Effects of phosphorylation and neuronal activity on thecontrol of synapse formation by synapsin I

Laura E. Perlini1,2,*, Francesca Botti1,*, Eugenio F. Fornasiero1, Maila Giannandrea1, Dario Bonanomi1,`,Mario Amendola1,3, Luigi Naldini1,3, Fabio Benfenati2,4 and Flavia Valtorta1,§

1San Raffaele Scientific Institute and Vita-Salute University, Via Olgettina 58, 20132 Milano, Italy2Department of Neuroscience and Brain Technologies, The Italian Institute of Technology, Via Morego 30, 16163 Genova, Italy3TIGET, Telethon Institute for Genetics and Medicine, Via Olgettina 58, 20132 Milano, Italy4Department of Experimental Medicine, Section of Physiology, University of Genoa and National Institute of Neuroscience,Viale Benedetto XV, 3, 16132 Genova, Italy

*These authors contributed equally to this work`Present address: Salk Institute, 10010 North Torrey Pines Road, La Jolla, CA 92037§Author for correspondence (valtorta.flavia@hsr.it)

Accepted 12 June 2011Journal of Cell Science 124, 3643–3653� 2011. Published by The Company of Biologists Ltddoi: 10.1242/jcs.086223

SummarySynapsins are synaptic vesicle (SV)-associated proteins that regulate synaptic transmission and neuronal differentiation. At early stages,Syn I and II phosphorylation at Ser9 by cAMP-dependent protein kinase (PKA) and Ca2+/calmodulin-dependent protein kinase I/IV

modulates axon elongation and SV-precursor dynamics. We evaluated the requirement of Syn I for synapse formation by siRNA-mediated knockdown as well as by overexpression of either its wild-type (WT) form or its phosphorylation mutants. Syn1 knockdown at14 days in vitro caused a decrease in the number of synapses, accompanied by a reduction of SV recycling. Although overexpression of

WT Syn I was ineffective, overexpression of its phosphorylation mutants resulted in a complex temporal regulation of synapse density.At early stages of synaptogenesis, phosphomimetic Syn I S9E significantly increased the number of synapses. Conversely,dephosphomimetic Syn I S9A decreased synapse number at more advanced stages. Overexpression of either WT Syn I or its

phosphomimetic S9E mutant rescued the decrease in synapse number caused by chronic treatment with tetrodotoxin at early stages,suggesting that Syn I participates in an alternative PKA-dependent mechanism that can compensate for the impairment of the activity-dependent synaptogenic pathway. Altogether these results indicate that Syn I is an important regulator of synapse formation, whichadjusts synapse number in response to extracellular signals.

Key words: Synaptic vesicle, cAMP-dependent protein kinase (PKA), Tetrodotoxin (TTX), Hippocampal neuron, Lentiviral vector

IntroductionThe synapsins are evolutionarily conserved neuron-specific

phosphoproteins (Kao et al., 1999) that localise at the

presynaptic level, where they interact with synaptic vesicle

(SV) proteins, phospholipids and actin, regulating SV

homeostasis (Cesca et al., 2010). In mammals synapsins are

encoded by three genes: Syn1, Syn2 and Syn3. Alternative

splicing of these genes gives rise to several protein isoforms that

share common domains (De Camilli et al., 1990; Kao et al., 1998;

Sudhof et al., 1989). Synapsin I (Syn I; also known as synapsin-1),

the best-characterized member of the family, has been shown to

control the availability of SVs for exocytosis by linking them to

each other and to the actin cytoskeleton of the nerve terminal,

thereby regulating SV functional pools and tuning

neurotransmitter release (Benfenati et al., 1992; Ceccaldi et al.,

1995; Valtorta et al., 1992).

Although Syn I has been studied for more than 30 years, some

aspects of its function are still not fully understood. The most

controversial role of Syn I concerns its involvement in

neuritogenesis and synaptogenesis (Fornasiero et al., 2010). A

large body of literature indicates that synapsins exert several

roles in neuronal development. Firstly, the expression levels of

synapsins correlate with the structural and functional maturation

of neurites and synaptic contacts (Chin et al., 1995; Ferreira et al.,

1994; Lu et al., 1992; Valtorta et al., 1995). Secondly,

overexpression of Syn I or Syn II in Xenopus motor neurons

increases quantal secretion properties of newly formed synapses,

which is paralleled by an increased SV density, SV clustering and

development of synaptic specializations (Lu et al., 1992;

Schaeffer et al., 1994; Valtorta et al., 1995). Syn I, II, III, I/II

or I/II/III knockout (KO) mice develop normally and display an

overall normal brain cytoarchitecture, neuronal morphology and

synapse number (Gitler et al., 2004a; Rosahl et al., 1995).

Primary cultures of neurons derived from Syn I, II or I/II KO

mice have a delay in synapse formation, although no differences

in synapse density were observed in mature cultures (Chin et al.,

1995; Rosahl et al., 1995; Ferreira et al., 1998).

Synapsins integrate intracellular signals, as prominent

substrates of protein kinases. Among these, cAMP-dependent

protein kinase (PKA) and Ca2+/calmodulin-dependent protein

kinase I/IV (CaMK I/IV), phosphorylate all Syn isoforms at a

single site (site 1, Ser9 in Syn I), which is highly conserved

across vertebrate and invertebrate species (Kao et al., 1999;

Candiani et al., 2010). Phosphorylation of site 1 is involved in

precocious stages of neuron development, as seen in Xenopus

spinal neurons, where phosphorylation of Syn II at this site

stimulates neurite outgrowth both in vitro and in vivo (Kao et al.,

2002). In rat hippocampal neurons phosphorylation of the same

Research Article 3643

Journ

alof

Cell

Scie

nce

site in Syn I regulates the localisation of SVs and their rate of

recycling in axonal growth cones and mature SVs (Bonanomi

et al., 2005). It is unclear if the latter effect modulates the

putative synaptogenic activity of Syn I; however it is known that

the regulation of exo-endocytotic cycles of SV precursors at the

nascent synapse is important for synapse establishment (Matteoli

et al., 1992). Taken together, these observations suggest that

synapsins might regulate synaptogenesis in a phosphorylation-

dependent manner, although the exact role and regulation of Syn

I in formation of synaptic contacts has never been carefully

investigated.

In this study, we investigated the effects of changes in the

levels of expression of Syn I and in its state of phosphorylation at

site 1 in the regulation of synaptogenesis. We show that

physiological levels of expression are necessary for synapse

formation, and that the ratio between phosphorylated and

dephosphorylated Syn I appears to be more important for

synapse formation than the absolute levels of the protein.

Moreover, we report that Syn I phosphorylation rescues the

decrease in synapse number caused by chronic treatment with

tetrodotoxin at early stages of development, thus compensating

for the defective synapse formation caused by blockade of

electrical activity.

ResultsThe knockdown of Syn1 by siRNA decreases the

expression of an array of synaptic proteins involved in

synaptic transmission

To investigate the function of Syn I during synapse formation

in hippocampal cultures we used RNA interference (RNAi)

technology. We identified a Syn I target-specific small interfering

RNA (siRNA) sequence and tested its efficacy in HeLa cells by

co-transfecting Syn Ia and IIa expression plasmids with the

duplex. To minimize the possibility of nonspecific off-target

effects, siRNAs were applied at low concentration (20 nM). As a

control, we used luciferase (luc) siRNA. Syn1 siRNA specifically

and efficiently downregulated Syn Ia without affecting Syn IIa

expression (supplementary material Fig. S1A). The same siRNA

was able to downregulate the endogenously expressed protein in

primary hippocampal neurons (supplementary material Fig.

S1B). To evaluate Syn1 siRNA efficacy at single synaptic

boutons, we co-transfected Syn1 siRNA and a plasmid encoding

synaptophysin (Syp)–YFP and measured Syn I fluorescence

intensity at single YFP-positive synapses. Syn I levels were

decreased in at least 63% of Syp–YFP-positive synapses

compared with the control and 58% compared with luc siRNA

synapses (supplementary material Fig. S1C). Moreover, we

realized that the transfection of neurons with Syn1 siRNA

induced a decrease in Syn I level also in Syp–YFP-negative

synapses. In detail, Syn I level was decreased in 38% of total

synapses compared with the control and 36% compared with the

luc siRNA (data not shown). Owing to this widespread effect on

single synapses in our cultures, we decided to evaluate the effect

of Syn1 siRNA on the total number of synapses, even though this

procedure underestimates the effects of siRNAs.

We then analyzed the effect of acute Syn1 knockdown on

synaptic protein expression in primary hippocampal neurons.

Immunoblot analysis revealed that treatment of 14 days in vitro

(DIV) hippocampal neurons for 7 days with Syn1 siRNA

effectively decreased the levels of Syn Ia/Ib as well as the

levels of several synaptic proteins, including other SV proteins

such as Syn IIa/IIb, synaptotagmin I (Syt I), Syp I, vesicular

glutamate transporter-1 (VGLUT-1), vesicular c-aminobutyric

acid (GABA) transporter (VGAT) and Rab3a, the endosomal

protein Rab5, the t-SNARE protein syntaxin 1 (Syx 1) and the

postsynaptic density protein PSD 95 (Fig. 1A,B). Noticeably, for

all synaptic proteins the extent of the decrease was similar to that

of Syn I. The reduction in the levels of both SV proteins and

presynaptic and postsynaptic proteins suggests a decrease in

synapse formation and/or stability upon Syn1 knockdown.

Fig. 1. siRNA-mediated knockdown of Syn1

in hippocampal neurons decreases several

components of synapses. Representative

western blot (A) and quantitative analysis (B) of

the expression levels of synaptic proteins in total

extracts of primary hippocampal neurons (21

DIV) that had been transfected at 14 DIV with

Syn1 siRNA. Data were normalized to the values

of samples transfected with luc siRNA (dashed

line) and analyzed using two-way ANOVA

followed by Tukey’s post-hoc test. *P,0.01 vs

the levels of the same protein in luc-siRNA-

treated samples; ˚P,0.01 vs Syn I levels in

Syn1-siRNA-treated samples (n53).

Journal of Cell Science 124 (21)3644

Journ

alof

Cell

Scie

nce

The knockdown of Syn1 by siRNA induces a parallel

decrease of glutamatergic and GABAergic synapses

The simultaneous presence of an SV population and a

presynaptic scaffold that stabilizes the presynaptic terminal is

an important feature that defines a synapse. We used double

staining for the synapse scaffold (bassoon; Bsn) and SV clusters

(either Syt I, VGLUT1 or VGAT) to identify bona fide synapses

in developing neurons. Because differential alterations at

glutamatergic and GABAergic synapses have been described in

Syn1 null mice (Gitler et al., 2004a), we separately analyzed

excitatory and inhibitory synapse formation in order to identify

any differential effect of Syn1 knockdown or overexpression. To

this aim, we used the colocalisation between VGLUT1 and Bsn

to define glutamatergic synapses and the colocalisation between

VGAT and Bsn to define GABAergic synapses.

To investigate whether Syn I regulates the number of synaptic

contacts, we transfected Syn1 siRNA in hippocampal neurons at

14 DIV, when neuritogenesis is virtually complete and

synaptogenesis is the predominant ongoing neurodevelopmental

process, and, after 7 days, we counted the number of VGLUT1-

and VGAT-positive puncta present in the first 30 mm of dendrites

proximal to the cell body (see Materials and Methods; Fig. 2A).

We found that acute knockdown of Syn1 caused a significant

decrease in the number of both glutamatergic (228.1¡3.2% vs

non-transfected neurons; 226.6¡2.8% vs neurons transfected

with luc siRNA) and GABAergic synapses (217.9¡2.4% vs

non-transfected neurons; 218.6¡2.8% vs neurons transfected

with luc siRNA; Fig. 2B).

To test whether SV exo-endocytosis at single synapses was

also impaired after acute Syn1 knockdown, we monitored SV

recycling using the fluorescent styryl dye FM 4-64 (Betz et al.,

1996). The entire pool of recycling SVs was loaded using high

K+ depolarization of hippocampal neurons for 1 minute (Fig. 3A)

and the quantification of FM 4-64 uptake was performed by

measuring fluorescence intensity at single synaptic boutons.

Despite a broad distribution of the fluorescence intensity values

due to variability among individual synapses, we found that

treatment of neurons with Syn1 siRNA significantly reduced

depolarization-induced FM 4-64 uptake by 20.2¡1.8%

(Fig. 3B,C) suggesting a decrease in the number of active

synapses.

The overexpressed phosphorylation site mutants of Syn I

are correctly targeted to presynaptic sites in primary

hippocampal neurons

Because knockdown of Syn1 decreases the formation of synapses,

we sought to determine whether synapse number is dependent on

the expression levels of Syn1. To this aim, we overexpressed the

enhanced cyan fluorescent protein (ECFP)-tagged WT protein

and its site 1 phosphorylation mutants in rat primary hippocampal

neurons. As a control of infection we used ECFP–VAMP2, i.e.

the fluorescent chimera of an SV protein that was previously

shown not to have detectable effects on synaptogenesis.

Hippocampal neurons were transduced at 3 DIV and fixed at

13 DIV, allowing transgene overexpression during the most

intense period of synapse formation. To exclude a bias due to a

heterogeneous protein expression, cultures were infected with

high efficiency (supplementary material Fig. S2), and the

presence of the tagged protein was verified at all analyzed

synapses. VAMP2, WT Syn I and Syn I S9A localised at the

presynaptic site in mature hippocampal neurons, as previously

described (Bonanomi et al., 2005; Pennuto et al., 2002). The

pseudophosphorylated mutant Syn I S9E was also efficiently

expressed and localised in the presynaptic compartment similarly

to the WT and S9A isoforms, as confirmed by its colocalisation

with the presynaptic markers Syt I and Bsn (Fig. 4A). All the

exogenously expressed Syn I isoforms were expressed in

comparable amounts and exhibited a similar nerve terminal

targeting (Fig. 4B,C). Thus, the localisation of Syn I was not

affected by mutagenesis of phosphorylation site 1, confirming

that phosphorylation of Syn I at this site does not affect nerve

terminal targeting (Gitler et al., 2004b), and allowing direct

comparisons between the effects of the phosphorylation mutants

on synapse formation.

Fig. 2. Acute Syn1 knockdown decreases the total number of glutamatergic and GABAergic synapses in mature hippocampal neurons. (A) Fluorescence

images of hippocampal neurons transfected with either luc-siRNA, Syn1 siRNA or left untreated (Ctrl) at 14 DIV and analyzed at 21 DIV. Glutamatergic and

GABAergic synaptic boutons were identified by double immunostaining with Bsn (red in the merge panels), and either VGLUT-1 or VGAT (green in the merge

panels). The colocalisation panels (Col) highlight the puncta double positive for Bsn and either VGLUT-1 or VGAT that were identified as glutamatergic and

GABAergic synapses, respectively (shown in black; for further details, see Materials and Methods). Scale bar: 10 mm. (B) Quantitative analysis of the number

(means ¡ s.e.m.) of identified glutamatergic (upper panel) and GABAergic (lower panel) synapses counted on a 30 mm dendrite length proximal to the neuron

cell body in untreated neurons and neurons transfected with either luc siRNA or Syn1 siRNA. Statistical analysis was performed using the Kruskal–Wallis and

Tukey’s post-hoc tests applied to the whole data sample. *P,0.05; **P,0.01 vs all other groups; n540 dendrites (total length .3600 mm)/experimental

condition for each of three independent experiments.

Synapsin I controls synaptogenesis 3645

Journ

alof

Cell

Scie

nce

Overexpression of WT Syn I does not further increase the

number of synapses in mature hippocampal neurons

Statistical analysis of the number of infected presynaptic nerve

terminals present in the first 30 mm of dendrites proximal to the

cell body revealed that overexpression of WT Syn I did not affect

the total number of synapses compared with uninfected neurons

and control VAMP2-infected neurons (Fig. 5A,B). This indicates

that, although a decrease in Syn I impairs synaptogenesis, further

Fig. 3. Reduction of FM 4-64 uptake after Syn1 knockdown. (A) Representative fluorescence images of rat hippocampal neurons transfected at 14 DIV with

either luc siRNA, Syn1 siRNA or left untreated (Ctrl) and loaded with FM 4-64 at 21 DIV by exposure for 1 minute to KRH with 55 mM KCl (depolarizing

solution). Scale bar: 10 mm. (B,C) Quantitative analysis of FM 4-64 uptake in single synaptic boutons. The distribution of classes of FM 4-64 fluorescence, shown

as smoothened lines (C), and the histograms of mean (¡ s.e.m.) fluorescence values (B; arbitrary units, AU) are shown for neurons transfected with Syn1 siRNA

(red line), luc siRNA (grey line), or left untreated (ctrl; black line). Statistical analysis was performed using the Kruskal–Wallis and Tukey’s post-hoc tests applied

to the whole data sample. **P,0.01 vs all other groups (n53 independent experiments, .100,000 synaptic boutons per experimental condition).

Fig. 4. ECFP–Syn localise at presynaptic terminals in rat

hippocampal neurons. (A) E18 rat hippocampal neurons were

infected at 3 DIV with the lentiviral constructs encoding ECFP

chimeras of WT Syn I, Syn I S9A or Syn I S9E and processed for

immunofluorescence at 13 DIV. In the merge panels, ECFP–Syn

I is shown in blue, Syt I is shown in green and Bsn is shown in

red. All expressed synapsins strictly colocalise with the two

presynaptic markers at presynaptic terminals. Scale bar: 10 mm.

(B) Targeting factor (means ¡ s.e.m. of three independent

experiments) calculated for WT Syn I, Syn I S9A and Syn I S9E

as the ratio between the total number of Syn-I-positive puncta

and the number of puncta double positive for Syt I and Syn I. The

three transgene products have a very similar nerve terminal

targeting [n.s., not significant; n510 dendrites (total length

.900 mm)/experimental condition for each of three independent

experiments)]. (C) The expression levels of WT Syn I, Syn I S9A

and Syn I S9E were analyzed at 13 DIV by western blotting using

anti-Syn I antibody (G177) to recognize both the endogenous and

transgenic Syn I (85 and 105 kDa apparent molecular mass,

respectively), anti-GFP antibody to detect only transgenic Syn I

and anti-actin antibody to control equal loading.

Journal of Cell Science 124 (21)3646

Journ

alof

Cell

Scie

nce

increases in Syn I expression above physiological levels are

ineffective, suggesting that physiological levels of Syn I are

sufficient to fully exert its synaptogenic potential.

Syn I phosphorylation at site 1 regulates the formation of

both glutamatergic and GABAergic synapses at various

developmental stages

If Syn I controls synapse formation, then its state of

phosphorylation might influence its activity. To test this, in the

same set of experiments we evaluated the synaptogenic potential

of site 1 phosphorylation mutants. Indeed, overexpression of the

dephosphomimetic Syn I S9A mutant caused a significant

decrease (217.7¡3.7%) in the total number of synapses at 13

DIV (Fig. 5A,B), whereas the phosphomimetic S9E mutant had

no effect. These results suggest that phosphorylation of Syn I at

site 1 plays an important role in the control of synapse formation.

To investigate in detail the time course of this control during

neuronal development, neurons infected at 3 DIV were followed

and analyzed at 7, 10 and 13 DIV by counting separately

glutamatergic and GABAergic synaptic boutons, as described

above (Fig. 6A). The time courses show that the effects of Syn I

mutants were very similar in both glutamatergic and GABAergic

synapse populations (Fig. 6B). At the beginning of

synaptogenesis (7 DIV), only the pseudophosphorylated mutant

caused a significant increase in synapse number with respect to

uninfected or VAMP2-infected neurons (+20.8¡5.4% in

glutamatergic synapses and +13.5¡3.4% in GABAergic

synapses), whereas overexpression of either WT Syn I or Syn I

S9A was ineffective (Fig. 6B). By contrast, at later times (10 and

13 DIV), Syn I S9A was seen to be exerting its inhibitory

influence and caused a significant decrease in the number of both

glutamatergic and GABAergic synapses with respect to

uninfected or VAMP2-infected neurons (214.5¡2.0% in

glutamatergic and 215.5¡5.1% in GABAergic synapses at 10

DIV; 224.6¡5.1% in glutamatergic and 212.6¡1.3% in

GABAergic synapses at 13 DIV). At these developmental

times, overexpression of either WT Syn I or Syn I S9E was

ineffective (Fig. 6B).

Overexpression of WT or pseudophosphorylated Syn I

rescues the early impairment of synaptogenesis caused by

blockade of synaptic activity

Because Syn I has a role in the transduction of extracellular

signals into regulation of neurotransmitter release, we next

examined whether the effects of Syn I in the regulation of

synaptogenesis require synaptic activity. We chronically treated

rat hippocampal neuronal cultures with tetrodotoxin (TTX) to

prevent action potential-evoked synaptic activity, starting from 4

DIV. Because the effects of Syn I mutants on the glutamatergic

and GABAergic synapse populations had very similar trends, we

analyzed these effects on the whole synapse population using the

colocalisation between Syt I (a common marker of excitatory and

inhibitory synapses) and Bsn to identify synaptic contacts, as

described above (Fig. 7A,C).

Chronic exposure to TTX led to an almost complete blockade

of synaptic activity (supplementary material Fig. S3) and

significantly impaired synapse formation during the early

stages of synaptogenesis (218.7¡5.7% and 221.4¡2.6% at 7

and 10 DIV, respectively). However, this inhibition was

spontaneously overcome at later stages of development (13

DIV), indicating that the influence of synaptic activity on

synaptogenesis at early stages is spontaneously rescued at later

time points (Fig. 7A,B). To test whether the delay in

synaptogenesis after TTX block involves Syn I, we

overexpressed Syn I and its phosphorylation mutants in TTX-

treated cells. Interestingly, at 7 DIV, the TTX-induced

impairment in synapse formation was rescued by

overexpression of either WT Syn I or Syn I S9E, whereas it

was unaffected by overexpression of either VAMP2

(219.1¡2.6) or Syn I S9A (222.6¡8.3%; Fig. 7C,D). Similar

effects were observed at 10 DIV, although the TTX-induced

decrease in synapse number was fully and significantly rescued

only by overexpression of Syn I S9E, but not by overexpression

of VAMP2 (224.28¡3.8%), WT Syn I (210.7¡3.8 %) or Syn I

S9A (220.7¡3.9%; Fig. 7C,D). At 13 DIV, when the inhibitory

effect of activity blockade spontaneously subsided, the

overexpressed proteins had no detectable effects, except for the

Fig. 5. Overexpression of Syn I S9A decreases the

total number of synapses in mature hippocampal

neurons. (A) Fluorescence images of hippocampal

neurons at 13 DIV infected at 3 DIV with lentiviral

constructs encoding the ECFP chimeras of WT Syn I,

Syn I S9A, Syn I S9E or the control SV protein

VAMP2. In the merge panels, the overexpressed

protein is shown in blue, Syt I is shown in green and

Bsn is shown in red. In the colocalisation (Col.)

panels, only terminals that were triple positive for

ECFP–Syn I, Syt I and Bsn are shown in black (for

further details, see Materials and Methods). Scale bar:

10 mm. (B) The number of infected synapses (means

¡ s.e.m. of three independent experiments) counted

on a 30 mm dendrite length proximal to the neuron

cell body. Statistical analysis was performed using the

Kruskal–Wallis and post-hoc Tukey’s tests applied to

the whole data sample. **P,0.01 vs all other groups;

n540 dendrites (total length .3600 mm)/

experimental condition for each of three

independent experiments).

Synapsin I controls synaptogenesis 3647

Journ

alof

Cell

Scie

nce

Syn I S9A mutant, which still had the ability to inhibit synapse

formation (218.6¡0.7%; Fig. 7C,D), similar to what was

observed in untreated neurons at the same developmental stage

(see above).

DiscussionA large body of experimental data suggests that synapsins

modulate neuronal development. Syn I loaded into early

blastomeres of Xenopus embryos increased both the functional

and morphological maturation of developing synaptic contacts

(Lu et al., 1992; Valtorta et al., 1995). Consistently, ectopic

expression of synapsins in non-neuronal cells dramatically

promoted the extension of neurite-like processes and the

formation of synaptic-like varicosities (Han et al., 1991).

Constitutive Syn I knockout neurons had delayed neurite

outgrowth and synapse formation in vitro, although

development had almost caught up at mature stages (Chin et al.,

1995; Ferreira et al., 1998; Rosahl et al., 1995). However,

hippocampal mossy fibre terminals and cerebellar granule cell

terminals from Syn12/2 mice showed presynaptic defects that

could be linked to impaired synapse formation (Takei et al.,

1995).

To better understand the role of synapsins in synaptogenesis,

we characterized the effects of acute knockdown and

overexpression of Syn1 and its site 1 phosphorylation mutants

in rat hippocampal neurons in culture. Cultured embryonic

hippocampal neurons represent an excellent in vitro model to

study synapse formation (Dotti et al., 1988). The whole processes

Fig. 6. Overexpression of pseudophosphorylated Syn I accelerates the early phase of synaptogenesis whereas nonphosphorylatable Syn I prevents the

addition of new synapses in both glutamatergic and GABAergic synapse populations. (A) Representative fluorescence images of glutamatergic synapses

formed on hippocampal neuron dendrites at 7, 10 and 13 DIV after infection at 3 DIV with the lentiviral constructs encoding CFP chimeras of WT Syn I, Syn I S9A,

Syn I S9E or the control SV protein VAMP2. In the merge panel the overexpressed protein is shown in blue, VGLUT-1 is shown in green and Bsn is shown in red. In

the Col. panels, only terminals that were triple positive for ECFP–Syn I, VGLUT-1 and Bsn are shown in black. Scale bar: 10 mm. (B) The number of infected

glutamatergic or GABAergic synapses counted on a 30 mm dendrite length (means ¡ s.e.m. of three or four independent experiments) is shown as a function of the

developmental time (7, 10 and 13 DIV). Total dendrite length for condition analyzed .3600 mm. Statistical analysis was performed using the Kruskal–Wallis and

post-hoc Tukey’s tests applied to the whole data sample. ˚P,0.05 vs uninfected or VAMP2; *P,0.05; **P,0.01 vs all other conditions; n540 dendrites [(total

length .3600 mm)/experimental condition for each of three independent experiments for VGLUT-1; n540 dendrites (total length .4800 mm)/experimental

condition for each of four independent experiments for VGAT].

Journal of Cell Science 124 (21)3648

Journ

alof

Cell

Scie

nce

of differentiation and synaptogenesis have been extensively

studied and well characterized in this system: neurons start to

display synapses at around 5 DIV (Fletcher et al., 1994;

Bonanomi et al., 2008) and synaptogenesis proceeds at least

until 16–19 DIV, when dendritic protrusions are stabilized and

become spines (Ziv and Smith, 1996). However, a study of the

effects of acutely decreased Syn I expression on synapse

formation without affecting the early stages of neurite

outgrowth has never been performed. In this paper, we

investigated the effects of an acute knockdown of Syn1 by

siRNA technology and of the overexpression of distinct Syn I

conformers. In particular, we overexpressed WT Syn I or its

phosphomimetic and dephosphomimetic mutants of Ser9 (site 1),

the major Syn phosphorylation site that has been shown to be

directly involved in the early stages of neuronal development

(Bonanomi et al., 2005; Kao et al., 2002).

Specific knockdown of Syn1 by siRNA performed at the time

of synaptogenesis was associated with a decrease in a large array

of synaptic proteins, including proteins of SVs, as well as

proteins of the presynaptic and postsynaptic membranes. These

effects are only partially consistent with those observed in Syn I

KO mice in which the main impairment was on the levels of SV

Fig. 7. Chronic treatment with TTX and overexpression of Syn I phosphorylation mutants have complex effects in the various stages of synaptogenesis.

(A) Representative fluorescence images of hippocampal neurons at 10 DIV chronically treated with TTX starting at 4 DIV. Syt I is shown in green and Bsn is

shown in red. For further details, see legend to Fig. 2. Scale bar: 10 mm. (B) The number of infected synapses counted on a 30 mm dendrite length (means ¡

s.e.m.) is shown as a function of the developmental time (7, 10 and 13 DIV) in TTX-treated (red trace) and untreated (blue trace) neurons. Kruskal–Wallis and

Tukey’s post-hoc tests were applied on each time point. *P,0.05; *P,0.05 TTX treated vs untreated. (C) Representative fluorescence images of hippocampal

neurons at 10 DIV that had been infected at 3 DIV with the lentiviral constructs encoding for CFP chimeras of VAMP2, WT Syn I, Syn I S9A or Syn I S9E and

chronically treated with TTX. The overexpressed protein is shown in blue, Syt I is shown in green and Bsn is shown in red. For further details, see legend to Fig. 2.

(D) Number of infected synapses counted on a 30 mm dendrite length (means ¡ s.e.m. of three independent experiments). The blue dashed line represents the

mean synapse number for uninfected and untreated neurons, and the red dashed line represents the mean synapse number for uninfected TTX-treated neurons

(values taken from B). WT Syn I and Syn I S9E, but not Syn I S9A, rescued the effect of TTX treatment, whereas Syn I S9A did not affect the TTX effect at early

stages of development and decreased synapse number at later stages when the TTX effect was overcome. Statistical analysis was performed using the Kruskal–

Wallis and post-hoc Tukey’s tests applied to the whole data sample. **P,0.01 vs untreated/uninfected; ˚P,0.05; ˚˚P,0.01 vs TTX-treated uninfected; n540

dendrites (total length .3600 mm)/experimental condition for each of three independent experiments.

Synapsin I controls synaptogenesis 3649

Journ

alof

Cell

Scie

nce

proteins. A specific decrease in the levels of plasma membraneand scaffolding proteins suggests a decreased number of synaptic

contacts, in addition to a decrease in the number of SVs persynapse. This was demonstrated by a parallel decrease in bothglutamatergic and GABAergic synapses and by a decreaseduptake of the styryl dye FM 4-64, which indicates that the

decrease involved actively recycling synapses.

When Syn I was overexpressed by viral infection at earlystages of neuronal development, all Syn I conformers correctly

localised at presynaptic sites, and targeting was not affected bymutagenesis of phosphorylation site 1, consistent with previousobservations (Gitler et al., 2004b; Menegon et al., 2006). This

allowed us to investigate the dependence of synapse formation onthe expression levels and phosphorylation state of Syn I.Overexpression of WT Syn I did not affect glutamatergic orGABAergic synapses between 7 and 13 DIV, suggesting that

under normal conditions the absolute amount of Syn I abovephysiological levels is not a major determinant for synapsedevelopment. By contrast, the state of phosphorylation of Syn I

on site 1 was important for synapse development: theoverexpression of Syn I S9E accelerated synapse formation asobserved at 7 DIV whereas, at later stages, the presence of Syn I

S9A impaired a further increase in the number of synapses. Thiseffect was more pronounced in glutamatergic synapses, but wasalso present in GABAergic synapses.

It is known that phosphorylation of Syn I on site 1 regulates

Syn I–SVs and Syn I–actin interactions (Benfenati et al., 1992;Chieregatti et al., 1996; Fesce et al., 1992; Hosaka et al., 1999;Stefani et al., 1997; Valtorta et al., 1992). Thus, the

pseudophosphorylated mutant should scarcely interact withactin and SVs, similarly to the PKA-phosphorylated WT Syn I.By contrast, the non-phosphorylatable mutant should strongly

bind to both actin and SVs, similar to dephosphorylated WT SynI. Under overexpression conditions, the mutants, present inhigher concentration, should compete with the endogenous

protein. Previous data demonstrated that, when overexpressedin a WT background, the Syn I S9A mutant was not able todiffuse into the axonal shaft after PKA activation, in contrast toWT Syn I (Menegon et al., 2006). Furthermore, neurons

expressing Syn I S9A took up less FM 4-64 and exhibited ahigher synaptic depression than WT-Syn-I-expressing neurons(Menegon et al., 2006). The pseudophosphorylated mutant might

act in the opposite way. Experiments carried out in Aplysia

demonstrated that, in a model of impaired neurotransmitterrelease, the injection of pseudophosphorylated Syn I mutant

induced a significant enhancement of neurotransmitter release,suggesting that, upon a stimulus, the excess ofpseudophosphorylated protein in a WT background favours SV

release (Fiumara et al., 2004). Our data are consistent with thehypothesis that, at precocious stages of neuronal development,the presence of Syn I S9E makes the immature SVs more proneto undergo exocytosis, thus accelerating the exo-endocytotic

activity, release of neurotransmitter and synapse formation and/ormaturation. On the contrary, the presence of the non-phosphorylatable mutant at later stages, by interfering with the

release of SVs, might prevent the formation of new stablesynapses or promote the regression of preformed ones.

Of particular interest is the interaction between the Syn I

phosphorylation mutants and the chronic blockade of electricalactivity during development. The results found in the literatureon the role of electrical activity in synapse formation are quite

controversial, partly because of the different models andmethodologies used to address the question and, more

importantly, to the developmental stages at which the activityis blocked (Bouwman et al., 2004; Lauri et al., 2003; Murthyet al., 2001; van Huizen et al., 1985; Verhage et al., 2000). Thegeneral idea that emerges is that electrical activity acts in

different ways on developing and mature synapses. The datasuggest that synaptogenesis is enhanced by spontaneous activity,whereas activity-dependent selection is thought to maintain the

balance between synaptic assembly and disassembly at laterstages of development and in mature networks. The decreasednumber of synapses that we observed after chronic treatment with

TTX at early stages of development is consistent with this view.At later stages, the blockade of synaptic activity andneurotransmitter release is believed to alter synapsemaintenance, causing either an increase or a decrease in the

number of selected synaptic populations. Thus, it is not surprisingthat we found that the total number of synapses was unaltered byTTX treatment at 13 DIV. Moreover, our data are consistent with

the observation that blockade of spiking activity reduced thedensity of GABAergic terminals in developing hippocampalneurons, but did not cause any change in GABAergic synapse

density later in development (Hartman et al., 2006).

Considering the substantial lack of effect of the overexpressionof Syn I S9A on glutamatergic and GABAergic synapses at 7DIV, the data on TTX-treated neurons at 7 DIV suggest that, at

early stages, the blockade of synaptic activity impairs synapseformation through a mechanism that is independent ofendogenous Syn I. Nonetheless, this impairment could be

rescued by either WT or pseudophosphorylated Syn I. Thisindicates that, at this stage of development, Syn I is probably partof an alternative synaptogenic pathway that is PKA, but not

activity, dependent, suggesting that the two parallel (activity- andPKA-dependent) pathways act additively and that activation ofPKA is not exclusively coupled to electrical activity at this stage.

This is also consistent with the lack of effect of Syn I S9E inincreasing synapse number above control levels under the chroniceffect of TTX. At 10 DIV the TTX chronic treatment decreasedthe number of synapses to a similar level to that observed for Syn

I S9A overexpression in untreated neurons. Overexpression of thepseudophosphorylated mutant, but not of WT Syn I or its non-phosphorylatable mutant, rescued this effect, implying that, at

this developmental stage, the effect of Syn I on synapsedevelopment is related to neuronal activity and that PKAactivation has become activity dependent. At 13 DIV, a stage

in which the TTX chronic treatment had no effect on synapsenumber, the Syn I S9A mutant retained its capacity to impairsynaptogenesis, indicating that at later stages of synaptogenesis

Syn I acts through mechanisms that are not linked to neuronalactivity.

In conclusion, the evidence that synapsins exhibit actin-binding and SV-clustering activities in vitro (Bahler and

Greengard, 1987; Valtorta et al., 1992) led to the hypothesisthat these activities underlie neurodevelopmental processes.However, results obtained from the constitutive Syn knockout

indicate that synapsins are more modulators rather thandeterminants of synapse formation. An obvious mechanism forregulation of synaptogenesis could be the switch between a

positive and negative form of the protein, which could occurthrough post-translational modifications such as phosphorylation.A similar mechanism has been demonstrated for Syn II during

Journal of Cell Science 124 (21)3650

Journ

alof

Cell

Scie

nce

morphological development of neurons because the

dephosphomimetic mutant inhibits, whereas its phosphomimetic

mutant increases neurite outgrowth (Kao et al., 2002). Our data

support the idea that in addition to a basal amount of Syn being

required for synapse formation, the pattern of active Syn

conformers could be a key factor in the regulation of

synaptogenesis. However, because Syn exhibits complex

molecular interactions (Cesca et al., 2010) and not all its

functional properties are phosphorylation dependent (see

Bonanomi et al., 2005), phosphorylation might have a positive

or a negative role depending on the interactions involved. Thus,

dephosphorylation mutant and phosphomimetic forms of the

protein do not necessarily exhibit opposite effects under all

circumstances.

Electrical activity activates neurotransmitter release and

several signal transduction pathways, including those triggered

by cAMP and Ca2+. In particular, a large body of experimental

data suggests that PKA activity and neurotransmitter-induced

signalling are correlated and strongly involved in synaptogenesis.

In fact, the repeated activation of PKA by either a cAMP

analogue or forskolin in primary rat hippocampal neurons

(Yamamoto et al., 2005) or by glutamate in hippocampal slices

(Tominaga-Yoshino et al., 2008) leads to a slowly developing

long-lasting synaptogenesis. However, the expression of

dominant-negative PKA in zebrafish olfactory sensory neurons

in vivo prevents the formation of SV clusters in axon terminals

during synaptogenesis (Yoshida and Mishina, 2005). The data

available so far indicate that the phosphorylation of synapsins by

PKA plays a key role in the regulation of neurite outgrowth (Kao

et al., 2002), SV dynamics in the growth cone (Bonanomi et al.,

2005) and formation of excitatory and inhibitory synapses (this

paper). These results further confirm the central role in neural

development of the activation of cAMP-dependent protein kinase

(Funte and Haydon, 1993) and the integration between the

cAMP- and Ca2+/calmodulin-linked signal tranduction pathways

occurring at phosphorylation site 1 of synapsins.

Materials and MethodsMaterials

The mouse monoclonal antibody anti-Syn I/II (clone 19.11) and the polyclonal

antibody against Syp were made in-house. The rabbit antiserum VGLUT1, VGAT,

Rab5 and the monoclonal antibodies against syntaxin 1, PSD95, Rab3a and the Syt

I luminal epitope were purchased from Synaptic Systems (Goettingen, Germany).

The monoclonal antibody against the bassoon (Bsn) epitope was from StressgenBioreagents (Victoria, BC, USA). The monoclonal antibody against tubulin bIII

(TuJI) was from Covance (Dedham, MA, USA) and the polyclonal antibody

against GFP was from Invitrogen (Carlsbad, CA, USA). Tetramethyl Rhodamine

iso-thiocyanate (TRITC)- and fluorescein isothiocyanate (FITC)-conjugated

secondary antibodies were from Jackson ImmunoResearch. Hoechst 33342 was

from Sigma-Aldrich (Saint Louis, MI, USA). FM 4-64 was from Sigma-Aldrich

and citrate-free TTX was from Latoxan (Valence, France). All other reagents were

purchased from standard commercial suppliers. Sprague Dawley rats from Charles

River (Calco, Italy) were housed (two to four per cage) under constant temperature(22¡1˚C) and humidity (50%) conditions with a 12 hour light–dark cycle, and

were provided with food and water ad libitum. All experiments were carried out in

accordance with the guidelines established by the European Communities Council

(Directive of November 24th, 1986) and the National Council on Animal Care of

the Ministry of Health.

siRNA constructs

siRNAs (21-nucleotide) specific for Syn1 were generated using the NCBI free

software Sfold (software for statistical folding and rational design of nucleic acids)

and scored on the basis of the criteria listed by Reynolds and co-workers (Reynolds

et al., 2004). The following siRNA was selected: 59-UGUCUAACGCG-

GUCAAACATT-39. The selected siRNA sequence was also submitted to aBLAST search against the rat genome sequence to ensure that only one gene of the

rat genome was targeted. To minimize the possibility of nonspecific off-targeteffects, siRNAs were applied at low concentration (20 nM).

Lentiviral constructs and virus production

Lentiviral vectors were used to induce the Syn1 transgene expression in neurons(Jakobsson and Lundberg, 2006). Sequences encoding the transgenes were insertedinto a lentiviral vector derived from pRRL-Sin18, a human immunodeficiencyvirus (HIV)-based lentiviral self-inactivating vector that is able to transduceneurons at very high efficiency (Dull et al., 1998). The ‘a’ isoform of Syn I wastagged with ECFP, as described previously (Bonanomi et al., 2005; Chi et al.,2003; Gitler et al., 2004a). The lentiviral constructs pRRL-ECFP-Syn Ia(pgk) andpRRL-ECFP-Syn I S9A(pgk) were previously described (Bonanomi et al., 2005).pRRL-ECFP-VAMP2(pgk) was generated from the plasmid pECFP-VAMP2 aspreviously described (Pennuto et al., 2002). The sequence encoding the chimericprotein ECFP–VAMP2 was amplified by PCR and AgeI and SalI restriction siteswere introduced. The resultant AgeI–SalI PCR fragment was ligated intoRRL.ppt.hPGK.GFP.pre.sin-18, plasmid digested with the same enzymes. ThepRRL-ECFP-Syn Ia S9E(pgk) was generated from the pEGFP-Syn Ia S9Econstruct, which was a kind gift from Huang-Teh Kao (Brown University, RhodeIsland, USA). The S9E mutation was produced through PCR mutagenesis using theQuickChange Mutagenesis kit (Stratagene-Medical, Milano, Italy) at TheRockefeller University, New York, USA (Gitler et al., 2004b). First the EGFPwas swapped for ECFP cutting ECFP coding sequence with AgeI–BsrGI andligating it in the Syn-I-S9E-containing vector cut with the same restrictionenzymes. The resulting plasmid was then cut with the AgeI and SalI restrictionenzymes, and the obtained fragment was ligated into RRL.ppt.hPGK.GFP.pre.sin-18 plasmid digested with the same enzymes. The lentiviral construct sequenceswere verified by automated sequencing at the PRIMM Facility (Milano, Italy).Viral stocks were prepared as previously described (Lotti et al., 2002) by transientco-transfection of 293T cells with the following vectors: the pCMV deleted R8.74plasmid encoding Gag, Pol, Tat and Rev; the pMD.G plasmid encoding thevesicular stomatitis virus G-protein (VSV-G) for pseudotyping, and the transfervector. Viral titres were increased by ultracentrifugation. Because ECFPfluorescence was not detectable by flow cytometry, viruses were used totransduce 293T cells that were subsequently lysed, loaded on an SDS-PAGE geland blotted on a nitrocellulose membrane. The membrane was hybridized withan anti-GFP antibody and the signal was revealed through enhancedchemiluminescence (ECL; GE Healthcare, Chalfont St Giles, Buckinghamshire,UK). The signal was quantified and the ratio between the efficiencies of infectionof the viruses was calculated. The amount of lentiviral vectors used to transduceneurons was chosen in order to obtain a 1:1 ratio between the vectors and a 100%efficiency of infection (supplementary material Fig. S2).

Cell culture procedures

HeLa cell cultureHeLa cells were grown in Dulbecco’s modified Eagle’s medium (DMEM)supplemented with 10% foetal bovine serum (FBS), penicillin and streptomycinand were regularly passaged to maintain exponential growth.

Astroglial cell culturesAstroglial cultures were prepared from brains of Sprague Dawley rats (CharlesRiver) at postnatal day 2 as previously described (Kaech and Banker, 2006).Briefly, rat pups were sacrificed by decapitation, brains were extracted from theskull, and then cerebellum, meninges and all subcortical structures were removedto leave only the cerebral hemispheres. The tissue was chopped with a blade intosmall pieces, washed with Hanks’ balanced salt solution (HBSS; GibcoHInvitrogen) to remove the smallest debris, then trypsinised in a solutioncontaining 2.5% trypsin (Sigma-Aldrich) and 1 mg/ml DNAse (Sigma-Aldrich)for 15 minutes at 37˚C. After the incubation cells were washed three times withHBSS and then dissociated. The cell suspension was let stand for 2 minutes inorder to settle undissociated piece of tissues by gravity. Then the cells contained inthe supernatant were plated in glial medium [modified Eagle’s medium; MEM(GibcoH Invitrogen) containing 10% horse serum (GibcoH Invitrogen), 0.6%glucose (Sigma-Aldrich), 2 mM glutamine (Lonza, Basel, Switzerland), 100 IU/mlpenicillin and 100 mg/ml streptomycin (Lonza)] and maintained at 37˚C in a 5%CO2 humidified atmosphere. On the day before preparing the hippocampal culture,the glial medium was removed and replaced with hippocampal medium [MEMsupplemented with 1% N2 supplement (Invitrogen), 2 mM glutamine, 1 mMsodium pyruvate (Sigma-Aldrich) and 4 mM glucose].

E18 hippocampal neuronal culturesPrimary neuronal cultures were prepared from the hippocampi of embryonic day18 embryos from Sprague Dawley rats as previously described (Banker andCowan, 1977). The pregnant rat was killed by CO2 asphyxiation and the uterus wasremoved. Foetuses were extracted and decapitated. Skulls were opened and brainswere dissected out and placed into HBSS. Hippocampi were removed under adissecting microscope, and collected. After 15 minutes of incubation with 0.25%

Synapsin I controls synaptogenesis 3651

Journ

alof

Cell

Scie

nce

trypsin in HBSS at 37˚C, the whole hippocampi were washed three times withHBSS to remove trypsin and then mechanically dissociated. Neurons stained withvital dye (Trypan Blue; Sigma-Aldrich) were counted using a Burker chamber.Neurons were plated on poly-L-lysine (Sigma-Aldrich; 0.1 mg/ml)-treated glasscoverslips in order to obtain 100,000 cells/coverslip (low density culture). Cellswere plated in plating medium [MEM supplemented with 10% horse serum,3.3 mM glucose and 2 mM glutamine] and incubated for 4 hours at 37˚C in a 5%CO2 humidified atmosphere to allow adhesion to the substrate. Given that thedensity of neurons in culture influences synapse number (Fletcher et al., 1994), wealways plated the same number of neurons (100,000 neurons on 24 mm coverslip).After plating, coverslips were transferred into a cell-culture dish covered with aglia monolayer prepared as described above containing hippocampal mediumconditioned for at least 24 hours. Coverslips were turned upside down, withneurons facing down, toward the glia.

Transfection and transduction procedures

HeLa cell transfectionThe day before transfection, cells were trypsinised, suspended in fresh mediumwithout antibiotics and plated in 35-mm Petri dishes. Transfection was carried outwith Lipofectamine 2000 (Invitrogen) according to manufacturer’s instructionsusing 2 mg ECFPSynIa or ECFPSynIIa with 20 nM siRNA per 35-mm dish. Cellswere lysed 48 hours after transfection.

Neuron transfectionTransfections were performed with Lipofectamine 2000 (Invitrogen) according tomanufacturer’s instructions using 20 nM siRNA oligonucleotides. Neurons at 14DIV were placed in a clean dish and incubated for 1 hour with hippocampalmedium containing Lipofectamine and siRNA, then returned to the original dishesand analyzed at 21 DIV.

Neuron transductionInfection was carried out on 3 DIV neurons as previously described (Bonanomiet al., 2005). Coverslips were placed in clean dishes containing glia-conditionedhippocampal medium in the presence of viral supernatant for 12–15 hours at 37˚Cin a 5% CO2 humidified atmosphere. After transduction, neurons were returned tothe original dishes and maintained in culture in glia-conditioned medium untilprocessing.

Pharmacological treatmentTTX was resuspended in 10 mM sodium acetate, pH 4.5, and used at a finalconcentration of 1 mM. The medium was changed every 3 days and fresh drug wasadded.

Cell labelling protocols

Standard immunofluorescence experiments were carried out as previouslydescribed (Pennuto et al., 2003). Briefly, cells were rinsed once with Krebs–Ringer’s solution (KRH)–EGTA (130 mM NaCl, 5 mM KCl, 1.2 mM KH2PO4,1.2 mM MgSO4, 2 mM MgCl2, 2 mM EGTA, 25 mM Hepes, 6 mM glucose,pH 7.4), fixed for 20 minutes with 4% paraformaldehyde, 4% sucrose in 120 mMsodium phosphate buffer, pH 7.4, supplemented with 2 mM EGTA. Coverslipswere rinsed twice with phosphate-buffered saline (PBS), once with high salt buffer(HS; 500 mM NaCl, 20 mM sodium phosphate buffer, pH 7.4), and then incubatedovernight at 4˚C into a humidified chamber with the primary monoclonal antibodyappropriately diluted in goat serum dilution buffer (GSDB; 15% goat serum,450 mM NaCl, 0.3% Triton X-100, 20 mM sodium phosphate buffer, pH 7.4).Incubation with the polyclonal primary antibodies was carried out for 2 hours atroom temperature. Cells were washed twice within 20 minutes with HS and thenincubated with the appropriate secondary antibodies for 90 minutes at roomtemperature. After three washes with HS for 30 minutes each and one 10 minutewash with PBS, coverslips were mounted with 70% glycerol in PBS supplementedwith phenylenediamine (1 mg/ml) as an anti-bleaching agent. All the reagents usedwere from Sigma-Aldrich.

For FM 4-64 uptake experiments, 21 DIV hippocampal neurons were stimulatedfor 1 minute with 55 mM KCl in KRH, in the presence of FM 4-64 (10 mM) and 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX; 10 mM) to block recurrent activity.After FM 4-64 loading, neurons were washed by three complete mediumsubstitutions followed by perfusion for 10 minutes with warmed KRH (37˚C)supplemented with TTX (1 mM) and CNQX (10 mM).

Image acquisition, processing and statistical analysis

Specimens were viewed with an Axiovert 135 inverted microscope (Zeiss,Oberkochen, Germany) equipped with epifluorescence optics. Images wererecorded with a C4742-98 ORCA II cooled charge-coupled device camera(Hamamatsu Photonics, Hamamatsu City, Japan) and processed using Image ProPlus 4.5 (Media Cybernetics, Silver Spring, MD) and Adobe Photoshop 6.0(Adobe System, San Jose, CA). To obtain RGB images we acquired a single greyimage for every colour channel, then the three grey images were merged and

processed using Adobe Photoshop CS (Adobe Systems). To better visualizecolocalised puncta, after excluding the cell body, we adjusted the fluorescencelevel by applying the auto-level function, and then inverted the colours. The nerveterminal targeting of the exogenously expressed synapsins was determined bycalculating the ‘targeting factor’, i.e. the ratio between the total number of Syn-I-positive puncta and the number of Syt-I-, Syn-I-double-positive puncta. In theliterature, synapses in culture are often defined simply as discrete puncta of singlepresynaptic proteins (Chin et al., 1995; Ferreira et al., 1998; Paganoni et al., 2010).However, to avoid counting structures other than synapses (e.g. transport packets),we evaluated the simultaneous presence of a scaffolding component (Bsn) with SVproteins. Bsn and SVs are transported by two distinct families of transport vesiclesthat probably utilize different molecular motors and/or adaptors (Zhai et al., 2001;McAllister, 2007). To test the validity of our method, we counted the singlepositive Syt, Bas and overexpressed protein puncta of a sample experiment(supplementary material Fig. S4). Indeed, no significant difference between thenumber of puncta labelled for either Syt or overexpressed protein (either VAMP2,WT or mutant synapsin) was detected. By contrast, there were significantly moreBsn-positive puncta under all experimental conditions, suggesting that structuresother than synapses were labelled.

In inverted images we counted the number of blue (double colocalisation) orblack (triple colocalisation) puncta present within a 30 mm dendrite tract startingfrom the cell body. Measuring a longer tract of dendrite was troublesome, becauseoften dendrites do not follow linear routes. During the analysis we considered onlypuncta characterized by a signal to noise (S/N) ratio greater than 3 and an area inthe range of 0.3–0.8 mm2. We measured at least 20 neurons for each sample andselected at least two dendrites for each neuron.

To analyze FM4-64 uptake at synaptic boutons, we analyzed three separatecoverslips of cultured neurons and acquired 10 images per condition in eachexperiment. Using the Gran Filter Plugin of NIH ImageJ software, we created amask to select synapses with an area of 0.3–0.8 mm2 and we measured FM 4-64fluorescence intensity in each terminal. Statistical analysis was performed usingthe Kruskal–Wallis one-way analysis of variance on ranks followed by Tukey’spost-hoc test.

Miscellaneous procedures

Protein concentrations were determined by the Bradford or BCA assays (ThermoScientific, Rockford, IL, USA). SDS-PAGE was performed according to Laemmli(Laemmli, 1970). Immunoblotting was performed using peroxidase-conjugatedsecondary antibodies coupled with the ECL chemiluminescence detection system.Statistical analysis of data obtained from western blot quantification wasperformed by two-way analysis of variance or Kruskal–Wallis test, followed bythe Tukey’s post-hoc test.

AcknowledgementsWe thank Elena Monzani for technical support.

FundingThis study was supported by research grants from the Italian Ministryof University and Research [PRIN grant number 2008T4ZCNL toF.B. and F.V.]; the Compagnia di San Paolo, Torino [grant number2005-1964 to F.B. and 2008-2207 to F.V.]; and Telethon-Italy [grantnumber GGP09134 to F.B. and F.V.].

Supplementary material available online at

http://jcs.biologists.org/lookup/suppl/doi:10.1242/jcs.086223/-/DC1

ReferencesBahler, M. and Greengard, P. (1987). Synapsin I bundles F-actin in a phosphorylation-

dependent manner. Nature 326, 704-707.

Banker, G. A. and Cowan, W. M. (1977). Rat hippocampal neurons in dispersed cellculture. Brain Res. 126, 397-442.

Benfenati, F., Valtorta, F., Chieregatti, E. and Greengard, P. (1992). Interaction offree and synaptic vesicle-bound synapsin I with F-actin. Neuron 8, 377-386.

Betz, W. J., Mao, F. and Smith, C. B. (1996). Imaging exocytosis and endocytosis.Curr. Opin. Neurobiol. 6, 365-371.

Bonanomi, D., Menegon, A., Miccio, A., Ferrari, G., Corradi, A., Kao, H. T.,Benfenati, F. and Valtorta, F. (2005). Phosphorylation of synapsin I by cAMP-dependent protein kinase controls synaptic vesicle dynamics in developing neurons. J.

Neurosci. 25, 7299-7308.

Bonanomi, D., Fornasiero, E. F., Valdez, G., Halegoua, S., Benfenati, F., Menegon,

A. and Valtorta, F. (2008). Identification of a developmentally regulated pathway ofmembrane retrieval in neuronal growth cones. J. Cell Sci. 121, 3757-3769.

Bouwman, J., Maia, A. S., Camoletto, P. G., Posthuma, G., Roubos, E. W.,

Oorschot, V. M., Klumperman, J. and Verhage, M. (2004). Quantification ofsynapse formation and maintenance in vivo in the absence of synaptic release.Neuroscience 126, 115-126.

Journal of Cell Science 124 (21)3652

Journ

alof

Cell

Scie

nce

Candiani, S., Moronti, L., Pennati, R., De Bernardi, F., Benfenati, F. and Pestarino,M. (2010) The synapsin gene family in basal chordates: evolutionary perspectives inmetazoans. BMC Evol. Biol. 10, 32.

Ceccaldi, P. E., Grohovaz, F., Benfenati, F., Chieregatti, E., Greengard, P. and

Valtorta, F. (1995). Dephosphorylated synapsin I anchors synaptic vesicles to actincytoskeleton: an analysis by videomicroscopy. J. Cell Biol. 128, 905-912.

Cesca, F., Baldelli, P., Valtorta, F. and Benfenati, F. (2010). The synapsins: Keyactors of synapse function and plasticity. Prog Neurobiol 91, 313-348.

Chi, P., Greengard, P. and Ryan, T. A. (2003). Synaptic vesicle mobilization isregulated by distinct synapsin I phosphorylation pathways at different frequencies.Neuron 38, 69-78.

Chieregatti, E., Ceccaldi, P. E., Benfenati, F. and Valtorta, F. (1996). Effects ofsynaptic vesicles on actin polymerization. FEBS Lett. 398, 211-216.

Chin, L. S., Li, L., Ferreira, A., Kosik, K. S. and Greengard, P. (1995). Impairmentof axonal development and of synaptogenesis in hippocampal neurons of synapsin I-deficient mice. Proc. Natl. Acad. Sci. USA 92, 9230-9234.

De Camilli, P., Benfenati, F., Valtorta, F. and Greengard, P. (1990). The synapsins.Annu. Rev. Cell Biol. 6, 433-460.

Dotti, C. G., Sullivan, C. A. and Banker, G. A. (1988). The establishment of polarityby hippocampal neurons in culture. J. Neurosci. 8, 1454-1468.

Dull, T., Zufferey, R., Kelly, M., Mandel, R. J., Nguyen, M., Trono, D. and Naldini,

L. (1998). A third-generation lentivirus vector with a conditional packaging system. J.

Virol. 72, 8463-8471.Ferreira, A., Kosik, K. S., Greengard, P. and Han, H. Q. (1994). Aberrant neurites

and synaptic vesicle protein deficiency in synapsin II-depleted neurons. Science 264,977-979.

Ferreira, A., Chin, L. S., Li, L., Lanier, L. M., Kosik, K. S. and Greengard, P.(1998). Distinct roles of synapsin I and synapsin II during neuronal development.Mol. Med. 4, 22-28.

Fesce, R., Benfenati, F., Greengard, P. and Valtorta, F. (1992). Effects of theneuronal phosphoprotein synapsin I on actin polymerization. II. Analyticalinterpretation of kinetic curves. J. Biol. Chem. 267, 11289-11299.

Fiumara, F., Giovedi, S., Menegon, A., Milanese, C., Merlo, D., Montarolo, P. G.,

Valtorta, F., Benfenati, F. and Ghirardi, M. (2004). Phosphorylation by cAMP-dependent protein kinase is essential for synapsin-induced enhancement ofneurotransmitter release in invertebrate neurons. J. Cell Sci. 117, 5145-5154.

Fletcher, T. L., De Camilli, P. and Banker, G. (1994). Synaptogenesis in hippocampalcultures: evidence indicating that axons and dendrites become competent to formsynapses at different stages of neuronal development. J. Neurosci. 14, 6695-6706.

Fornasiero, E. F., Bonanomi, D., Benfenati, F. and Valtorta, F. (2010). The role ofsynapsins in neuronal development. Cell Mol. Life Sci. 67, 1383-1396.

Funte, L. R. and Haydon, P. G. (1993). Synaptic target contact enhances presynapticcalcium influx by activating cAMP-dependent protein kinase during synaptogenesis.Neuron 10, 1069-1078.

Gitler, D., Takagishi, Y., Feng, J., Ren, Y., Rodriguiz, R. M., Wetsel, W. C.,

Greengard, P. and Augustine, G. J. (2004a). Different presynaptic roles ofsynapsins at excitatory and inhibitory synapses. J. Neurosci. 24, 11368-11380.

Gitler, D., Xu, Y., Kao, H. T., Lin, D., Lim, S., Feng, J., Greengard, P. and

Augustine, G. J. (2004b). Molecular determinants of synapsin targeting topresynaptic terminals. J. Neurosci. 24, 3711-3720.

Han, H. Q., Nichols, R. A., Rubin, M. R., Bahler, M. and Greengard, P. (1991).Induction of formation of presynaptic terminals in neuroblastoma cells by synapsinIIb. Nature 349, 697-700.

Hartman, K. N., Pal, S. K., Burrone, J. and Murthy, V. N. (2006). Activity-dependentregulation of inhibitory synaptic transmission in hippocampal neurons. Nat. Neurosci.

9, 642-649.Hosaka, M., Hammer, R. E. and Sudhof, T. C. (1999). A phospho-switch controls the

dynamic association of synapsins with synaptic vesicles. Neuron 24, 377-387.Jakobsson, J. and Lundberg, C. (2006). Lentiviral vectors for use in the central

nervous system. Mol. Ther. 13, 484-493.Kaech, S. and Banker, G. (2006). Culturing hippocampal neurons. Nat. Protoc. 1,

2406-2415.Kao, H. T., Porton, B., Czernik, A. J., Feng, J., Yiu, G., Haring, M., Benfenati, F.

and Greengard, P. (1998). A third member of the synapsin gene family. Proc. Natl.

Acad. Sci. USA 95, 4667-4672.Kao, H. T., Porton, B., Hilfiker, S., Stefani, G., Pieribone, V. A., DeSalle, R. and

Greengard, P. (1999). Molecular evolution of the synapsin gene family. J. Exp. Zool.

285, 360-377.Kao, H. T., Song, H. J., Porton, B., Ming, G. L., Hoh, J., Abraham, M., Czernik, A.

J., Pieribone, V. A., Poo, M. M. and Greengard, P. (2002). A protein kinase A-dependent molecular switch in synapsins regulates neurite outgrowth. Nat. Neurosci.

5, 431-437.Laemmli, U. K. (1970). Cleavage of structural proteins during the assembly of the head

of bacteriophage T4. Nature 227, 680-685.

Lauri, S. E., Lamsa, K., Pavlov, I., Riekki, R., Johnson, B. E., Molnar, E., Rauvala,H. and Taira, T. (2003). Activity blockade increases the number of functionalsynapses in the hippocampus of newborn rats. Mol. Cell Neurosci. 22, 107-117.

Lotti, F., Menguzzato, E., Rossi, C., Naldini, L., Ailles, L., Mavilio, F. and Ferrari,

G. (2002). Transcriptional targeting of lentiviral vectors by long terminal repeatenhancer replacement. J. Virol. 76, 3996-4007.

Lu, B., Greengard, P. and Poo, M. M. (1992). Exogenous synapsin I promotesfunctional maturation of developing neuromuscular synapses. Neuron 8, 521-529.

Matteoli, M., Takei, K., Perin, M. S., Sudhof, T. C. and De Camilli, P. (1992). Exo-endocytotic recycling of synaptic vesicles in developing processes of culturedhippocampal neurons. J. Cell Biol. 117, 849-861.

Menegon, A., Bonanomi, D., Albertinazzi, C., Lotti, F., Ferrari, G., Kao, H. T.,

Benfenati, F., Baldelli, P. and Valtorta, F. (2006). Protein kinase A-mediatedsynapsin I phosphorylation is a central modulator of Ca2+-dependent synapticactivity. J. Neurosci. 26, 11670-11681.

McAllister, A. K. (2007). Dynamic aspects of CNS synapse formation. Annu. Rev.

Neurosci. 30, 425-450.

Murthy, V. N., Schikorski, T., Stevens, C. F. and Zhu, Y. (2001). Inactivity producesincreases in neurotransmitter release and synapse size. Neuron 32, 673-682.

Paganoni, S., Bernstein, J. and Ferreira, A. (2010). Ror1-Ror2 complexes modulatesynapse formation in hippocampal neurons. Neuroscience 165, 1261-1274.

Pennuto, M., Dunlap, D., Contestabile, A., Benfenati, F. and Valtorta, F. (2002).Fluorescence resonance energy transfer detection of synaptophysin I and vesicle-associated membrane protein 2 interactions during exocytosis from single livesynapses. Mol. Biol. Cell 13, 2706-2717.

Pennuto, M., Bonanomi, D., Benfenati, F. and Valtorta, F. (2003). Synaptophysin Icontrols the targeting of VAMP2/synaptobrevin II to synaptic vesicles. Mol. Biol. Cell

14, 4909-4919.

Rosahl, T. W., Spillane, D., Missler, M., Herz, J., Selig, D. K., Wolff, J. R., Hammer,R. E., Malenka, R. C. and Sudhof, T. C. (1995). Essential functions of synapsins Iand II in synaptic vesicle regulation. Nature 375, 488-493.

Schaeffer, E., Alder, J., Greengard, P. and Poo, M. M. (1994). Synapsin IIaaccelerates functional development of neuromuscular synapses. Proc. Natl. Acad. Sci.

USA 91, 3882-3886.

Stefani, G., Onofri, F., Valtorta, F., Vaccaro, P., Greengard, P. and Benfenati, F.

(1997). Kinetic analysis of the phosphorylation-dependent interactions of synapsin Iwith rat brain synaptic vesicles. J. Physiol. 504, 501-515.

Sudhof, T. C., Czernik, A. J., Kao, H. T., Takei, K., Johnston, P. A., Horiuchi, A.,

Kanazir, S. D., Wagner, M. A., Perin, M. S., De Camilli, P. et al. (1989).Synapsins: mosaics of shared and individual domains in a family of synaptic vesiclephosphoproteins. Science 245, 1474-1480.

Takei, Y., Harada, A., Takeda, S., Kobayashi, K., Terada, S., Noda, T., Takahashi,

T. and Hirokawa, N. (1995). Synapsin I deficiency results in the structural change inthe presynaptic terminals in the murine nervous system. J. Cell Biol. 131, 1789-1800.

Tominaga-Yoshino, K., Urakubo, T., Okada, M., Matsuda, H. and Ogura, A. (2008).Repetitive induction of late-phase LTP produces long-lasting synaptic enhancementaccompanied by synaptogenesis in cultured hippocampal slices. Hippocampus 18,281-293.

Valtorta, F., Greengard, P., Fesce, R., Chieregatti, E. and Benfenati, F. (1992).Effects of the neuronal phosphoprotein synapsin I on actin polymerization. I.Evidence for a phosphorylation-dependent nucleating effect. J. Biol. Chem. 267,11281-11288.

Valtorta, F., Iezzi, N., Benfenati, F., Lu, B., Poo, M. M. and Greengard, P. (1995).Accelerated structural maturation induced by synapsin I at developing neuromuscularsynapses of Xenopus laevis. Eur. J. Neurosci. 7, 261-270.

van Huizen, F., Romijn, H. J. and Habets, A. M. (1985). Synaptogenesis in ratcerebral cortex cultures is affected during chronic blockade of spontaneous bioelectricactivity by tetrodotoxin. Brain Res. 351, 67-80.

Verhage, M., Maia, A. S., Plomp, J. J., Brussaard, A. B., Heeroma, J. H., Vermeer,

H., Toonen, R. F., Hammer, R. E., van den Berg, T. K., Missler, M. et al. (2000).Synaptic assembly of the brain in the absence of neurotransmitter secretion. Science

287, 864-869.

Yamamoto, M., Urakubo, T., Tominaga-Yoshino, K. and Ogura, A. (2005). Long-lasting synapse formation in cultured rat hippocampal neurons after repeated PKAactivation. Brain Res. 1042, 6-16.

Yoshida, T. and Mishina, M. (2005). Distinct roles of calcineurin-nuclear factor ofactivated T-cells and protein kinase A-cAMP response element-binding proteinsignaling in presynaptic differentiation. J. Neurosci. 25, 3067-3079.

Zhai, R. G., Vardinon-Friedman, H., Cases-Langhoff, C., Becker, B., Gundelfinger,

E. D., Ziv, N. E. and Garner, C. C. (2001). Assembling the presynaptic active zone:a characterization of an active one precursor vesicle. Neuron 29, 131-143.

Ziv, N. E. and Smith, S. J. (1996). Evidence for a role of dendritic filopodia insynaptogenesis and spine formation. Neuron 17, 91-102.

Synapsin I controls synaptogenesis 3653

Journ

alof

Cell

Scie

nce