Compartmentalized versus Global Synaptic Plasticity on Dendrites Controlled by Experience

Neuron

Article Compartmentalized versus Global Synaptic Plasticity on Dendrites Controlled by Experience Hiroshi Makino1,2 and Roberto Malinow1,2,* 1Center for Neural Circuits and Behavior, Section of Neurobiology, Division of Biology and Department of Neuroscience, University of California, San Diego, La Jolla, CA 92093, USA 2Watson School of Biological Sciences, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY 11724, USA *Correspondence: [email protected] DOI 10.1016/j.neuron.2011.09.036

SUMMARY

Synapses in the brain are continuously modified by experience, but the mechanisms are poorly understood. In vitro and theoretical studies suggest threshold-lowering interactions between nearby synapses that favor clustering of synaptic plasticity within a dendritic branch. Here, a fluorescently tagged AMPA receptor-based optical approach was developed permitting detection of single-synapse plasticity in mouse cortex. Sensory experience preferentially produced synaptic potentiation onto nearby dendritic synapses. Such clustering was significantly reduced by expression of a phosphomutant AMPA receptor that is insensitive to threshold-lowering modulation for plasticity-driven synaptic incorporation. In contrast to experience, sensory deprivation caused homeostatic synaptic enhancement globally on dendrites. Clustered synaptic potentiation produced by experience could bind behaviorally relevant information onto dendritic subcompartments; global synaptic upscaling by deprivation could equally sensitize all dendritic regions for future synaptic input.

INTRODUCTION Cortical circuits display fine functional and structural organization (Feldmeyer et al., 2002; Lefort et al., 2009; Petreanu et al., 2009) that is carefully established and tuned by sensory experience (Bender et al., 2003; Buonomano and Merzenich, 1998; Feldman and Brecht, 2005; Stern et al., 2001). Modification of synapses includes Hebbian plasticity mechanisms where correlated (or uncorrelated) activity leads to structural as well as functional alternations, such as changes in spine morphology (Alvarez and Sabatini, 2007), or synaptic insertion or removal of AMPA receptors (Kessels and Malinow, 2009; Malenka and Bear, 2004; Newpher and Ehlers, 2008; Nicoll et al., 2006). In parallel to such Hebbian mechanisms, neurons are also equipped with homeostatic-scaling machinery that may serve to avoid instability problems of network activity (Turrigiano and Nelson, 2004). Such scaling can globally regulate synaptic strength by

altering the number of AMPA receptors in individual synapses (Turrigiano et al., 1998). Although a number of molecular and cellular mechanisms underlying these plasticity mechanisms have been identified, how synapses on a dendritic branch cooperate with each other to drive such plasticity is not well understood. Accumulating in vitro and theoretical evidence suggests that there exists biochemical compartmentalization on dendrites that leads to clustered synaptic plasticity (Branco and Ha¨usser, 2010; Govindarajan et al., 2006; Ha¨usser and Mel, 2003; Iannella and Tanaka, 2006; Larkum and Nevian, 2008). For example NMDA receptor-dependent Ca2+ influx caused by a dendritic spike (Golding et al., 2002; Schiller et al., 2000; Wei et al., 2001), spread of Ras activity during long-term potentiation (LTP) (Harvey et al., 2008), and exocytosis of AMPA receptors into dendritic membrane during LTP (Lin et al., 2009; Makino and Malinow, 2009; Patterson et al., 2010; Petrini et al., 2009) all occur locally on short stretches of a dendrite and could contribute to synaptic potentiation at nearby synapses. Indeed, in hippocampus, LTP at one synapse reduces the threshold for LTP induction at neighboring synapses (Govindarajan et al., 2011; Harvey and Svoboda, 2007). Moreover, there is a trend that newly formed spines in hippocampal cultures appear in close proximity to activated spines during LTP (De Roo et al., 2008), potentially leading to clustering of synaptic enhancement. Such clustered synaptic potentiation could bind behaviorally relevant inputs onto dendritic subcompartments and improve storage capacity of individual neurons (Poirazi and Mel, 2001). Despite such studies, direct evidence for clustered synaptic plasticity in vivo is still lacking, owing to difficulties in online or retrospective identification of synaptic plasticity at individual synapses. In this study, we have developed an AMPA receptor-based optical approach to monitor recent history of synaptic plasticity induced in vivo through sensory experience or deprivation. We show that synaptic potentiation, revealed by experience-driven GluR1 incorporation into synapses, is clustered on short stretches of dendrites. Such clustered synaptic potentiation is effectively eliminated when animals are deprived of sensory experience or by expressing AMPA receptors insensitive to modulation for plasticity-driven incorporation into synapses. In contrast, homeostatic plasticity, revealed by synaptic GluR2 incorporation caused by sensory deprivation, occurs globally on dendrites, showing little evidence for clustering. Such coordinated modification of synapses could implement a framework for circuit development and refinement. Neuron 72, 1001–1011, December 22, 2011 ª2011 Elsevier Inc. 1001

Neuron Compartmentalized and Global Synaptic Plasticity

A Constructs pCALNL-SEP-GluR1 pCALNL-SEP-GluR2

CAG promoter

SEP AMPA receptor polyA

CAG promoter SEP AMPA receptor polyA

STOP loxP CAG promoter

pCALNL-DsRed

loxP DsRed polyA STOP

loxP CAG promoter ERT2

loxP Cre ERT2 polyA

loxP CAG promoter DsRed polyA 4-Hydroxytamoxifen (4OHT)

loxP

pCAG-ERT2CreERT2

B

E15 In utero electroporation

P11 P11 4OHT Whisker manipulation

C

P13 Image

L2/3

L4 L5

D

5 m

Whisker intact Whisker trimmed

1.2 1 0.8 0.6 0.4 0.2 0

GluR1 p < 0.001 0

0.4 0.8 1.2 1.6 Enrichment value

F 1.2 1 0.8 0.6 0.4 0.2 0

SEP-GluR2 DsRed Whisker intact Whisker trimmed

GluR2 p < 0.001

Enriched spine Non-enriched spine

G Enrichment value

Cumulative probability

E

Enriched spine Non-enriched spine

Cumulative probability

SEP-GluR1 DsRed

5 m

1.6

Whisker intact Whisker trimmed

1.2 0.8 0

0 0.5 1 1.5 2 2.5 3 Enrichment value

RESULTS Identification of Experience-Driven Synaptic Plasticity at Individual Spines To examine experience-dependent plasticity at individual synapses, we monitored the synaptic incorporation of fluorescently tagged AMPA receptor subunits, GluR1 and GluR2. To achieve acute expression of recombinant genes in a small number of neurons, we used a Cre/loxP-mediated inducible expression system where the transcription of genes of interest is regulated by a floxed stop cassette (Matsuda and Cepko, 2007). In this system, Cre expression is dependent on 4-hydroxytamoxifen (4-OHT). Once expressed, Cre drives removal of the (floxed) stop cassettes, permitting expression of genes of interest (Figure 1A). We used in utero electroporation to deliver three DNA constructs into layer 2/3 pyramidal neurons of the developing mouse barrel field: (1) a floxed stop cassette followed by the gene for GluR1 (or GluR2) tagged with a pH-sensitive form of green fluorescent protein (Super Ecliptic pHluorin, SEP) on the N terminus; (2) a floxed stop cassette followed by the gene for 1002 Neuron 72, 1001–1011, December 22, 2011 ª2011 Elsevier Inc.

GluR1

Figure 1. Identification of ExperienceDriven Synaptic Plasticity at Individual Spines of L2/3 Pyramidal Cells in the Barrel Cortex (A) Expression system. Intraperitoneal injection of 4-OHT drives expression of Cre recombinase, which induces expression of AMPA receptor subunits and DsRed (cytoplasmic marker) by removing the transcriptional stop cassettes located in front of the genes of interest. (B) Experimental design. The DNA plasmids were in utero electroporated at E15 in the right barrel cortex. When animals reached P11, 4-OHT was injected i.p., and whiskers were either left intact or all trimmed daily. At P13, acute slices were prepared, and basal dendrites of L2/3 pyramidal cells were imaged with two-photon microscopy. (C) Example of sparse expression of SEP-GluR1 and DsRed in L2/3 pyramidal cells in the barrel cortex. (D) Left panels show examples of SEP-GluR1 and DsRed-expressing neurons in whisker-intact and whisker-trimmed animals. Right panels are examples of SEP-GluR2 and DsRed-expressing neurons in whisker-intact and whisker-trimmed animals. Arrowheads indicate enriched spines, whereas arrows indicate nonenriched spines. (E) Spine enrichment values for SEP-GluR1 in whisker-intact (n = 2,701 spines, 23 cells, 11 animals) and whisker-trimmed (n = 1,878 spines, 17 cells, 7 animals) animals (p < 10
17, Kolmogorov-Smirnov test). (F) Spine enrichment values for SEP-GluR2 in whisker-intact (n = 1,057 spines, 8 cells, 5 animals) and whisker-trimmed (n = 1,226 spines, 8 cells, 4 animals) animals (p < 10
9, Kolmogorov-Smirnov test). (G) Spine enrichment values obtained from (E) and (F) to illustrate the difference between SEP-GluR1 and SEP-GluR2 spine enrichment (mean ± SEM).

GluR2

DsRed, a red cytoplasmic marker; and (3) the 4-OHT-dependent Cre recombinase-expressing plasmid, pCAG-ERT2CreERT2. Animals were injected intraperitoneally (i.p.) with 4-OHT at postnatal day (P) 11, and coronal brain slices were prepared at P13 (Figure 1B). A small number of neurons (<1% of layer 2/3 neurons) displayed expression of SEP-GluR1 (or SEP-GluR2) and DsRed (Figure 1C). Animals that did not receive 4-OHT showed no detectable expression (four animals; data not shown), indicating little leak in the expression system. Sensory experience, controlled by trimming or leaving intact an animal’s whiskers (Feldman and Brecht, 2005), can drive GluR1 into synapses between layer 4 and layer 2/3 neurons through an LTP-like process (Clem and Barth, 2006; Takahashi et al., 2003). We wished to determine whether synaptic incorporation of SEP-GluR1 can be monitored optically using dualchannel two-photon microscopy. We measured SEP-GluR1 enrichment in dendritic spines, which is the spine SEP signal normalized for spine area and for neuronal expression level of the SEP-tagged protein (see Experimental Procedures). We focused on basal dendrites of layer 2/3 pyramidal neurons

Neuron Compartmentalized and Global Synaptic Plasticity

because they receive the majority of synaptic inputs (Feldmeyer et al., 2002; Petreanu et al., 2009). Consistent with electrophysiological studies (Clem and Barth, 2006; Takahashi et al., 2003), following 2 days of 4-OHT-driven expression, SEP-GluR1 spine enrichment was higher in animals with whiskers intact (0.84 ± 0.005, n = 2701 spines) compared with animals with whiskers trimmed (0.77 ± 0.006, p < 10
17, n = 1878 spines; Figures 1D, 1E, and 1G). Although LTP is thought to depend on the GluR1 AMPA receptor subunit, GluR2 is not required for LTP (Hayashi et al., 2000; Jia et al., 1996; Zamanillo et al., 1999) but is required for homeostatic plasticity produced by deprivation of activity or sensory input (Gainey et al., 2009). We examined the synaptic incorporation of SEP-GluR2 under similar (2 day expression) conditions. In contrast to SEP-GluR1, following 2 days of 4OHT-driven expression, whisker-trimmed animals had increased spine enrichment of SEP-GluR2 (1.43 ± 0.01, n = 1226 spines) compared to whisker-intact animals (1.30 ± 0.01, p < 10
9, n = 1057 spines; Figures 1D, 1F, and 1G), consistent with the view that reduced input activity produces homeostatic synaptic strengthening that is controlled by GluR2 (Gainey et al., 2009). Spine Enrichment of AMPA Receptors as an Indicator for Plasticity To test if spine enrichment of SEP-tagged AMPA receptors was a good estimate of their synaptic incorporation, we used fluorescence recovery after photobleaching (Makino and Malinow, 2009). Because synaptic receptors are relatively immobile (Heine et al., 2008; Makino and Malinow, 2009), the recovery of fluorescence after photobleaching a spine containing synaptic SEP-tagged AMPA receptors is incomplete. Following 2 days of 4-OHT-driven expression, the fraction of SEP-GluR1 spine fluorescence that failed to recover (immobile fraction) correlated well with the SEP-GluR1 spine enrichment (r = 0.58, p < 0.001, n = 29 spines; Figures 2A and 2B). In contrast, immobile fractions of spine SEP-GluR1 were not correlated with spine size (r = 0.12, p = 0.53, n = 29 spines; Figure 2C), consistent with the view that spine size is a consequence of plasticity integrated over a period longer than the 2 day expression period of recombinant receptors. Indeed, expression of SEP-GluR1 for longer periods (e.g., 4 days) produced spines in which the SEP-GluR1 spine enrichment was correlated with spine size (see Figures S1A–S1D available online). Similar to SEP-GluR1, following 2 days of 4-OHT-driven more expression, there was a strong relation between SEP-GluR2 immobile fractions and SEP-GluR2 spine enrichment (r = 0.66, p < 0.003, n = 19 spines; Figure 2D), but not with spine size (r = 0.14, p = 0.56, n = 19 spines; Figure 2E). These results indicate that experience- or deprivation-driven synaptic plasticity can be detected using fluorescently tagged AMPA receptors. To test further the view that spine enrichment of SEP-tagged AMPA receptors serves as an indication of their synaptic incorporation, we performed glutamate uncaging onto spines that had various levels of SEP-GluR1 enrichment. We obtained whole-cell recordings from neurons expressing recombinant receptors and measured AMPA receptor-mediated responses from focally applied glutamate on spines (Figure 2F; see Experimental Procedures). We recorded responses at positive

(VH = +40mV) and negative (VH =
60mV) holding potentials; their ratio (current at VH = +40 mV/current at VH =
60 mV) is the rectification index. Because recombinant receptors form homomeric receptors, they display little outward current at positive potentials and, thus, a low rectification index. We found a correlation between rectification indices and enrichment values for different spines (r =
0.59, p < 0.03, n = 14 spines; Figure 2G), consistent with the view that enrichment value is a good measure for synaptic incorporation of recombinant SEP-tagged AMPA receptors. Experience-Dependent Clustering of Synaptic Potentiation on Dendrites To examine if nearby spines on individual dendrites displayed similar levels of plasticity, we calculated the correlation coefficient of SEP-GluR1 spine enrichment for neighboring spines (see Experimental Procedures; Figure 3A) following 2 day transient expression. Neighboring spines showed a significant positive correlation value (0.14 ± 0.03, p < 10
5, n = 95 dendrites) in dendrites from animals with whiskers intact (Figures 3B–3D and S2A). This correlation value between neighboring spines was significantly greater than that observed in whisker-trimmed animals (0.003 ± 0.03, p < 0.009 with Bonferroni correction, n = 68 dendrites; Figures 3D and S2A). These results indicate that sensory experience drives coordinated potentiation onto nearby synapses. It is possible that some of the dendritic segments examined received little plasticity during the period of SEP-GluR1 expression (see below). Thus, we wished to determine what fraction of dendritic segments showed a significant correlation in the enrichment values of neighboring spines. For each dendritic segment we calculated the correlation coefficient of neighboring enrichment values and compared this to a value obtained by random shuffling of the enrichment values for that dendritic segment. If the correlation coefficient for the dendritic segment was greater than 95% of the correlation coefficients obtained from randomly shuffled enrichment values, that dendrite was deemed to have a significant correlation of nearby enrichment values. In dendrites obtained from animals with whiskers intact, 28 of 95 (29%) dendrites displayed a significant correlation between neighboring spine enrichment values (Figure 3E). The correlation coefficient for enrichment values in neighboring spines in dendrites with significant correlation was 0.36 ± 0.04 (Figure 3F). In dendrites obtained from animals with whiskers trimmed, only 5 of 68 (7%) were significant. The fraction of dendrites with significant correlation with nearby spines was greater in those obtained with whiskers intact (p < 0.0007, Fisher’s exact test). Sensory Deprivation Causes Global Synaptic Upscaling on Dendrites Inactivity or sensory deprivation produces homeostatic synaptic upscaling that is global throughout a cell and depends on GluR2 (Gainey et al., 2009; Turrigiano, 2008). We, thus, tested the effect of sensory deprivation on the correlation of enrichment values in spines from cortical neurons expressing SEP-GluR2, using the same temporally regulated expression system. In animals with whiskers trimmed for 2 days, nearby spines failed to show significant positive correlation (0.02 ± 0.03, p = 0.46, Neuron 72, 1001–1011, December 22, 2011 ª2011 Elsevier Inc. 1003

Neuron Compartmentalized and Global Synaptic Plasticity

GluR1 immobile fraction

5 m

0.8 0.6 0.4 0.2 0 -0.2 -0.4

b

0.2

a

0.1 0

0.2 0.4 0.6 0.8

1

Enrichment value

0 0.5 1 1.5 2 2.5 Enrichment value

Dodt image SEP-GluR1 DsRed

Uncaging pulse

r = 0.66 p < 0.003

0.6 0.5 0.4 0.3 0.2 0.1 0

a

a b

r = -0.59 p < 0.03 0

Whisker intact Whisker trimmed

0.8 0.6 0.4 0.2 0 -0.2 -0.4

r = 0.12 p = 0.53 0

0.5 1 Spine size

E

0 0.5 1 1.5 2 2.5 Enrichment value

G 5 m

Whisker intact Whisker trimmed

0.8 0.6 0.4 0.2 0 -0.2 -0.4

Rectification (+40mV / -60mV)

F

C

r = 0.58 p < 0.001

D

0.3

50 m

Whisker intact Whisker trimmed

GluR1 immobile fraction

b a

B

GluR2 immobile fraction

+25 min

GluR1 immobile fraction

SEP-GluR1 DsRed +3 min

-10 min

GluR2 immobile fraction

A

1 2 3 Enrichment value

1.5

Whisker intact Whisker trimmed

0.8 0.6 0.4 0.2 0 -0.2 -0.4

r = 0.14 p = 0.56 0

0.5 1 Spine size

1.5

b +40 mV

-60 mV 10 pA 20 ms

Figure 2. Spine Enrichment of AMPA Receptors as an Indicator for Their Synaptic Localization (A) Example of fluorescence recovery after photobleaching of spines expressing SEP-GluR1. Top panels show two spines (‘‘a’’ and ‘‘b,’’ indicated as arrowheads) that were simultaneously photobleached, and recovery of the SEP-GluR1 fluorescence was monitored at 25 min to measure the immobility of the receptor. Bottom graph shows enrichment values and GluR1 immobile fractions of the two spines (‘‘a’’ and ‘‘b’’). (B) Correlation between SEP-GluR1 spine enrichment values and immobile fractions of spine SEP-GluR1 (r = 0.58, p < 0.001, n = 16 spines, 4 cells, 4 animals for whisker-intact; and n = 13 spines, 3 cells, 2 animals for whisker-trimmed animals). (C) No correlation between spine size and immobile fractions of spine SEP-GluR1 (r = 0.12, p = 0.53, n = same as B). (D) Correlation between SEP-GluR2 spine enrichment values and immobile fractions of spine SEP-GluR2 (r = 0.66, p < 0.003, n = 10 spines, 3 cells, 3 animals for whisker-intact; and n = 9 spines, 2 cells, 2 animals for whisker-trimmed animals). (E) No correlation between spine size and immobile fractions of spine SEP-GluR2 (r = 0.14, p = 0.56, n = same as D). (F) Example of a whole-cell recording and glutamate uncaging at a spine with high SEP-GluR1 enrichment. Uncaging-evoked AMPA receptor-mediated postsynaptic currents are shown in (G) spine b. (G) Left view shows correlation between SEP-GluR1 spine enrichment values and rectification indices (r =
0.59, p < 0.03, n = 14 spines, 9 cells, 5 animals). Rectification indices were measured as amplitude of AMPA current at +40mV/amplitude of AMPA current at
60mV. Right view shows example traces of glutamate uncaging-evoked AMPA receptor-mediated postsynaptic currents at spines with different SEP-GluR1 enrichment values.

n = 45 dendrites; Figures 3D and S2B); this value was significantly different from that found in animals with whiskers intact expressing SEP-GluR1 (p < 0.05 with Bonferroni correction; n = 95 dendrites) but not different from that observed in animals with whiskers intact expressing SEP-GluR2 (
0.05 ± 0.03, p = 0.11, n = 44 dendrites; Figures 3D and S2B). These results indicate that synaptic incorporation of GluR2 caused by homeostatic plasticity occurs globally on dendrites with little compartmentalization. Reconstruction of Single Neurons To gain more insight into the distribution of clustered plasticity in a whole neuron, we measured enrichment values for all identifi1004 Neuron 72, 1001–1011, December 22, 2011 ª2011 Elsevier Inc.

able spines in individual neurons (Figures 4A, 4B, S3A, and S3B). For a neuron expressing SEP-GluR1 in a whisker-intact animal, of the 1,078 spines we considered the spines with the highest 15% of enrichment values. Spines with these values appeared not to be randomly distributed. Many of the highly enriched spines were seen at the very tip of dendrites (p < 0.0003, n = 161 spines, compared to nonenriched spines, n = 917 spines; Figures 4A and 4C), suggesting that terminal dendritic segments were particularly sensitive to plasticity. Indeed, when we examined all of the data obtained from individual dendritic segments expressing GluR1, we noted an increase in enrichment as a function of distance from cell body (Figure S3C). We wished to test if the occurrence of highly enriched spines was more

Neuron Compartmentalized and Global Synaptic Plasticity

A

B Clustered

SEP-GluR1 DsRed Enrichment value

Distributed

or

D 0.2

**

ns

E

0.1 0

p > 0.05 p < 0.05

15 10 5 0

-0.1

0 5 10 15 20 25 30 35 Spine number

F 20

Frequency

Correlaiton coefficient

Enriched spine

1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0

5 m

ed m im Tr t c ta d In me im Tr t c ta

In

-0.6 0 0.6 Correlation coefficient at 1st neighbor

Correlaiton coefficient

Potentiated spine *

Figure 3. Experience-Dependent Clustering of Synaptic Potentiation and Global Synaptic Upscaling Driven by Sensory Deprivation

C

0.4 0.3 0.2 0.1 0 -0.1 -0.2

GluR1 GluR2

likely to occur in neighboring spines. In this neuron, of the 161 spines showing the highest 15% enrichment, 50 were neighboring spines. When the enrichment values were randomly shuffled, there was on average 24 pairs of neighboring spines with enrichment values in the top 15% (p < 0.001; Figure 4D). We conducted similar analysis considering the top 5% or 10% of enrichment values. In all cases the number of neighboring spines with highly enriched values in the analyzed neuron was significantly greater than what was observed when the enrichment values were randomly shuffled (Figures S3A and S3B). We next examined SEP-GluR2 on a fully reconstructed neuron from a whisker-trimmed animal. In this case highly enriched spines were not found on distal regions, as was the case for SEP-GluR1 (Figure 4B). There was a tendency for highly enriched spines (n = 150) to be proximal relative to nonenriched spines (p < 0.005, n = 851 spines; Figures 4C, S3B, and S3C). We also noted that neighboring spines were no more likely to have high enrichment values than randomly shuffled values (p = 0.29; Figures 4E, S3A, and S3B). Taken together, these results suggest that there are distinct trafficking patterns produced by experience-driven synaptic potentiation and deprivation-driven synaptic upscaling. Clustered Synaptic Potentiation with Heteromeric AMPA Receptors The data above suggest that the clustering of plasticity is observed for GluR1, but not GluR2, consistent with their dependence on LTP and experience (Hayashi et al., 2000; Takahashi et al., 2003; Zamanillo et al., 1999). However, when expressed alone, these AMPA receptor subunits form homomeric receptors,

0

2 4 6 8 10 Neighboring spine

(A) Schematic of distributed synaptic potentiation and clustered synaptic potentiation. (B) Example of clustered synaptic SEP-GluR1 enrichment in a basal dendrite of a L2/3 pyramidal cell in a whisker-intact animal. (C) Profile of SEP-GluR1 spine enrichment along the dendrite shown in (B). Line indicates a running average. (D) Correlation coefficients at neighboring synapses of dendrites expressing SEP-GluR1 in whiskerintact (n = 95 dendrites, 23 cells, 11 animals) and whisker-trimmed (n = 68 dendrites, 17 cells, 7 animals) animals (**p < 0.009, t test with Bonferroni correction; mean ± SEM), and SEP-GluR2 in whisker-intact (n = 44 dendrites, 8 cells, 5 animals) and whisker-trimmed (n = 45 dendrites, 8 cells, 4 animals) animals (*p < 0.05, t test with Bonferroni correction; ns, nonsignificant; mean ± SEM). (E) Histogram of correlation coefficients at neighboring synapses for dendrites expressing SEPGluR1 in whisker-intact animals (n = 95 dendrites, 23 cells, 11 animals). For each dendrite, p value for its correlation coefficient was obtained by calculating the likelihood of obtaining such a correlation coefficient from randomly shuffled spines of that dendrite. (F) Autocorrelation of dendrites with p < 0.05 in (E) as a function of spine lag (mean ± SEM).

which normally comprise a small proportion of endogenously expressed receptors (Wenthold et al., 1996). To examine the trafficking of heteromeric receptors, which constitute the predominant species of receptors (Wenthold et al., 1996), we transiently coexpressed SEP-GluR1 with untagged-GluR2, or untaggedGluR2 and SEP-GluR3 (see Experimental Procedures). We first confirmed, using electrophysiological measures, that heteromeric receptors were formed when expressing SEP-GluR1 with GluR2. We obtained whole-cell recordings from neurons expressing recombinant receptors and measured responses from focally applied glutamate on spines (Figure 5A; see Experimental Procedures). Homomeric receptors display inward rectification, which was observed in neurons expressing SEP-GluR1 (0.28 ± 0.02, n = 15 spines; Figure 5B). However, no such inward rectification was observed from neurons expressing SEP-GluR1 and GluR2 (0.49 ± 0.03, p < 0.00003, n = 13 spines; Figure 5B), indicating that heteromeric receptors were formed. We examined in animals with whiskers intact the spine enrichment values in neurons transiently expressing SEP-GluR1 and GluR2 (Figures 5C and 5E). Spine enrichment of SEP-GluR1/ GluR2 heteromeric receptors (0.84 ± 0.006, n = 1865 spines) did not differ from that of SEP-GluR1 homomeric receptors (0.84 ± 0.005, p = 0.70, n = 2701 spines; Figures 5C, 5E, S4A, and S4C). Similarly, spine enrichment of GluR2/SEP-GluR3 (1.29 ± 0.01, n = 1390 spines) was not different from that of SEP-GluR2 (1.30 ± 0.01, p = 0.08, n = 1057 spines; Figures 5D, 5E, S4B, and S4C). As with homomeric receptors, the immobile fractions of heteromeric receptors correlated with spine enrichment (r = 0.72, p < 10
5, n = 31 spines; Figure 5F), but not with spine size (r = 0.04, p = 0.85, n = 31 spines; Figure 5G). Neuron 72, 1001–1011, December 22, 2011 ª2011 Elsevier Inc. 1005

Neuron Compartmentalized and Global Synaptic Plasticity

A

Figure 4. Reconstruction of Single Neurons

SEP-GluR1 & DsRed from whisker-intact animal 1

GluR1

3 1

2

2

3 4

5 m

B

50 m

4

SEP-GluR2 & DsRed from whisker-trimmed animal 1

GluR2

3

1

3

4 2

4

(A) Left, reconstruction of a neuron expressing SEP-GluR1 from a whisker-intact animal. Insets are examples of dendritic segments. Right, spines with the highest 15% enrichment are shown in red (n = 161 spines) and the rest in gray (n = 917 spines). (B) Left, reconstruction of a neuron expressing SEP-GluR2 from a whisker-trimmed animal. Insets are examples of dendritic segments. Right, spines with the highest 15% enrichment are shown in red (n = 150 spines) and the rest in gray (n = 851 spines). (C) Absolute distance from the soma of enriched and nonenriched spines for the GluR1 (***p < 0.0003, Kolmogorov-Smirnov test) and GluR2expressing neurons (**p < 0.005, KolmogorovSmirnov test). (D) Histogram of the number of neighbor pairs with high SEP-GluR1 enrichment obtained from shuffled spines. There are 50 such pairs in the original data (dashed line; p < 0.001). (E) Histogram of the number of neighbor pairs with high SEP-GluR2 enrichment obtained from shuffled spines. There are 24 such pairs in the original data (dashed line; p = 0.29).

2

5 m

GluR2 enriched ** GluR2 non-enriched

GluR1 non-enriched GluR1 enriched ***

0 100 200 Distance from soma ( m )

D

Raw data 0.5

GluR1

0.4 0.3

Shuffled

0.2 0.1

E Probability

1.2 1 0.8 0.6 0.4 0.2 0

Probability

Cumulative probability

C

50 m

p < 0.001

0

50 0 # Enriched spine pair

For neurons expressing SEP-GluR1 and GluR2, there was a significant positive correlation in enrichment values between neighboring spines in animals with whiskers intact (0.12 ± 0.03, p < 0.0005, n = 59 dendrites; Figures 5C, 5H, 5I, and S2C). Of 59 dendrites, 12 (20%) showed significant near-neighbor correlations (Figure 5J), which reached a value of 0.32 ± 0.04 (Figure 5K). For neurons expressing GluR2 and SEP-GluR3, the distribution of enrichment values mirrored that found in neurons expressing SEP-GluR2: neighboring spines displayed no significant correlation in enrichment values (
0.005 ± 0.02, p = 0.85, n = 47 dendrites; Figures 5I and S2C). These results indicate that the effect of experience on the distribution of heteromeric SEPGluR1/GluR2 and GluR2/SEP-GluR3 receptors is similar to that observed in homomeric SEP-GluR1 or SEP-GluR2 receptors. Receptor Modulation Sensitivity Controls Clustered Synaptic Potentiation The results presented above indicate that neural activity patterns onto cortical neurons driven by sensory experience

produce clustered potentiation of nearby synapses. Such patterning could be Raw data produced by LTP-like processes, which have been shown in in vitro systems to 0.5 GluR2 lower threshold of nearby spines for plas0.4 ticity (Govindarajan et al., 2011; Harvey 0.3 Shuffled and Svoboda, 2007; Harvey et al., 0.2 2008). One model to explain such nearby 0.1 p = 0.29 threshold lowering is the following: nor0 0 50 mally, an individual synapse is potenti# Enriched spine pair ated (and accumulates GluR1) when it receives sufficient presynaptic activity paired with postsynaptic depolarization (the latter provided by close or distant synapses). Such point potentiation would activate intracellular signal transduction pathways (e.g., Ras; Harvey et al., 2008) that could activate downstream kinases leading to phosphorylation of GluR1 at nearby regions (within 5 mm). Receptors at these nearby regions would now have lower threshold for becoming incorporated into synapses (for as long as GluR1 maintains a phosphorylated status). To test for this possibility, we expressed SEPGluR1 with mutations at two phosphorylation sites (S831A and S845A) in the cytoplasmic segment (designated GluR1AA; Figure 6A). These mutations on GluR1 render the receptor insensitive to modulation by protein kinases at these sites. Phosphorylation at these sites is known to lower the threshold for GluR1 incorporation into synapses during LTP (Hu et al., 2007). We examined the distribution of spine enrichment values in animals with whiskers intact transiently expressing SEP-GluR1AA. The average spine enrichment of SEP-GluR1AA (0.84 ± 0.007, n = 1584 spines) was similar to that of SEP-GluR1 (0.84 ± 0.005, p = 0.14, n = 2701 spines; Figure 6B). This is consistent with

1006 Neuron 72, 1001–1011, December 22, 2011 ª2011 Elsevier Inc.

Neuron Compartmentalized and Global Synaptic Plasticity

B GluR1/2 +40 mV

-60 mV 12

D

E GluR2 SEP-GluR3 DsRed

10 m

5 m

I

p < 0.001

1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0 0 10 20 30 40 50 60 Spine number

F 0.8 0.6 0.4 0.2 0 -0.2 -0.4

J **

0.2 0.1 0 -0.1

GluR1/2 GluR2/3

***

15

13

GluR1/1 GluR1/2 GluR1/2 GluR2/3

r = 0.72 p < 0.001 0 0.5 1 1.5 2 2.5 Enrichment value

0 0.5 1 1.5 2 2.5 3 Enrichment value

Correlaiton coefficient

Enrichment value

GluR1/2 GluR2/3

1.2 1 0.8 0.6 0.4 0.2 0

H

SEP-GluR1 GluR2 DsRed

10 pA 20 ms

20

Frequency

C

11

Uncaging pulse

Cumulative probability

SEP-GluR1 GluR2 Glu uR2 R DsRed

0.6 0.5 0.4 0.3 0.2 0.1 0

p > 0.05 p < 0.05

15 10 5 0

-0.6 0 0.6 Correlation coefficient at 1st neighbor

G Immobile fraction

GluR1

5 m

GluR1/2 GluR2/3

0.8 0.6 0.4 0.2 0 -0.2 -0.4

r = 0.04 p = 0.85 0 0.2 0.4 0.6 0.8 1 Spine size

K Correlaiton coefficient

50 m

Rectification (+40mV / -60mV)

Dodt Dod o t image od e

Immobile fraction

A

0.4 0.3 0.2 0.1 0 -0.1 -0.2 0

2 4 6 8 10 Neighboring spine

Figure 5. Clustered Synaptic Potentiation with Heteromeric AMPA Receptors (A) Example of a whole-cell recording and glutamate uncaging at a spine containing GluR1/2 heteromeric AMPA receptors. (B) Left views show example traces of glutamate uncaging-evoked AMPA receptor-mediated postsynaptic currents at spines expressing GluR1/1 and GluR1/2. Right views illustrate rectification indices for GluR1/1 homomeric (n = 15 spines, 10 cells, 8 animals) and GluR1/2 heteromeric (n = 13 spines, 8 cells, 5 animals) AMPA receptors at single spines (***p < 0.00003, t test; mean ± SEM). Rectification indices were measured as amplitude of AMPA current at +40 mV/amplitude of AMPA current at
60mV. (C) Example of clustered synaptic potentiation with GluR1/2 heteromeric AMPA receptors in a whisker-intact animal. Arrowheads indicate enriched spines. (D) Example of a GluR2/3 heteromeric AMPA receptor-expressing neuron in a whisker-intact animal. (E) Spine enrichment values for GluR1/2 (n = 1,865 spines, 11 cells, 8 animals) and GluR2/3 (n = 1,390 spines, 14 cells, 5 animals) heteromeric AMPA receptors (p < 10
166, Kolmogorov-Smirnov test). (F) Correlation between spine enrichment values and spine immobile fractions of heteromeric AMPA receptors (r = 0.72, p < 10
5, n = 16 spines, 3 cells, 3 animals for GluR1/2; and n = 15 spines, 3 cells, 3 animals for GluR2/3). (G) No correlation between spine size and spine immobile fractions of heteromeric AMPA receptors (r = 0.04, p = 0.85, n = same as F). (H) Profile of SEP-GluR1/2 spine enrichment along the dendrite shown in (C). Line indicates a running average. (I) Correlation coefficients at neighboring synapses of dendrites expressing GluR1/2 (n = 59 dendrites, 11 cells, 8 animals) and GluR2/3 (n = 47 dendrites, 14 cells, 5 animals) heteromeric AMPA receptors (**p < 0.005, t test; mean ± SEM). (J) Histogram of correlation coefficients at neighboring synapses for dendrites expressing SEP-GluR1/2 in whisker-intact animals (n = 59 dendrites, 11 cells, 8 animals). p values are calculated as in Figure 3E. (K) Autocorrelation of dendrites with p < 0.05 in (J) as a function of spine lag (mean ± SEM).

the previous observation that mice in which GluR1 has been replaced with GluR1AA have the same number of synaptic AMPA receptors as wild-type mice (Lee et al., 2003). Apparently, the reduced synaptic incorporation resulting from the lost threshold-lowering effects of phosphorylation is offset by the reduced synaptic receptor removal produced by the absent LTD also described for this GluR1 mutant (Lee et al., 2003). Immobile fractions of SEP-GluR1AA were well correlated with its enrichment in spines (r = 0.87, p < 0.00003, n = 15 spines; Figure 6C), but not with spine size (r = 0.29, p = 0.29, n = 15 spines; Figure 6D). Unlike SEP-GluR1, the enrichment values at neighboring spines were not positively correlated (0.03 ± 0.03,

p = 0.41, n = 62 dendrites), and were significantly different from the correlation value displayed by neighboring spines in animals with whiskers intact expressing SEP-GluR1 (p < 0.04 with Bonferroni correction, n = 95 dendrites; Figures 6E and S2D). These data suggest that removing trafficking modulation signals on GluR1 effectively eliminates the dendritic clustering of synaptic potentiation displayed by SEP-GluR1. Finally, we examined if clustering of GluR1 synaptic delivery could be observed in older animals (Figures S5A–S5C). In this group of animals, electroporation was conducted in utero, and the induction (injection with 4-OHT) was initiated at P34 or P35. Two days later, brain slices were prepared and neurons Neuron 72, 1001–1011, December 22, 2011 ª2011 Elsevier Inc. 1007

Neuron Compartmentalized and Global Synaptic Plasticity

DISCUSSION

A Modulation sensitive region GluR1wt

Potentiation

Potentiation

GluR1AA

831 S 845 S

831 S->A 845 S->A

Low

0

0.4 0.8 1.2 1.6 Enrichment value

GluR1AA GluR1wt

0.8 0.6 0.4 0.2 0 -0.2 -0.4

r = 0.29 p = 0.29 0

0.5 1 Spine size

1.5

GluR1 immobile fraction

p = 0.14

GluR1AA GluR1wt

0.8 0.6 0.4 0.2 0 -0.2 -0.4

r = 0.87 p < 0.001 0

0.5 1 1.5 2 Enrichment value

E Correlaiton coefficient

Cumulative probability

GluR1AA GluR1wt

1.2 1 0.8 0.6 0.4 0.2 0

D GluR1 immobile fraction

High

C

B

0.2

*

0.1 0 -0.1

GluR1wt GluR1AA

Figure 6. Modulation Insensitivity Controls Clustered Synaptic Potentiation (A) Schematic of heterosynaptic threshold reduction for plasticity by singlespine potentiation. GluR1AA is insensitive to heterosynaptic biochemical signals, and thus, heterosynaptic threshold reduction is minimized. (B) Spine enrichment values for SEP-GluR1AA (n = 1,584 spines, 11 cells, 5 animals) and SEP-GluR1 (n = 2,701 spines, 23 cells, 11 animals) in whiskerintact animals (p = 0.14, Kolmogorov-Smirnov test). The ‘‘GluR1wt’’ data are from Figure 1E. (C) Correlation between SEP-GluR1AA spine enrichment values and immobile fractions of spine SEP-GluR1AA (r = 0.87, p < 0.00003, n = 15 spines, 3 cells, 2 animals). The ‘‘GluR1wt’’ data are from Figure 2B. The regression line was fitted for both conditions. (D) No correlation between spine size and immobile fractions of spine SEPGluR1AA (r = 0.29, p = 0.29, n = same as C). The ‘‘GluR1wt’’ data are from Figure 2C. The regression line was fitted for both conditions. (E) Correlation coefficients at neighboring synapses of dendrites expressing SEP-GluR1wt (n = 95 dendrites, 23 cells, 11 animals) and SEP-GluR1AA (n = 62 dendrites, 11 cells, 5 animals, *p < 0.04, t test with Bonferroni correction; mean ± SEM). The ‘‘GluR1wt’’ data are from Figure 3D.

were imaged (Figure S5A). Spine enrichment values were significantly higher (1.27 ± 0.01, n = 996 spines) than those seen in younger animals (0.84 ± 0.005, n = 2701 spines, p < 10
148; Figure S5B), due to a large reduction in SEP-GluR1 on dendritic membrane (data not shown). Correlation of enrichment values between neighboring spines was significantly different from zero (0.16 ± 0.04, p < 0.002, n = 24 dendrites; Figure S5C). Of 24 dendritic segments, 10 (42%) displayed significant nearneighbor correlations, which reached a value of 0.27 ± 0.04. These observations indicate that experience-driven clustering of synaptic potentiation also occurs in older animals. 1008 Neuron 72, 1001–1011, December 22, 2011 ª2011 Elsevier Inc.

In this study, we have examined the spatial distribution of plasticity on neuronal dendrites produced as a result of sensory experience. We used temporally restricted expression of SEPtagged glutamate receptors to identify individual synapses that had recently undergone plasticity in vivo. The spine enrichment correlated well with the immobile fraction as well as the electrophysiological property of tagged receptor, indicating that spine enrichment corresponds to synaptically incorporated receptors. Experience increased the synaptic enrichment of SEP-GluR1, whereas deprivation increased the synaptic enrichment of SEP-GluR2, supporting their use as indicators of plasticity. The trafficking of SEP-GluR1, which forms homomeric receptors, mirrored that of heteromeric SEP-GluR1/GluR2 receptors. Similarly, the trafficking of SEP-GluR2 paralleled that of heteromeric SEP-GluR3/GluR2. These findings, in addition to previous results indicating that the level of overexpression of transiently expressed recombinant AMPA receptors in dendritic regions is less than 50% above endogenous levels (Kessels et al., 2009), suggest that our optical approach to measure receptor incorporation into synapses can be used to analyze endogenous synaptic plasticity mechanisms. A number of in vitro and theoretical studies have examined the role of compartmentalized plasticity in neuronal function (Govindarajan et al., 2006; Larkum and Nevian, 2008; Poirazi and Mel, 2001; Polsky et al., 2004). Clustered plasticity could bind functionally relevant inputs onto dendrites and enhance storage capacity of individual neurons by locally recruiting nonlinear voltage-gated conductances (Poirazi and Mel, 2001). Furthermore, clustered plasticity can increase the probability of local spike initiation by enhancing excitability of dendrites (Frick et al., 2004), which in turn strengthens the coupling between a dendritic branch and the soma (Losonczy et al., 2008). Such branch strength potentiation permits temporally precise and robust somatic output, which is generally believed to be important for information processing by single neurons (Koch and Segev, 2000). Clustered synaptic plasticity could complement plasticity of dendritic excitability as mechanisms of experience-driven information storage (Makara et al., 2009). What cellular mechanisms could underlie such clustered synaptic plasticity? Based on simple simulations (see Figure S6), we found that our data with SEP-GluR1 (and GluR1/2) are consistent with a model in which the cluster of synaptic potentiation spans on average approximately four synapses, corresponding to 8 mm of dendrite. Notably, such a spatial scale is similar to the biochemical compartmentalization of dendritic plasticity machinery in vitro (Harvey et al., 2008; Makino and Malinow, 2009; Patterson et al., 2010; Schiller et al., 2000; Wei et al., 2001) as well as in vivo (Jia et al., 2010), suggesting that the local spread of intracellular signaling factors is important for the coordinated potentiation among nearby synapses. In this respect the GluR1AA mutant, which should be insensitive to heterosynaptic biochemical signals (e.g., Ras-driven protein kinase activation) and, thus, the effect of the heterosynaptic threshold reduction, showed no clustered spine enrichment. Our data cannot fully rule out the possibility that groups of presynaptic fibers with similar activity patterns, thereby driving

Neuron Compartmentalized and Global Synaptic Plasticity

similar levels of plasticity, make synapses on nearby regions of dendrites. However, a recent study in the auditory cortex argues against simple sensory activity providing such clustered inputs (Chen et al., 2011). Furthermore, a model in which the clustering is solely due to afferent coactivity is difficult to reconcile with the results observed with GluR1AA. Our data suggest that natural stimuli engage postsynaptic mechanisms leading to locally clustered enhancement of synapses. In conclusion our results support the view that experience can drive clustered synaptic enhancement onto neuronal dendritic subcompartments, providing fundamental architecture to circuit development and function. Sensory deprivation drives cell-wide synaptic enhancement that globally sensitizes a neuron. EXPERIMENTAL PROCEDURES Experiments were conducted according to National Institutes of Health guidelines for animal research and were approved by the Institutional Animal Care and Use Committee at Cold Spring Harbor Laboratory and University of California, San Diego. DNA Constructs SEP-GluR1, SEP-GluR1(S831A,S845A), SEP-GluR2(R586Q), untaggedGluR2(edited), and SEP-GluR3 from rat were PCR amplified and subcloned into an expression vector with a ubiquitous promoter CAG, pCALNL. pCALNL-DsRed and pCAG-ERT2CreERT2 were obtained from Addgene. All the DNA plasmids were amplified with the endotoxin-free Maxiprep kit (QIAGEN). For the formation of homomeric GluR2, SEP-GluR2(R586Q) was expressed. Heteromeric AMPA receptors were formed by coexpressing untagged-GluR2(edited) with either SEP-GluR1 or SEP-GluR3 at a 1:1 molar ratio. In Utero Electroporation L2/3 progenitor cells were transfected by in utero electroporation. E15 time pregnant C57BL/6J mice (Charles River) were anesthetized with an isoflurane-oxygen mixture (Lei Medical). Approximately 0.5 ml of DNA solution containing fast green was pressure injected through a pulled-glass capillary tube by mouth into the right lateral ventricle of each embryo. The head of each embryo was placed between tweezers electrodes with the anode contacting the right hemisphere. Electroporation was achieved with five square pulses (duration = 50 ms, frequency = 1 Hz, voltage = 25V; Harvard Apparatus). Cre Recombinase Activation by 4-OHT 4-OHT (Sigma-Aldrich) was dissolved in ethanol at a concentration of 20 mg/ml and diluted with 9 vol of corn oil (Sigma-Aldrich). Diluted 4-OHT (2 mg/ml) was i.p. injected into each mouse at P11 (100 ml per animal) or P34–P35 (300–450 ml per animal). Whisker Manipulation For sensory deprivation all the major whiskers were trimmed daily from P11. Whisker-intact animals were handled similarly to whisker-trimmed animals. Preparation Acute coronal brain slices (350 mm thick) from in utero electroporated mice at P13 or P36–P37 were prepared. Slices were cut in gassed (95% O2 and 5% CO2) ice-cold solution containing 25 mM NaHCO3, 1.25 mM NaH2PO4, 2.5 mM KCl, 0.5 mM CaCl2, 7 mM MgCl2, 25 mM D-glucose, 110 mM choline chloride, 11.4 mM sodium ascorbate, and 3.1 mM sodium pyruvate. Slices were then incubated in artificial cerebrospinal fluid (ACSF) containing 118 mM NaCl, 2.5 mM KCl, 26 mM NaHCO3, 1.2 mM NaH2PO4, 11 mM D-glucose, 4 mM MgCl2, and 4 mM CaCl2 at 35 C for 30 min and then at room temperature until used. All experiments were performed at 30 C.

Imaging We used a two-photon laser-scanning microscope (Prairie) to image L2/3 pyramidal cells of the mouse barrel cortex (403 0.8 NA objective lens and 1.4 NA oil condenser; Olympus) in a perfusion chamber containing ACSF. SEP and DsRed were excited at 910 nm with a Ti:sapphire laser (Coherent). Green and red fluorescence signals were separated by a set of dichroic mirrors and filters (Chroma). Both epifluorescence and transfluorescence signals were collected by photomultiplier tubes (PMTs), and they were summed. Individual spines were photobleached by scanning a single plane 50 times with higher intensity of the laser power, which took 0.5 s. Electrophysiology Whole-cell voltage-clamp recordings were obtained from L2/3 pyramidal cells expressing SEP-GluR1 (homomeric receptors) or SEP-GluR1 and untaggedGluR2(edited) (heteromeric receptors) for 4–6 days. Patch recording pipettes (3–6 MU) were filled with internal solution containing 115 mM Cs-methanesulfonate, 20 mM CsCl, 10 mM HEPES, 2.5 mM MgCl2, 4 mM Na2ATP, 0.4 mM Na3GTP, 10 mM Na-phosphocreatine, 0.6 mM EGTA (pH 7.2), and 0.1 mM Spermine (Sigma-Aldrich). A total of 2.5 mM MNI-caged-L-glutamate (Tocris), 1 mM tetrodotoxin (Ascent Scientific), and 0.1 mM APV (Tocris) was added to ACSF, and recordings were obtained at 30 C. Glutamate uncaging-evoked AMPA receptor-mediated postsynaptic currents were measured at individual spines located in basal dendrites in response to test stimuli (1 ms, 0.05 Hz) at
60mV and +40mV holding potentials (5–20 sweeps averaged). The intensity of the uncaging laser (Ti:sapphire laser tuned at 720 nm) was controlled with electro-optical modulators (Pockels cells; Conoptics). Data Analysis SEP and DsRed fluorescence in spines and dendrites was measured as integrated green and red fluorescence, respectively, after background and leak subtraction. To measure the density of spine surface AMPA receptors as an enrichment value, spine SEP fluorescence was normalized to: ð4  pÞ1=3 ð3  RSpine Þ2=3 ; where RSpine represents spine DsRed fluorescence (i.e., spine volume was converted to spine area assuming that spine heads are spherical). To compare across different cells, these values were then normalized to the fluorescence signal of common dendritic regions. Thus, spine enrichment values were calculated as: ),( ) ( GSpine GDendrite ; 1=3 2=3 1=3 2=3 ð4  pÞ ð3  RSpine Þ ð4  pÞ ð3  RDendrite Þ where GSpine and GDendrite represent spine and dendrite SEP fluorescence, respectively, and RDendrite dendrite DsRed fluorescence. Fluorescence recovery of spine SEP was measured at +25 and +30 min after photobleaching and compared to baseline fluorescence obtained at
10 and
5 min prior to photobleaching, and averaged. Immobility of AMPA receptors was calculated as: immobility = 1
fluorescence recovery. To measure autocorrelation functions, two factors were considered: fluctuations in spine enrichment values independent of distance-dependent changes and the distance-dependent changes in spine enrichment values. The fluctuations were obtained by subtracting regression lines (linear component) fitted for each dendrite as a function of spine lag. This allowed us to measure autocorrelation functions without contributions from the distancedependent changes we observed (Figure S3C). Autocorrelation coefficients of spine SEP enrichment were then calculated for each dendrite by the following equation, averaged across dendrites, and normalized so that the correlation coefficients at zero lag corresponds to 1.0.

CðmÞ =

N
m
1 X n=0

xn


N
1 1 X xi N i=0

! xðn + mÞ


! N
1 1 X xi : N i=0

Here, xn represents the spine SEP enrichment values at the nth spine, with N the number of spines for the dendrite, and m, the spine lag. The regression lines for spine enrichment values were used to examine distance-dependent changes.

Neuron 72, 1001–1011, December 22, 2011 ª2011 Elsevier Inc. 1009

Neuron Compartmentalized and Global Synaptic Plasticity

Simulation To determine the number of synapses in clusters that would produce the autocorrelation values we obtained, we performed simulations (depicted in Figure S6). The following procedure was conducted to generate dendrites with simulated enrichment values satisfying different cluster distributions. We considered a series of 40 spines per dendritic segment and assigned an initial enrichment value to each spine that varied randomly from 0 to 1. On top of these values, a cluster of enrichment-potentiated spines was added. Two cluster parameters were varied: cluster size and potentiation value of enrichment. A cluster size was characterized by Gaussian-distributed enrichment potentiation values along a dendrite with SD s = 0.8, 1.2, and 1.6. Enrichment potentiation p varied from 2 to 5.5. To simulate a dendrite with s cluster size and potentiation factor p, a Gaussian distribution with SD = s and maximum value p was multiplied by a random number (between 0 and 1) at each spine lag and added at a random location within the initial 40 spine enrichment values. By calculating an autocorrelation coefficient for each dendrite and repeating the same procedure 10,000 times, we derived an average autocorrelation curve for each parameter combination. By fitting the true data with the simulated data, we determined s and the potentiation factor p (i.e., the number of potentiated synapses) in the cluster. Simulations were carried out using MATLAB (MathWorks).

Frick, A., Magee, J., and Johnston, D. (2004). LTP is accompanied by an enhanced local excitability of pyramidal neuron dendrites. Nat. Neurosci. 7, 126–135.

SUPPLEMENTAL INFORMATION

Hayashi, Y., Shi, S.H., Esteban, J.A., Piccini, A., Poncer, J.C., and Malinow, R. (2000). Driving AMPA receptors into synapses by LTP and CaMKII: requirement for GluR1 and PDZ domain interaction. Science 287, 2262–2267.

Supplemental Information includes six figures and can be found with this article online at doi:10.1016/j.neuron.2011.09.036.

ACKNOWLEDGMENTS We thank C. Cepko for pCALNL-DsRed (Addgene 13769) and pCAG-ERT2 CreERT2 (Addgene 13777), W. Guo for cloning the DNA constructs, D. Bortone for technical advice, and J. Isaacson, T. Komiyama and M. Scanziani for critical comments on the manuscript. This study was supported by National Institutes of Health (to R.M.) and Elizabeth-Sloan Livingston Fellowship (to H.M.). Accepted: September 28, 2011 Published: December 21, 2011 REFERENCES Alvarez, V.A., and Sabatini, B.L. (2007). Anatomical and physiological plasticity of dendritic spines. Annu. Rev. Neurosci. 30, 79–97. Bender, K.J., Rangel, J., and Feldman, D.E. (2003). Development of columnar topography in the excitatory layer 4 to layer 2/3 projection in rat barrel cortex. J. Neurosci. 23, 8759–8770. Branco, T., and Ha¨usser, M. (2010). The single dendritic branch as a fundamental functional unit in the nervous system. Curr. Opin. Neurobiol. 20, 494–502. Buonomano, D.V., and Merzenich, M.M. (1998). Cortical plasticity: from synapses to maps. Annu. Rev. Neurosci. 21, 149–186. Chen, X., Leischner, U., Rochefort, N.L., Nelken, I., and Konnerth, A. (2011). Functional mapping of single spines in cortical neurons in vivo. Nature 475, 501–505. Clem, R.L., and Barth, A. (2006). Pathway-specific trafficking of native AMPARs by in vivo experience. Neuron 49, 663–670. De Roo, M., Klauser, P., and Muller, D. (2008). LTP promotes a selective longterm stabilization and clustering of dendritic spines. PLoS Biol. 6, e219. Feldman, D.E., and Brecht, M. (2005). Map plasticity in somatosensory cortex. Science 310, 810–815. Feldmeyer, D., Lu¨bke, J., Silver, R.A., and Sakmann, B. (2002). Synaptic connections between layer 4 spiny neurone-layer 2/3 pyramidal cell pairs in juvenile rat barrel cortex: physiology and anatomy of interlaminar signalling within a cortical column. J. Physiol. 538, 803–822.

1010 Neuron 72, 1001–1011, December 22, 2011 ª2011 Elsevier Inc.

Gainey, M.A., Hurvitz-Wolff, J.R., Lambo, M.E., and Turrigiano, G.G. (2009). Synaptic scaling requires the GluR2 subunit of the AMPA receptor. J. Neurosci. 29, 6479–6489. Golding, N.L., Staff, N.P., and Spruston, N. (2002). Dendritic spikes as a mechanism for cooperative long-term potentiation. Nature 418, 326–331. Govindarajan, A., Kelleher, R.J., and Tonegawa, S. (2006). A clustered plasticity model of long-term memory engrams. Nat. Rev. Neurosci. 7, 575–583. Govindarajan, A., Israely, I., Huang, S.Y., and Tonegawa, S. (2011). The dendritic branch is the preferred integrative unit for protein synthesis-dependent LTP. Neuron 69, 132–146. Harvey, C.D., and Svoboda, K. (2007). Locally dynamic synaptic learning rules in pyramidal neuron dendrites. Nature 450, 1195–1200. Harvey, C.D., Yasuda, R., Zhong, H., and Svoboda, K. (2008). The spread of Ras activity triggered by activation of a single dendritic spine. Science 321, 136–140. Ha¨usser, M., and Mel, B. (2003). Dendrites: bug or feature? Curr. Opin. Neurobiol. 13, 372–383.

Heine, M., Groc, L., Frischknecht, R., Be´ı¨que, J.C., Lounis, B., Rumbaugh, G., Huganir, R.L., Cognet, L., and Choquet, D. (2008). Surface mobility of postsynaptic AMPARs tunes synaptic transmission. Science 320, 201–205. Hu, H., Real, E., Takamiya, K., Kang, M.G., Ledoux, J., Huganir, R.L., and Malinow, R. (2007). Emotion enhances learning via norepinephrine regulation of AMPA-receptor trafficking. Cell 131, 160–173. Iannella, N., and Tanaka, S. (2006). Synaptic efficacy cluster formation across the dendrite via STDP. Neurosci. Lett. 403, 24–29. Jia, H., Rochefort, N.L., Chen, X., and Konnerth, A. (2010). Dendritic organization of sensory input to cortical neurons in vivo. Nature 464, 1307–1312. Jia, Z., Agopyan, N., Miu, P., Xiong, Z., Henderson, J., Gerlai, R., Taverna, F.A., Velumian, A., MacDonald, J., Carlen, P., et al. (1996). Enhanced LTP in mice deficient in the AMPA receptor GluR2. Neuron 17, 945–956. Kessels, H.W., and Malinow, R. (2009). Synaptic AMPA receptor plasticity and behavior. Neuron 61, 340–350. Kessels, H.W., Kopec, C., Klein, M., and Malinow, R. (2009). Roles of stargazin and phosphorylation in control of AMPA receptor subcellular distribution. Nat. Neurosci. 12, 888–896. Koch, C., and Segev, I. (2000). The role of single neurons in information processing. Nat. Neurosci. Suppl. 3, 1171–1177. Larkum, M.E., and Nevian, T. (2008). Synaptic clustering by dendritic signalling mechanisms. Curr. Opin. Neurobiol. 18, 321–331. Lee, H.K., Takamiya, K., Han, J.S., Man, H., Kim, C.H., Rumbaugh, G., Yu, S., Ding, L., He, C., Petralia, R.S., et al. (2003). Phosphorylation of the AMPA receptor GluR1 subunit is required for synaptic plasticity and retention of spatial memory. Cell 112, 631–643. Lefort, S., Tomm, C., Floyd Sarria, J.C., and Petersen, C.C. (2009). The excitatory neuronal network of the C2 barrel column in mouse primary somatosensory cortex. Neuron 61, 301–316. Lin, D.T., Makino, Y., Sharma, K., Hayashi, T., Neve, R., Takamiya, K., and Huganir, R.L. (2009). Regulation of AMPA receptor extrasynaptic insertion by 4.1N, phosphorylation and palmitoylation. Nat. Neurosci. 12, 879–887. Losonczy, A., Makara, J.K., and Magee, J.C. (2008). Compartmentalized dendritic plasticity and input feature storage in neurons. Nature 452, 436–441. Makara, J.K., Losonczy, A., Wen, Q., and Magee, J.C. (2009). Experiencedependent compartmentalized dendritic plasticity in rat hippocampal CA1 pyramidal neurons. Nat. Neurosci. 12, 1485–1487.

Neuron Compartmentalized and Global Synaptic Plasticity

Makino, H., and Malinow, R. (2009). AMPA receptor incorporation into synapses during LTP: the role of lateral movement and exocytosis. Neuron 64, 381–390. Malenka, R.C., and Bear, M.F. (2004). LTP and LTD: an embarrassment of riches. Neuron 44, 5–21. Matsuda, T., and Cepko, C.L. (2007). Controlled expression of transgenes introduced by in vivo electroporation. Proc. Natl. Acad. Sci. USA 104, 1027– 1032. Newpher, T.M., and Ehlers, M.D. (2008). Glutamate receptor dynamics in dendritic microdomains. Neuron 58, 472–497. Nicoll, R.A., Tomita, S., and Bredt, D.S. (2006). Auxiliary subunits assist AMPAtype glutamate receptors. Science 311, 1253–1256. Patterson, M.A., Szatmari, E.M., and Yasuda, R. (2010). AMPA receptors are exocytosed in stimulated spines and adjacent dendrites in a Ras-ERK-dependent manner during long-term potentiation. Proc. Natl. Acad. Sci. USA 107, 15951–15956.

Schiller, J., Major, G., Koester, H.J., and Schiller, Y. (2000). NMDA spikes in basal dendrites of cortical pyramidal neurons. Nature 404, 285–289. Stern, E.A., Maravall, M., and Svoboda, K. (2001). Rapid development and plasticity of layer 2/3 maps in rat barrel cortex in vivo. Neuron 31, 305–315. Takahashi, T., Svoboda, K., and Malinow, R. (2003). Experience strengthening transmission by driving AMPA receptors into synapses. Science 299, 1585– 1588. Turrigiano, G.G. (2008). The self-tuning neuron: synaptic scaling of excitatory synapses. Cell 135, 422–435. Turrigiano, G.G., and Nelson, S.B. (2004). Homeostatic plasticity in the developing nervous system. Nat. Rev. Neurosci. 5, 97–107. Turrigiano, G.G., Leslie, K.R., Desai, N.S., Rutherford, L.C., and Nelson, S.B. (1998). Activity-dependent scaling of quantal amplitude in neocortical neurons. Nature 391, 892–896.

Petreanu, L., Mao, T., Sternson, S.M., and Svoboda, K. (2009). The subcellular organization of neocortical excitatory connections. Nature 457, 1142–1145.

Wei, D.S., Mei, Y.A., Bagal, A., Kao, J.P., Thompson, S.M., and Tang, C.M. (2001). Compartmentalized and binary behavior of terminal dendrites in hippocampal pyramidal neurons. Science 293, 2272–2275.

Petrini, E.M., Lu, J., Cognet, L., Lounis, B., Ehlers, M.D., and Choquet, D. (2009). Endocytic trafficking and recycling maintain a pool of mobile surface AMPA receptors required for synaptic potentiation. Neuron 63, 92–105.

Wenthold, R.J., Petralia, R.S., Blahos J, I.I., and Niedzielski, A.S. (1996). Evidence for multiple AMPA receptor complexes in hippocampal CA1/CA2 neurons. J. Neurosci. 16, 1982–1989.

Poirazi, P., and Mel, B.W. (2001). Impact of active dendrites and structural plasticity on the memory capacity of neural tissue. Neuron 29, 779–796.

Zamanillo, D., Sprengel, R., Hvalby, O., Jensen, V., Burnashev, N., Rozov, A., Kaiser, K.M., Ko¨ster, H.J., Borchardt, T., Worley, P., et al. (1999). Importance of AMPA receptors for hippocampal synaptic plasticity but not for spatial learning. Science 284, 1805–1811.

Polsky, A., Mel, B.W., and Schiller, J. (2004). Computational subunits in thin dendrites of pyramidal cells. Nat. Neurosci. 7, 621–627.

Neuron 72, 1001–1011, December 22, 2011 ª2011 Elsevier Inc. 1011