four strongly la-beled proteins were found in the side compartments atamounts ranging between 30 and 80% of their counter-parts in the center compartment. These proteins ex-
ceeded the 10% threshold determined for proteins thathad reached the side compartment by diffusion. There-
fore, we concluded that these four proteins had to bederived from the neurites of the side compartment. Oneof them (Fig. 1E, arrowhead 1) was previously identi-fied as neuroserpin, an axonally secreted serine pro-tease inhibitor (Osterwalder et al., 1996). The other threeproteins were unknown. The protein with an apparentmolecular weight of 115 kDa and a pI of 5.9 to 6.3
FIG. 1. Identification of proteins released from the neurites of ven-tral spinal cord neurons with a compartmental cell culture system. Inthe compartmental cell culture system, the cell culture surface issubdivided into three compartments by a Teflon divider (C). Adjacentcompartments remain connected by a thin film of medium that pre-vents cells migrating between the compartments, but that is wideenough to allow outgrowing neurites to cross into the adjacent sidecompartment. When dissociated neurons from the ventral halves ofE6 chicken spinal cords were seeded into the center compartment (A),the first neurites reached the side compartments after 3.5 days. After6 days, when the neurites had extended up to 8 mm into the sidecompartments (B), the newly synthesized proteins were metabolicallylabeled by adding [35S]methionine to the medium of the center com-partment. Proteins released into the conditioned media of both thecentral and the side compartments were analyzed by two-dimen-sional SDS–PAGE and fluorography (D and E). The 55-kDa protein(arrowhead 1) is neuroserpin (Osterwalder et al., 1996). The protein of115 kDa and a pI between 5.9 and 6.3 (arrowhead 2), calsyntenin-1,
as isolated, cloned, and characterized.
(arrowhead 2 in Fig. 1E) was isolated and characterizedas reported in the following sections.
s
153Calsyntenin-1, a Regulated Postsynaptic Ca21-Binding Protein
For the purification of the 115-kDa protein, we usedthe conditioned medium of dissociated cultures of theventral halves of spinal cords from E6 chickens. Theproteins of the culture supernatant were separated byanion-exchange chromatography. Fractions containingthe 115-kDa protein were subjected to preparative two-dimensional SDS–PAGE. The amino acid sequences ofthe N-terminus and seven internal peptides were deter-mined and used to design degenerated primers forRT-PCR using total RNA from E14 chicken brain astemplate. The longest PCR product had a length of 2.2kb. It contained a single open reading frame (ORF) thatencoded all previously determined amino acid se-quences (Fig. 2). Screening an oligo(dT)-primed E14chicken brain cDNA library (Zuellig et al., 1992) withthis fragment as a probe yielded clones containing ad-ditional 39 sequence of the ORF and the 39-untranslatedregion. The composite cDNA contained an ORF of 2850nt (starting from the amino-terminus of the purifiedprotein). The hydropathy plot provided evidence for asingle transmembrane segment of 19 amino acids closeto the C-terminus (Fig. 2). Therefore, we concluded thatthe mature protein was composed of an extracellularN-terminal moiety of 831 amino acids, a transmem-brane segment of 19 amino acids, and a cytoplasmicmoiety of 100 amino acids. Based on the presumedstructural characteristics as a type I transmembraneprotein, the 115-kDa protein isolated from the superna-tant of E6 spinal cord cultures represents the proteolyti-cally cleaved N-terminal fragment of the full-lengthtransmembrane protein. The exact location of the cleav-age site within the sequence of full-length calsyntenin-1remains to be determined. Based on the location of thetryptic peptides (boxed in gray in Fig. 2), the releasedfragment isolated from the culture supernatant musthave a length of at least 750 amino acids (as countedfrom the N-terminus of the mature protein).
In a homology search we found a weak similarity oftwo adjacent N-terminal segments of calsyntenin-1 withthe extracellular domains of cadherins. However, thesesegments did not contain the complete signature pat-tern ([LIV]-x-[LIV]-x-D-x-N-D-[NH]-x-P) of the cad-herin domain (Takeichi, 1990). Thirty-nine of the 100amino acids of the cytoplasmic segment of calsynte-nin-1 are acidic. In the most acidic middle part, 18 of 20residues are acidic and the flanking sequences are en-riched in acidic residues as well. Similarly acidic seg-ments are characteristic for calsequestrin, calreticulin,and protein disulfide isomerase (Edman et al., 1985;Fliegel et al., 1987, 1989). These proteins are essential for
the storage of Ca21 in the sarcoplasmic reticulum ofkeletal muscle cells and the endoplasmic reticulum of
nonmuscle cells, due to their capacity to bind largenumbers of Ca21 ions with low affinity (Baksh andMichalak, 1991; Lebeche et al., 1994; Ohnishi and Re-ithmeier, 1987).
Species Homologues of Calsyntenin-1 in Humanand Mouse Exhibit a High Degree of StructuralConservation
The cDNA of mouse calsyntenin-1 was obtained byRT-PCR and subsequent screening of brain cDNA li-braries. Based on the sequence of overlapping clones, asingle ORF of 2937 nt, encoding a peptide of 979 aminoacids, was defined (Fig. 2). The major part of the cDNAof human calsyntenin-1 was found by searching theTHC (tentative human consensus sequence) databasewith the THC Blast program. Seven THCs (THC176438,THC178825, THC195843, THC200424, THC192325,THC211114, and THC211115) with homology to thecDNA of chicken calsyntenin-1 were identified andused to compose a partial sequence of the human cDNAlacking a segment of the 59 end and two internal seg-ments. The gaps were closed by RT-PCR. The putativetranslation start codon and a segment of 59 UTR se-quence were found by screening a human brain cDNAlibrary. Thus, we obtained a human cDNA sequencewith an ORF of 2943 bp that was 100% identical withKIAA0911, a cDNA resulting from a screen for brain-specific proteins that was not further characterized (Na-gase et al., 1998a).
The sequences of human and mouse calsyntenin-1starting with the amino acid 29 correspond to the se-quence of the N-terminal peptide of chicken calsynte-nin-1 (Fig. 2). The deduced amino acid sequences of thehuman and the mouse orthologs had an identity of 86.4and 84.9%, respectively, to chicken calsyntenin-1 (Table1). The first 28 amino acids contain a hydrophobic seg-ment conforming to the consensus signal sequences(von Heijne, 1983). The translation start site of humanand mouse calsyntenin-1 is preceded by nucleotides inagreement with the consensus sequence described byKozak (1987).
Related Sequences Found in Databases Suggest aCalsyntenin Gene Family in Vertebrates andCalsyntenin-like Genes in Drosophilamelanogaster and Caenorhabditis elegans
In a database search we found an unclassified humancDNA, KIAA0726 (Nagase et al., 1998b), with a se-
quence identity of 53.0, 52.3, and 54.5% with human,mouse, and chicken calsyntenin-1, respectively (Table
m
ous g
154 Vogt et al.
1). In genomic nucleotide databases, evidence for atleast one additional human gene related to calsynte-nin-1 was found (Accession No. AC022479). Therefore,
FIG. 2. The amino acid sequence deduced from the ORF of the cDNAORF in the human (hs), the mouse (mm), and the chicken (gg) cDNA apurified from the culture supernatant of embryonic chicken spinal cputative signal peptide (underlined). The N-terminal and the internalThe black box marks the putative transmembrane domain. Two segmalternative splicing, are printed in small letters. Stretches with contiguThe stop codons are marked with asterisks.
we concluded that calsyntenin-1 may be a representa-tive of a gene family.
Dl
Genes with a structural relationship to vertebratecalsyntenin-1 were also found in the databases for D.
elanogaster and C. elegans. In the Genome Annotation
alsyntenin-1. The amino acid sequences deduced from the single longgned. Based on the N-terminal sequence determined for calsyntenin-1eurons, the first 28 amino acids of the ORF are characterized as theide sequences of purified calsyntenin-1 are marked with gray boxes.of 10 and 17 amino acids, respectively, which probably are due to
lutamic acid residues in the cytoplasmic moiety are printed in italics.
of cre aliord n
peptents
atabase of Drosophila (GadFly), a single calsyntenin-1-ike gene was found (Accession No. GC11059). Based on
qc
N dpsipet
hhtdscec
155Calsyntenin-1, a Regulated Postsynaptic Ca21-Binding Protein
six overlapping ESTs (GM09293, HL03914, LD07408,LD 11689, GM10465, LD06216) we have determined thesequence of the corresponding cDNA. The deducedprotein exhibits an amino acid sequence identity ofapproximately 35% with vertebrate calsyntenin-1. Afurther calsyntenin-1-related gene, B0034.3 (Wilson etal., 1994) with Accession No. AAC38816, was found inC. elegans (see Table 1).
The Cytoplasmic Segment of Calsyntenin-1 BindsCalcium Ions
The clustered occurrence of acidic amino acids is atypical trait of high-capacity, low-affinity Ca21-bindingproteins found in vesicular Ca21 stores, such as calse-
uestrin (Yano and Zarain-Herzberg, 1994) and calreti-ulin (Krause and Michalak, 1997). To test for the Ca21-
binding capacity of calsyntenin-1, we generated afusion protein of its cytoplasmic segment with the in-tein tag. As shown in Fig. 3, a dose-dependent 45Casignal overlapping with the calsyntenin-1/intein fusionprotein was found when the nitrocellulose filters wereincubated with Ca21 concentrations of 0.5, 1, and 5 mM.
o Ca21 binding was observed with a fusion proteincomposed of the bacterial maltose-binding protein andintein (MBP/int in Fig. 3). The addition of 50 mM ormore Ca21 caused the soluble calsyntenin-1/intein, butnot MBP/intein, to precipitate (not shown). The Ca21-induced precipitation of calsyntenin-1/intein was pre-
Note. Percentage of amino acid identity between different speciesomologues of calsyntenin-1 and calsyntenin-1-related proteins. hs,uman calsyntenin-1; mm, mouse calsyntenin-1; gg, chicken calsyn-
enin-1; dm, calsyntenin-1-like protein of Drosophila melanogaster, asetermined form cDNA. CG11059, calsyntenin-1-like protein of Dro-ophila melanogaster, as deduced from genomic sequences. B0034.3,alsyntenin-1-like protein deduced from genomic sequences of C.legans. KIAA0726, calsyntenin-1-like protein deduced from brainDNA.
vented by the addition of an equimolar concentration ofEDTA. Because precipitation was not observed at Ca21
wi
concentrations below 50 mM, we concluded that cross-bridging between cytoplasmic domains of calsyntenin-1involves low-affinity binding of Ca21 to sites which areclearly distinct from the high-affinity sites detected by45Ca binding on the nitrocellulose membranes. In sum-mary, these results indicate that the cytoplasmic do-main of calsyntenin-1 exhibits both high-affinity andlow-affinity binding of Ca21.
In Situ Hybridization Reveals a PredominantlyNeuronal Expression of Calsyntenin-1
Northern blot analysis of poly(A)-enriched RNAfrom adult human tissues revealed a single species ofcalsyntenin-1 mRNA of approximately 5 kb (Fig. 4C).The highest expression of calsyntenin-1 mRNA wasobserved in brain. Low signals were detected in heart,placenta, skeletal muscle, and kidney. No transcriptwas found in lung and liver.
In situ hybridization on cryosections from a E18mouse revealed a strong cellular expression of calsyn-tenin-1 mRNA in the gray matter of the central and theperipheral nervous system (Fig. 4A). No calsyntenin-1mRNA was detected in nonneural tissues, except in thesubmandibular gland. Control sections processed withthe sense probe showed no staining (data not shown).In the adult mouse, calsyntenin-1 mRNA was abundantin all areas of the gray matter (Fig. 4B). Inspection athigher magnification indicated a neuronal expressionpattern in all areas of the CNS and the PNS. Most, if not
FIG. 3. Demonstration of the calcium-binding capacity of the cyto-plasmic moiety of calsyntenin-1. The Ca21-binding capacity of thecytoplasmic moiety of calsyntenin-1 was investigated by 45Ca autora-
iography. Recombinant cytoplasmic domain of calsyntenin-1 wasroduced in bacteria as a calsyntenin-1/intein fusion protein (Cal-yn/int). As a control for unspecific Ca21 binding, the maltose-bind-ng protein fused to the intein tag (MBP/int) was used. In the leftanel, the two fusion proteins are visualized by amido black afterlectrotransfer onto a nitrocellulose membrane. The three panels onhe right are autoradiographs of identical nitrocellulose membranes
ith adjacent lanes containing 5 mg MBP/intein and calsyntenin-1/ntein, respectively, after incubation with 0.5, 1, and 5 mM 45Ca21.
ab
156 Vogt et al.
FIG. 4. Expression patterns of calsyntenin-1 mRNA and protein. (A and B) Localization of calsyntenin-1 mRNA by in situ hybridizationwith digoxygenin-labeled complementary RNA. (A) Sagittal section of an E18 mouse. Labeled regions include, the neocortex (nc), thehippocampal formation (hi), the caudate putamen (cpu), the thalamus (th), the tectum (te), the hypothalamus (hyth), the cerebellum (ce),the pons (po), the trigeminal ganglion (tg), the dorsal root ganglia (drg), the olfactory epithelium (oe), the submandibular gland (sg), andthe intestine (in). The staining of the intestine (in) was found to be due to endogenous alkaline phosphatase activity. (B) Coronal sectionof an adult mouse brain. Note that calsyntenin-1 mRNA is found exclusively in cells of the gray matter. In the white matter, such as thecorpus callosum (cc), no calsyntenin-1 expression was found. (C) Northern blot analysis of calsyntenin-1 mRNA in adult human tissues.Two micrograms of purified poly(A)1 RNA per lane form heart (He), brain (Br), placenta (Pl), lung (Lu), liver (Li), skeletal muscle (Sm),
nd kidney (Ki) were analyzed with radiolabeled cDNA fragments of human calsyntenin-1. The molecular size scale is in kb. (D) Westernlot analysis of chicken calsyntenin-1 protein. 150 mg of tissue extract from adult chicken brain (Br), heart (He), liver (Li), testis (Te), and
cerebrospinal fluid (CSF) was subjected to SDS–PAGE and immunoblotting using polyclonal antibodies R63 (left) and R71 (right) againstcalsyntenin-1. The molecular weight scale is in kilodaltons. The two bands detected with antibody R63 in brain extract (Br) correspond tofull-length calsyntenin-1 (top band, 150 kDa) and the released cleavage product (bottom band, 115 kDa). In cerebrospinal fluid, only thereleased cleavage product was found. Antibody R71 recognized a protein migrating at an apparent MW of approximately 34 kDa. (E)Western blot analysis of chicken calsyntenin-1 protein heterologously expressed in HEK 293T cells. HEK 293T cells were transientlytransfected with pcDNA 3.1 containing the ORF of calsyntenin-1 and compared by Western blotting with mock-transfected cells. AntibodyR63 recognized full-length calsyntenin-1 and the 115-kDa released cleavage product in the cell lysates (ly). In the supernatant (sup),antibody R63 detected only a band at approximately 115 kDa, corresponding to the released N-terminal cleavage product. Antibody R71only recognized full-length calsyntenin-1 in the cell lysates. Note that the released 115-kDa fragment in the supernatant was not recognizedby antibody R71. (F) Schematic drawing indicating the proteolytic cleavage site (arrow) on the calsyntenin-1 protein and the location ofthe recombinant peptide segments used for raising the R63 (shadowed) and the R71 (hatched) antibodies in the complete sequence ofmature calsyntenin-1. Note that antibody R63 recognizes both the full-length form of calsyntenin-1 and the N-terminal cleavage product.
In contrast, antibody R71 recognizes the full-length form and transmembrane stump generated by the proteolytic cleavage of calsyntenin-1.The transmembrane domain (TM) is marked in black. Bars: A, 2.5 mm; B, 1.0 mm.
157Calsyntenin-1, a Regulated Postsynaptic Ca21-Binding Protein
all neurons, expressed calsyntenin-1 mRNA, yet consid-erable differences in the expression level were found.
Calsyntenin-1 Occurs as a Full-LengthTransmembrane Protein, a Membrane-BoundC-Terminal Cleavage Product, and a SolubleN-Terminal Cleavage Product
To analyze the tissue distribution of full-length cal-syntenin-1 and its cleavage products two antibodieswere raised. Antibody R63, raised against the N-termi-nal 267 amino acids of the mature protein, was de-signed to detect the full-length form of calsyntenin-1and the N-terminal part resulting from the proteolyticcleavage (Fig. 4F). Antibody R71, raised against a seg-ment of 87 amino acids located adjacent to the trans-membrane domain, was designed to detect the full-length form of calsyntenin-1 and the transmembranestump generated by the proteolytic cleavage (Fig. 4F).Both antibodies were affinity purified with their corre-sponding peptide antigens conjugated to CNBr-acti-vated Sepharose. The reactivity of the affinity-purifiedantibodies was tested by Western blotting of cell lysatesand culture supernatants of HEK 293T cells transientlytransfected with pcDNA 3.1 containing the ORF of cal-syntenin-1 and their mock-transfected counterparts(Fig. 4E). In cell lysates, antibody R63 recognized twobands at apparent MWs of 150 and 115 kDa, corre-sponding to full-length calsyntenin-1 and its N-terminalcleavage product, respectively. As expected, only theN-terminal cleavage product was found in the culturesupernatant. Antibody R71 recognized a single band of150 kDa, the putative full-length protein in the celllysates, but did not react with the released N-terminalfragment. The absence of aminoreactive bands in mock-transfected cells identifies the labeled protein bands inthe transfected cells unequivocally as components ofcalsyntenin-1. Although calsyntenin-1, expressed tran-siently in HEK 293T cells, was extensively cleaved, nofragment corresponding to the transmembrane stumpwas detected in the cell lysates with antibody R71.Because antibody R71 reacted strongly with the puri-fied peptide used as antigen for its generation (notshown), we concluded that the absence of the trans-membrane stump in HEK 293T cell lysates was due to arapid degradation.
In Western blots of tissue extracts, calsyntenin-1-im-munoreactive bands were found exclusively in brainextract and in the cerebrospinal fluid (CSF; Fig. 4D).Extracts of all the other tissues that were tested, includ-
ing heart, liver, and testis (Fig. 4D), as well as kidney,lung, and spleen (not shown) did not exhibit calsynte-
nin-1 immunoreactivity. In brain extracts of adult chick-ens, two bands with apparent MWs of 150 and 115 kDawere found with antibody R63, designed for detectingthe full-length form of calsyntenin-1 and its N-terminalcleavage product. The 115-kDa band comigrated withthe protein initially identified with the compartmentalculture system as a released protein of the axodendriticcompartment of spinal cord neurons. The 150-kDa bandrepresents most likely the full-length form of calsynte-nin-1, based on the estimated size of the released frag-ment and the length of the transmembrane and cyto-plasmic segments. In the cerebrospinal fluid, antibodyR63 recognized only a single band of 115 kDa thatcorresponds to the soluble N-terminal cleavage productof calsyntenin-1. With antibody R71, designed to detectthe full-length form of calsyntenin-1 and its transmem-brane stump generated by proteolytic cleavage, a singleband at 34 kDa was found in brain extracts. No immu-noreactivity was detected at 150 kDa, the putative sizeof full-length calsyntenin-1. Because R71 showed a rea-sonably strong reaction with full-length calsyntenin-1overexpressed in HEK 293T cells (Fig. 4E), we explainedthe absence of a detection of full-length calsyntenin-1with R71 in brain extracts with a relatively low avidityof the R71 antibody, compared with R63. Taken to-gether, these results indicate that full-length calsynte-nin-1 and its cleavage products coexist in brain tissue.The N-terminal 115-kDa fragment of calsyntenin-1 thatis solubilized after proteolytic cleavage is also found inthe CSF.
Cell-Surface-Bound Calsyntenin-1 Is Colocalizedwith Established Synaptic Marker Proteins
Immunoperoxidase staining of tissue sections of thehippocampus (Fig. 5A) and the cerebral cortex (notshown) revealed that calsyntenin-1 was abundant insynapse-rich regions. At higher magnification, a punc-tate appearance of the immunostaining in the neuropil(Fig. 5A, inset) was found, suggesting a synaptic local-ization of calsyntenin-1.
A detailed study of the subcellular location of cell-surface-associated calsyntenin-1 was performed in cul-tures of dissociated hippocampal neurons. Establishedsynaptic markers, such as synaptophysin, the a2 sub-unit of the GABAA receptor, and the GluR2 subunit ofthe AMPA receptor were used as markers for presyn-aptic terminals and postsynaptic membranes, respec-tively. As demonstrated in Figs. 5B–5D, calsyntenin-1immunoreactivity exhibited a patchy pattern along neu-
rite bundles. A very similar staining pattern was foundwith the antibodies against synaptophysin (Fig. 5B) and
wm lays (w
158 Vogt et al.
the GABAA receptor (Fig. 5C). In the overlay, the ma-jority of the large areas labeled with antibodies againstsynaptophysin and the GABAA receptor were at leastpartially superposed with the calsyntenin-1 immuno-reactivity. With a commercially available antibodyagainst a cytoplasmic epitope of GluR2, which requiredpermeablization of the cells and, therefore, stained bothsurface-exposed and internal AMPA receptors (Fig.5D), large immunoreactive patches were found in closeproximity to and sometimes partially overlapping withpatches of calsyntenin-1 immunoreactivity. Together,these results demonstrate the synaptic localization ofcalsyntenin-1.
Full-Length Calsyntenin-1 Is Restricted to thePostsynaptic Membrane, but It Is Not aComponent of the Postsynaptic Density
Preembedding immuno-EM with peroxidase-labeledantibodies located calsyntenin-1 in the postsynapticmembrane of synapses located on dendritic spines, ondendritic shafts, and on neuronal somas (Figs. 6A–6C).In some spine synapses, floccular immunoreactivitywas also found beneath the postsynaptic membrane.
FIG. 5. Synaptic localization of calsyntenin-1. (A) Calsyntenin-1 wasrat by immunolabeling with a peroxidase-conjugated secondary antibthe stratum radiatum, calsyntenin-1 was distributed in small discrete s(B–D) Colocalization of calsyntenin-1 with synaptic markers in dissociat E17, cultured for 3 weeks, and then double-labeled with antibosynaptophysin, the a2 subunit of the GABAA receptor, and the Ganti-calsyntenin-1 antibodies were visualized with FITC-labeled secoGABAA receptor (C), and AMPA receptor (D) were visualized with C
ith 0.1% saponin was used for the labeling of GluR2. Therefore, AMembranes were visualized. Colocalization was demonstrated in overith both FITC and Cy3. Bars: A, 50 mm; B–D, 5 mm.
Postembedding immunogold labeling of rat hippocam-pus embedded at low temperature confirmed the local-
ization of calsyntenin-1 in the postsynaptic membrane(Figs. 6D–6E). Both asymmetric synapses with roundvesicles and thick PSDs (Type 1 according to Gray(1959)) and a subpopulation of symmetric synapseswith pleomorphic vesicles and thin PSDs (Type 2) ex-hibited calsyntenin-1 immunoreactivity, confirming cal-syntenin-1 as a component of the postsynaptic mem-brane in both excitatory and inhibitory synapses.
The synaptic localization of calsyntenin-1 was furtherstudied by subcellular fractionation and the isolation ofsynaptosomes (Phelan and Gordon-Weeks, 1997). Asdemonstrated in Fig. 7, synaptosomes were enriched infull-length calsyntenin-1 and its cleavage products. Hy-potonic disruption of synaptosomes and treatment witha mild detergent resulted in the solubilization of allthree forms of calsyntenin-1. In contrast, typical mark-ers of the postsynaptic density, viz. PSD-95 and GluR1(O’Brien et al., 1998), remained in the particulate frac-tions (P4 and PSD, according to Phelan and Gordon-Weeks, 1997). The clearance of calsyntenin-1 from thePSD fraction indicates that its cytoplasmic segment isnot firmly associated with the subsynaptic molecularscaffold that corresponds to the PSD observed in theEM and that is operationally defined as the particulate
alized in a section of the CA1 region of the hippocampus of an adultA strong immunoreactivity was found in the dendritic layers. Withinalong dendritic or varicose fibers (inset, 15-fold higher magnification).ippocampal cultures. Hippocampal neurons were isolated from miceagainst calsyntenin-1 (top row) as well as antibodies specific for
subunit of the AMPA receptor, respectively (middle row). They antibodies (green). The antibodies against synaptophysin (B), thenjugated secondary antibodies (red). Note that the permeabilizationeceptors of the postsynaptic membrane and the internal subsynapticbottom panels) in which yellow color indicates simultaneous labeling
visuody.potsated hdiesluR2ndary3-coPA r
matter resulting after detergent treatment of synapto-somes.
ne
sa, srs A–
159Calsyntenin-1, a Regulated Postsynaptic Ca21-Binding Protein
The Transmembrane Fragment of ProteolyticallyCleaved Calsyntenin-1 Is Accumulated in theSpine Apparatus of Spine Synapses and theSubsynaptic Membranes of Shaft Synapses
To identify the fate of the transmembrane segment ofcalsyntenin-1 after proteolytic cleavage, we used R71, theantibody against the membrane-proximal segment for im-muno-EM with peroxidase- and gold-conjugated second-ary antibodies. With peroxidase the membranes of thespine apparatus in spine synapses were labeled (Figs. 8Aand 8B). In some synapses, weaker immunoreactivity wasalso found over the postsynaptic membrane (Fig. 8A).Similarly, a strong signal was found in the cisternal mem-branes found beneath a fraction of the synapses in den-dritic shafts and neuronal somas (not shown). With im-munogold, known as less sensitive, labeling was foundexclusively in association with the spine apparatus (Figs.8C–8E) and the subsynaptic cisternae of shaft synapses.Because no labeling of the subsynaptic membranous or-
FIG. 6. Ultrastructural localization of calsyntenin-1 in the postsynapof adult rat hippocampus was subjected to immuno-EM with the antiantibody for preembedding staining (A–C) or a gold particle-coupleperoxidase/preembedding method, calsyntenin-1 was found over thefloccular immunoreactivity was found in dendritic spines. In the popostsynaptic membranes of axospinous synapses. at, axon terminal;toward the postsynaptic membrane and the postsynaptic density. Ba
ganelles was found with the N-terminal antibody (Fig. 6),we concluded that the spine apparatus contained neither
m5
full-length calsyntenin-1 nor the N-terminal cleavageproduct. Therefore, the full-length as well as the 115-kDaform of calsyntenin-1 found in Western blots of synapto-somes (Fig. 7) can only be derived from the postsynapticmembranes. These results indicate that the proteolyticcleavage must occur at the cell surface, i.e., in the synapticcleft, and that the transmembrane stump is internalizedthereafter.
DISCUSSION
We have identified calsyntenin-1, a transmembraneprotein of the postsynaptic membrane of a subpopulationof both excitatory and inhibitory synapses in the CNS. Thecytoplasmic moiety of calsyntenin-1 binds Ca21 and,thereby, it may modulate and prolong Ca21 signals be-
eath the postsynaptic membrane. The extracellular moi-ty of calsyntenin-1 is proteolytically cleaved within the
embrane of spine synapses. The stratum radiatum of the CA1 regionntenin-1 antibody R63, using either a peroxidase-coupled secondaryondary antibody for postembedding labeling (D and E). Using theynaptic membrane of axospinous synapses (A–C). In some synapses,
bedding sections (D and E), the gold particles were associated withpine apparatus; pd, parental dendrite; sp, spine. The arrows pointC, 0.2 mm; D and E, 0.1 mm.
tic m-calsyd secpostsstem
embrane-proximal third at a site estimated to be located0–85 amino acids away from the transmembrane se-
dmot
160 Vogt et al.
quence. The resulting N-terminal segment, of at least 750amino acids and an apparent molecular weight of 115kDa, is released into the extracellular fluids of the CNS,including the cerebrospinal fluid. The transmembranestump is internalized and accumulates in the spine appa-ratus of spine synapses and in subsynaptic lamellar mem-branes of shaft synapses. This internalization after proteo-lytic cleavage of calsyntenin-1 removes its Ca21-binding
omain from the cytoplasmic side of the postsynapticembrane and translocates it to the cytoplasmic surface
FIG. 7. Subcellular fractionation localizes calsyntenin-1 in synapto-somes, but not in postsynaptic densities. Synaptosomes and postsynapticdensities were prepared from chicken and mouse brains by subcellularfractionation. Identical blots of SDS–PAGE gels of the subcellular frac-tions of adult chicken brains were probed with the antibodies R63,directed against the N-terminal, and R71, directed against the mem-brane-proximal segment, of the extracellular moiety of calsyntenin-1.From the subcellular fractions of adult mouse brains, identical blots wereprobed with antibody R63 against calsyntenin-1, an antiserum againstthe GluR1 subunit of the AMPA receptor, and an antibody againstPSD95, an established marker for the postsynaptic density. Note thatsynaptosomes are enriched for full-length calsyntenin-1 as well as forboth cleavage products. The detergent treatment (Triton I and Triton II,according to Phelan and Gordon-Weeks (1997)) used for the isolation ofpostsynaptic densities resulted in a complete loss of calsyntenin-1 fromthe particulate matter (lanes P4 and PSD), indicating that calsyntenin-1 isnot tightly associated with the postsynaptic density.
f the spine apparatus (Fig. 9). In homology screenshrough genomic and cDNA databases we found evi-
dence for two related genes. Calsyntenin-1 may thereforebe a representative of a novel gene family of membrane-associated Ca21-binding proteins.
The Cytoplasmic Domain of Calsyntenin-1 BindsCa21 at Concentrations Occurring duringPostsynaptic Ca21 Influx, Suggesting Calsyntenin-1as a Modulator of Postsynaptic Ca21 Signals
Our studies provide evidence for the presence ofhigh-affinity Ca21-binding sites, because we found Ca21
binding to the cytoplasmic domain of calsyntenin-1 at aconcentration as low as 0.5 mM. In parallel to the high-capacity, low-affinity Ca21-binding function of calse-
FIG. 8. Ultrastructural localization of the transmembrane fragmentof proteolytically cleaved calsyntenin-1 over the spine apparatus ofspine synapses. The stratum radiatum of the CA1 region of adult rathippocampus was subjected to immuno-EM with the anti-calsynte-nin-1 antibody R71, using either peroxidase-coupled secondary anti-bodies for preembedding staining (A and B) or gold-particle-coupledsecondary antibodies for postembedding (C) and preembedding la-beling (D and E). After preembedding staining peroxidase reactionproduct was found over the postsynaptic membrane and density (A)and associated with the spine apparatus (A and B). Postembedding(C) and preembedding (D and E) immunogold localization demon-strated the transmembrane fragment of proteolytically cleaved cal-syntenin-1 over the membranes of the spine apparatus of spine syn-apses. Note that no calsyntenin-1 immunoreactivity was observedover the internal membranes of the spine apparatus with antibodyR63, which recognizes full-length calsyntenin-1 (Fig. 6). at, axon ter-minal; sa, spine apparatus; pd, parental dendrite. The arrows point
toward the postsynaptic membrane and the postsynaptic density.Bars, A–E, 0.2 mm.
t
a
saS
NC
n
bm
t
161Calsyntenin-1, a Regulated Postsynaptic Ca21-Binding Protein
questrin, which exhibits a similar clustering of acidicresidues cumulating in a contiguous stretch of 14 acidicresidues, the cytoplasmic domain of calsyntenin-1 mayalso have the capacity for low-affinity Ca21 binding. TheCa21-binding capacity of peptides with contiguousacidic residues has been linked to a general cation-binding capacity rather than specific Ca21 sites. A com-parison of proteins with such acidic stretches suggestedthat the Ca21-binding capacity was proportional to thecontent of acidic residues (Lucero et al., 1994). X-ray crys-tallography suggested that the low-affinity binding ofCa21 occurred via intercalation of Ca21 between the acidicC-terminal segments of calsequestrin dimers (MacLennanand Reithmeier, 1998; Wang et al., 1998). Similarly, calsyn-enin-1 may bind Ca21 by intercalation between its cyto-
plasmic moieties which are held in an ordered parallelorientation by transmembrane anchorage.
Based on the localization in the postsynaptic mem-brane and the calcium-binding capacity of the cytoplas-mic domain, we speculate that calsyntenin-1 may play a
FIG. 9. Diagram of the protease-dependent translocation of thepostsynaptic Ca21 binding of calsyntenin-1. Full-length calsyntenin-1is found exclusively in the postsynaptic membrane. After proteolyticcleavage by an extracellular protease in the synaptic cleft, the trans-membrane stump is internalized and accumulated in the membraneof the spine apparatus. Internalization, thus, translocates calsyntenin-1-mediated Ca21 binding from the area directly beneath the postsyn-aptic membrane to the cytoplasmic surface of the spine apparatus.
role in Ca21-mediated signals beneath the postsynapticmembrane. Due to its anchorage in the postsynaptic
membrane, the cytoplasmic domain of calsyntenin-1becomes restricted to the zone immediately beneath thepostsynaptic membrane. Thus, it may bind calcium im-mediately after its influx via postsynaptic Ca21 channelsnd participate in postsynaptic Ca21 signaling, either as
a Ca21 sensor, transducing a transient Ca21 elevationinto a changed postsynaptic enzyme activity, or as aCa21 buffer, modulating the postsynaptic Ca21 tran-ient. In both functions, the restriction to the postsyn-ptic membrane may be an essential characteristic.tudies with Ca21 buffers have revealed that fixed buff-
ers, in contrast to mobile buffers, restrict the diffusion ofCa21 (Allbritton et al., 1992; Kasai and Petersen, 1994;
eher, 1998). They also decrease the peak values of freea21 and, by delayed release of Ca21, prolong Ca21
elevations. Therefore, as a fixed Ca21 buffer, calsynte-in-1 may temporarily retain Ca21 in the subsynaptic
zone and retard its dissipation.In either role, calsyntenin-1 may potentially be a
modulatory element in synaptic processes where tran-sient increases in intracellular Ca21 are of crucial impor-tance, such as LTP (Bliss and Collingridge, 1993) andLTD (Linden and Connor, 1995), as well as in coinci-dence detection within dendritic spines (for a reviewsee Zucker, 1999).
Internalization of the Transmembrane StumpMoves the Ca21-Binding, Cytoplasmic Domainfrom the Postsynaptic Membrane to the Surfaceof the Spine Apparatus
Proteolytic cleavage in the extracellular segment re-sults in the release of the major extracellular portion ofcalsyntenin-1. This soluble fragment of calsyntenin-1spreads in the extracellular fluids, as demonstrated byits accumulation in the cerebrospinal fluid. The remain-ing transmembrane stump is internalized into the spineapparatus. Due to its membrane topology, the Ca21-binding domain of the internalized transmembranestump covers the cytoplasmic surface of the spine ap-paratus. Thus, internalization may translocate the Ca21-
inding domain of calsyntenin-1 from the postsynapticembrane to the spine apparatus.Recently, the release of Ca21 from intracellular stores,
i.e., the ER and the spine apparatus, via activation of theIP3 receptors, was identified as an important contribu-ion to the Ca21 signal within dendritic spines (Finch
and Augustine, 1998; Takechi et al., 1998). The IP3-mediated Ca21 release is regulated by cytoplasmic Ca21
in a biphasic mode (Taylor, 1998). Release is low at both
low and high Ca21 concentrations, but favored at inter-mediate concentrations of 200–300 nM. By its capacity
l
Cb
ylofpfotsoTmrt
cC
Tm(L
PP
ihacctCcctp
t
162 Vogt et al.
to bind Ca21 the cytoplasmic domain of calsyntenin-1may modulate Ca21 effects on IP3-mediated Ca21 re-ease.
alsyntenin-1 Internalization May Be Regulatedy Proteolytic Cleavage
The comparison of the results of the immuno-EM anal-sis and subcellular fractionation indicated that the full-
ength form and both cleavage products of calsyntenin-1ccur in the postsynaptic membrane. In contrast, neitherull-length calsyntenin-1 nor the N-terminal cleavageroduct, but exclusively the transmembrane stump, was
ound in internal membranes. This complete segregationf the N-terminal and the C-terminal cleavage products tohe interstitial fluids and to the internal membranes, re-pectively, can only be explained by extracellular cleavagef calsyntenin-1 by a protease located in the synaptic cleft.he selectivity of the internalization process for the trans-embrane stump of calsyntenin-1 implicates a regulatory
ole of the proteolytic cleavage in the synaptic cleft for theranslocation of the Ca21-binding domain of calsyntenin-1
from the postsynaptic membrane to the surface of thespine apparatus (Fig. 9).
Neither the nature of the protease that cleaves cal-syntenin-1 nor the mechanism conferring selective in-ternalization of the transmembrane stump of calsynte-nin-1 are currently known. Several extracellular pro-teases have been reported to be expressed in thenervous system, including tPA, thrombin, neurotryp-sin, and neuropsin. For two of them, namely tPA andneuropsin, transcription has been reported to be regu-lated by neuronal activity (Chen et al., 1995; Qian et al.,1993). They are intriguing candidates for regulators ofcalsyntenin-1 internalization.
CONCLUDING REMARKS
Calsyntenin-1 is a transmembrane protein of thepostsynaptic membrane of both excitatory and inhibi-tory synapses in the CNS. By binding Ca21 with itsytoplasmic moiety, it may function as a modulator ofa21-mediated processes beneath the postsynaptic
membrane. Extracellular proteolytic cleavage of calsyn-tenin-1 results in internalization of its transmembranestump. Thereby, the Ca21-binding domain is translo-cated from the cytoplasmic side of the postsynapticmembrane to the surface of the spine apparatus (Fig. 9).Depending on its subcellular localization, which in turn
appears to be regulated via proteolytic cleavage by anextracellular protease residing in the synaptic cleft, cal-
Nt
syntenin-1 may modulate Ca21 transients locally eitherbeneath the postsynaptic membrane or around intracel-lular Ca21 stores. Because Ca21 influx and Ca21 releasefrom internal stores cooperate as the fast and the de-layed component of Ca21 transients in many synapticprocesses (Finch and Augustine, 1998; Takechi et al.,1998), it is conceivable that the proteolysis-dependenttranslocation of the Ca21 buffer domain of calsyntenin-1from the postsynaptic membrane to the spine apparatusrepresents a distinct downstream event in protease-mediated synaptic plasticity.
EXPERIMENTAL PROCEDURES
Identification of Proteins Released from Neurites
The compartmental cell culture system was set up asdescribed by Campenot (1979). Dissociated cells fromthe ventral halves of spinal cords of E6 chicken embryoswere cultivated in the center compartment of the com-partmental culture system as described previously(Sonderegger et al., 1984). The newly synthesized pro-teins were metabolically labeled by addition of [35S]me-thionine to the medium of the center compartment for40 h (Sonderegger et al., 1983, 1984; Stoeckli et al., 1989).
he culture medium of the center and the side compart-ents was subjected to two-dimensional SDS–PAGE
O’Farrell, 1975) followed by fluorography (Bonner andaskey, 1974).
urification and Microsequencing of Calsyntenin-1rotein
Spinal cord neurons from E6 chicken were cultivatedn 60-mm collagen-coated culture dishes for 7 days. Toarvest released proteins, the cells were washed twicend grown for 2–3 days in serum-free medium. Theonditioned medium was subjected to anion-exchangehromatography and the fractions were analyzed bywo-dimensional SDS–PAGE (O’Farrell, 1975). Afteroomassie blue staining, protein spots with the gel
oordinates of calsyntenin-1 were excised and pro-essed by SDS–PAGE using the funnel-well concentra-ion system (Lombard-Platet and Jalinot, 1993). Thus, arotein band containing approximately 20 mg of calsyn-
tenin-1 was obtained. After excision and destainingwith 40% n-propanol (LichroSolv grade, Merck), calsyn-tenin-1 was extracted with 0.2 M NH4HCO3, 50% ace-onitrile, and dried in a Speed Vac. For sequencing the
-terminus, concentrated calsyntenin-1 was electro-ransferred from the gel onto a PVDF membrane (Im-
wN
polprcsc
N
wc
163Calsyntenin-1, a Regulated Postsynaptic Ca21-Binding Protein
mobilon P, Millipore). The calsyntenin-1-containingarea on the PVDF membrane was localized by autora-diography and excised. The sequence was determinedby Edman degradation on a protein sequencer (Model477 A, Applied Biosystems, Inc.). For sequencing ofinternal peptides, tryptic digestion was carried outwithin the gel pieces obtained by the funnel-wellmethod (Jeno et al., 1995). After digestion, the peptides
ere extracted from the gel pieces with 100 mMH4HCO3, pH 8.0, containing 80% acetonitrile and
0.1% trifluoroacetic acid and separated by reversed-phase HPLC (Vydac C8, 5-mm particle size). The peaksof the chromatograms were analyzed by mass spec-trometry (API-III, PE Sciex) and fractions containing asingle peptide were chosen for sequencing (Model G1005 A protein sequencer, Hewlett–Packard).
Cloning and Sequencing of Calsyntenin-1 cDNA
Total RNA was prepared from E14 chicken brain andfrom P10 mouse cerebellum (Chomczynski and Sacchi,1987). Oligo(dT)- and random-primed cDNA was pro-duced using M-MLV reverse transcriptase (Promega).For PCR, degenerated primers corresponding to theamino-terminus and four internal peptides were syn-thesized (sense primers, 59-GTIAAMAAGCAYAA-GCCITGGAT-39 and 59-CATGGIATHGTIACIGA-GAATGATAA-39; antisense primers, 59-CCIGTICCIG-GCTCATACTCDAT-39, 59-GTATCIACICCITADAT-DATCATICC-39, 59-ACICCATCIGCDATCTGIAA-ATC-39 and 59-GCATCAAACTCIGCCTCCTTATA-AAA-39). PCR was performed using Taq DNA polymer-ase (Promega). PCR fragments were sequenced andused for screening cDNA libraries. Approximately2.5 3 106 plaques of an oligo(dT)-primed E14 chickenbrain cDNA library (Zuellig et al., 1992), an oligo(dT)-
rimed P20 mouse brain cDNA library (Stratagene), anligo(dT)- and random-primed E15 mouse brain cDNAibrary (Clontech), and an oligo(dT)- and random-rimed fetal human brain cDNA library (Clontech),espectively, were screened by hybridization with theorresponding radiolabeled PCR fragments under hightringency conditions (Sambrook et al., 1989). Positivelones were further characterized.
orthern Blot Analysis and in Situ Hybridization
A human multiple-tissue Northern blot (Clontech)
as hybridized with a 2.8-kb cDNA fragment of human
alsyntenin-1 labeled with [a-32P]-dCTP (Amersham)
using the Prime-It II random primer labeling kit (Strat-agene). Hybridization was performed at 42°C overnightand the hybridization signals were analyzed with aPhosphorImager (Molecular Dynamics). In situ hybrid-ization was performed as described previously (Schae-ren-Wiemers and Gerfin-Moser, 1993).
Antibodies
The anti-calsyntenin-1 antibodies R63 and R71 wereraised in rabbits. The immunogen for the R63 antiserumconsisted of a 267-amino-acid peptide starting at theN-terminus of chicken calsyntenin-1. The immunogenfor the R71 antiserum consisted of an 87-amino-acidpeptide located immediately outside of the transmem-brane segment of chicken calsyntenin-1. Both fragmentswere expressed with a His-tag in bacteria and purifiedusing a Ni–NTA column (Qiagen). The anti-calsynte-nin-1 antibodies were affinity purified by a passageover a protein G column followed by an antigen-conju-gated column. The antibody against the GABAA recep-tor subunit a2 was provided by J.-M. Fritschy. Theantibodies against synaptophysin, PSD95, GluR1, andGluR2 were from Roche, PharMingen, and Chemicon,respectively.
Isolation of Synaptosomes and PostsynapticDensities
For subcellular fractionation, brains of 200 adult miceand 20 adult chickens, respectively, were homogenizedwith a Dounce homogenizer in 5 vol of 10 mM Hepes,0.32 M sucrose supplemented with the Mini Completeinhibitor mix (Roche). The subcellular fractionation wasperformed as described by (Phelan and Gordon-Weeks,1997). For Western blot analysis with the antibodies R63and R71, 100 mg total protein was loaded per lane. Forthe immunodetection of GluR1 and PSD95, only 50 mgtotal protein was loaded per lane.
Cell suspensions of hippocampi dissected frombrains of E17 mice were prepared by digestion withtrypsin (0.25% for 10 min at 37°C) and trituration usinga blue Gilson tip. Cells were then plated onto acid-washed, poly-l-lysine-treated glass coverslips or poly-l-lysine-treated plastic dishes in DMEM supplementedwith B27 (Gibco/Life Technologies), 0.25 mg/ml Albu-
max (Gibco/Life Technologies), 2 mM glutamine, and0.1 M sodium pyruvate. Cultures were maintained for
B
B
B
B
C
C
C
164 Vogt et al.
up to 4 weeks in a humidified incubator with 5% CO2 at37°C.
Cells were fixed in 4% paraformaldehyde and 4%sucrose in PBS for 30 min at 37°C. After rinsing withPBS, cells were preincubated in 10% fetal calf serumand 0.1% glycine in PBS at room temperature for 1 hbefore incubation with the primary antibody in 3% fetalcalf serum in PBS at 4°C for 24–48 h. For the double-labeling experiments, primary antibodies were incu-bated together. Cells were washed for at least 30 min inthree changes of PBS. For secondary antibodies FITC-conjugated goat anti-rabbit IgG (Cappel) and Cy3-con-jugated donkey anti-mouse IgG (Jackson ImmunoRe-search Laboratories, Inc.) were used. For stainings withanti-GluR2, the cells were permeabilized with 0.1% sa-ponin.
Pre- and Postembedding Immunocytochemistry
Eight adult Wistar and OFA line rats (200–250 g) andfive C57Bl6 mice of both sexes were deeply anesthe-tized with metiofane (methoxyflurane, Pitman-MooreInc.) and perfused through the ascending aorta for15–25 min with the fixative composed of 3.5–4% para-formaldehyde, 0.015–0.05% glutaraldehyde, and 0.2%picric acid in 0.1 M sodium phosphate, pH 7.4. Brainswere removed from the skull and sectioned in a coronalplane. Seventy-micrometer-thick sections were pro-cessed for preembedding immunoperoxidase stainingand immunogold labeling, postfixation, and embed-ding for electron microscopy as described previously(Meskenaite, 1997). The freeze substitution and the low-temperature embedding procedure was described ear-lier (Baude et al., 1993). Hippocampal slices of 500 mmthickness from two rats were dehydrated in methanol at280°C and embedded in Lowicryl HM 20 (ChemischeWerke Lowi GmBH, Germany). Eighty-nanometer-thick sections from Lowicryl-embedded blocks werepicked up on nickel grids and incubated overnight inprimary antibodies (16–24 mg/ml), followed by incuba-tion for 2 h on drops of goat anti-rabbit IgG coupled to1.4-nm gold particles (1:80; Nanoprobes Inc.) diluted inTris-buffered saline containing 1% BSA, 0.1% cold-wa-ter fish skin gelatine (Sigma), 5% NGS, and 0.1% TritonX-100. After silver enhancement with an HQ kit (Nano-probes Inc.) for 3–5 min, the sections were subjected tocontrast enhancement for electron microscopy with sat-urated aqueous uranyl acetate followed by lead citrate.For double-sided immunoreaction, the sections were
etched with sodium ethanolate for 2–3 s prior to immu-noincubation (Matsubara et al., 1996).
45Ca Autoradiography on Nitrocellulose Membrane
A fusion protein of the cytoplasmic segment ofmouse calsyntenin-1 and an N-terminal intein tag wasexpressed in bacteria using the Impact-CN system(New England Biolabs Inc.). As a control, maltose-bind-ing protein fused to the intein tag was used. For the 45Caautoradiography, 5 mg of each purified fusion proteinwas blotted on a nitrocellulose membrane. The Ca21-binding assay was performed as described previously(Maruyama et al., 1984). The blots were analyzed with aPhosphorImager (Molecular Dynamics).
ACKNOWLEDGMENTS
We thank Dr. Paul Jeno for helping with in-gel digestion, Dr. PeterHunziker and Neil Birchler for the mass spectrometry and proteinsequencing, Dr. Jean-Marc Fritschy for providing GABAA receptorantibody, and Dr. Peter Somogyi for supplying blocks for immuno-electron microscopy. This work was supported by the Swiss NationalScience Foundation, the Novartis Foundation, the Olga Mayenfisch-Foundation, the Jubilaumsstiftung Rentenanstalt/Swisslife, and theEMDO-Foundation.
REFERENCES
Allbritton, N. L., Meyer, T., and Stryer, L. (1992). Range of messengeraction of calcium ion and inositol 1,4,5-trisphosphate. Science 258:1812–1815.
Baksh, S., and Michalak, M. (1991). Expression of calreticulin inEscherichia coli and identification of its Ca21 binding domains.J. Biol. Chem. 266: 21458–21465.
aranes, D., Lederfein, D., Huang, Y. Y., Chen, M., Bailey, C. H., andKandel, E. R. (1998). Tissue plasminogen activator contributes tothe late phase of LTP and to synaptic growth in the hippocampalmossy fiber pathway. Neuron 21: 813–825.
aude, A., Nusser, Z., Roberts, J. D., Mulvihill, E., McIlhinney, R. A.,and Somogyi, P. (1993). The metabotropic glutamate receptor(mGluR1 alpha) is concentrated at perisynaptic membrane of neu-ronal subpopulations as detected by immunogold reaction. Neuron11: 771–787.
liss, T. V., and Collingridge, G. L. (1993). A synaptic model ofmemory: Long-term potentiation in the hippocampus. Nature 361:31–39.
onner, W. M., and Laskey, R. A. (1974). A film detection method fortritium-labelled proteins and nucleic acids in polyacrylamide gels.Eur. J. Biochem. 46: 83–88.
ampenot, R. B. (1979). Independent control of the local environmentof somas and neurites. Methods Enzymol. 58: 302–307.
hen, Z. L., Yoshida, S., Kato, K., Momota, Y., Suzuki, J., Tanaka, T.,Ito, J., Nishino, H., Aimoto, S., Kiyama, H., et al. (1995). Expressionand activity-dependent changes of a novel limbic-serine proteasegene in the hippocampus. J. Neurosci. 15: 5088–5097.
homczynski, P., and Sacchi, N. (1987). Single-step method of RNA
isolation by acid guanidinium thiocyanate–phenol–chloroform ex-traction. Anal. Biochem. 162: 156–159.
F
G
KL
L
L
L
M
M
M
M
M
N
N
N
O
O
O
O
P
Q
S
S
S
S
S
S
T
165Calsyntenin-1, a Regulated Postsynaptic Ca21-Binding Protein
Edman, J. C., Ellis, L., Blacher, R. W., Roth, R. A., and Rutter, W. J.(1985). Sequence of protein disulphide isomerase and implicationsof its relationship to thioredoxin. Nature 317: 267–270.
Finch, E. A., and Augustine, G. J. (1998). Local calcium signalling byinositol-1,4,5-trisphosphate in Purkinje cell dendrites. Nature 396:753–756.
Fliegel, L., Burns, K., MacLennan, D. H., Reithmeier, R. A., andMichalak, M. (1989). Molecular cloning of the high affinity calcium-binding protein (calreticulin) of skeletal muscle sarcoplasmic retic-ulum. J. Biol. Chem. 264: 21522–21528.
liegel, L., Ohnishi, M., Carpenter, M. R., Khanna, V. K., Reithmeier,R. A., and MacLennan, D. H. (1987). Amino acid sequence of rabbitfast-twitch skeletal muscle calsequestrin deduced from cDNA andpeptide sequencing. Proc. Natl. Acad. Sci. USA 84: 1167–1171.
Frey, U., Muller, M., and Kuhl, D. (1996). A different form of long-lasting potentiation revealed in tissue plasminogen activator mu-tant mice. J. Neurosci. 16: 2057–2063.ray, E. G. (1959). Axo-somatic and axo-dendritic synapses of thecerebral cortex: An elecron microscope study. J. Anat. 93: 420–433.
Huang, Y. Y., Bach, M. E., Lipp, H. P., Zhuo, M., Wolfer, D. P.,Hawkins, R. D., Schoonjans, L., Kandel, E. R., Godfraind, J. M.,Mulligan, R., Collen, D., and Carmeliet, P. (1996). Mice lacking thegene encoding tissue-type plasminogen activator show a selectiveinterference with late-phase long-term potentiation in both Schaffercollateral and mossy fiber pathways. Proc. Natl. Acad. Sci. USA 93:8699–8704.
Jeno, P., Mini, T., Moes, S., Hintermann, E., and Horst, M. (1995).Internal sequences from proteins digested in polyacrylamide gels.Anal. Biochem. 224: 75–82.
Kasai, H., and Petersen, O. H. (1994). Spatial dynamics of secondmessengers: IP3 and cAMP as long-range and associative messen-gers. Trends Neurosci. 17: 95–101.
Kozak, M. (1987). An analysis of 59-noncoding sequences from 699vertebrate messenger RNAs. Nucleic Acids Res. 15: 8125–8148.
rause, K. H., and Michalak, M. (1997). Calreticulin. Cell 88: 439–443.ebeche, D., Lucero, H. A., and Kaminer, B. (1994). Calcium bindingproperties of rabbit liver protein disulfide isomerase. Biochem. Bio-phys. Res. Commun. 202: 556–561.
inden, D. J., and Connor, J. A. (1995). Long-term synaptic depres-sion. Annu. Rev. Neurosci. 18: 319–357.
ombard-Platet, G., and Jalinot, P. (1993). Funnel-well SDS–PAGE: Arapid technique for obtaining sufficient quantities of low-abun-dance proteins for internal sequence analysis. Biotechniques 15: 668–670, 672.
ucero, H. A., Lebeche, D., and Kaminer, B. (1994). ERcalcistorin/protein disulfide isomerase (PDI): Sequence determination and ex-pression of a cDNA clone encoding a calcium storage protein withPDI activity from endoplasmic reticulum of the sea urchin egg[published erratum appears in J Biol Chem 1995 May 12;270(19):11701]. J. Biol. Chem. 269: 23112–23119.acLennan, D. H., and Reithmeier, R. A. (1998). Ion tamers [news;comment]. Nature Struct. Biol. 5: 409–411.adani, R., Hulo, S., Toni, N., Madani, H., Steimer, T., Muller, D., andVassalli, J. D. (1999). Enhanced hippocampal long-term potentia-tion and learning by increased neuronal expression of tissue-typeplasminogen activator in transgenic mice. EMBO J. 18: 3007–3012.aruyama, K., Mikawa, T., and Ebashi, S. (1984). Detection of calciumbinding proteins by 45Ca autoradiography on nitrocellulose mem-brane after sodium dodecyl sulfate gel electrophoresis. J. Biochem.95: 511–519.
atsubara, A., Laake, J. H., Davanger, S., Usami, S., and Ottersen,O. P. (1996). Organization of AMPA receptor subunits at a gluta-
T
mate synapse: A quantitative immunogold analysis of hair cellsynapses in the rat organ of Corti. J. Neurosci. 16: 4457–4467.eskenaite, V. (1997). Calretinin-immunoreactive local circuit neu-rons in area 17 of the cynomolgus monkey, Macaca fascicularis.J. Comp. Neurol. 379: 113–132.agase, T., Ishikawa, K., Suyama, M., Kikuno, R., Hirosawa, M.,Miyajima, N., Tanaka, A., Kotani, H., Nomura, N., and Ohara, O.(1998a). Prediction of the coding sequences of unidentified humangenes. XII. The complete sequences of 100 new cDNA clones frombrain which code for large proteins in vitro. DNA Res. 5: 355–364.agase, T., Ishikawa, K., Suyama, M., Kikuno, R., Miyajima, N.,Tanaka, A., Kotani, H., Nomura, N., and Ohara, O. (1998b). Predic-tion of the coding sequences of unidentified human genes. XI. Thecomplete sequences of 100 new cDNA clones from brain whichcode for large proteins in vitro. DNA Res. 5: 277–286.eher, E. (1998). Vesicle pools and Ca21 microdomains: New toolsfor understanding their roles in neurotransmitter release. Neuron20: 389–399.’Brien, R. J., Lau, L. F., and Huganir, R. L. (1998). Molecular mech-anisms of glutamate receptor clustering at excitatory synapses.Curr. Opin. Neurobiol. 8: 364–369.’Farrell, P. H. (1975). High resolution two-dimensional electro-phoresis of proteins. J. Biol. Chem. 250: 4007–4021.hnishi, M., and Reithmeier, R. A. (1987). Fragmentation of rabbitskeletal muscle calsequestrin: Spectral and ion binding propertiesof the carboxyl-terminal region. Biochemistry 26: 7458–7465.sterwalder, T., Contartese, J., Stoeckli, E. T., Kuhn, T. B., and Son-deregger, P. (1996). Neuroserpin, an axonally secreted serine pro-tease inhibitor. EMBO J. 15: 2944–2953.
helan, P., and Gordon-Weeks, P. R. (1997). Isolation of synapto-somes, growth cones, and their subcellular components. In Neuro-chemistry (A. J. Turner and H. S. Bachelard, Eds.), Vol. 172, pp. 1–38.Oxford Univ. Press, New York.ian, Z., Gilbert, M. E., Colicos, M. A., Kandel, E. R., and Kuhl, D.(1993). Tissue-plasminogen activator is induced as an immediate-early gene during seizure, kindling and long-term potentiation.Nature 361: 453–457.
ambrook, J., Fritsch, E. F., and Maniatis, T. (1989). Molecular Cloning.Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.
chaeren-Wiemers, N., and Gerfin-Moser, A. (1993). A single protocolto detect transcripts of various types and expression levels in neuraltissue and cultured cells: In situ hybridization using digoxigenin-labelled cRNA probes. Histochemistry 100: 431–440.
eeds, N. W., Williams, B. L., and Bickford, P. C. (1995). Tissueplasminogen activator induction in Purkinje neurons after cerebel-lar motor learning. Science 270: 1992–1994.
onderegger, P., Fishman, M. C., Bokoum, M., Bauer, H. C., Neale,E. A., and Nelson, P. G. (1984). A few axonal proteins distinguishventral spinal cord neurons from dorsal root ganglion neurons.J. Cell Biol. 98: 364–368.
onderegger, P., Fishman, M. C., Bokoum, M., Bauer, H. C., andNelson, P. G. (1983). Axonal proteins of presynaptic neurons duringsynaptogenesis. Science 221: 1294–1297.
toeckli, E. T., Lemkin, P. F., Kuhn, T. B., Ruegg, M. A., Heller, M.,and Sonderegger, P. (1989). Identification of proteins secreted fromaxons of embryonic dorsal-root-ganglia neurons. Eur. J. Biochem.180: 249–258.
akechi, H., Eilers, J., and Konnerth, A. (1998). A new class of synapticresponse involving calcium release in dendritic spines. Nature 396:757–760.
akeichi, M. (1990). Cadherins: A molecular family important inselective cell–cell adhesion. Annu. Rev. Biochem. 59: 237–252.
v
W
W
Y
Z
Z
166 Vogt et al.
Taylor, C. W. (1998). Inositol trisphosphate receptors: Ca21-modu-lated intracellular Ca21 channels. Biochim. Biophys. Acta 1436: 19–33.
on Heijne, G. (1983). Patterns of amino acids near signal-sequencecleavage sites. Eur. J. Biochem. 133: 17–21.ang, S., Trumble, W. R., Liao, H., Wesson, C. R., Dunker, A. K., andKang, C. H. (1998). Crystal structure of calsequestrin from rabbitskeletal muscle sarcoplasmic reticulum. Nature Struct. Biol. 5: 476–483.ilson, R., Ainscough, R., Anderson, K., Baynes, C., Berks, M., Bon-field, J., Burton, J., Connell, M., Copsey, T., Cooper, J., et al. (1994).
2.2 Mb of contiguous nucleotide sequence from chromosome III of
ano, K., and Zarain-Herzberg, A. (1994). Sarcoplasmic reticulumcalsequestrins: structural and functional properties. Mol. Cell. Bio-chem. 135: 61–70.
ucker, R. S. (1999). Calcium- and activity-dependent synaptic plas-ticity. Curr. Opin. Neurobiol. 9: 305–13.
uellig, R. A., Rader, C., Schroeder, A., Kalousek, M. B., VonBohlen und Halbach, F., Osterwalder, T., Inan, C., Stoeckli, E. T.,Affolter, H. U., Fritz, A., et al. (1992). The axonally secretedcell adhesion molecule, axonin-1: Primary structure,immunoglobulin-like and fibronectin-type-III-like domains and
glycosyl-phosphatidylinositol anchorage. Eur. J. Biochem. 204:
C. elegans. Nature 368: 32–38. 453– 463.
Received September 4, 2000Revised October 19, 2000
