Synaptotagmin-1 functions as a Ca2+ sensor forspontaneous release

Jun Xu1–4,6, Zhiping P Pang1–4, Ok-Ho Shin3,6 & Thomas C Sudhof1–5

Spontaneous ‘mini’ release occurs at all synapses, but its nature remains enigmatic. We found that > 95% of spontaneous release

in murine cortical neurons was induced by Ca21-binding to synaptotagmin-1 (Syt1), the Ca21 sensor for fast synchronous

neurotransmitter release. Thus, spontaneous and evoked release used the same Ca21-dependent release mechanism. As a

consequence, Syt1 mutations that altered its Ca21 affinity altered spontaneous and evoked release correspondingly. Paradoxically,

Syt1 deletions (as opposed to point mutations) massively increased spontaneous release. This increased spontaneous release

remained Ca21 dependent but was activated at lower Ca21 concentrations and with a lower Ca21 cooperativity than

synaptotagmin-driven spontaneous release. Thus, in addition to serving as a Ca21 sensor for spontaneous and evoked release,

Syt1 clamped a second, more sensitive Ca21 sensor for spontaneous release that resembles the Ca21 sensor for evoked

asynchronous release. These data suggest that Syt1 controls both evoked and spontaneous release at a synapse as a simultaneous

Ca21-dependent activator and clamp of exocytosis.

Spontaneous miniature release (minis) was observed in initial record-ings of synaptic activity1 and has engaged neuroscientists ever since2–5.Under physiological conditions, every synapse spontaneously releasesneurotransmitter quanta at a low frequency. In a typical central neuron,these quanta add up to a sizable signal because thousands of synapsesare present on the neuron. A major question has been whetherspontaneous neurotransmitter release is biologically meaningful oraccidental. Indeed, the very nature of spontaneous release remainsunclear; is it truly spontaneous or is it regulated2,3? Does it operate withthe same synaptic machinery as evoked release, or does it represent adifferent type of exocytosis that involves a special synaptic vesiclepool5,6? Presynaptic mini release produces multiple postsynapticeffects; for example, spontaneous release maintains dendritic spines7,regulates cerebellar interneuron firing8 and suppresses local dendriticprotein synthesis9. Moreover, many neurotransmitters and psycho-active compounds, such as endocannabinoids, caffeine and nicotine,modulate spontaneous release by activating presynaptic receptors10–16,indicating that spontaneous minis perform a biological function.Consistent with this notion, a substantial component of spontaneousrelease appears to be Ca2+ dependent and may be triggered bypresynaptic Ca2+ sparks17,18. An additional, sizable component ofspontaneous minis, however, was found to be Ca2+ independent inat least some experiments2–4, although the mechanisms involvedremain unclear.

Evoked neurotransmitter release is mediated by a synaptic mem-brane fusion machinery composed of soluble NSF-attachment protein

receptors (SNAREs) and Sec1/Munc18-like proteins, and is controlledby Ca2+ binding to Syt1, synaptotagmin-2 (Syt2) or synaptotagmin-9,which perform similar interchangeable functions in evoked release19–22.Deletions of SNAREs and Sec1/Munc18-like proteins block evoked andspontaneous release, suggesting that the generic synaptic membranefusion machinery mediates both types of release23–25. Deletions of Syt1and Syt2, however, produce increases in spontaneous mini release inDrosophila neuromuscular junctions and in vertebrate centralsynapses26–28 (although not in vertebrate autapses29). This findingled to the hypothesis that Syt1 is not only the Ca2+ sensor for evokedrelease but is also a clamp of SNAREs that limits spontaneous release,and that minis are accidental ‘leaks’ of this clamping function20,21. Theclamping hypothesis, however, argues against the notion that sponta-neous release may be biologically meaningful, as it is difficult toimagine how an accidental byproduct of evoked release could controla physiological process. Moreover, the clamping hypothesis fails toexplain why at least some mini release is Ca2+ dependent. We exploredhow Ca2+ regulates spontaneous release and examined the relationshipof this mechanism to the clamping of spontaneous release by synapto-tagmins. We found that Ca2+ triggered most spontaneous release, as itdid most evoked release, by binding to synaptotagmin and that deletionof synaptotagmin activated a second Ca2+ sensor for spontaneousrelease that has a higher apparent Ca2+ affinity than synaptotagmin. Asa result, deletion of synaptotagmin led to a paradoxical increase inspontaneous release mediated by this second Ca2+ sensor, which isnormally clamped by synaptotagmin.

Received 17 December 2008; accepted 24 March 2009; published online 3 May 2009; doi:10.1038/nn.2320

1Department of Molecular & Cellular Physiology, and 2Howard Hughes Medical Institute, Stanford University, Palo Alto, California, USA. 3Department of Neuroscience,4Howard Hughes Medical Institute, and 5Department of Molecular Genetics, The University of Texas Southwestern Medical Center, Dallas, Texas, USA. 6Present address:GlaxoSmithKline (China) R&D Co. Pudong, Shanghai, China (J.X.), and Department of Neuroscience & Cell Biology, University of Texas Medical Branch, Galveston, Texas,USA (O.-H.S.). Correspondence should be addressed to T.C.S. (tcs1@stanford.edu).

NATURE NEUROSCIENCE VOLUME 12 [ NUMBER 6 [ JUNE 2009 759

ART ICLES

©20

09 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

RESULTS

Spontaneous release is Ca2+ dependent

Using cultured cortical neurons, we measured evoked inhibitorypostsynaptic currents (IPSCs) and spontaneous miniature IPSCs(mIPSCs) in extracellular medium containing or lacking Ca2+.Removal of extracellular Ca2+ only partly depressed the frequency ofspontaneous mIPSCs, but almost completely blocked evoked IPSCs(Fig. 1a,b and data not shown). Thus, as previously shown, sponta-neous mini release is less sensitive than evoked release to reductions inextracellular Ca2+ concentration1–4. However, when we pre-incubatedcultured neurons with 1,2-bis-(o-aminophenoxy)-ethane-N,N,N¢,N¢-tetraacetic acid, tetraacetoxymethyl ester (BAPTA-AM), a membrane-permeable Ca2+ chelator, almost all mIPSCs (495%) were blocked(Fig. 1a,b and Supplementary Fig. 1 online). The addition of thapsi-gargin to neurons in Ca2+-free medium strongly enhanced the minifrequency, but pre-incubation with BAPTA-AM again blocked sponta-neous mini release (Fig. 1a,b). Caffeine, which acts to increasepresynaptic Ca2+ concentrations30, also increased the frequency ofspontaneous release (Supplementary Fig. 2 online).

As described above, cortical synapses lacking Syt1 show almost noevoked release but instead have a paradoxical increase in spontaneousrelease27, which is thought to reflect a clamping of SNARE complexesby Syt1 (refs. 20,21). This increased spontaneous release, together withthe loss of evoked release, could be rescued by the expression ofexogenous Syt1 and was therefore not developmentally induced(Fig. 1c,d). However, the Ca2+ requirement of mini release in wild-type synapses raises the question whether the increased minis in Syt1-deficient synapses may also be Ca2+ dependent. Indeed, we found that

495% of mIPSCs in Syt1-deficient synapses were suppressed byBAPTA-AM (Fig. 1c,d), indicating that minis are Ca2+ triggered andcannot be a result of simple ‘popping’ of the SNARE assemblyfor fusion.

A potential concern for these experiments is that Ca2+ may berequired for the viability of neurons and the integrity of synapses. Toensure that the unexpectedly strong suppression of mIPSCs by BAPTA-AM specifically reflects the effect of withdrawing Ca2+ on mIPSCs andis not a result of a simple loss of all synaptic function, we measured thereadily releasable pool (RRP) of vesicles in synapses that were pre-incubated with BAPTA-AM. As estimated by application of hypertonicsucrose31, we found no decrease in the size of the RRP, which wemeasured as the total synaptic charge transfer that was induced byhypertonic sucrose after pre-incubation with BAPTA-AM (Fig. 1e,f).This result also eliminated the possibility of defects in postsynapticGABAA receptors after treatment with BAPTA-AM.

Spontaneous release in Syt1 knockout synapses

These results raise several questions. First, are minis in wild-type andSyt1-deficient synapses produced by an identical Ca2+-dependentprocess that is simply enhanced by the Syt1 knockout or is there afundamental difference in the Ca2+ triggering of minis between wild-type and Syt1-deficient synapses? Second, do these results apply equallyto excitatory and inhibitory synapses? Third, what is/are the Ca2+

sensor(s) for spontaneous mini release?To address the first two questions, we measured the Ca2+ dependence

of spontaneous release in wild-type and Syt1-deficient synapses in bothexcitatory and inhibitory synapses (Fig. 2). We analyzed culturedcortical neurons from littermate wild-type and Syt1�/� mice atincreasing concentrations of extracellular Ca2+ and monitored minia-ture excitatory postsynaptic currents (mEPSCs) and mIPSCs in tetro-dotoxin (TTX) after pharmacological isolation. We observed a similarbehavior of spontaneous mini release in excitatory and inhibitorysynapses, but there was a marked difference between wild-typeand Syt1�/� synapses. At all of the Ca2+ concentrations, the mEPSCand mIPSC frequencies were strongly increased in Syt1�/� synapses,with an apparent left shift in the Ca2+-concentration dependence(Fig. 2b,c). To quantify this, we fitted the results of individualexperiments to a Hill function and determined the apparent Ca2+

cooperativity (Hill coefficient) and apparent Ca2+ affinity (Kd) ofextracellular Ca2+ for miniature release. Analysis of multiple indepen-dent experiments revealed a significant decrease in the Ca2+ coopera-tivity and a large increase in the Ca2+-affinity (that is, a 4threefold

Control BAPTA-AM

0RR

P (

syna

ptic

cha

rge

tran

sfer

in n

C)

4

8

12

e f

WT

mIP

SC

freq

uenc

y (H

z)

0

2

12

2 s

50 pA

Syt1–/–

Syt1–/– + Syt1 rescue

Syt1–/– + BAPTA-AM

16

14

d

***

***

n.s.

Sucrose

5 s0.5 nA

Sucrose Sucrose

Sucrose

n.s.

WT Syt1 KO

WT Syt1–/–

WT

Syt1–/– Control BAPTA-AM

50 pA2 s

mIP

SC

freq

uenc

y (H

z)

0

1

2

3

Thapsigargin:BAPTA-AM:

b2 mM Ca2+

0 mM Ca2+ + thapsigargin

0 mM Ca2+

0 mM Ca2+ + thapsigargin+ BAPTA-AM

0 mM Ca2+ + BAPTA-AM

Ca2+ (mM):

a

2 0 0 0 0– – + – +– – +– +

***

*********

c

Contro

l

Contro

l

BAPTA-A

M

Contro

l

BAPTA-A

M

Contro

l

BAPTA-A

M

Syt1 re

scue

Figure 1 Spontaneous miniature release is Ca2+ dependent in wild-type and

Syt1 knockout synapses. (a–d) Representative traces (a,c) and summary

graphs (b,d) of mIPSCs monitored by whole-cell recordings in cultured

cortical neurons from wild-type (WT) and Syt1�/� mice under the indicated

conditions (10 mM BAPTA-AM applied at 37 1C for 1 h before and during

recordings; 1 mM thapsigargin added 5 min after stable recordings were

established). All recordings from cultured neurons were obtained at 14–16 d

in vitro (DIV), all mini recordings were produced in 1 mM TTX, and mIPSCswere monitored in the presence of 50 mM D(-)-2-amino-5-phosphonovaleric

acid (AP5) and 10 mM 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX). Rescue

of Syt1�/� neurons was performed by lentiviral infection at 4 DIV22.

(e,f) Representative traces (e) and summary graphs (f) of IPSCs induced by

puffing 0.5 M sucrose onto neurons from WT and Syt1�/� mice without or

with pretreatment with BAPTA-AM. Recording conditions were the same as

in a–d. Summary graphs depict the synaptic charge transfer integrated over

30 s to estimate the size of the RRP. Data shown in panels b,d and f are

means ± s.e.m. (see Supplementary Table 1 online for all numerical values).

*** P o 0.001, n.s. ¼ nonsignificant (P 4 0.05).

760 VOLUME 12 [ NUMBER 6 [ JUNE 2009 NATURE NEUROSCIENCE

ART ICLES

©20

09 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

decrease in Kd) of spontaneous mini release in Syt1�/� synapses (P o0.001, note that the apparent Ca2+ cooperativity and Ca2+ affinity ofrelease determined by varying the extracellular Ca2+-concentration, asperformed for this study, underestimates the true cooperativity andaffinity as measured by intracellular Ca2+, and is only meant as arelative measure of these parameters, which cannot be translateddirectly into intracellular Ca2+ affinities and cooperativities;Fig. 2d,e). It is notable that even though the relative increase inmini release produced by Syt1 knockout differed markedly betweenexcitatory and inhibitory synapses (mIPSCs, Btwofold increase;mEPSCs, Bfourfold increase), the apparent Ca2+ affinity andCa2+ cooperativity in wild-type and Syt1�/� mice were almost

identical between the two types of synapses. These results, amongothers, validate our approach.

Our results answer the first two of the three questions that we posedabove; Ca2+ triggering of spontaneous release is indeed fundamentallydifferent between wild-type and Syt1-deficient synapses, and this issimilarly true for excitatory and inhibitory synapses. In regards to thethird question, the identity of the Ca2+ sensor(s) involved, it is notablethat the Ca2+-dependent properties of spontaneous and evoked releasein wild-type versus Syt1�/� synapses appear to be very similar.Specifically, both spontaneous and evoked release showed a lowerapparent Ca2+ affinity, but had a higher apparent Ca2+ cooperativityin wild-type synapses than in Syt1�/� synapses. Because Syt1 serves asthe primary Ca2+ sensor for evoked release in cortical wild-typesynapses, whereas an as yet unidentified Ca2+ sensor for asynchronousrelease assumes this role in Syt1�/� synapses27,29, it is tempting tospeculate that Syt1 also serves as the primary Ca2+ sensor for sponta-neous release and that the unknown asynchronous Ca2+ sensor adoptsthis role in the Syt1�/� synapses. This hypothesis implies that Syt1 andthe asynchronous Ca2+ sensor both act on primed vesicles and that theformer normally clamps the latter (Supplementary Fig. 3 online). Thishypothesis predicts that the Ca2+ affinity of Syt1 should dictate themagnitude of spontaneous and evoked release in wild-type synapses, aprediction that can be readily tested.

Knockin mutations in Syt1 Ca2+-binding sites

To test the hypothesis that mIPSCs and mEPSCs are triggered by Ca2+

binding to Syt1, we employed mice carrying knockin point mutationsin the Ca2+-binding site of the Syt1 C2A domain32,33. Three differentmutations were analyzed: D232N, D238N and R233Q (referred to asD2N, D8N, and R3Q, respectively; Fig. 3a). Previous studies havedemonstrated that these mutations selectively alter the Ca2+-bindingproperties of Syt1. Specifically, the D2N mutation enhancesCa2+-dependent binding of Syt1 to SNARE complexes without alteringits phospholipid-binding properties33, whereas the D8N and R3Qmutations moderately (D8N) or severely (R3Q) decrease the apparentCa2+ affinity of phospholipid binding by Syt1 without alteringSNARE complex binding32,33. Moreover, the R3Q mutation wasshown to alter the apparent Ca2+ affinity of release correspondingly,an observation that provided the formal proof of the function of Syt1as Ca2+ sensor for release32, whereas the effects of the other mutationson the apparent Ca2+ affinity and Ca2+ cooperativity of release werenot tested. Thus, the various knockin mice carrying the threedifferent point mutations in Syt1 provide a tool for testing whetherSyt1 has equivalent functions as a Ca2+ sensor in evoked and inspontaneous release.

We first measured the effects of the three knockin mutations onevoked release by comparing neurons from mutant and littermatecontrol mice in the same experiments. We found that the three knockinpoint mutations had distinct effects on the apparent Ca2+ affinity ofrelease that correlated with their biochemical effects; the D2N mutation

1 s

50 pA

a

Ca2+

coop

erat

ivity

0

1

2

Ca2+

affin

ity K

d (m

M)

0

4

8

12

0.01 0.1 1 10

Min

i fre

quen

cy (

Hz)

b c

d

Syt1–/– Syt1–/–

WT

Syt1–/

–W

T

WT

0.01 0.1 1 10

WT

e mEPSCsmIPSCs

mEPSCsmIPSCs

0

1

2

Syt1–/

–

WT

[Ca2+] (mM) [Ca2+] (mM)

0.5 mM Ca2+

1 mM Ca2+

2 mM Ca2+

5 mM Ca2+

10 mM Ca2+

0.2 mM Ca2+

0 mM Ca2+ + 1 mM EGTASyt1–/–

mIPSCs

Ca2+

coop

erat

ivity

Ca2+

affin

ity K

d (m

M)

******

******

Syt1–/

–W

T

Syt1–/

–

WT

1.0

2.0

1.0

2.0

1.51.5

0

5

10

15

20

WTscaled

WTscaled

Figure 2 Enhanced frequency, increased apparent Ca2+ affinity and

decreased Ca2+ cooperativity of spontaneous release in Syt1�/� synapses.

(a–c) Ca2+ dependence of the frequency of mIPSCs and mEPSCs.

Representative mIPSC traces (a, see Supplementary Fig. 2 for mEPSC

traces) and summary graphs (b,c) of the Ca2+ dependence of the mIPSC and

mEPSC frequency are shown (continuous lines ¼ Hill function fits; dotted

lines ¼ scaled WT Ca2+ dependence). (d,e) Mean apparent Ca2+ cooperativity

and Ca2+ affinity of mini release at inhibitory (d) and excitatory (e) synapsesdetermined by Hill function fitting to individual experiments. Data shown in

b–e are means ± s.e.m. (see Supplementary Table 1 for all numerical values).

*** P o 0.001.

NATURE NEUROSCIENCE VOLUME 12 [ NUMBER 6 [ JUNE 2009 761

ART ICLES

©20

09 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

increased the apparent Ca2+ affinity (that is, decreased the apparentKd for Ca2+), whereas the D8N and the R3Q mutations decreased theapparent Ca2+ affinity of release (that is, increased the apparent Kd forCa2+; Fig. 3b–d). None of the mutations significantly altered theapparent Ca2+ cooperativity of release, as expected (P4 0.05; Supple-mentary Fig. 4 online). We then measured the effects of the threeknockin mutations on spontaneous release and observed the sameresult, indicating that miniature release is triggered by Ca2+ binding toSyt1 (Fig. 4a–c and Supplementary Fig. 4). The increase in minifrequency that we observed in knockin mice containing the D2N Syt1mutation that increases its apparent Ca2+ affinity could be fully blockedby pre-incubation with BAPTA-AM, demonstrating that the corre-sponding mini release conforms to the same overall mechanism as inwild-type synapses (Supplementary Fig. 5 online). Thus, spontaneousrelease is normally driven by Ca2+ binding to Syt1 in culturedcortical neurons.

Effect of Syt1 knockin mutations in slices

A potential concern of the experiments described above is that theywere performed in cultured neurons in which abnormal Ca2+ transientscould be generated. To address this concern, we measured the effects ofthe three knockin mutations on spontaneous mini release in acutebrain slices from Syt1 knockin mutant mice and their littermate wild-type controls (Fig. 5a,b). We observed the same effect of the knockinmutations as in cultured neurons, indicating that Syt1 functions as theCa2+ sensor for spontaneous release in wild-type synapses.

Finally, to investigate whether the role of Syt1 as a Ca2+ sensor forspontaneous release operates during a physiologically meaningfulregulation of spontaneous release, we monitored the effect of nicotineon spontaneous release in brain slices from wild-type and knockinmice. Nicotine is known to bind to, among others, presynapticacetylcholine receptors and thereby regulate spontaneous release11–14,which we confirmed for wild-type synapses (Fig. 5c,d). In synapsesfrom Syt1 knockin mutant mice, spontaneous release was still increasedby nicotine, but in the context of the overall regulation of release bySyt1; that is, the overall amount of spontaneous release continued to bedictated by the apparent Ca2+ affinity of Syt1 that was present in thesynapses examined (Fig. 5c,d). This result suggests that the Syt1-dependent regulation of spontaneous release operates under physio-logical conditions.

Syt1 Ca2+-binding sites clamp spontaneous release

How does deletion of Syt1 activate (that is, derepress) Ca2+ triggeringof mini release at Ca2+ concentrations that are lower than those thatactivate Syt1 itself? To explore this question, we infected cultured

0.2 1 10

1

2

3

0.5 5[Ca2+] (mM)

Kd

for

Ca2+

(m

M)

0

D2N D8N R3Q

c

0

2

4

6

8

mIP

SC

freq

uenc

y (H

z)

b

**

**

**

D8N mt

D2N mt

R3Q mt

MutantWT

[Ca2+] D8N WTD2N WT R3Q mtR3Q WT

a

50 pA5 s

0.01 mM

0.2 mM

0.5 mM

1 mM

2 mM

5 mM

10 mM

D8N mtD2N mt

D8N WTD2N WTR3Q WT

*Figure 4 Knockin mutations altering the apparent Ca2+ affinity of Syt1 cause

corresponding changes in the apparent Ca2+ affinity of spontaneous release.

(a) Representative mIPSC traces monitored at different Ca2+ concentrations

in neurons cultured from littermate WT and knockin mice with the indicated

mutations. (b) Mean mIPSC frequency plotted as a function of the extra-

cellular Ca2+ concentration. (c) Mean Kd for extracellular Ca2+ for mIPSCs as

determined by Ca2+ titrations. Data are means ± s.e.m. (see Supplementary

Table 1 for all numerical values). ** P o 0.01, *** P o 0.001.

1 10

0

2

4

6

7

IPS

C a

mpl

itude

(nA

)

0.5 mM 1 mM 2 mM 5 mM 10 mM

1

2

3

Kd

for

Ca2+

(m

M)

a

0

[Ca2+] (mM)

[Ca2+]:

0.5 5

1

3

5

D8N mt

D8N WT

D2N mt

D2N WT

R3Q mt

R3Q WT

MutantWT

2 nA250 ms

b

**

**

**

Ca2+ Ca2+

Mem

bran

e

C2ACN C2B

Link

er

R3Q = R233Q

dc

D8N mt

D8N WTD2N mt

D2N WT

R3Q mt

R3Q WT

Synaptotagmin-1

D2N = D232ND8N = D238N

D2N D8N R3Q

Figure 3 Knockin mutations altering the apparent Ca2+ affinity of Syt1 cause

corresponding changes in the apparent Ca2+ affinity of evoked release.

(a) Point mutations introduced into the C2A domain of mouse Syt1 by

knockin32,33. The D2N mutation increases Ca2+-dependent binding of Syt1

to SNARE complexes33, whereas the D8N and R3Q mutations decrease the

apparent Ca2+ affinity of Syt1-phospholipid membrane complexes32,33.

(b) Representative traces of evoked IPSCs monitored in cultured neurons

from littermate WT and Syt1 D2N, D8N and R3Q knockin mice. mt, mutant.(c) Mean IPSC amplitudes in neurons from WT and knockin mice plotted as a

function of the extracellular Ca2+ concentration. (d) Mean Kd for extracellular

Ca2+ of IPSCs, determined by Hill function fits to individual Ca2+-titration

experiments as shown in c and d. In all experiments with knockin mice, each

mutant was analyzed with its own separate littermate WT control. Data

are means ± s.e.m. (see Supplementary Table 1 for all numerical values,

for other parameters, see Supplementary Fig. 3). ** P o 0.01.

762 VOLUME 12 [ NUMBER 6 [ JUNE 2009 NATURE NEUROSCIENCE

ART ICLES

©20

09 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

neurons from Syt1�/� mice with control lentivirus or lentivirusesexpressing wild-type or mutant Syt1 in which the Ca2+-binding sitesof either the C2A domain, the C2B domain or both of the C2 domainsof Syt1 were mutated (Fig. 6a). In these mutants, three aspartateresidues in the respective Ca2+-binding sites of the C2A and/or the C2Bdomain were exchanged for alanine residues, thereby abolishing Ca2+

binding to the C2 domains. The mutant C2 domains were expressed atsimilar levels in the infected neurons and still folded well (Supplemen-tary Fig. 6 online and data not shown), allowing us to determine theeffect of Ca2+ binding to the individual C2 domains on the Ca2+

triggering and clamping of spontaneous exocytosis.As expected27, evoked IPSCs were impaired in Syt1�/� neurons but

were fully rescued by expression of wild-type Syt1 (Fig. 6b). Mutationof the Ca2+-binding sites of the C2A domain partially blocked rescue ofevoked responses (50% decrease), whereas mutation of the C2Bdomain or of both C2 domains completely blocked rescue of evokedrelease (Fig. 6b). This result is consistent with previous studies thatfound that Ca2+ binding to both the C2A and the C2B domain isinvolved in the Ca2+ triggering of synchronous release, with the C2Bdomain being relatively more important31,34–36. Notably, the mutationsof the C2 domain Ca2+-binding sites also impaired the ability of Syt1 toclamp spontaneous release (Fig. 6c and Supplementary Fig. 7 online).However, the relative efficacy of different mutants differed between

evoked and spontaneous release in that the C2A domain mutation wasmore deleterious than the C2B domain mutation for the clampingfunction of Syt1 (spontaneous mini frequency: C2B domain mutant,B6 Hz; C2A mutant, B10 Hz), whereas the opposite was observed forthe release function of Syt1 (Fig. 6c and Supplementary Fig. 7).

To more precisely quantify the similarity of the effect of the C2A andC2B mutations on miniature release, we measured the Ca2+ depen-dence of the mIPSC frequency in Syt1 knockout neurons expressingwild-type Syt1 or the C2A or C2B domain mutant of Syt1 (Fig. 6d–fand Supplementary Fig. 8 online). Quantitation of the apparent Ca2+

affinity and Ca2+ cooperativity of spontaneous release revealed thatneither the C2A nor the C2B mutation rescued the change in eitherparameter (Fig. 6e,f), despite similar expression levels. These resultssuggest that, in addition to triggering synchronous and spontaneousrelease, the Ca2+-binding sites of both C2 domains are involved inclamping the asynchronous Ca2+ sensor, with the C2A domain beingmore effective than the C2B domain at clamping the second Ca2+

sensor and the C2B domain being more effective than the C2A domainat triggering release.

Ca2+ triggering and clamping of release

A plausible hypothesis to account for the requirement of the C2domain Ca2+-binding sites for clamping mini release is that thesesites clamp mini release in a Ca2+-independent manner but activateevoked and spontaneous release in a Ca2+-dependent manner. It islikely that the ability of Syt1 to evoke synchronous release requires aclose proximity of the Syt1 C2 domains to the membrane in which theSNAREs are located because, as confirmed by our experiments on theknockin mutations (Figs. 3–5), Syt1 functions by simultaneous, rapidand Ca2+-dependent binding to SNARE complexes and to phospho-lipids. Indeed, a most notable feature of synaptotagmin-triggeredrelease is the speed with which it operates during an action potential,which can best be accounted for by a lack of conformational changesinvolved and by short reaction distances. This argument should notapply to the function of Syt1 as a clamp of the asynchronous Ca2+

sensor or as a Ca2+ sensor for spontaneous release, as these functions donot involve transient Ca2+ fluxes that are induced by action potentials.

To test these predictions, we examined the effect of increasing thedistance of the Syt1 C2 domains from the vesicle membrane onspontaneous and evoked release (Fig. 7a). We expressed Syt1 with aduplicated linker sequence between the C2 domains and transmem-brane region. The double-linker mutant rescued only 50% of evokedCa2+ triggered release, but did not alter the apparent Ca2+ affinity orCa2+ cooperativity of evoked release as expected, as the C2 domainswere not changed (Fig. 7b,c, and Supplementary Fig. 9 online).Moreover, the double-linker mutation did not alter the clampingfunction of Syt1 (Fig. 7d,e), as we predicted. Thus, although bothevoked and spontaneous release normally utilize Syt1 as a Ca2+ sensor,only the former requires rapid coupling of the C2 domains to the

Nicotine

Control

50 pA

2 s

2

4

0

WT R3Q

0

WT

0

WT

1

2

3

4

2

6

4

mIP

SC

freq

uenc

y (H

z)

D2N D8N

c

d

Nicotine: – + – + – + – + – + – +

D8N mt

D8N WT

D2N mt

D2N WT

R3Q mt

R3Q WT

1

5

3

5

1

5

3

66

Nicotine

Control

0

mIP

SC

freq

uenc

y (H

z)

2

4

5

**

3

1

b

***

*

1 s

25 pA

Mutant

WTa

D8N mt

D8N WT

D2N mt

D2N WT

R3Q mt

R3Q WT

MutantWT

D2N D8N R3Q

Figure 5 Knockin mutations in Syt1 alter spontaneous release monitored in

acute slices. (a,b) Representative traces (a, calibration bars at the bottom

apply to all traces above the bars) and summary graphs (b) of mIPSCs

recorded in acute cortical slices from littermate WT and Syt1 knockin mice.

Note that the increased frequency of spontaneous release in D2N mutant

mice was fully blocked by BAPTA-AM (Supplementary Fig. 3). (c,d)

Representative traces (c) and frequency summary graphs (d) of mIPSCs

monitored in acute cortical slices WT and Syt1 knockin mice before and afteraddition of 100 mM nicotine, applied at a perfusion rate of 2 ml min�1 at

5 min after stable recordings were established. Data are means ± s.e.m. (see

Supplementary Table 1 for all numerical values and statistical analyses of d).

* P o 0.05, ** P o 0.01, *** P o 0.001.

NATURE NEUROSCIENCE VOLUME 12 [ NUMBER 6 [ JUNE 2009 763

ART ICLES

©20

09 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

membrane because it is dependent on the Ca2+ dynamics induced by anaction potential, whereas the latter operates via a random sensing ofCa2+ sparks and clamping of the asynchronous Ca2+ sensor.

DISCUSSION

Evoked and spontaneous neurotransmitter release are generally con-sidered to represent distinct types of release that are differentiallyregulated1–3. Their distinct natures are evidenced by the fact that

spontaneous release is maintained in the presence of the sodium-channel inhibitor TTX, which abolishes action potentials and evokedrelease. We found, however, that despite their differential regulation,these two types of release are mechanistically identical in that they bothare triggered by Ca2+ binding to Syt1 (Figs. 1–5). The major evidencefor this conclusion rests on the three Syt1 knockin mutations thatwe used. We previously demonstrated that these mutations eitherincrease Ca2+-dependent binding of Syt1 to SNARE complexes(D2N) or decrease the apparent Ca2+ affinity of Syt1 binding tophospholipids (D8N and R3Q)32,33. In a direct comparison of allthree knockin mutations, we found that they cause a correspondingchange in evoked release and a precisely equivalent change in sponta-neous release (Figs. 3–5).

The finding that Syt1 functions as a Ca2+ sensor for both sponta-neous and evoked release was surprising to us because Syt1 is thoughtto be physiologically activated only by Ca2+ influx during an actionpotential, as the Ca2+ affinity of Syt1 is too low for it to be activated bysubmicromolar resting Ca2+ concentrations32. Indeed, it is likely thatspontaneous release is triggered not by the resting Ca2+ concentrationin the presynaptic terminal, but by Ca2+ sparks17,18. Ca2+ sparks mayarise from internal stores or Ca2+ influx, as indicated by the increase inspontaneous release induced by thapsigargin, nicotine and caffeine andthe dependence of spontaneous release on extra- and intracellular Ca2+

(Figs. 1, 2 and 5 and Supplementary Fig. 2). The Syt1 knockinmutations caused similar changes in spontaneous release in neuronsin culture (Figs. 3 and 4) and in acute brain slices (Fig. 5), suggestingthat the role of Syt1 as a Ca2+ sensor for spontaneous release is notpeculiar to cultured neurons, but is generally applicable. Moreover, ourresults support previous suggestions7–9 that spontaneous release isphysiologically important. Ca2+ regulation generally implies a physio-logically controlled function; thus, the finding that spontaneous releaseis controlled by Ca2+ binding to Syt1 implies a physiological role, as isalso supported by the observation that the potentiation of spontaneousrelease by nicotine operated via Ca2+ binding to Syt1 (Fig. 5). Manyneurotransmitters and neuromodulators act by increasing or decreas-ing presynaptical Ca2+ concentrations10–16, suggesting that these agentsmay control synaptic circuits, at least in part, by regulating Syt1-dependent spontaneous release without triggering action potentials.

A major argument against the notion that Syt1 and other synapto-tagmins are Ca2+ sensors for spontaneous release was the finding that

0

8

16m

IPS

C fr

eque

ncy

(Hz)

0

1

2

3

c

e f

Evo

ked

IPS

C a

mpl

. (nA

)

d

0.01 0.1 1 10

0

4

8

12

16

mIP

SC

freq

uenc

y (H

z)

Ca2+ concentration (mM)

0

0.5

1.0

1.5

2.0

Syt1–/–

WT

Syt1–/–

+ C2A3DA

Syt1–/–

+ C2B3DA

WTSyt1–/–

Syt1–/–

WTSyt1–/–

WT

4

12

2

6

10

14

0.05 0.5 5

WTSyt1–/–b

Syt1–/–

rescue:

Syt1–/–

rescue:Syt1–/–

rescue:

Syt1–/–

rescue:

***

***

WTCtrl

Syt1

Syt1 C

2A3D

A

Syt1 C

2B3D

A

Syt1 C

2AB6D

ACtrl

Ctrl

Syt1

Syt1 C

2A3D

A

Syt1 C

2A3D

A

Syt1 C

2B3D

A

Syt1 C

2B3D

ACtrl

Syt1 C

2A3D

A

Syt1 C

2B3D

A

Syt1 C

2AB6D

AWT

***

***

***

***

*** ***

***

******

1.75

1.50

2.00

1.25Ca2+

coo

pera

tivity ***

******

1.00

3DAC2A3DA = D178A, D230A, D232AC2B3DA = D309A, D363A, D365A

C2AB6DA = C2A3DA + C2B3DA

a

Ca2+ Ca2+

Mem

bran

e

C2ACN

C2B

Link

er

Kd

for

Ca2+

(m

M)

Figure 6 Syt1 clamping of spontaneous release requires intact Syt1

Ca2+-binding sites. (a) Diagram of mutations in the C2A and/or the C2B

domain of Syt1 that abolish Ca2+ binding to individual C2 domains. Each

mutation consists of three separate aspartate-to-alanine substitutions in the

Ca2+-binding site (referred to as DA mutations), with the precise substitutions

listed on the right. Mutations were designed on the basis of the atomic

structures of the C2A and C2B domains42,43. (b,c) Comparison of the evoked

IPSC amplitude (b) and spontaneous mIPSC frequency (c) in neurons fromlittermate WT or Syt1�/� mice. Syt1�/� neurons were infected with either

control lentivirus (Ctrl), lentivirus expressing WT Syt1 (Syt1) or lentivirus

expressing mutant Syt1 (Syt1 C2A3DA, mutant lacking the C2A domain

Ca2+-binding sites; Syt1 C2B3DA, mutant lacking the C2B domain

Ca2+-binding sites; Syt1 C2AB6DA, mutant lacking the C2A and the C2B

domain Ca2+-binding sites; see Supplementary Figs. 4–6 for immunoblots

and representative traces). (d) Mean spontaneous mIPSC frequency as a

function of the extracellular Ca2+ concentration in WT or Syt1�/� neurons

expressing C2A3DA or C2B3DA mutant Syt1 (continuous lines, fittings with

Hill functions; dotted lines, scaled fittings; see Supplementary Fig. 6 for

representative traces). (e,f) Summary graphs of the mean Ca2+ cooperativity

(e) and Kd for extracellular Ca2+ (f) determined from the data shown in d.

Data shown are means ± s.e.m. (see Supplementary Table 1 for all numerical

values). *** P o 0.001, n.s. ¼ nonsignificant (P 4 0.05).

764 VOLUME 12 [ NUMBER 6 [ JUNE 2009 NATURE NEUROSCIENCE

ART ICLES

©20

09 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

deletions of Syt1 and of Syt2 paradoxically increase spontaneousrelease26–28. We found that the paradoxical increase in spontaneousrelease induced by deletion of Syt1 was a result of the unclamping, notof SNARE complexes, but of an alternative second Ca2+ sensor. Thisfinding argues against the notion that Syt1 acts as a clamp of SNAREcomplexes20,21, a notion that was plausible given the binding of Syt1 toSNARE complexes and the fact that SNARE complexes are probablyassembled during priming, ready to go for release19, but is contradictedby our data. Clearly, SNARE complexes are tight and do not simply popbecause otherwise spontaneous fusion would not be Ca2+ dependent.

The second Ca2+ sensor for spontaneous release that is unclampedby the Syt1�/� has a substantially higher apparent Ca2+ affinity and asubstantially lower apparent Ca2+ cooperativity than Syt1. Theseproperties correspond to those of the second Ca2+ sensor for releasethat drives asynchronous exocytosis and that we previously character-ized biophysically in the calyx of Held synapse37. In the calyx of Heldsynapse, Syt2 functions as Ca2+ sensor for synchronous evoked releaseinstead of Syt1, and asynchronous release persists in this synapse afterknockout of Syt2, similar to asynchronous release in cortical synapsesafter knockout of Syt1 (refs. 22,28,37). The calyx of Held synapse allowsprecise manipulations and measurements of presynaptic Ca2+ concen-trations and of exocytosis38,39, which differs from other synapses,permitting us to define the biophysical properties of Syt2 as the Ca2+

sensor for synchronous release and of the second Ca2+ sensor thatmediates asynchronous release, although the identity of this Ca2+

sensor remains unknown37. The second Ca2+ sensor shows a lowerCa2+ cooperativity than synaptotagmins a as Ca2+ sensor for synchro-nous release but, as a result of this lower Ca2+ cooperativity, shows ahigher apparent Ca2+ affinity37. These findings led to the dual Ca2+

sensor of neurotransmitter release model, according to which evoked

release is normally mediated by synaptotagmins, whereas the secondCa2+ sensor generally makes only a minor contribution to release, but isactivated during high-frequency stimulation27,29,37. The most parsi-monious explanation of our results is that the same ‘second’ Ca2+

sensor is also responsible for the increased spontaneous release in Syt1and Syt2 knockout synapses. At present, we cannot rule out thepossibility that more than two types of Ca2+ sensors mediate release(that is, that evoked asynchronous release under physiological condi-tions and the increased spontaneous release after Syt1 or Syt2 deletionsare driven by distinct Ca2+ sensors), but the similarity between thedifferent forms of release argues against this possibility. Thus, it appearsprobable that the dual Ca2+ sensor model is generally applicable tospontaneous and evoked release and that all release is under the controlof only two competing Ca2+ sensors: synaptotagmins that dominateunder physiological conditions and an unknown second Ca2+-sensorthat kicks in when synaptotagmin is deleted or when synapses arestimulated repeatedly (Supplementary Fig. 3).

Although it remains unclear as to how Syt1 and Syt2 clamp thesecond Ca2+ sensor, our results demonstrate that the Ca2+-binding sitesof Syt1 themselves are essential for clamping. Moreover, we found thatthe C2B domain of Syt1 was more important than the C2A domain forevoked exocytosis (Fig. 6b, as reported previously34,35), but that theC2A domain was more important for clamping spontaneous releasethan the C2B-domain (Fig. 6c,d). Moreover, clamping did not dependon the distance of the C2 domains from the membrane, whereas evokedrelease did (Fig. 7). Thus, two lines of evidence support the notion thatthe two functions of Syt1, Ca2+ sensing versus Ca2+ clamping, aremediated by related, but distinct, mechanisms.

Our data also raise new questions. Most notably, what is the identityof the second Ca2+ sensor and how does it interact with synaptotag-mins? How do the Ca2+-binding sites of Syt1 clamp the second Ca2+

sensor? Because the second Ca2+ sensor operates with an apparentlyhigher Ca2+ affinity, this effect is unlikely to involve Ca2+ binding toSyt1. Clamping must either be mediated by the top C2 domainsequences independently of Ca2+ or by C2 domains with partiallyoccupied Ca2+-binding sites. In the calyx synapse, we have observedcompetition between the synchronous and asynchronous Ca2+ sen-sors37, as did previous studies in hippocampal synapses40,41. In con-trast, our current results and our previous studies on cortical synapses27

provide evidence for clamping of asynchronous release by the syn-chronous Ca2+ sensor. This issue may be related to the puzzling absenceof increased mini release in Syt1-deficient autapses29,32; resolving it willrequire the identification of the asynchronous Ca2+ sensor. Anotherquestion is whether there is a physiological regulation of Syt1 thatinactivates it (and evokes release), thereby activating spontaneous minirelease, a possibility that would confer a new dynamic dimension ontosynapses and account for the physiological role of the mechanisms thatwe describe here. Answering this question again requires the identifica-tion of the asynchronous Ca2+ sensor.

mIPSCs

a

d e

b0.5 mM 1 mM 2 mM

Syt12× linker

WT

Ca2+ Ca2+

C2ACN

C2B

Link

er

Ca2+ Ca2+

C2ACN

C2B

Mem

bran

e

50 pA

1 s

0

1

2

mIP

SC

freq

uenc

y (H

z)

5 mM

2 nA1 s

10 mM [Ca2+]Li

nker

Link

er

Syt1 2× linker

WT

Syt12× linker

WT

IPSCs

Syt1 2

×LWT

0

2

4

6

eIP

SC

am

plitu

de (

nA)

1 10

c

[Ca2+] (mM)

Syt12× linker

WT

***

Figure 7 Membrane proximity of Syt1 C2 domains contributes to evoked,

but not spontaneous, release. (a) Diagram of insertion mutation in Syt1

that duplicates the linker separating C2 domains from the membrane.

(b) Representative traces of evoked IPSCs in WT neurons and Syt1�/�

neurons expressing mutant Syt1 with a double linker. Responses were

monitored at different extracellular Ca2+ concentrations. (c) Plot of the mean

IPSC amplitude as a function of the extracellular Ca2+ concentration (see

Supplementary Fig. 9 for summary graphs). (d,e) Representative traces (d)and summary graph of the frequency (e) of mIPSCs monitored in WT neurons

and in Syt1�/� neurons expressing the double-linker mutant of Syt1. Data

shown in all panels are means ± s.e.m. (see Supplementary Table 1 for all

numerical values). *** P o 0.001, n.s. ¼ nonsignificant.

NATURE NEUROSCIENCE VOLUME 12 [ NUMBER 6 [ JUNE 2009 765

ART ICLES

©20

09 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

METHODS

Methods and any associated references are available in the onlineversion of the paper at http://www.nature.com/natureneuroscience/.

Note: Supplementary information is available on the Nature Neuroscience website.

ACKNOWLEDGMENTSWe thank I. Kornblum, J. Mitchell, L. Fan and A. Roth for excellent technicalassistance, and A. Maximov and C. Zhang for advice.

AUTHOR CONTRIBUTIONSJ.X., Z.P.P. and O.-H.S. planned, performed and analyzed the experiments. T.C.S.conceived the project, supervised the experiments and wrote the paper.

Published online at http://www.nature.com/natureneuroscience/

Reprints and permissions information is available online at http://npg.nature.com/

reprintsandpermissions/

1. Fatt, P. & Katz, B. Spontaneous subthreshold activity at motor nerve endings. J. Physiol.(Lond.) 117, 109–128 (1952).

2. Glitsch, M.D. Spontaneous neurotransmitter release and Ca2+—how spontaneous isspontaneous neurotransmitter release? Cell Calcium 43, 9–15 (2008).

3. Chung, C. & Kavalali, E.T. Seeking a function for spontaneous neurotransmission. Nat.Neurosci. 9, 989–990 (2006).

4. Deitcher, D.L. et al. Distinct requirements for evoked and spontaneous release ofneurotransmitter are revealed by mutations in the Drosophila gene neuronal-synapto-brevin. J. Neurosci. 18, 2028–2039 (1998).

5. Atasoy, D. et al. Spontaneous and evoked glutamate release activates two populations ofNMDA receptors with limited overlap. J. Neurosci. 28, 10151–10166 (2008).

6. Sara, Y., Virmani, T., Deak, F., Liu, X. & Kavalali, E.T. An isolated pool of vesiclesrecycles at rest and drives spontaneous neurotransmission. Neuron 45, 563–573(2005).

7. McKinney, R.A., Capogna, M., Durr, R., Gahwiler, B.H. & Thompson, S.M. Miniaturesynaptic events maintain dendritic spines via AMPA receptor activation. Nat. Neurosci.2, 44–49 (1999).

8. Carter, A.G. & Regehr, W.G. Quantal events shape cerebellar interneuron firing. Nat.Neurosci. 5, 1309–1318 (2002).

9. Sutton, M.A. et al. Miniature neurotransmission stabilizes synaptic function viatonic suppression of local dendritic protein synthesis. Cell 125, 785–799(2006).

10. Prestwich, S.A., Forda, S.R. & Dolphin, A.C. Adenosine antagonists increase sponta-neous and evoked transmitter release from neuronal cells in culture. Brain Res. 405,130–139 (1987).

11. Lena, C., Changeux, J.P. & Mulle, C. Evidence for ‘‘preterminal’’ nicotinic receptors onGABAergic axons in the rat interpeduncular nucleus. J. Neurosci. 13, 2680–2688(1993).

12. Barazangi, N. & Role, L.W. Nicotine-induced enhancement of glutamatergic andGABAergic synaptic transmission in the mouse amygdala. J. Neurophysiol. 86,463–474 (2001).

13. Lambe, E.K., Picciotto, M.R. & Aghajanian, G.K. Nicotine induces glutamate releasefrom thalamocortical terminals in prefrontal cortex. Neuropsychopharmacology 28,216–225 (2003).

14. Radcliffe, K.A., Fisher, J.L., Gray, R. & Dani, J.A. Nicotinic modulation of glutamateand GABA synaptic transmission of hippocampal neurons. Ann. NY Acad. Sci. 868,591–610 (1999).

15. Simkus, C.R. & Stricker, C. The contribution of intracellular calcium stores to mEPSCsrecorded in layer II neurons of rat barrel cortex. J. Physiol. (Lond.) 545, 521–535(2002).

16. Grillner, P., Berretta, N., Bernardi, G., Svensson, T.H. & Mercuri, N.B. Muscarinicreceptors depress GABAergic synaptic transmission in rat midbrain dopamine neurons.Neuroscience 96, 299–307 (2000).

17. Llano, I. et al. Presynaptic calcium stores underlie large-amplitude miniature IPSCs andspontaneous calcium transients. Nat. Neurosci. 3, 1256–1265 (2000).

18. Emptage, N.J., Reid, C.A. & Fine, A. Calcium stores in hippocampal synaptic boutonsmediate short-term plasticity, store-operated Ca2+ entry and spontaneous transmitterrelease. Neuron 29, 197–208 (2001).

19. Rizo, J. & Rosenmund, C. Synaptic vesicle fusion. Nat. Struct. Mol. Biol. 15, 665–674(2008).

20. Yoshihara, M., Adolfsen, B. & Littleton, J.T. Is synaptotagmin the calcium sensor? Curr.Opin. Neurobiol. 13, 315–323 (2003).

21. Sollner, T.H. Regulated exocytosis and SNARE function. Mol. Membr. Biol. 20,209–220 (2003).

22. Xu, J., Mashimo, T. & Sudhof, T.C. Synaptotagmin-1, -2 and -9: Ca2+ sensors for fastrelease that specify distinct presynaptic properties in subsets of neurons. Neuron 54,567–581 (2007).

23. Verhage, M. et al. Synaptic assembly of the brain in the absence of neurotransmittersecretion. Science 287, 864–869 (2000).

24. Schoch, S. et al. SNARE function analyzed in synaptobrevin/VAMP knockout mice.Science 294, 1117–1122 (2001).

25. Delgado-Martınez, I., Nehring, R.B. & Sørensen, J.B. Differential abilities of SNAP-25homologs to support neuronal function. J. Neurosci. 27, 9380–9391 (2007).

26. Littleton, J.T., Stern, M., Perin, M. & Bellen, H.J. Calcium dependence of neurotrans-mitter release and rate of spontaneous vesicle fusions are altered in Drosophilasynaptotagmin mutants. Proc. Natl. Acad. Sci. USA 91, 10888–10892 (1994).

27. Maximov, A. & Sudhof, T.C. Autonomous function of synaptotagmin 1 in triggeringasynchronous release independent of asynchronous release. Neuron 48, 547–554(2005).

28. Pang, Z.P., Sun, J., Rizo, J., Maximov, A. & Sudhof, T.C. Genetic analysis of synapto-tagmin 2 in spontaneous and Ca2+-triggered neurotransmitter release. EMBO J. 25,2039–2050 (2006).

29. Geppert, M. et al. Synaptotagmin I: a major Ca2+ sensor for transmitter release at acentral synapse. Cell 79, 717–727 (1994).

30. Elmqvist, D. & Feldman, D.S. Calcium dependence of spontaneous acetyl-choline release at mammalian motor nerve terminals. J. Physiol. (Lond.) 181,487–497 (1965).

31. Rosenmund, C. & Stevens, C.F. Definition of the readily releasable pool of vesicles athippocampal synapses. Neuron 16, 1197–1207 (1996).

32. Fernandez-Chacon, R. et al. Synaptotagmin I functions as a calcium regulator of releaseprobability. Nature 410, 41–49 (2001).

33. Pang, Z.P., Shin, O.-H., Meyer, A.C., Rosenmund, C. & Sudhof, T.C. A gain-of-functionmutation in synaptotagmin-1 reveals a critical role of Ca2+-dependent SNARE-complexbinding in synaptic exocytosis. J. Neurosci. 26, 12556–12565 (2006).

34. Mackler, J.M., Drummond, J.A., Loewen, C.A., Robinson, I.M. & Reist, N.E. The C2BCa2+-binding motif of synaptotagmin is required for synaptic transmission in vivo.Nature 418, 340–344 (2002).

35. Nishiki, T. & Augustine, G.J. Dual roles of the C2B domain of synaptotagmin I insynchronizing Ca2+-dependent neurotransmitter release. J. Neurosci. 24, 8542–8550(2004).

36. Stevens, C.F. & Sullivan, J.M. The synaptotagmin C2A domain is part of the calciumsensor controlling fast synaptic transmission. Neuron 39, 299–308 (2003).

37. Sun, J. et al. A dual Ca2+ sensor model for neurotransmitter release in a central synapse.Nature 450, 676–682 (2007).

38. Schneggenburger, R. & Neher, E. Intracellular calcium dependence of transmitterrelease rates at a fast central synapse. Nature 406, 889–893 (2000).

39. Bollmann, J.H., Sakmann, B. & Borst, J.G. Calcium sensitivity of glutamate release in acalyx-type terminal. Science 289, 953–957 (2000).

40. Otsu, Y. et al. Competition between phasic and asynchronous release for recoveredsynaptic vesicles at developing hippocampal autaptic synapses. J. Neurosci. 24,420–433 (2004).

41. Hagler, D.J. Jr. & Goda, Y. Properties of synchronous and asynchronous release duringpulse train depression in cultured hippocampal neurons. J. Neurophysiol. 85,2324–2334 (2001).

42. Ubach, J., Zhang, X., Shao, X., Sudhof, T.C. & Rizo, J. Ca2+ binding to synaptotagmin:how many Ca2+ ions bind to the tip of a C2-domain? EMBO J. 17, 3921–3930 (1998).

43. Fernandez, I. et al. Three-dimensional structure of the synaptotagmin 1 C2B-domain:synaptotagmin 1 as a phospholipid binding machine. Neuron 32, 1057–1069 (2001).

766 VOLUME 12 [ NUMBER 6 [ JUNE 2009 NATURE NEUROSCIENCE

ART ICLES

©20

09 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.

ONLINE METHODSReagents, mice and neuronal cultures. We obtained AP5, CNQX, bicuculline

and CGP55845 from Tocris, nicotine and caffeine from Sigma, and EGTA-AM

and BAPTA-AM from CalBioChem. Neurons were cultured as described

previously22,27 in modified eagle medium (Invitrogen) supplemented with

B27 (Invitrogen), glucose, transferrin, fetal bovine serum and cytosine

b-D-arabinofuranoside (Sigma). Syt1 knockout and knockin mice were

described previously29,32,33. All experiments were performed on either cortical

neurons cultured from littermate wild-type and mutant offspring from

heterozygous crossings (for Syt1 knockin mice) or neurons from the same

culture infected with the various test and control viruses (for Syt1 knockout

mice rescue experiments).

Generation and use of recombinant lentivirus. The lentiviral constructs that

we used were based on the pFUW vector (modified from pFUGW vector44), a

shuttle vector containing the gene of interest and recombination arms for

incorporating into the mammalian genome. The vector includes the HIV-1 flap

sequence, the human polyubiquitin promoter-C, a multicloning site and a

WRE element22. Recombinant lentiviruses were produced by transfecting HEK

293T cells with pFUW, pVSVg and pCMVD8.9 using FuGENE. Viruses were

harvested 48 h after transfection by collecting the medium from transfected

cells and a 0.45-mm filter was used to remove cellular debris. Protein expression

was tested in the transfected 293T cells to ensure that viruses were working

before they were applied to neurons. Neurons were infected with 500 ml of

conditioned cell medium for each 24-well of high-density neurons at 4 DIV, the

medium was exchanged for normal growth medium at 6 DIV, and cells were

analyzed at 14–18 DIV. Syt1 point mutations were generated using standard

procedures45. The Syt1 2� linker constructs were made by PCR using two sets

of primers (JX051: 5¢-CGG AAT TCA TGG TGA GTG CCA GTC ATC CT-3¢and JX065: 5¢-GGC AGG CAC TGC AGA AGG ACG-3¢; JX066: 5¢-GAA AGT

ATA GGA ACT TCG TCG ATC GAC CTC G-3¢ and OH05303: 5¢-GCT CTA

GAT TAC TTC TTG ACA GCC AGC ATG GC-3¢).

Electrophysiological recordings from cultured cortical neurons. We carried

out recordings essentially as described22,33,46. mIPSCs and eIPSCs were recorded

in the presence of 10 mM CNQX and 50 mM AP5. mEPSCs were recorded in the

presence of 30 mM bicuculline and 5 mM CGP55845. EPSCs were recorded with

a pipette solution 123 mM potassium gluconate, 10 mM KCl, 1 mM MgCl2,

10 mM HEPES, 1 mM EGTA, 0.1 mM CaCl2, 1 mM K2ATP, 0.2 mM Na4GTP

and 4 mM glucose, pH adjusted to 7.2 with KOH. mIPSCs and mEPSCs were

analyzed by Mini Analysis Software (Synaptosoft; mIPSC search parameters:

gain, 20; blocks, 3,940; threshold, 10 pA; period to search for a local maximum,

20,000 ms; time before a peak for baseline, 10,000 ms; period to search a decay

time, 20,000; fraction of peak to find a decay time, 0.5; period to average a

baseline, 2,000 ms; area threshold, 30; number of points to average for peak, 3;

direction of peak, negative; mEPSC search parameters: gain, 20; blocks, 3,940;

threshold, 10 pA; period to search for a local maximum, 20,000 ms; time before a

peak for baseline, 5,000 ms; period to search a decay time, 5,000; fraction of peak

to find a decay time, 0.5; period to average a baseline, 2,000 ms; area threshold,

10; number of points to average for peak, 3; direction of peak, negative). The

initial analysis was done automatically by the software with subsequent visual

proofreading. Application of 0.5 M sucrose (in extracellular solution) was

effected with a picospritzer for 30 s (pressure ¼ 10 psi), with puffing pipette

tip placed 50 mm from the soma. For Ca2+-titration experiments, IPSCs, mIPSCs

and mEPSC were recorded in bath solutions containing the various Ca2+

concentrations (0 + 1 mM EGTA, 0.2, 0.5, 1, 2, 5 and 10 for mIPSC or mEPSC;

0.5, 1, 2, 5 and 10 mM for IPSC), and the frequency of mIPSCs or mEPSCs or

the amplitude of the evoked IPSCs was plotted as a function of the Ca2+

concentration on a logarithm axis. The curves were fitted with a Hill function

(Y ¼ VmaxXn

Kn+Xn , where Vmax is the maximal response, K is the Ca2+ affinity and n is

the cooperativity32), using IGOR Pro 4.03 software (Wavemetrics) for the

fitting, and the maximal responses, the apparent Kd for Ca2+ and the Hill

coefficient (that is, apparent Ca2+ cooperativity) were derived from the fits32.

Electrophysiological recordings from acute cortical slices. Cortical slices

(300 mm) were prepared from male littermate mice aged 13–16 d. Anesthetized

mice were decapitated, the brains were removed and placed into ice cold

dissection buffer (87 mM NaCl, 3 mM KCl, 1.25 mM NaH2PO4, 7 mM MgSO4,

26 mM NaHCO3, 20 mM D-glucose, 75 mM sucrose, 1.3 mM ascorbic acid and

0.5 mM CaCl2), and slices were cut with a Leica vibratome. The slices were

incubated at 34 1C in artificial cerebrospinal fluid (126 mM NaCl, 3 mM KCl,

1.25 mM NaH2PO4, 2 mM MgSO4, 26 mM NaHCO3, 25 mM D-glucose and

2 mM CaCl2) gassed with 95%O2/5%CO2 for 1 h, and then kept at 24 1C under

the same conditions until they were transferred to the recording chamber,

which was perfused at 2 ml min�1 with carbogenated artificial cerebrospinal

fluid containing 20 mM CNQX, 50 mM AP5 and 1 mM TTX to measure mIPSC.

Slices were equilibrated for 10 min before recordings. All recordings were

performed in layer 2/3 pyramidal neurons of the somatosensory cortex;

pyramidal neurons were identified by their size and single apical dendrite.

Recordings were obtained in voltage-clamp whole-cell mode using a Multi-

clamp 700B amplifier and a holding potential of �70 mV. The whole-cell

pipette solution contained 145 mM KCl, 5 mM NaCl, 10 mM HEPES, 10 mM

EGTA, 0.3 mM Na2GTP, 4 mM MgATP and 10 mM QX-314. The pipettes that

we used for whole-cell recording had a resistance of 3–5 MO; neurons with a

series resistance of 420 MO or a leak current of 4200 pA were discarded.

Spontaneous miniature postsynaptic currents were monitored over a 5-min

period and the series resistance was monitored before and after each recording.

Data were analyzed offline using Mini-Analysis. At least three littermate

pairs of wild-type and D2N, D8N and R3Q knockin mice were analyzed for

each parameter.

Statistical analyses. All statistical comparisons were made using Student’s t test

or a two-way ANOVA as described in Supplementary Table 1.

44. Lois, C., Hong, E.J., Pease, S., Brown, E.J. & Baltimore, D. Germline transmission andtissue-specific expression of transgenes delivered by lentiviral vectors. Science 295,868–872 (2002).

45. Han, W. et al. N-glycosylation is essential for vesicular targeting of synaptotagmin 1.Neuron 41, 85–99 (2004).

46. Maximov, A., Pang, Z.P., Tervo, D.G. & Sudhof, T.C. Monitoring synaptic transmission inprimary neuronal cultures using local extracellular stimulation. J. Neurosci. Methods161, 75–87 (2007).

doi:10.1038/nn.2320 NATURE NEUROSCIENCE

©20

09 N

atu

re A

mer

ica,

Inc.

All

rig

hts

res

erve

d.