Proc. Natl. Acad. Sci. USAVol. 93, pp. 4760-4764, May 1996Neurobiology

Synaptophysin, a major synaptic vesicle protein, is not essentialfor neurotransmitter releaseHARVEY T. MCMAHON*t, VADIM Y. BOLSHAKOVt, ROGER JANZ§, ROBERT E. HAMMER*¶, STEVEN A. SIEGELBAUMt,AND THOMAS C. SIiDHOF*¶II*Howard Hughes Medical Institute and Departments of §Molecular Genetics and 1Biochemistry, The University of Texas Southwestern Medical School, Dallas,TX 75235; tMedical Research Council Laboratory of Molecular Biology, Neurobiology Division, Hills Road, Cambridge, CB2 2QH, England; and tHowardHughes Medical Institute and Department of Pharmacology, Center for Neurobiology and Behavior, Columbia University, New York, NY 10032

Communicated by Eric R. Kandel, Columbia University, New York, NY, January 16, 1996 (received for review December 5, 1995)

ABSTRACT Synaptophysin (syp I) is a synaptic vesiclemembrane protein that constitutes -7% of the total vesicleprotein. Multiple lines of evidence implicate syp I in a numberof nerve terminal functions. To test these, we have disruptedthe murine syp I gene. Mutant mice lacking syp I were viableand fertile. No changes in the structure and protein compo-sition of the mutant brains were observed except for a decreasein synaptobrevin/VAMP II. Synaptic transmission was nor-mal with no detectable changes in synaptic plasticity or theprobability of release. Our data demonstrate that one of themajor synaptic vesicle membrane proteins is not essential forsynaptic transmission, suggesting that its function is eitherredundant or that it has a more subtle function not apparentin the assays used.

Synaptophysin (syp I) constitutes an abundant synaptic vesiclemembrane protein that contains four transmembrane regions(1-4). Syp I .forms a high molecular weight complex in thevesicle membrane containing at least four syp I subunits anda low molecular protein of 18 kDa (5) recently identified assynaptobrevin/VAMP (6, 7). Syp I is phosphorylated on itscytoplasmic C-terminus by both serine/threonine and tyrosinekinases (8-10). In the mature nerve terminal it represents oneof the major proteins carrying phosphotyrosine (11).

In addition to syp I, synaptic vesicles contain a highlyhomologous membrane protein named synaptoporin or syn-aptophysin II (syp II) (12). Syp I and syp II exhibit distinctdistributions in brain: Syp I is ubiquitously expressed andpresent in virtually all synapses whereas syp II is expressed athigh levels only in selected neurons, suggesting that syp II hasa more specialized function (13, 14). In the hippocampus, highlevels of syp II are detected in mossy fiber terminals, with othersynapses containing much lower amounts. Although syp IIexhibits a heteromultimeric subunit structure similar to that ofsyp I, the two syps are present in distinct complexes and do notform heteromultimers (14).The abundance, uniform synaptic expression, and subunit

structure of syp I suggested that it may have an importantfunction in synaptic vesicle exocytosis.. This hypothesis wassupported by the following series of studies: (i) Purified syp Iassociates into multimers that form channels in black lipidmembranes (15). However, it is unclear if these channelsrepresent a physiologically activated state of syp I because thedisulfide bonding and subunit structure of purified syp Idiffered from those of native syp I. (ii) Syp I exhibited Ca2+binding activity in a blotting assay (16), suggesting it may be aCa-2 sensor, although Ca2+ binding was not observed byequilibrium dialysis (17). (iii) Injection of total mRNA from ratcerebellum into Xenopus oocytes resulted in Ca2+-dependentglutamate release that was inhibited by coinjection of syp I

antisense RNA, suggesting that syp I is essential for release(18). However, previous studies using injection of mRNA intooocytes indicated that the Ca2+-dependent neurotransmitterrelease observed is nonvesicular (19). (iv) Injection of anti-bodies to syp I into Xenopus blastomeres inhibited transmittersecretion at neuromuscular synapses (20). (v) Transfection ofsyp I into nonneuronal cells resulted in its targeting to thereceptor-mediated endocytosis pathway (21) or even in thecreation of a novel type of vesicles (22). (vi) Overexpression ofsyp I in Xenopus oocytes caused enhanced neurotransmittersecretion from the resulting neuromuscular synapses (23).Together these results support the notion that syp I has an

important role in the nerve terminal. However, the functionssuggested by the different experiments are diverse, rangingfrom that of a vesicular Ca2+ sensor or channel to a role insynaptic vesicle biogenesis. In an effort to elucidate theessential functions of syp I we have now studied synapticfunctions in mice lacking syp I.

MATERIALS AND METHODS

Genomic Cloning and Sequencing and Construction of aKnockout Vector. A knockout vector was constructed fromtwo nonoverlapping genomic clones encoding exons I-III orIV-VII of syp I isolated by standard techniques (3, 24). In theknockout vector most of the sequences between exons II andIV are replaced by a neomycin gene cassette (Fig. 1) as aselectable marker. Embryonic stem cells were electroporatedwith the linearized vector and subjected to positive (0.19 g/literof G415 on days 2-9) and negative selection (0.1 mg/literFIAU on days 3-6). Clones were analysed for homologousrecombination using PCR with oligonucleotides priming eitherin the neomycin resistance gene (primer D in Fig. 1; sequence:GAGCGCGCGCGGCGGAGTTGTTGAC) and the syp Igene outside of the knockout vector (primer C, sequenceATCTGGCAGGTAGTCCTCTCCTCA), or in the wild-typeallele (primers A and B, sequences CTTCGTGAAGGTGC-TGCAGTGG and CGCACTAGTATCGATGGGTACT-CAAATTCGACTTC). Embryonic stem cell clones carryinghomologously recombined syp I genes were injected intomouse blastocysts and chimeric offspring were bred, resultingin mutant mice derived from two independent homologousrecombinations, one of which was analyzed in detail.Immunoblotting and Immunocytochemistry Analysis.

SDS/PAGE was carried out as described (25). Proteins weredetected after blotting by either enhanced chemiluminescence(Amersham) or 1251-labeled secondary antibodies followed byquantitation on a Molecular Dynamics Phosphorlmager es-sentially as described (26, 27). Cryostat brain sections fromperfusion-fixed adult mice were reacted with Nissl stain or withantibodies (serum at 1:500-1:2000 dilution). Antibody reac-

Abbreviations: syp I and II, synaptophysin I and II; EPSC, excitatory post-synaptic current; mEPSC, miniature EPSC.IITo whom reprint requests should be addressed.

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

4760

Proc. Natl. Acad. Sci. USA 93 (1996) 4761

A BA E IV\ vI Vil

X S K H E EEE1 hEI(x s k ~Ht H E E EEX2h K X ,

B pm38L4-1 / prn3SL3-6I If IV \ VI VII

TK :TK 4_4 Ne 1 -NeS K:|E X:i E E E E XEKXE

C E I 11 I V VI VIIl

XSK' E X E E E E Xh E K X

C D

FIG. 1. Mutation of the murine syp I gene. (A) Map of thewild-type gene as deduced from the genomic subclones pm38L4-1 andpm38L3-6. The parts of the gene not analyzed are shown as a dashedline. (B) Structure of the knockout vector containing two copies of theherpes simplex virus thymidine kinase gene (TK) and one copy of theneomycin gene cassette (Neo). (C) Map of the mutant gene generatedby homologous recombination. Positions of PCR primers used todetect wild-type (A and B) and mutant (C and D) alleles are indicatedby arrows. Locations of exons are indicated by Roman numerals andrestriction enzyme sites by letters (E, EcoRI; H, HindIII; K, KpnI; X,XbaI; S, SpeI; X, XhoI).

tions were detected using the peroxidase-antiperoxidase tech-nique and heavy metal enhancement (26).

Electrophysiological Recordings. Hippocampal slices (250-300 ,uM thick) were obtained from 2.5- to 4-week-old mice andwhole-cell recordings were obtained from CAl pyramidalneurons as described (28). Compound excitatory postsynapticcurrents (EPSCs) were evoked by stimulating the Schaffercollateral pathway. Unitary EPSCs were evoked by extracel-lular stimulation of single CA3 neurons (28). Holding potentialof the CAl neuron was -70 mV. Miniature EPSCs (mEPSCs)were recorded in the presence of 1 ,uM tetrodotoxin and 0.1mM picrotoxin and analyzed as described (29) with or without0.5 M sucrose or 1 nM a-latrotoxin in the bath solution.

RESULTSStructure of the Murine Syp I Gene. The murine syp I gene

was cloned from a genomic 129SV library on two nonover-

Table 1. Levels of synaptic proteins in wild-type and syp Iknockout mice (shown as % of wild-type levels)

Protein

Synaptophysin ISynaptophysin IISynaptogyrinSNAP25Synaptobrevin IISyntaxinComplexins I and IISynaptotagmin Ia-SNAPDynaminVCPMunc-18Synapsins I and IINa/K ATPaseRablBRab3ARab3CTransf. Recept.

Wild type100+ 9100± 9100+ 5100± 0100± 5100± 1100 ± 19100 ± 21100+ 1100 + 24100 ± 15100 + 15100+ 6100+ 4100 ± 10100 ± 36100 + 26100 + 15

Heterozygous

59 ± 13109 ± 1199± 8104± 284± 5100 ± 13103 ± 10104± 3104± 2100± 8114 ± 11105 ± 1294± 487 ± 24114 ± 1289 ± 2390 ± 19111 ± 5

Knockouts0± 0

108± 8113 ± 11100 ± 1282± 5100 ± 10110 ± 2099 ± 1298± 6112 ± 16114 ± 18104 ± 11104 ± 16103± 2107 ± 1594 ± 1593± 9104± 3

Proteins were quantitated by immunoblots using 125I-labeled sec-

ondary antibodies and Phospholmager detection. Data shown aregiven as ± SD from triplicate determinations except for syp I and II

and synaptogyrin where N = 6. The differences between wild type,heterozygotes, and knockouts are not statistically significant at the 1%level in the Student's t-test except for syp I and synaptobrevin II.

lapping A clones. Sequencing showed that the exon-intronstructure of the murine gene was identical to that of the humangene (Fig. 1; ref. 30). It is unclear if the murine or human genescontain an intron in the 5' noncoding region, but we will referto the identified exons as exons I-VII to facilitate discussion.The 5' end of the mouse syp I mRNA as determined from thegenomic sequence contains an in-frame methionine codon 18bp upstream of the putative initiator methionine (3). Thisresults in an extended N-terminal sequence of MLLLADM-DVVN for murine syp I that is also present in human syp 1 (30).An interesting conserved feature of the syp I gene (Fig. 1; ref.30) is the presence of an intron in the 3'-untranslated regionimmediately 5' to an A+T-rich sequence. This results in a 3'exon, exon VII, which is entirely composed of 3'-untranslatedregion. The high degree of conservation of the 3'- untranslated

A 2.01 t - i.I e-_

& 1.5-

1.0-

B

C

g 2.0-

a)

-

1.5-

10 1.0- -0

0.5- r-

3.01~0_2.5-

v2.0-'10

a) 1.5

._.q

1.0-z

0.5 J

o - wiia type0 - mutant

0 50 100 150 200 250Interstimulus interval (ms)

300

II

10 20 30EPSCs number in the train

A - wild typeA - mutant

.. m..-w...r~0 25 50

Time (min)

FIG. 2. Three types of synaptic plasticity in wild-type and sypI-deficient mice. (A) Paired-pulse facilitation. Ratios of the peakEPSCs for the second versus the first stimulus are plotted as a functionof the interstimulus interval. EPSC pairs (8-10) were averaged foreach cell and plotted as a function of the interstimulus interval from24 cells from 7 wild-type mice and from 31 cells from 9 mutant mice.(B) Frequency facilitation. EPSCs were recorded in response to a trainof extracellular presynaptic stimuli (14 Hz) and normalized by theresponse to the first stimulus. Three trains were averaged for eachneuron from 19 cells from 7 wild-type mice and from 23 cells from 9mutant mice. EPSCs to successive stimuli exhibit facilitation followedby depression. (C) Long-term potentiation (LTP). Graph depicts theaverage response to an LTP induction protocol (2 x 100 Hz for 1 secwith a 20-sec interval applied with current-clamped postsynapticneurons). EPSCs were normalized to EPSC amplitude prior to LTPinduction. Data were averaged from 12 cells from 7 wild-type mice and16 cells from 8 mutant mice. Error bars are SEM.

Neurobiology: McMahon et al.

4762 Neurobiology: McMahon et al.

region and of its genomic structure support the hypothesis thatthe A+T-rich sequence may serve as a regulatory signal.

Disruption of the Syp I Gene. A knockout vector was con-structed from the genomic clones consisting of a short armcontaining exons I and II, a neomycin gene cassette as a selectablemarker, and a long arm containing exons IV-VII (Fig. 1). Tlwocopies of the herpes simplex virus thymidine kinase gene wereinserted to allow negative selection. Homologous recombinationof this vector with the endogenous syp I gene results in a mutantgene in which exon III and large portions of introns 2 and 3 arereplaced by the- neomycin gene. After electroporation of thevector into embryonic stem cells, 20% of the clones survivingpositive and negative selection were found to have undergonehomologous recombination (Fig. 1). Two lines of mice carryingdisrupted syp I genes were generated from independent embry-onic stem cell clones, one of which was analyzed in detail.The exon deleted by homologous recombination encodes

residues 29-70 of syp I that include half of the first transmem-brane region and a cysteine residue that forms a disulfide bond(5, 25), suggesting that deletion of this exon should result in anull phenotype. This expectation was confirmed in immuno-blotting experiments with a series of independent antibodiesagainst the C terminus of syp I that failed to detect immuno-reactive syp I protein.

A

CD

0c

.0

Ez-

28

14

O L-8

.S i^*I Iwp ,7I"T-'Iy

Phenotype of Syp I Deficient Mice. Mice lacking syp I did notsuffer from premature morbidity or mortality as analyzed upto the age of 28 months. The mice were fertile and exhibitedno significant difference in fecundity. Thus, syp I is not anessential gene. To test for more subtle changes, we studied thestructure of the nervous system of the mutant mice. Macro-scopically and microscopically, the brains were normal (datanot shown). Antibodies against syp II, synaptotagmin I, syn-aptobrevin II, synaptogyrin, SNAP-25, synapsins I and II,rab3A, and rabphilin were tested by immunocytochemistry andfailed to detect differences. Even for syp II, no change indistribution was observed. Syp II was found to exhibit the sameheterogeneous expression in mouse brains as previously ob-served for rats (14), suggesting that the expression of syp II isnot subject to major changes in the syp I knockout.Immunoblots for a series of synaptic proteins revealed no

major changes in expression. Quantification of protein levelsusing 125I-labeled secondary antibodies showed no significantdifferences between wild-type and mutant mice except forsynaptobrevin II. This protein [that interacts with syp I (6, 7)]exhibited a moderate statistically significant decrease (Table1). Together these data suggest that deficiency of syp I does notcause a major restructuring of the brain architecture or itsprotein composition.

C

- wt- mut

5 pA

20 ms

-4 0 4 8 12(-pA)

-8 0 8(-pA)

16

D

0.75

X 0.50-

0.25 -

-4 0 4 8 12(-pA)

o 0 - P- quantal amplil

0

0

0 A

- 5.0tude

*A

4.0<*0 , '- 4.0 _

0 3

wt

FIG. 3. Quantal analysis of synaptic transmission in wild-type and syp I-deficient mice. EPSCs were recorded from postsynaptic CAl neuronsunder whole-cell voltage clamp in response to extracellular patch pipette stimulation of single presynaptic CA3 neurons. (A) Binned histogramsummarizing the number of events observed in wild-type mice for each EPSC amplitude (bin size = 0.2 pA; n = 374 total). The peaks of thehistogram were fitted by two Gaussian functions. The first component (mean = -0.3 pA; SD = 0.8 pA) represents background noise andtransmission failures. The second component (mean = -3.8 pA; SD = 1.22 pA) represents successful transmissions. The probability of release(Pr) calculated from the relative peak areas was 0.49. (Inset) Ten superimposed unitary EPSCs illustrating failed and successful transmissions. (B)Analysis of EPSC amplitude histograms recorded from mutant mice as in (A). Calculated parameters are similar to those obtained from wild-typemice (first component: mean = -0.24 pA; SD = 0.73 pA; second component: mean = -4.39 pA; SD= 1.28 pA; n = 652 total; Pr = 0.56). (C)Comparison of histograms for unitary EPSC density estimates from wild-type (solid line) and mutant mice (dotted line). The ordinate values ofthe second peaks representing successful transmissions were matched to facilitate comparison. (D) Summary of quantal release parameters forwild-type and mutant mice. Each symbol represents data obtained from a separate cell. Pr, release probability.

B

a)

a)

CD?

Ez

36

18 -

0 --8

Proc. Natl. Acad. Sci. USA 93 (1996)

Proc. Natl. Acad. Sci. USA 93 (1996) 4763

Synaptic Transmission in Syp I-Deficient Mice. To explorethe possible effects of deleting syp I on synaptic transmission,we recorded composite EPSCs in hippocampal CAl pyramidalneurons stimulated via the Schaffer collateral/commissuralpathway. The mutant mice displayed normal-size EPSCs andfailed to exhibit changes in several forms of synaptic plasticity,including paired-pulse facilitation, frequency facilitation dur-ing repetitive stimulation, and long-term potentiation (Fig. 2).

Quantal analysis of synaptic transmission between singlepresynaptic CA3 and postsynaptic CAl neurons also revealedno changes in synaptic function (Fig. 3). As previously reportedfor rat hippocampal slices (28, 31), stimulation of a singlemouse hippocampal CA3 neuron either failed to elicit anEPSC in the CAl neuron or elicited an EPSC with anamplitude around -4 pA. Amplitude histograms of theseresponses displayed only two prominent peaks, one at 0 pAcorresponding to transmission failures, and one at -4 pAcorresponding to successful transmissions. The amplitude his-tograms were fitted well by the sum of two Gaussian functionsand can be be interpreted in terms of the probabilistic releaseof a single quantum of transmitter during synaptic transmissionbetween a given CA3 neuron and a single CAl neuron (28).Both the release probability (Pr, relative area under success

peak) and the quantal amplitude (Qr, position of success peak)were similar between mutant and wild-type mice. Pr was 0.50± 0.02 (mean ± SEM, n = 6) and 0.48 ± 0.05 (n = 5),respectively, for wild-type and mutant mice (not significantlydifferent by t test, t = 0.35). Qr was -3.82 ± 0.03 pA inwild-type mice (n = 4) and -4.08 + 0.09 pA in mutant mice(n = 5). The difference in Qr between wild-type and mutantmice was statistically significant (t = 2.43, p < 0.05) butamounted to less than 10%. If the increase in Qr was causedby an increase in the size of synaptic vesicles, only a 2%increase in diameter would be required that is not measurablewith current techniques.

These results indicate that deletion of syp I does not have amajor effect on Ca2+-dependent transmitter release. Wetherefore examined the role of syp I in mEPSCs that reflectspontaneous release events that are at least partly Ca2+-independent. The frequency of mEPSCs recorded in thepresence of tetrodotoxin (1 ,uM) to block action potentials wasunchanged in mutant mice (3.9 ± 0.9 events/min in wild-typemice, n = 8 cells, versus 3.3 ± 0.8 events/min in mutant mice,n = 12 cells). Stimulation of neurotransmitter release by twoagents that appear to act independent of Ca2+, hypertonicsucrose solution and a-latrotoxin, also resulted in apparentlynormal responses in mutant mice indistinguishable from wild-type mice (Fig. 4 and data not shown). Thus, syp I is notrequired for either Ca2+-dependent or Ca2+-independent re-lease.

DISCUSSIONOur data demonstrate that syp I, a major intrinsic membraneprotein of synaptic vesicles, is not essential for synaptic func-tion. The only changes observed consisted of a small decreasein synaptobrevin II (<20%) and an even smaller increase inquantal amplitude (<10%). The decrease in synaptobrevin IIis consistent with an in vivo interaction of synaptobrevin andsyp I (7, 8), but is much smaller than for example the decreasein rabphilin after knockout of rab3A (32).

In view of its abundance and uniform expression, the lack ofan essential function for syp I is surprising. It contrasts withprevious studies in which the essential nature of syp I wasprobed more indirectly by antibody or antisense techniques(18, 20). There are several possible interpretations of ourresults: (i) Syp I is functionally redundant. With syp II, a secondisoform of synaptophysin is present in the mutant mice thatmay substitute for syp I. However, syp I and syp II have distinctexpression patterns (see Introduction). In the hippocampus,

A 75-

> 50-

0

0--0

25-7

B

+-j

UU

0:z

75-

50-

25-

0 -

mut SUCROSEcontrol

jsucrose__^

Wt SUCROSE

control

sucrose

.8 ... . .

4 80 12Time (min)

FIG. 4. Analysis of mEPSCs evoked by hypertonic solution. Graphsdepict histograms of the mEPSC frequency (bin size = 10 sec) inresponse to 0.5 M sucrose (solid bar) from six cells from mutant mice(A) or three cells from wild-type mice (B). Insets depict records ofrecordings used for the histograms (calibration bars: 5 pA, 50 ,usec).Parallel experiments using a-latrotoxin to stimulate mEPSCs gavesimilar results.

high levels of syp II are observed in mossy-fiber terminals butonly low levels are seen at the CA3/CA1 synapses studied herephysiologically (14). Therefore, if the lack of a phenotype inthe syp I knockout mice is due to functional redundancybetween syp I and syp II, low levels of syp II must be able tofully perform that function. (di) Deletion of syp I expressioncan be compensated for by changes in other proteins. Thispossibility is a specialized form of redundancy. However, in thesyp I knockout we observed no significant changes in the levelsof other synaptic proteins except for synaptobrevin II, espe-cially not in syp II or synaptogyrin, vesicle proteins that areclosely or distantly related to syp I (12, 33). (iii) Syp I has asubtle essential function that was not detected in the testsapplied in the current study. Although this possibility cannotbe excluded, the tests applied to the syp I knockout in thecurrent study were extensive. (iv) Syp I represents an evolu-tionary remnant without function. Although many argumentshave been advanced against such a possibility, such as theabundance, evolutionary conservation, and protein-proteininteractions of syp I, it cannot be ruled out entirely.How can the results of previous studies indicating an essen-

tial function of syp I be explained in view of the current data?Because syp I is an abundant protein, injected antibodies maydecorate synaptic vesicles and thus inhibit exocytosis (20). Ina second set of experiments, antisense RNA inhibited Ca2+-

Neurobiology: McMahon et aL

Il

4764 Neurobiology: McMahon et al.

dependent glutamate secretion from Xenopus oocytes injectedwith rat cerebellar mRNA (18). This is a puzzling result sincesyp II is expressed in cerebellum (14). It is possible that syp Iand syp II have distinct functions or that adaptive changes mayoccur in the knockouts but not in the oocyte system tocompensate for the deletion of syp I. Alternatively, the neu-rotransmitter release reconstituted in oocytes may differ fromthat of synapses (19) and not reflect synaptic vesicle functions.Although it is impossible to obtain a conclusive answer to

these questions at present, it seems most likely that syp I isfunctionally redundant with either syp II (in spite of its moreheterogeneous and restricted distribution) or with with syn-aptogyrin [in spite of its low degree of homology (33)]. Futureexperiments will test these hypotheses in double and tripleknockouts of syp I with syp II and/or synaptogyrin.

We thank I. Leznicki, A. Roth, and E. Borowicz for excellenttechnical assistance; Dr. M. Yanagisawa for the use of his microscope;Drs. R. Jahn, I. Trowbridge, and W. Balch for antibodies; Dr. M.Missler for helping with the morphological analysis; and Drs. M. S.Brown and J. L. Goldstein for advice. This study was supported by afellowship to R.J. from the Deutsche Forschungsgemeinschaft and byGrant MH50733 to S.A.S.

1. Jahn, R., Schiebler, W., Ouimet, C. & Greengard, P. (1985) Proc.Natl. Acad. Sci. USA 82, 4137-4141.

2. Wiedenmann, B. & Franke, W. W. (1985) Cell 41, 1017-1028.3. Sudhof, T. C., Lottspeich, F., Greengard, P., Mehl, E. & Jahn, R.

(1987) Science 238, 1142-1144.4. Leube, R. E., Kaiser, P., Seiter, A., Zembelmann, R., Franke,

W. W., Rehm, H., Knaus, P. & Betz, H. (1987) EMBO J. 6,3261-3268.

5. Johnston, P. A. & Sudhof, T. C. (1990) J. Biol. Chem. 265,7849-7852.

6. Calakos, N. & Scheller, R. (1994) J. Biol. Chem. 269, 24534-24537.

7. Edelmann, L., Hanos, P. I., Chapman, E. R. & Jahn, R. (1995)EMBO J. 14, 225-231.

8. Barnekow, A., Jahn R. & Schartl, M. (1990) Oncogene 5, 1019-1024.

9. Rubenstein, J. L., Greengard, P. & Czernik, A. J. (1993) Synapse13, 161-172.

10. Linstedt, A. D., Vetter, M. L., Bishop, J. M. & Kelly, R. B. (1992)J. Cell Biol. 117, 1077-1084.

11. Pang, D. R., Wang, J. K. T., Valtorta, F., Benfenati, F. &Greengard, P. (1988) Proc. Natl. Acad. Sci. USA 85, 762-766.

12. Knaus, P., Marqueze-Pouey, B., Scherer, H. & Betz, H. (1990)Neuron 5, 453-462.

13. Marqueze-Pouey, B., Wisden, W., Malosio, M. L. & Betz, H.(1991) J. Neurosci. 11, 3388-3397.

14. Fykse, E. M., Takei, K., Walch-Solimena, C., Geppert, M., Jahn,R., De Camilli, P. & Sudhof, T. C. (1993) J. Neurosci. 13,4997-5007.

15. Thomas, L., Hartung, K., Langlosch, D., Rehm, H., Bamberg, E.,Franke, W. W. & Betz, H. (1988) Science 242, 1050-1053.

16. Rehm, H., Wiedenmann, B. & Betz, H. (1986) EMBO J. 5,535-541.

17. Brose, N., Petrenko, A. G., Sudhof, T. C. & Jahn, R. (1992)Science 256, 1021-1025.

18. Alder, J., Lu, B., Valtorta, F., Greengard, P. & Poo, M.-m. (1992)Science 257, 657-661.

19. Cavalli, A., Eder-Colli, L., Dunant, Y., Loctin, F. & Morel, N.(1991) EMBO J. 10, 1671-1675.

20. Alder, J., Xie, Z.-P., Valtorta, F., Greengard, P. & Poo, M.-m.(1992) Neuron 9, 759-768.

21. Johnston, P. A., Cameron, P. L., Stukenbrok, H., Jahn, R., De-camilli, P. & Sudhof, T. C. (1989) EMBO J. 8, 2863-2872.

22. Leube, R. E., Wiedenmann, B. & Franke, W. W. (1989) Cell 59,433-446.

23. Alder, J., Kanki, H., Valtorta, F., Greengard, P. & Poo, M.-m.(1995) J. Neurosci. 15, 511-519.

24. Sambrook, J., Fritsch, E. F. & Maniatis, T. (1989) MolecularCloning: A Laboratory Manual (Cold Spring Harbor Lab. Press,Plainview, NY), 2nd Ed.

25. Johnston, P. A., Jahn, R. & Sudhof, T. C. (1989) J. Biol. Chem.264, 1268-1273.

26. Rosahl, T. W., Spillane, D., Missler, M., Herz, J., Selig, D. K.,Wolff, J. R., Hammer, R. E., Malenka, R. C. & Sudhof, T. C.(1995) Nature (London) 375, 488-493.

27. McMahon, H. T. & Sudhof, T. C. (1995) J. Biol. Chem. 270,2213-2217.

28. Bolshakov, V. Y. & Siegelbaum, S. A. (1995) Science 269, 1730-1734.

29. Bolshakov, V. Y. & Siegelbaum, S. A. (1994) Science 264, 1148-1152.

30. Ozcelik, T., Lafreniere, G., Archer, B. T., Johnston, P. A., Wil-lard, H. F., Francke, U. & Sudhof, T. C. (1990) Am. J. Hum.Genet. 47, 551-561.

31. Stevens, C. F. & Wang, Y. (1995) Neuron 14, 795-802.32. Geppert, M., Bolshakov, V., Siegelbaum, S. A., Takei, K., De

Camilli, P., Hammer, R. E. & Siidhof, T. C. (1994) Nature(London) 369, 493-497.

33. Stenius, K., Janz, R., Sudhof, T. C. & Jahn, R. (1995) J. Cell Biol.131, 1801-1809.

Proc. Natl. Acad. Sci. USA 93 (1996)