calcium transient prevalence across the dendritic arbour...

LETTERdoi:10.1038/nature13871

Calcium transient prevalence across the dendriticarbour predicts place field propertiesMark E. J. Sheffield1 & Daniel A. Dombeck1

Establishing the hippocampal cellular ensemble that represents ananimal’s environment involves the emergence and disappearance ofplace fields in specific CA1 pyramidal neurons1–4, and the acquisitionof different spatial firing properties across the active population5.While such firing flexibility and diversity have been linked to spatialmemory, attention and task performance6,7, the cellular and networkorigin of these place cell features is unknown. Basic integrate-and-fire models of place firing propose that such features result solely fromvarying inputs to place cells8,9, but recent studies3,10 suggest insteadthat place cells themselves may play an active role through regener-ative dendritic events. However, owing to the difficulty of performingfunctional recordings from place cell dendrites, no direct evidenceof regenerative dendritic events exists, leaving any possible connec-tion to place coding unknown. Using multi-plane two-photon cal-cium imaging of CA1 place cell somata, axons and dendrites in micenavigating a virtual environment, here we show that regenerative den-dritic events do exist in place cells of behaving mice, and, surprisingly,their prevalence throughout the arbour is highly spatiotemporallyvariable. Furthermore, we show that the prevalence of such eventspredicts the spatial precision and persistence or disappearance ofplace fields. This suggests that the dynamics of spiking throughoutthe dendritic arbour may play a key role in forming the hippocam-pal representation of space.

CA1 pyramidal cell dendrites contain voltage-gated calcium andsodium channels along with NMDA (N-methyl-D-aspartate) receptorsthat allow them to produce nonlinear, regenerative (spiking) events. Thespatial extent and site of generation of dendritic regenerative events canvary from widespread back-propagation of somatic action potentials(bAPs) into the arbour11,12 and multi-dendrite calcium spikes10, to morespatially heterogeneous processes such as partial bAP propagation13,14

and local spike generation15–17. Such events can provide amplificationof synaptic input15–17 and the depolarization necessary for Hebbian plas-ticity induction13,18,19, both of which may be important for place fieldfiring20. However, no measurements of regenerative dendritic activityin place cells have been made during behaviour, when network statesaffecting dendritic excitability are intact and relevant.

To study regenerative dendritic activity in the hippocampus duringbehaviour we co-acquired time-series movies through a chronic imagingwindow of calcium transients from dendrites, axons and somata of CA1place cells sparsely labelled with a genetically-encoded calcium indicator21

(GCaMP6f) while head-restrained mice navigated a virtual linear track1,22

(Fig. 1a, b). One imaging plane was focused on the soma while the otherwas focused in the dendritic arbour, slicing through several branches.Many of the labelled neurons were identified as place cells by somaticcalcium transients repeatedly occurring during traversals of the sametrack location (place field significance P , 0.05 from bootstrapping).Unless otherwise stated, our analysis focused only on these cells (33 placefields, 28 place cells, 19.3 6 13.2 min per place cell imaging session,8 mice), their basal arbours (170 total branches), their axon (visible in4 place cells, 5 place fields), and on activity observed during place fieldtraversals. The dendritic fields of view (,145 3 75mm) on average con-tained 5 6 3.5 (range, 2–18) basal arbour branches connected to the

co-imaged place cell soma. The imaged branch sections had a mean lengthof 10 6 3mm (range, 3–23mm), were positioned a mean of 74 6 15%(range, 38– 99%) of the distance along the dendritic length (from somato dendrite tips), a mean of 2.8 6 0.9 branch points (range, 1– 6) and130 6 44mm (range, 58–284mm) from the soma, and a mean of 3.8 6 2.0branch points (range, 1–9) and 191 6 83mm (range, 20–440mm) fromeach other. Our recordings of ,5 basal branches would typically represent

1Department of Neurobiology, Northwestern University, Evanston, Illinois 60208, USA.

a

c

Frame 1

(0 ms)

Frame 2

(33 ms)

Frame 3

(66 ms)

Frame 4

(99 ms)

Focal plane 1 Focal plane 2 Focal plane 1 Focal plane 2Continued

Electriclens

Soma

Focal plane 1

1

2

3

Axon

Focal plane 2

20 μm

180 cm

Right

Left

Topxx

x

xx x

xx x

Branch 1:

199 μm/99%

Soma

Axon: 158 μm

10 s

1,000% ΔF/F

Trackposition

0 cm

180 cm

* *

*

* *

* *

* *

* * * * * * * *

** *

*

*

* *

* *

* *

* *

Branch 2:

192 μm/97%

Branch 3:

152 μm/90%

b

Figure 1 | Co-acquired time-series of CA1 place cell somata, dendrites andaxons during virtual navigation. a, Two-photon microscopy with anelectric lens (top) rapidly switches between two focal planes to generateco-acquired images of soma, dendrites and axon (bottom). b, Left, expandedview of place cell shown in a. The time-series was acquired while the mousenavigated a virtual linear track (right). c, Top, mouse position along lineartrack; bottom, fluorescence change over baseline (DF/F) traces from the soma(red), axon (brown) and dendrites (green, blue, cyan) of place cell shown in aduring place field traversals (grey columns). Note the absence of detectablebranch spikes during some somatic firing. Branch distance to soma and percent distance from soma to dendrite tip shown for each branch. *P , 0.001from bootstrapping.

0 0 M O N T H 2 0 1 4 | V O L 0 0 0 | N A T U R E | 1

Macmillan Publishers Limited. All rights reserved©2014

www.nature.com/doifinder/10.1038/nature13871

about one-third of all basal dendritic branches at their branching depth(,2.8 branch points from the soma).

During track traversals we often found significant calcium transients(detectable transients with ,0.1% false positive error rates, indicated withan asterisk in figures; Methods) in the soma, axon (mean of 139 6 26mmfrom soma) and basal dendrites in the cell’s somatic place field (Fig. 1b, c;activity in these structures along the track but outside of the place fieldwas rarely observed). Somatic calcium transients (Extended Data Fig. 1)were used as a surrogate measure of action potential firing1,21, and mul-tiple lines of evidence (Methods) showed that somatic action potentialfiring occurred throughout nearly the entire somatic calcium-transient-defined place field. In cells in which we were able to record from theaxon along with the soma and dendrites, we found that axonal tran-sients co-occurred with somatic transients 100% of the time (130 tran-sients; n 5 4 place cells; Supplementary Video 1). Because axonal calciumtransients are tightly coupled to somatic firing, they provided an inde-pendent indicator of action potential firing (Methods). In dendrites ofCA1 pyramidal neurons of navigating mice (not necessarily place cells),we observed calcium transients restricted to single spines, with no detect-able shaft transient; however, we more frequently observed calcium tran-sients invading all visible pixels of the recorded branch (both shaft andspines) that were larger in amplitude and spatial extent than single-spinetransients. These transients were considered non-regenerative (excit-atory input to a single spine) or regenerative (bAPs or dendriticallygenerated spikes (dspikes) such as Na1, NMDA and calcium spikes)depolarizations, respectively (Methods and Extended Data Fig. 2d, e),and we focused our analysis on the latter.

Here, we refer to these regenerative dendritic events as ‘branch spikes’— events caused by either dspikes or bAPs or both. Furthermore, becausesingle-spine transients are detectable under our recording conditions,we presume that our measurements are sensitive enough to detect mostbranch spiking, and therefore the absence of branch spiking addition-ally indicates that dspikes and bAPs probably did not occur.

The ability to measure spiking in both the soma and in multiple den-drites of CA1 neurons during navigation allowed us to investigate spe-cifically whether somatic place field firing was associated with branchspiking. During many place field traversals with somatic firing, dendriticbranch spikes were observed (often with onset latencies with respect tosomatic firing, Fig. 2 and Extended Data Fig. 1d), providing the first directevidence of their existence in place cells. Furthermore, from traversal-to-traversal, branch spiking was found to be highly spatially variable acrossthe arbour. For example, in many place field traversals, all observed den-dritic branches in our imaging field (representing a subset of all the cell’sbranches) spiked along with the soma (Fig. 2a and Extended Data Figs 5and 6). However, in some cases, only a subset of the observed branchesdisplayed detectable spikes (Figs 1c, 2b and Extended Data Figs 3 and 4a).Finally, in many cases, none of the observed branches displayed detect-able spikes during place field traversals while the soma (and axon) fired(Fig. 2c and Extended Data Figs 3–6). From 747 somatic place field tran-sients (from all 28 place cells), 395 (52.9%) showed co-occurring branchspikes in all imaged branches while 154 (20.6%) showed at least onebranch with and one branch without detectable spiking, and 198 somatictransients (26.5%) failed to show co-occurring detectable spikes in anyof the imaged branches. Importantly, these three observation classes wereoften seen in the same place cell, but on different place field traversals(Fig. 2e, Extended Data Figs 3 and 4 and Supplementary Video 1).

Characterizing somato-dendritic events as one of these three observa-tion classes did not depend on the recording distance of the dendritic imag-ing plane from the soma or the number of observed dendritic branches(Supplementary Information and Extended Data Fig. 5). Thus, record-ing from a subset of branches in a single dendritic imaging plane pro-vides a reasonably accurate characterization of dendritic branch spikingthroughout a large portion of the basal (and apical) arbour (Supplemen-tary Information and Extended Data Figs 5 and 6), implying some levelof cooperativity between the spiking in different branches (ExtendedData Fig. 7).

A fourth class of observation was also made: branch spiking in theabsence of detectable somatic spiking (Fig. 2d and Extended Data Fig. 4b),representing probable evidence of dspikes in place cells. These localizeddspikes were rarely observed in the absence of somatic spiking (6 totalin or around the mean somatic place field, during 8.3 recording hours; 2other dspikes were observed outside of place fields or in non-place cells;Extended Data Fig. 2a–c, f). However, dspikes could contribute to moreglobal branch spiking, making isolated dspikes difficult to detect.

Place cells exhibited different degrees and patterns of branch-spikingheterogeneity. As a measure of this heterogeneity we calculated the prev-alence of branch spiking (fraction of branches with detectable branchspikes) during each place field traversal, defined as branch-spike prev-alence (BSP, see Methods for calculation). Figure 3 shows a place fieldwith low average BSP (the average BSP across all place field traversals,Fig. 3a), one with moderate average BSP (Fig. 3b), and one with consis-tent somato-dendritic unison (high average BSP, Fig. 3c). Further, weobserved that average BSP could differ significantly between differentplace fields of the same cell and that in-field average BSP was significantlygreater than out-of-field BSP (Supplementary Information, ExtendedData Fig. 8). Together, these results demonstrate that average BSP variesbetween different place cells and place fields and is not solely determinedby cellular properties, such as the general degree of excitability, but isalso synaptic-input specific.

e

250% ΔF/F2.5 s

Trackposition

180 cm

0 cm

Soma

*

*

*

*

*

*

*

*

*

1

13

2

156/93

134/71

71/43

213/

202/

246/

210/90

180 cm

0 cm

Soma 2 s

a

*

**

**

* *

*

*

*

*

*

*

*

*

Branch 1:

206 μm/86%

Branch 2:

284 μm/91%

Branch 3:

204 μm/81%

Branch 1: 143 μm/67%




Latency

97

97

88

500% ΔF/F

143/67

149/77

115/65

80/56

68/47

83/78

12

3

4

5

23

4

41

2

3

20 μm

41

2

3

μm/%

b c d

Figure 2 | Variability of dendritic branch spiking during somatic place fieldfiring. a–d, Top, co-acquired images of place cell soma and dendrites (a–c aredifferent place cells from three mice, a and d are the same cell). Bottom,DF/F traces from the (co-acquired) soma and numbered dendritic branchesduring a place field traversal (grey columns). Co-occurring somatic firing anddetectable branch spiking was observed in all (a), some (b) or none (c) ofthe imaged dendrites. Branch spiking in the absence of detectable somatic firingwas also observed (d). e, Variability in branch spiking was often observed inthe same place cell (top) during different place field traversals of the samesession. Branch spikes also showed variable onset times with respect to somaticfiring onset (black arrows). *P , 0.001 from bootstrapping.

RESEARCH LETTER

2 | N A T U R E | V O L 0 0 0 | 0 0 M O N T H 2 0 1 4


The above observations demonstrate spatiotemporal variability in theregenerative dendritic events that are widely believed to provide ampli-fication of synaptic input and the depolarization necessary for Hebbianplasticity (defining plasticity ‘windows’). Many possible mechanismscould generate our observed patterns of somatic action potential firingand branch spiking: (1) bAP-induced branch spiking could be spatiallyand temporally modulated23 by inhibition, synaptic boosting or sodiumchannel inactivation; (2) clustered synaptic input could generate branchspiking24,25 in the form of widespread calcium spikes and/or more localdspikes10,25,26; or (3) dspikes and bAPs could co-occur. However, regard-less of spike initiation site(s), the known involvement of branch spikingin plasticity and input amplification suggests that differences betweenplace cells and their firing fields may be related to BSP.

To investigate the relationship between branch spiking and placefield firing, we looked at two features of place fields previously linked toNMDA-receptor mediated plasticity: precision and stability. Place fieldspatial precision refers to a place cell’s somatic spatial firing consistencyover many place field traversals, and has been shown to be related tospatial memory task performance6,7. Long-term place field stability, whichrefers to the persistence of a place field over days, is thought to repre-sent the storage of spatial memories4.

We found a significant correlation between a place field’s average BSPand its spatial precision index (defined in Methods; Fig. 4a–c; Spearman’srank correlation coefficient: P 5 0.0038; r 5 0.6442; significant positivelinear slope within 95% confidence bounds; 26 place fields). The ninefields with the largest average BSP were also the most precise, indicat-ing that average BSP can be used to predict precision. Notably, when wecompared a place field’s somatic firing intensity (integral of somatictransients, Methods) to its spatial precision, little to no correlation wasobserved (Fig. 4c; Spearman’s rank correlation coefficient: P 5 0.01;r 5 0.29; linear slope not significantly positive). Thus average BSP, butnot somatic firing intensity, is a predictor of a place field’s spatial precision.

We next assessed the relationship between average BSP and placefield stability (Fig. 4d–f) by monitoring somatic and dendritic activityin the same cells over the course of 2 days. Place field average BSP wasmeasured on day 1 and then the same cell was imaged the next day todetermine whether it had the same somatic place field (Fig. 4d). Weclassified place fields that persisted to day 2 as stable place fields andthose that disappeared by day 2 as transitory place fields (Fig. 4e). Stablefields had significantly greater average BSP than transitory fields (Fig. 4f;0.83 6 0.06 versus 0.37 6 0.07; t-test, P 5 0.0006). When the somatic

firing intensity of transitory and stable place fields was compared no signi-ficant difference was observed (Fig. 4f; 0.3746 0.036 versus 0.3686 0.057;t-test, P 5 0.92). This indicates that average BSP, but not somatic firing

Place field traversal

Branch spiking

No branch spiking

Somatic spiking

0.6 0.8 1.00.40.200

6

Average BSP = 0.3

BSP

No

. o

f events

SomaB1

B8

5 10 15 20

5 10 15 20

B11

B1Soma 0.6 0.8 1.00.40.20

0

6

a

b

c

0

35

SomaB1

B7

5 10 15 20 25 30 35 0.6 0.8 1.00.40.20

Average BSP = 0.56

Average BSP = 0.95

No

. o

f events

No

. o

f events

Figure 3 | Place field branch-spiking heterogeneity. a–c, Coloured plots(left; three different place cells from two mice) show occurrence of detectablespiking in each branch (blue or black; different branches indicated by ‘B’followed by the branch number) during somatic place field firing (red). Middle,histograms of BSP on each traversal. Cartoons (far right) do not representreal data.

Tra

vers

al n

o.

1

5

10

COM

Cell with high average

BSP

0

0.4

0.8

P = 0.0006

Day 1

avera

ge B

SP

P = 0.92

Day 1

so

matic fi

ring

inte

nsity

Cell with high average BSPon day 1

Mean

ΔF/F200/20%

Day 1

Day 2

Day 1

Day 2

B1

B4

Soma5 10 15

Cell with low average BSPon day 1

5 10 15 20

B5

B1

Soma



Day 1 Day 2

Same cell

Stable somatic

place field

Transitory somatic

place field

Plane 1Plane 2

Plane 1

Plane 2

Stable somatic place field

Transitory somatic place field

Precise somatic

place field

a

5 10

B1

B12

Soma

Single day

Branch spiking

No branch spiking

Somatic spiking

Track position

0 cm 180 cm

0

0.4

0.8

% ΔF/F

0

1600

Cell with low average BSP

B1

B4

Soma

Imprecise somaticplace field

1

5

10

Track position 0 cm 180 cm

5 10

0.06

0.12

0.18

0 0.5 10

Average BSP Somatic firing intensity

0

0.1

0.2

0.3

0.5 10

Pla

ce fi

eld

sp

atial

pre

cis

ion ind

ex (cm

–1)

P = 0.0038 P = 0.01

NS

*

b

c

d e

f

Figure 4 | Dendritic BSP predicts place field spatial precision and long-termstability. a, Schematic of imaging planes. b, Example cells with high (top left)and low (bottom left) average BSP fields; right, somatic DF/F versus tracklocation raster plots with each traversal’s transient centre of mass (COM) (blackcircles). c, Left, plot of precision versus average BSP for each place field (n 5 26place fields from n 5 8 mice); right, plot of precision versus somatic firing(includes single-cell and population imaging experiments; n 5 74 place fieldsfrom n 5 11 mice). Arrows indicate examples from b. NS, not significant.P values from Spearman’s rank correlation test; linear fit shown in red; asteriskindicates a significantly positive linear slope. d, Schematic of imaging planeson days 1 and 2. e, Example cells with high (top left) and low (bottom left)day 1 average BSP; right, associated place fields on days 1 and 2 (scale barrepresents 200% and 20% DF/F for top and bottom traces, respectively).f, Average BSP (left) and somatic firing intensity (SFI; right, includes single-celland population imaging experiments) for stable (n 5 8 place fields from n 5 3mice for average BSP group; n 5 21 place fields from n 5 6 mice for SFIgroup) and transitory (n 5 5 place fields from n 5 2 mice for average BSPgroup; n 5 40 place fields from n 5 5 mice for SFI group) place fields; error barsrepresent s.e.m.; P values from Student’s unpaired t-test. Black arrows in findicate the example cells shown in e.

LETTER RESEARCH

0 0 M O N T H 2 0 1 4 | V O L 0 0 0 | N A T U R E | 3


intensity, is a predictor of the long-term stability of a place cell’s placefield. Together, the above results indicate that place field spatial pre-cision, persistence and disappearance can be predicted based on theprevalence of regenerative events in place cell dendrites, but not basedon somatic firing alone (precision and stability were also positively cor-related, Supplementary Information and Extended Data Fig. 9).

Branch spikes can act to amplify synaptic input (through dspikes15–17)to drive somatic firing and can also provide the post-synaptic signalrequired for Hebbian plasticity13,18,19. In this context, our observationthat BSP is variable implies that spatially heterogeneous and tempor-ally variable input amplification and synaptic plasticity have a role inplace field precision and stability. Somewhat paradoxically, however,increased branch spiking did not strongly correlate with an increasedaverage number of somatic action potentials during place field traversals(Extended Data Fig. 10), and fields associated with more ‘open plasticitywindows’ (more branch spiking possibly associated with greater plas-ticity) were actually more stable and precise.

It is somewhat puzzling then how synaptic plasticity and amplifica-tion mechanisms might be used by place cells to develop their represen-tations of space. A ‘winner-takes-all’ scenario incorporating synapticHebbian potentiation and somatic firing homeostasis provides a possi-ble answer. Previous models27,28 examined spatially imprecise place fieldsgenerated by different combinations of synaptic input, each with differ-ent spatial tuning. The combinations which happened to overlap more(in space and time) generated greater post-synaptic responses and moreHebbian potentiation of the contributing inputs. This increased theeffectiveness of the potentiated inputs to drive somatic firing relative toother inputs with non-overlapping spatial tuning. A repetition of thisprocess caused the cell to respond mainly to the subset of strong inputswith spatial tuning overlap, leading to more precise place field firing.The increased effective drive to the cell caused by the potentiated inputsthen triggered a homeostatic mechanism that lowered overall cellularexcitability (through synaptic or intrinsic excitability renormalization)29.This increased the contribution from the strong inputs (in relation to theweak inputs) in driving somatic firing, thus further increasing spatialprecision, but maintaining the average firing rate of the place cell.

Assuming that branch spikes represent the postsynaptic responsesthat signal Hebbian potentiation, our observation of increased spatialprecision of somatic firing with greater average BSP fits with the abovemodel. Also, the sets of surviving, stronger synapses in the model wouldbe expected to persist longer in time, consistent with our observationof greater place field stability with increased average BSP. The use of asomatic action potential firing homeostasis mechanism also explainsthe weak relationship observed between BSP and action potential firing,with the implication that firing may be driven more by regenerative den-dritic activity in high average BSP fields, and less in low average BSPfields. Furthermore, given the above scenario, in silent cells, strong setsof clustered synapses formed through potentiation in different environ-ments may be partially activated in the animal’s current environment,but just below dspike threshold and therefore sensitive to small changesin somatic depolarization, as recently observed3.

In the context of synaptic plasticity (see Supplementary Informationfor further discussion), while our methods do not directly reveal if branchspikes induce long-term potentiation, long-term depression or no weightchange30, the above model implies that in place fields, branch spikingpotentiates synapses, is the result of previous potentiation, and/or main-tains the strength of previously potentiated synapses through positivefeedback. Our measurements demonstrate multiple levels of dissoci-ation between action potential firing and the mechanisms widely believedto signal dendritic plasticity, providing the first clues regarding the spatialand temporal scales at which associative Hebbian learning rules20 mayoperate in a behaving animal performing a task known to engage inte-gration and plasticity in the cells being studied. One implication is thatthe process by which place cells orchestrate plasticity signals (dendriticbranch spiking) may be variably, rather than consistently, linked to theirmode of information transmission (the somato-axonal action potential).

Online Content Methods, along with any additional Extended Data display itemsandSourceData, are available in the online version of the paper; references uniqueto these sections appear only in the online paper.

Received 29 January; accepted 18 September 2014.

Published online 26 October 2014.

1. Dombeck, D. A., Harvey, C. D., Tian, L., Looger, L. L. & Tank,D. W. Functional imagingof hippocampal place cells at cellular resolution during virtual navigation. NatureNeurosci. 13, 1433–1440 (2010).

2. Kentros, C. et al. Abolition of long-term stability of new hippocampal place cellmaps by NMDA receptor blockade. Science 280, 2121–2126 (1998).

3. Lee, D., Lin, B. J. & Lee, A. K. Hippocampal place fields emerge upon single-cellmanipulation of excitability during behavior. Science 337, 849–853 (2012).

4. Ziv, Y. et al. Long-term dynamics of CA1 hippocampal place codes. NatureNeurosci. 16, 264–266 (2013).

5. Mizuseki, K., Royer, S., Diba, K. & Buzsaki, G. Activity dynamics and behavioralcorrelates of CA3 and CA1 hippocampal pyramidal neurons. Hippocampus 22,1659–1680 (2012).

6. Kentros, C. G., Agnihotri, N. T., Streater, S., Hawkins, R. D. & Kandel, E. R. Increasedattention to spatial context increases both place field stability and spatial memory.Neuron 42, 283–295 (2004).

7. Rotenberg, A., Mayford, M., Hawkins, R. D., Kandel, E. R. & Muller, R. U. Miceexpressing activated CaMKII lack low frequency LTP and do not form stable placecells in the CA1 region of the hippocampus. Cell 87, 1351–1361 (1996).

8. Burgess, N. & O’Keefe, J. Neuronal computations underlying the firing of placecells and their role in navigation. Hippocampus 6, 749–762 (1996).

9. Solstad, T.,Moser, E. I. &Einevoll, G. T. Fromgridcells toplacecells: amathematicalmodel. Hippocampus 16, 1026–1031 (2006).

10. Grienberger, C., Chen, X. & Konnerth, A. NMDA receptor-dependent multidendriteCa21 spikes required for hippocampal burst firing in vivo. Neuron 81, 1274–1281(2014).

11. Hill, D. N., Varga, Z., Jia, H., Sakmann, B.& Konnerth, A. Multibranch activity in basaland tuft dendrites during firing of layer 5 cortical neurons in vivo. Proc. Natl Acad.Sci. USA 110, 13618–13623 (2013).

12. Zhou, W. L., Yan, P., Wuskell, J. P., Loew, L. M. & Antic, S. D. Dynamics of actionpotential backpropagation in basal dendrites of prefrontal cortical pyramidalneurons. Eur. J. Neurosci. 27, 923–936 (2008).

13. Magee, J. C. & Johnston,D. A synaptically controlled, associative signal for Hebbianplasticity in hippocampal neurons. Science 275, 209–213 (1997).

14. Spruston, N., Schiller, Y., Stuart, G. & Sakmann, B. Activity-dependent actionpotential invasion and calcium influx into hippocampal CA1 dendrites. Science268, 297–300 (1995).

15. Ariav, G., Polsky, A. & Schiller, J. Submillisecond precision of the input–outputtransformation function mediated by fast sodium dendritic spikes in basaldendrites of CA1 pyramidal neurons. J. Neurosci. 23, 7750–7758 (2003).

16. Gasparini, S., Migliore, M. & Magee, J. C. On the initiation and propagation ofdendritic spikes inCA1pyramidal neurons. J.Neurosci. 24, 11046–11056 (2004).

17. Losonczy, A. & Magee, J. C. Integrative properties of radial oblique dendrites inhippocampal CA1 pyramidal neurons. Neuron 50, 291–307 (2006).

18. Golding, N. L., Staff, N. P. & Spruston, N. Dendritic spikes as a mechanism forcooperative long-term potentiation. Nature 418, 326–331 (2002).

19. Schiller, J., Schiller, Y.& Clapham,D. E.NMDAreceptorsamplify calcium influx intodendritic spines during associative pre- and postsynaptic activation. NatureNeurosci. 1, 114–118 (1998).

20. Wu, X. E. & Mel, B. W. Capacity-enhancing synaptic learning rules in a medialtemporal lobe online learning model. Neuron 62, 31–41 (2009).

21. Chen, T. W. et al. Ultrasensitive fluorescent proteins for imaging neuronal activity.Nature 499, 295–300 (2013).

22. Harvey, C. D., Collman, F., Dombeck, D. A. & Tank, D. W. Intracellular dynamics ofhippocampal place cells during virtual navigation. Nature 461, 941–946 (2009).

23. Svoboda,K., Denk, W., Kleinfeld,D.& Tank, D.W. In vivodendritic calciumdynamicsin neocortical pyramidal neurons. Nature 385, 161–165 (1997).

24. Kamondi, A., Acsady, L.& Buzsaki,G.Dendritic spikes are enhanced bycooperativenetwork activity in the intact hippocampus. J. Neurosci. 18, 3919–3928 (1998).

25. Smith, S. L., Smith, I. T., Branco, T. &Hausser,M.Dendritic spikes enhance stimulusselectivity in cortical neurons in vivo. Nature 503, 115–120 (2013).

26. Palmer, L. M. et al. NMDA spikes enhance action potential generation duringsensory input. Nature Neurosci. 17, 383–390 (2014).

27. Barry, C. & Burgess, N. Learning in a geometric model of place cell firing.Hippocampus 17, 786–800 (2007).

28. McHugh, T. J., Blum, K. I., Tsien, J. Z., Tonegawa, S. & Wilson, M. A. Impairedhippocampal representation of space in CA1-specific NMDAR1 knockout mice.Cell 87, 1339–1349 (1996).

29. Miller, K. D. Synaptic economics: competition and cooperation in synapticplasticity. Neuron 17, 371–374 (1996).

30. Gordon, U., Polsky, A. & Schiller, J. Plasticity compartments in basal dendrites ofneocortical pyramidal neurons. J. Neurosci. 26, 12717–12726 (2006).

Supplementary Information is available in the online version of the paper.

Acknowledgements We are grateful to B. Mensh for discussions on dataconceptualization, interpretation and presentation. We thank C. Harvey, B. Mensh,N. Spruston, C. Woolley, W. Kath and L. Looger for comments on the manuscript, E. Hanand P. Boueri for technical assistance, and V. Jayaraman, R. Kerr, D. Kim, L. Looger,K.Svoboda fromtheGENIEProject (Janelia Farm,HowardHughesMedical Institute) for

RESEARCH LETTER

4 | N A T U R E | V O L 0 0 0 | 0 0 M O N T H 2 0 1 4




GCaMP6. This work was supported by The Klingenstein Foundation, The WhitehallFoundation, The Chicago Biomedical Consortium with support from the Searle Fundsat The Chicago Community Trust, Northwestern University, The National Institutes ofHealth (1R01MH101297), and M.S. is an Ellison Medical Foundation Fellow of the LifeSciences Research Foundation.

Author Contributions M.S. performed the experiments, D.D. built the experimentalapparatus, M.S. performed data analysis with strategy suggestions from D.D. Both

authors conceived and designed the experiments, interpreted the data and wrote thepaper.

Author Information Reprints and permissions information is available atwww.nature.com/reprints. The authors declare no competing financial interests.Readers are welcome to comment on the online version of the paper.Correspondence and requests for materials should be addressed toD.D. ([email protected]).

LETTER RESEARCH

0 0 M O N T H 2 0 1 4 | V O L 0 0 0 | N A T U R E | 5


www.nature.com/reprints


mailto:[email protected]

METHODSMouse surgery, virtual reality, and behaviour training. All experiments wereapproved by the Northwestern University Animal Care and Use Committee. MaleC57BL/6 mice (postnatal day ,70) were anaesthetized (,1–2% isoflurane) and asmall (,0.521.0 mm) craniotomy was made over the hippocampus (1.8 mm lateral,2.4 mm caudal of Bregma). For single-cell dendritic imaging a low titre Cre virus(AAV2/1–CamkII–Cre, 1.5 3 108 GC ml21, all virus from University of Pennsyl-vania Vector Core) was injected (1 injection of ,30 nl at a depth of ,1,250mm belowthe dura surface using a bevelled glass micropipette, ,1–2 MV after bevelling) incombination with a high titre of flexed–GCaMP6 virus (AAV1–Syn–flex–GCaMP6f,1.4 3 1013 GC ml21) leading to expression of GCaMP6f (ref. 21) in a sparse CA1pyramidal neuron population. For population imaging (Extended Data Fig. 9),dense labelling was performed the same as sparse except AAV1–Syn–GCaMP6f(1.5 3 1013 GC ml21) was injected. Mouse water scheduling began the next day(0.8–1.0 ml per day) followed either the next day or ,7 days later by a hippocampalwindow and head-plate implantation surgery (as described in ref. 1), for popula-tion or single-cell dendritic imaging respectively. Training in a 1.8-m virtual lineartrack (one ,40–60 min session per mouse per day) began ,7 days after windowimplantation and continued until mice routinely ran back and forth along the lineartrack to achieve a high reward rate (.,2 rewards per minute); rewards consistedof water (4ml) delivered as described previously1,22. Once this criterion was reached(,7–14 days of virtual reality training), place cell imaging commenced and wasrepeated daily for ,1–4 weeks. During the imaging sessions included in our ana-lysis, mice received an average of 4.4 6 1.1 rewards per min. GCaMP6f expressionreached a somewhat steady state level ,21 days post-injection.

Our virtual reality and spherical treadmill system were similar to those previ-ously described1,22,31 but with the following differences. A curved screen monitor(CRVD, Ostendo) was used for displaying the virtual reality environment. The screenwas adjusted for low light emission (brightness, 15/100; contrast, 50/100; all RGBintensities, 1/100) and was placed ,14 inches in front of the mouse covering 133uand 48u of the mouse’s horizontal and vertical field of view, respectively. The Quake2video game engine described previously22 was used here for the virtual simulation,with minor modifications: the virtual environment was rendered with 2,880 3 900pixels, and the horizontal field of view in the virtual environment was 105u. Mouselocomotion speed and direction on the spherical treadmill were read using a G-400optical computer mouse (Logitech) and forward and yaw movements of the tread-mill were used to update forward position and view angle in the virtual environ-ment, respectively, as described previously22. The forward movement gain was setsuch that the full length of the virtual track was traversed by ,2.8 rotations of theball (180 cm of linear distance) and the yaw gain was set such that ,7.5 rotations ofthe ball resulted in one full field of view rotation (360u) in the virtual environment.Two-photon imaging of place cell soma, axon and dendrites. We customized aMoveable Objective Microscope (Sutter Instruments) for our imaging experiments.The microscope consisted of water cooled 6215H galvos with 3-mm B1 coated mirrors(Cambridge technology), a 3 40/0.8 numerical aperture (NA) objective (LUMPlanFLN3 40/0.8 W, Olympus) and enhanced collection optics. Green GCaMP6f fluores-cence was routed to a GaAsP PMT (H10770PA-40) using a series of dichroic mirrorsand band-pass filters (in order after leaving the back aperture; Semrock): FF665-Di02long pass dichroic, FF01-680/sp. short pass filter, FF560-Di01 long pass dichroic,FF01-510/84 band-pass filter. Stray light from the virtual reality monitor was blockedusing a custom box surrounding the top of the microscope objective, the electric lensand the overlying dichroic mirror (not including the tube lens, scan lens, galvan-ometers or routeing mirrors). This box had one hole on top, for entry of the exci-tation beam, which was covered with a coloured glass filter (FGL780, Thorlabs)and one hole on bottom for the microscope objective. This bottom hole was sealedusing the same loose black rubber tube and tight fitting metal rings describedpreviously1. ScanImage 3.8 was used for microscope control and acquisition32.Ti:Sapphire laser (Chameleon Ultra II, Coherent) light at 920 nm was used as theexcitation source. Laser average power at the sample (after the objective) was 10–75 mW. An electric lens33 (EL-10-30-C-VIS-LD, Optotune; f 5 2100 mm offset lens)was used to switch rapidly between different focal planes (planes up to ,500mm apartwere possible). The electric lens was mounted in close proximity to the microscopeback aperture and the focal plane was rapidly switched after each collected image ofthe time-series by changing the applied steady-state current (LD1255R current driver,Thorlabs). Simultaneous with focal plane switching, the average laser power wasmodulated, using a pockels cell (350-80-LA-BK-02, 302RM driver, Conoptics), toaccount for different amounts of tissue penetration (more power for deeper planes).

Images (256 3 64 pixels, 0.5 ms per line field of view of ,145 3 75mm) in eachplane were acquired at 31.3 Hz for single-plane, 15.6 Hz for 2-plane and 10.4 Hz for3-plane acquisitions. The resulting co-acquired time-series movies contain inter-leaved frames fast enough to provide sufficient sampling of transients in each plane.Note that while the effective frame rate at each plane changed depending on thenumber of acquired planes, the time to acquire each individual frame (32 ms) did

not change. Time-series imaging sessions lasted 19.3 6 13.2 min. Potential placecells were initially selected online during behaviour by comparing changes in fluo-rescence with track location. These cells were then examined to ensure that theirdendrites could be clearly separated from dendrites and axons of other cells. Multi-plane imaging was then performed on cells passing these criteria by selecting adendritic plane(s) with multiple basal branches in the region above (dorsal to) thesoma. Time-series acquisition was initiated during behaviour periods of high-reward rate on the linear track task.

A Digidata1440A (Molecular Devices) data acquisition system was used to record(Clampex 10.2) and synchronize position in the linear track, reward timing, and two-photon image frame timing (using the command signal to the slow galvanometer).

After time-series imaging, z series were acquired from each cell from the externalcapsule fibre surface through the proximal apical dendrite ,300mm deep using thefollowing parameters: 10 frames (short time-series) per z plane; 2mm steps betweenz planes; 256 3 128 pixels per frame, ,145 3 75mm; 1 ms per line. Cell bodies weretypically 100–150mm below the surface.Data analysis and statistics. Analysis was performed using ImageJ (1.45) and customscripts written in MatLab (R2011b). All data in the text and figures are presentedas mean 6 s.d., except in reference to the plots shown in Fig. 4f, Extended DataFigs 9d and 10b where error represents s.e.m. Sample sizes were chosen to reliablymeasure experimental parameters while remaining in compliance with ethical guide-lines to minimize the number of animals used. Experiments did not involve ran-domization or blinding because no place cell or animal groups were predefined.

Time-series movies for multi-plane recording were acquired using interleavedframes (that is, every other frame was from the same plane for 2-plane imaging). Theelectric lens settling time of ,5 ms created distortions in the first few lines of eachframe of the movie; these lines were therefore removed before subsequent analysis.Each multi-plane time-series was then split into separate time-series movies, one foreach acquired plane. Each single-plane time-series was then independently motioncorrected using whole frame cross-correlation, as described previously1,34.

Morphology analysis and identification of axons and dendrites. Because z serieswere acquired during behaviour, motion correction was required to correct forbrain movements. Each 10-frame time-series acquired at each z plane was motioncorrected independently using whole-frame cross-correlation34. A mean image (timeprojection) was made from each motion-corrected time-series at each z plane. Themean images were then registered with respect to each other by cross-correlating34

the mean image at plane 2 with the mean image at plane 1, then plane 3 with 2, andso on until all images were registered.

Dendrite and axon branches were traced (in 19 of 28 place cells; brain motionduring high-resolution z series prevented accurate reconstruction in 9 cases) frommotion-corrected z series using Simple Neurite Tracer in Fiji (ImageJ)35. Distancesand branch numbers were calculated from the resulting wire segments using cus-tom MatLab scripts. All dendritic and axonal distances from soma represent dis-tance travelled along the neurite; these numbers are provided for each branch inthe Figures along with the per cent distance from the soma to the end of the den-drite (dendritic tip).

Dendrites belonging to the co-imaged soma were identified offline by tracingthem to the soma in the z series; additional verification was provided by their oftenco-occurring significant calcium transients (see below) with the soma. Axon seg-ments were identified offline by their morphology and signal change. They weresmaller in diameter than dendritic branches and also had a lower resting fluores-cence level. Furthermore, their calcium transients were longer in duration and largerin signal change compared to dendritic branches.

Region of interest selection and calcium transient analysis. For single-cell imag-ing (sparse labelling), regions of interest (ROIs) were selected by hand on the meansoma or dendrite images (mean time projection of all frames in the motion-correctedtime-series at each plane). ROIs were drawn to closely follow the outline of thestructure of interest (soma, axon or dendrite).

For population imaging, ROIs were defined as previously described36 (mu 5 0.5,150 principal components, 100 independent components, s.d. threshold 5 1.5, s.d.smoothing width 5 1.5, 100 pixels , area of ROI , 400 pixels; see Mukamel et al.36

for parameter definitions). As seen previously1, ROIs nearly always defined singlecell regions.

From single-cell and population time-series, DF/F versus time traces were gen-erated for each ROI as previously described1. In brief, slow changes in the fluores-cence traces were removed by examining the distribution of fluorescence in a ,8 s(for single-cell/sparse labelling) or ,3.2 s (for population imaging) interval aroundeach sample in the trace and normalized by the 8th percentile value. These baseline-corrected soma, axon or dendrite fluorescence traces were then subjected to theanalysis of the ratio of positive- to negative-deflecting transients of various ampli-tudes and durations described previously31. We used this analysis to identify sig-nificant transients with ,0.1% false positive error rates; these identified significanttransients were used in the subsequent analysis and are marked with an asterisk in

RESEARCH LETTER


the figures. Note that brain movements can cause small fluorescence transients inthe positive- and negative-deflecting directions (equally) during behaviour. BecauseGCaMP6f indicates activity with positive-deflecting transients only, the negative-deflecting transients are assumed to be due to brain movements. The above citedanalysis detects positive-deflecting transients of a duration and amplitude that almostnever occurs in the negative-deflecting direction (,0.1% false positive rate). Thus,most of the small positive-deflecting transients not detected as significant by thisanalysis are probably induced by brain motion, since events of the same durationand amplitude often occur in the negative-deflecting direction; however some ofthese small positive-deflecting transients may be due to neural activity (though thenumber is likely small, see for example our ability to record single-spine events inExtended Data Fig. 2d, e).

The mean duration of somatic, axonal and dendritic calcium transients werecalculated as the full duration of the significant transients: soma, 2.89 6 0.59 s; axon,2.0 6 1.2 s; dendrites, 1.07 6 0.65 s. These durations were calculated from sparselylabelled cells only, where all 3 structures could be identified (not from populationlabelling). Only transients occurring during consistent runs through the place field,in which a long running period covered .90% of the place field (see below), wereincluded.

Previous studies have consistently found that somatic calcium transients arecaused by action potential firing where the number of underlying action potentialscorrelates with the change in fluorescence21,37–39. Thus, somatic DF/F fluorescencetraces are often used as a surrogate measure of action potential firing activity1,22.The following three related measurements of somatic and axonal calcium transi-ents in the present study are consistent with conclusions of prior studies assertingthat somatic calcium transients are caused by action potential firing:

(1) An axonal calcium transient co-occurred with the soma 100% of the time (130transients; n 5 4 place cells); axonal transients are tightly coupled to somatic firing40,41

and thus provide an independent indicator that action potential firing causes oursomatic calcium transients.

(2) Averaging our smallest-amplitude somatic calcium transients produced atrace (Extended Data Fig. 1a) nearly identical in amplitude, shape and duration towhat is expected from a single somatic action potential based on previous com-bined cell-attached and imaging measurements21.

(3) Place fields defined by somatic calcium transients are highly similar to thosedefined by somatic action potential firing in a comparable virtual track1,22.

Taken together, these three points indicate that somatic action potential firingoccurs throughout nearly the entire somatic calcium transient-defined place field.Further, somatic calcium transients varied in amplitude, consistent with a differencein the number of underlying action potentials, and varied in duration, consistentwith the summation of multiple transients (Extended Data Fig. 1).

Dendritic calcium transients42,43 occur as a result of either non-regenerative orregenerative depolarizations that involve calcium influx through local NMDA andvoltage-gated calcium channels44. Excitatory input to a single spine leads to a non-regenerative post-synaptic depolarization typically detected as a spine-head-restricted(that is, restricted to the head of the spine) calcium transient with little or no shaftcomponent21,44–47. Dendritically generated spikes (such as Na1, NMDA and calciumspikes, collectively referred to here as dspikes) and bAPs can also lead to dendriticcalcium influx, but by contrast, these events are regenerative depolarizations oftendetected as both shaft- and spine-invading calcium transients larger in both ampli-tude and spatial extent than transients associated with single-spine input17,44,46,47.In CA1 pyramidal neurons of navigating mice (not necessarily place cells), we observedcalcium transients restricted to single spines, with no detectable shaft transient;however, we more frequently observed calcium transients invading all visible pixelsof the recorded branch (both shaft and spines) that were larger in amplitude andspatial extent than single-spine transients (Extended Data Fig. 2d, e). These tran-sients were considered non-regenerative or regenerative depolarizations, respec-tively. The regenerative dendritic events were defined here as ‘branch spikes’ —events caused by either dspikes or bAPs or both.

Defining place fields. Place fields were identified and defined as described previously1,with minor changes outlined below. Place fields were defined solely based on somatic(not dendritic) calcium transients. First, long running periods were defined in eachdirection in which mouse movement along the virtual track consisted of virtualvelocity .,7 cm s21 and run length .40 cm (straight run without changing direc-tion or hitting the end of the track). These long run periods were first categorizedbased on the running direction (positive or negative direction) and then furthersubdivided into two categories based on the animal’s current task performance.Segments of time between two rewards in which long running periods of only onedirection occurred were defined as high-reward-rate periods, all other long run-ning periods were defined as low-reward-rate periods. Only high-reward-rate periodswere analysed and all time-series data sets included here had at least 20 (mean of50) long running segments during high-reward-rate periods in each of the positiveand negative directions. For each running direction for each cell, the mean somatic

DF/F was calculated as a function of virtual track position for 80 position bins andthis mean fluorescence versus position plot was then averaged over 3 adjacent points.Potential place fields were first identified as contiguous regions of this plot in whichall of the points were greater than 25% of the difference between the peak somaticDF/F value (for all 80 bins) and the baseline value (mean of the lowest 20 out of 80somatic DF/F values). These potential place field regions then had to satisfy thefollowing criteria: 1. The field must be .13 cm in width; 2. The field must have onevalue of at least 10% meanDF/F; 3. The mean in fieldDF/F value must be .3 timesthe mean out of field DF/F value; and 4. Significant calcium transients must bepresent .20% of the time the mouse spent in the place field. Potential place fieldregions that met these criteria were then defined as place fields if their P value frombootstrapping was ,0.05, as described previously1 and their mean widths were,125 cm.

For calculation of BSP, somatic firing intensity, spatial precision and centre ofmass (see below) only consistent runs through the place field, in which a long run-ning period covered .90% of the place field, were considered. Because place fieldspatial precision might be artificially increased in place fields occurring at the trackends due to edge effects, we only included place fields in which neither edge of theidentified field was at a track end. Additionally, we determined that place field spa-tial precision (for the included fields) does not correlate with distance from trackend (not shown).

Place field centre of mass and spatial precision index. To calculate the somatictransient centre of mass (COM) on each traversal (n) along the linear track when asomatic transient occurred in the defined place field, we first split location into4.5 cm bins (i) and measured somaticDF/F in each bin. We then used the followingequation to calculate the COM for each traversal (COMn):

COMn~

Pi DFixiP

i DFi

Where DFi is the somatic DF/F in bin i and xi is the distance of bin i from the startof the track. We then calculated the peak DF/F weighted mean COM (COMw)from all traversals n (COMn for each traversal was weighted by the peak transientDF/F on that traversal (An)):

COMw~

Pn AnCOMnP

n An

Spatial precision (SP) was then calculated as the inverse of the peak DF/F weightedCOMn standard deviation as follows:

SP~1ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiX

nAn COMn{COMwð Þ2P

nAn

vuut

Branch spiking analysis. The images shown (at left) in Figs 3a–c, 4b, e and ExtendedData Fig. 8a were generated as follows: somatic and branch spiking events thatoccurred in the cell’s place field (if a cell had more than one place field, each fieldwas treated separately) were converted to ones (a significant transient occurred)and zeros (no significant transient was detected).

Branch-spike prevalence (BSP) was then calculated for each place field traversalwith a somatic transient by dividing the sum of the number of branches with adetectable significant transient by the total number of recorded branches. The BSPdistribution for each place field in each session could then be examined (Fig. 3a, b, c,Extended Data Figs 7 and 8 histograms) or used to calculate the average BSP for thefield (the average of all BSP values for the session). Note that average BSP andresting fluorescence level for each place cell were not statistically significantly related(Extended Data Fig. 5g).

The branch spikes all co-occurred with somatic firing (except where stated other-wise; that is, example in Fig. 2d, and in Extended Data Figs 2 and 4b). All BSP num-bers presented are from dendritic activity that occurred during somatic place fieldfiring, except for specified values in Extended Data Fig. 8b (green dots).

To compare in-place field to out-of-place field BSP we examined 5 cells (ExtendedData Fig. 8b green dots) that had place fields and also had at least 3 somatic calciumtransients occurring randomly along the track (typically during running), but out-side of the place field and the reward zones. Note that out-of-place field somatic anddendritic firing along the track (outside of reward zones) was extremely rare.

For each recorded dendritic branch, the onset of each branch spike with respectto the co-occurring somatic transient was measured as the time difference betweenthe onset of the somatic and dendritic significant transients (Fig. 2e and ExtendedData Fig. 1d). If multiple branch spikes occurred during the somatic transient, theonset of each branch spike relative to the soma onset was measured.

To calculate how widespread each of the three main types of somato-dendriticobservations were (Fig. 2a–c), we classified each event type using one dendritic

LETTER RESEARCH


imaging plane and then included all of the branches from a more distal imagingplane (Extended Data Fig. 5a, b) and calculated the percentage of time our classi-fication was correct before and after the distal plane was included.

Somatic firing intensity. The somatic firing intensity is the mean somatic calciumtransient integral, calculated by summing the integrals of each somatic calcium tran-sient that occurred in the place field and dividing the sum by the total number ofplace field somatic calcium transients.

Branch-spike prevalence analysis for stable and transitory place cells. Stable placecells were defined as cells with a mean place field centre of mass (COMw) on day 1and day 2 that occurred within 15 cm of each other. In rare cases when place fieldsshifted between 15 and 50 cm, they were excluded from the analysis. Shifts of.50 cm were considered distinct place fields and so day 1 fields were classifiedas transitory. Place fields were excluded if they had fewer than 5 place field tran-sients. Day 1 and day 2 were consecutive days. Average branch-spike prevalenceand mean somatic firing intensity were measured on day 1 for both stable and tran-sitory place fields.

Population analysis for stable and transitory place cells (Extended Data Fig. 9).Place cells were excluded if they had fewer than 5 place field transients, had a placefield with one edge at a track end or if the field of view on day 2 did not include thecell. Day 1 and day 2 were consecutive days. Day 8 was seven days after day 1 andno imaging or training took place between day 2 and day 8.

31. Dombeck, D. A., Khabbaz, A. N., Collman, F., Adelman, T. L. & Tank, D. W. Imaginglarge-scale neural activity with cellular resolution in awake, mobile mice. Neuron56, 43–57 (2007).

32. Pologruto, T. A., Sabatini, B. L. & Svoboda, K. ScanImage: flexible software foroperating laser scanning microscopes. Biomed. Eng. Online 2, 13 (2003).

33. Grewe, B. F., Voigt, F. F., van ’t Hoff, M. & Helmchen, F. Fast two-layer two-photonimaging of neuronal cell populations using an electrically tunable lens. Biomed.Opt. Express 2, 2035–2046 (2011).

34. Miri, A., Daie, K., Burdine, R. D., Aksay, E. & Tank, D. W. Regression-basedidentification of behavior-encoding neurons during large-scale optical imaging ofneural activity at cellular resolution. J. Neurophysiol. 105, 964–980 (2011).

35. Longair, M. H., Baker, D. A. & Armstrong, J. D. Simple Neurite Tracer: open sourcesoftware for reconstruction, visualization and analysis of neuronal processes.Bioinformatics 27, 2453–2454 (2011).

36. Mukamel, E. A., Nimmerjahn, A. & Schnitzer, M. J. Automated analysis of cellularsignals from large-scale calcium imaging data. Neuron 63, 747–760 (2009).

37. Greenberg, D. S., Houweling, A. R. & Kerr, J. N. Population imaging of ongoingneuronal activity in the visual cortex of awake rats. Nature Neurosci. 11, 749–751(2008).

38. Kerr, J. N., Greenberg, D. & Helmchen, F. Imaging input and output of neocorticalnetworks in vivo. Proc. Natl Acad. Sci. USA 102, 14063–14068 (2005).

39. Tian,L.et al. Imaging neural activity inworms, flies andmicewith improvedGCaMPcalcium indicators. Nature Methods 6, 875–881 (2009).

40. Cox, C. L., Denk, W., Tank, D. W. & Svoboda, K. Action potentials reliably invadeaxonal arbors of rat neocortical neurons. Proc. Natl Acad. Sci. USA 97, 9724–9728(2000).

41. Petreanu, L. et al. Activity in motor-sensory projections reveals distributed codingin somatosensation. Nature 489, 299–303 (2012).

42. Helmchen, F., Svoboda,K.,Denk,W.&Tank,D.W. Invivodendritic calciumdynamicsin deep-layer cortical pyramidal neurons. Nature Neurosci. 2, 989–996 (1999).

43. Mittmann, W. et al. Two-photon calcium imaging of evoked activity from L5somatosensory neurons in vivo. Nature Neurosci. 14, 1089–1093 (2011).

44. Higley, M. J. & Sabatini, B. L. Calcium signaling in dendrites and spines: practicaland functional considerations. Neuron 59, 902–913 (2008).

45. Chen, X., Leischner, U., Rochefort, N. L., Nelken, I. & Konnerth, A. Functionalmapping of single spines in cortical neurons in vivo. Nature 475, 501–505 (2011).

46. Sabatini, B. L., Oertner, T. G. & Svoboda, K. The life cycle of Ca21 ions in dendriticspines. Neuron 33, 439–452 (2002).

47. Yuste, R. & Denk, W. Dendritic spines as basic functional units of neuronalintegration. Nature 375, 682–684 (1995).

48. Gobel, W., Kampa, B. M. & Helmchen, F. Imaging cellular network dynamics inthree dimensions using fast 3D laser scanning. Nature Methods 4, 73–79 (2007).

RESEARCH LETTER


Num

ber o

f som

atic

cal

cium

tran

sien

ts

Somatic calcium transient peak (%∆F/F)

Num

ber o

f den

driti

c ca

lciu

m tr

ansi

ents

a bNu

mbe

r of d

endr

itic c

alciu

m tr

ansie

nts

Branch spike peak ∆F/F normalized to co-occurringsomatic transient peak

Mean = 788 ± 624% ∆F/F

c

20%

∆F/F

200 ms

smallest 2%

of transients

*

Branch spike peak (%∆F/F)

Mean = 1048 ± 773% ∆F/F

Mean = 1.39 ± 1.18

# B

ranc

h sp

ikes

Branch spike onset delay (ms)

0 400 8000

20

40

60

80

Mean = 247.7 ± 232.3 ms

d

0 2000 4000 6000 80000

20

40

60

80

100

120

140

160

180

200

0 2 4 6 8 10 12 140

100

200

300

400

500

600

0

50

100

150

200

250

300

350

0 2000 4000 6000 8000

Extended Data Figure 1 | Somatic and dendritic place field Ca21-transientamplitudes and onset times. a–c, Histograms of somatic transient peakDF/F (a), branch-spike peak DF/F (b) and ratio of branch-spike peak DF/F tosomatic transient peak DF/F (c). Inset in a shows the smallest significantamplitudes of somatic transients detected (grey) and their mean (blue; averagetriggered when grey transients were first .2 s.d. above the baseline). Thesesmall transients are nearly identical in amplitude, shape and duration to what is

expected from single somatic action potentials based on previous in vivocombined cell-attached and imaging measurements in the visual cortex usingGCaMP6f (ref. 21). d, Histogram showing the distribution of branch-spikeonset time relative to somatic firing onset. Note that branch-spike onset leadingsomatic firing onset was not observed, and when branch spiking occurred inmultiple branches, their onsets were nearly always simultaneous with respect toeach other. Mean 6 s.d. is shown for a–d.

LETTER RESEARCH


unclassified dendritic events occurringwithout detectable somatic firing

classified single spine transients

Mea

n ΔF

/F (%

)

Area of fluorescence (μm2)

a b

Single spine transients

putative dendritically generated spikes

# tra

nsie

nts

c

1 23

1

2

3

1

2

3

e f

0

90

% ΔF/F

4

4

4

150

0

5 μm20 μm

% ΔF/F

100%ΔF/F

1 s

Large areabranchspiking

Large area branch spiking Putative dendriticallygenerated spike

Low res High res

Spine

Shaft

Spine

Shaft

5 s

250%ΔF/F

Spine

Shaft

Spine

Shaft

**

*

d

Putative dendriticallygeneratedspike* *

*

*

*

*

*

*

*

*

*

Spine

Shaft

*

*

400%ΔF/F

Spine

Shaft

*

*

1200

0

% ΔF/F

5 μm

2 s

Non

-reg

ener

ativ

eR

egen

erat

ive

5 μ m

0 10 20 30 400

200

400

600

800

Mea

n ΔF

/F (%

)

Area of fluorescence (μm2)0 10 20 30 40

0

200

400

600

800

single spinetransients

putative dendritically generated spikes

0 20 40 60 80 1000

2

4

6

8

10

12

Activity area index

RESEARCH LETTER


Extended Data Figure 2 | Identifying putative dspikes and discriminatingbetween regenerative and non-regenerative dendritic events. a, Mean DF/Fof significant calcium transients localized to a single branch (using the sameimaging parameters to measure co-occurring somatic firing and branch spikesin Figs 1–4), plotted against the area of significant DF/F increase (.3 s.d.).The mean DF/F and the area of significant fluorescence change was calculatedfor each transient as follows. DF/F movies were generated where each pixelvalue in each frame of the movie represents the change in fluorescence withrespect to the baseline mean for that pixel. The frames during the transient ofinterest were averaged together and the number of pixels with DF/F . 3 werecounted, converted to mm2 and used as the area of significant fluorescencechange. The mean DF/F value for the transient was then calculated as the meanvalue of the pixels with DF/F . 3. Black circles represent known single-spinecalcium transients acquired using low resolution time-series acquisition (thesame resolution used to identify branch spiking), but confirmed as spines usinghigher-resolution time-series acquisitions where calcium transients wererestricted to the spine head (see d and e). Green circles represent calciumtransients restricted to a single dendritic branch and occurring in the absence ofsomatic firing. Because no high-resolution time-series were acquired fromthese structures, it was unknown whether they represent transients restricted tosingle spines or branch spiking. The panels and analysis presented in b and cindicate that a majority of these events are due to branch spiking and notsingle-spine transients. b, As a combined metric of mean DF/F and area offluorescence change, we normalized mean DF/F and area of fluorescence totheir maximums and measured the distance from the origin (the normalizedEuclidean distance of each point in a); we refer to this metric as the activity areaindex (AAI). Histogram showing that known spine transients all fall into the

lowest AAI bins (black bars; AAI , 40), and most unclassified events (putativedspikes) in a have higher AAIs (green bars). The events in the larger AAI bins(with greater mean DF/F covering a larger area; AAI . 40; separated fromthe lower AAI bins by the dashed line) fit known characteristics of dspikes.c, Using the AAI threshold defined in b, most of the unclassified transients fallinto a separate group (red) from the known single-spine calcium transients(8 of the 13 unclassified transients had distinctly larger AAIs compared to spinetransients, blue) and were therefore considered branch spikes (putativedspikes). d, Example calcium transients restricted to a single spine head(bottom left) and invading both spine head and shaft (bottom right) in a placecell. Mean somatic place field is indicated by the grey dashed line. e, Example ofa stretch of dendrite imaged at low (left) and high (right) resolution in anavigating mouse. The red box and ROIs indicate the same structures that wereimaged at both low and high resolution. The same spine head is indicatedby arrows at different resolutions in all images. Calcium transients restricted tothe same single spine head are shown at both low and high resolution by thecolour-coded per cent DF/F map and by the per cent DF/F traces labellednon-regenerative; note that the shaft ROI includes other non-active spines.Calcium transients invading the branch and all spines are shown at both lowand high resolution by the colour-coded per cent DF/F map and by the per centDF/F traces at the bottom, labelled regenerative. f, Image of a dendritic branchsplit into four ROIs. Calcium transients are seen in all parts of the branchduring large area branch spiking (see both colour-coded map and traces ofper centDF/F). A putative dspike, during navigation, in the same branch causesa significant increase in fluorescence in only part of the branch (ROI 2), whichincludes the shaft.

LETTER RESEARCH


Soma10 s

1000%∆F/F * *

*

* *

* *

* *

** *

** * * *

* * *

**

* *

* *

* ** *

*

* *

*

*

*

*

*

1 s

100%∆F/F

1000%∆F/F

Branch 1: 199 μm

Axon: 158 μm

Branch 2: 192 μm

Branch 3: 152 μm

Extended Data Figure 3 | Expanded views of traces showing the variabilityin detectable events in the soma, dendrites and axon. Traces from Fig. 1bshowing variable branch-spike prevalence in the same cell. Three events are

shown amplified (for the amplified traces, the bottom y-axis scale bar refers tothe green, blue and cyan dendritic traces, and the top scale bar refers to thered and brown soma and axon traces).

RESEARCH LETTER


1 s

Soma

143 μm/67%

149 μm/77%

115 μm/65%

80 μm/56%

Trackposition

180 cm

0 cm

*

100%∆F/F

100%∆F/F

2 s

a

b

* *

**

*

* *

*

14 2

3

Branch 3:119 μm/61%




41

2

3

50%∆F/F

2 s2 s 2 s

Extended Data Figure 4 | Expanded views of traces showing variability ofdendritic branch spiking during somatic place field firing. a, Traces fromFig. 2 are shown here amplified (each trace has a y-axis scale bar representing100% DF/F). b, Two examples of branch spikes detected in single branches in

the absence of detectable somatic firing and branch spiking in other dendrites.In each case the soma y axis is amplified relative to the dendrite traces toshow the absence of detectable somatic transients.

LETTER RESEARCH


Soma

1000%

2 s

101112

Dendritic plane 1

Dendritic plane 2

2

4

6

1

3

5

7

8

9

6

35

78

9

Soma plane

Linear

track

position

180 cm

0 cm

Soma

a

Soma

Soma plane

Dendritic plane 2

Dendritic plane 1

24a

4b3

Linear

track

position

180 cm

0 cm

400%

2

4

1

3

2

3

4a

4b

b

Soma

162 μm/86%

140 μm/80%

119 μm/61%139 μm/67%

114 μm/65%124 μm/78%

131 μm/71%

109 μm/71%

109 μm/75%106 μm/68%109 μm/68%123 μm/71%

172 μm/92%

142 μm/76%

134 μm/87%

138 μm/73%

125 μm/66%

80 μm/43%

80 μm/52%

80 μm/42%

*

*

*

*

*

**

*

*

*

***

***

*

**

*

Rat

io o

f den

d tra

nsie

nt

peak

nor

mal

ized

to s

oma

Branch distance from soma (μm)

c

167 μm/86%

147 μm/84%153 μm/96%169 μm/92%138 μm/90%

144 μm/99%

d

1

23

4 56 8

79

1011 12

3

56

7

8

9

1 234

f

e

Normalized resting fluorescence intensity

Ave

rage

bra

nch

spik

e pr

eval

ence

g

0.2

0.4

0.6

0.8RHO = 0.135P = 5.4X10-12

0 50 100 150 200 250 3000

2

4

6

8

10

12

14RHO = -0.128P = 0.16

0 50 100 150 200 250 300

1

0

Bra

nch

spik

e pr

eval

ence

for

indi

vidu

al b

ranc

hes

0

10

20

30

40

Num

ber o

f bra

nche

s

0 0.2 0.4 0.6 0.8 1

0 0.2 0.4 0.6 0.8 10

0.2

0.4

0.6

0.8

1

RHO = -0.034P = 0.9

RHO = -0.057P = 0.75

Number of branches sampled in field0 2 4 6 8 10 12 14 16 18

Ave

rage

bra

nch

spik

e pr

eval

ence

0

0.2

0.4

0.6

0.8

1

*

∆F/F

∆F/F

Branch distance from soma (μm) Branch spike prevalence for individual branches

RESEARCH LETTER


Extended Data Figure 5 | Branch spiking in a single dendritic plane isrepresentative of activity throughout a large portion of the dendritic arbourand average branch-spike prevalence is independent of the number ofsampled branches in the field, their distance from the soma and the restingfluorescence level of the soma. a, An example of spiking throughout allimaged branches during 3-plane imaging. Co-acquired somatic and dendritictime-series were recorded using three planes; dendritic plane 1 is approximatelymid distance between the soma and dendrite tips, dendritic plane 2 is nearthe branch tips. Numbered arrows indicate branches connected to the imagedsoma with the same numbers in dendritic plane 1 and 2 indicating the samebranch. An example run through the cell’s somatic place field (grey) andcorresponding DF/F traces from the soma and numbered branches from thetwo dendritic planes is shown. Distance along dendrite between branch andsoma, and per cent distance from soma to dendritic tip, are on the right of eachtrace. b, Same as a, except a different cell where branch spiking is absentthroughout all imaged branches in both dendritic planes. c, Scatter plotshowing the average branch-spike prevalence for individual branches duringsomatic firing as a function of branch distance from the soma (each pointrepresents one branch). The average branch-spike prevalence for individual

branches is not significantly related to the distance from the soma (Spearman’srank correlation coefficient: P 5 0.16; r 5 20.128). d, Branch-spike peaksnormalized to co-occurring somatic peaks during place field traversals fromall place cells and branches plotted against branch distance from soma.Spearman’s rank correlation coefficient shows a significant correlation(P 5 5.4 3 10212, r 5 0.135) and a linear fit shows a significant positive slopewithin 95% confidence bounds. e, Histogram showing the branch-spikeprevalence for individual dendritic branches taken from all cells. f, Averagebranch-spike prevalence (for each place field) plotted against the number ofbranches sampled in the imaging field shows no significant correlation(Spearman’s rank correlation coefficient: P 5 0.75; r 5 20.057). g, Averagebranch-spike prevalence (for each place field) plotted against normalizedresting fluorescence intensity of the soma. Relative resting fluorescencebetween cells was calculated by dividing the mean measured fluorescence ofeach soma (not during transients; excluding nucleus) by the squared laserpower arriving at the soma (which was estimated based on the soma depthbelow the surface48). Spearman’s rank correlation coefficient shows nosignificant correlation (P 5 0.9, r 5 20.034).

LETTER RESEARCH


Extended Data Figure 6 | Place field branch spikes in basal and proximalapical dendrites often co-occur. Co-acquired somatic, basal dendritic andapical dendritic (depicted in cartoon in the centre) time-series from twoexample place cells showing somatic spiking with co-occurring branch spikes(top) and somatic spiking in the absence of detectable branch spikes (bottom)

in the basal and main apical dendrites (89.6 6 19.5% correlation betweenspiking in the basals and main apical; mean distance of apical site tosoma, 109 6 38mm; n 5 6 place fields; mean 6 s.d.) during place field traversals(grey). Note that our findings from the basal and proximal apical dendrites maynot extend to the oblique dendrites or apical tuft.

RESEARCH LETTER


Average branch spike prevalence = 0.56Number of branches sampled in field = 8

0.6 0.8 10.40.200

0.3

Branch spike prevalence

Frac

tion

of e

vent

s


0.6 0.8 10.40.200

0.25P = 0.034

P = 0.012 0.9

0.6 0.8 10.40.200


P = 0.0009

0.6 0.8 10.40.20

0.6

0


0.35 P = 0.024

0.6 0.8 10.40.20


0.6 0.8 10.40.20

0.6


P = 1 X 10-7

P = 5 X 10-12

P = 0.019

0.6 0.8 10.40.20

0.6


Extended Data Figure 7 | The distribution of branch-spike prevalencediffers from a model in which each branch fires independently with aspecific probability during place field traversals. Seven histograms fromseven example place fields showing the distribution of branch-spike prevalencein each field for real data (grey bars) and modelled data (red bars). Themodelled data was generated for each place field example as follows. Theprobability (Pi) that each dendritic branch (i) in the imaging field would spikein the place field was defined as the branch-spike prevalence for the individual

branch (total number of traversals in which branch i spiked divided by the totalnumber of traversals; from real data). For modelled/mock place field traversals,each branch fired with its random probability Pi and the branch-spikeprevalence (fraction of the total number of branches with spikes during thetraversal) was calculated. The distribution of branch-spike prevalence wasgenerated for 1,000 modelled/mock place field traversals (red bars). A two-sample Kolmogorov–Smirnov test was used to compare real and modelleddistributions (P values shown in each plot).

LETTER RESEARCH


a

Place field A average branch spike prevalence

0 0.5 10

0.5

1b

0

0.5

1

Average branch spikeprevalence= 0.48

Average branch spikeprevalence = 0.88

Branch spike prevalence

# ev

ents

0.6 0.8 10.40.200

25

0

25

0.6 0.8 10.40.20soma

b1b2

Place field A

10 20 30 40

somab1b2

Place field B

10 20

Pla

ce fi

eld

B a

vera

ge

bran

ch s

pike

pre

vale

nce

Out

of p

lace

fiel

d av

erag

e br

anch

spi

ke p

reva

lenc

e


Extended Data Figure 8 | Average branch-spike prevalence can differbetween different place fields of the same cell and also between in-place fieldand out-of-place field somatic firing. a, Coloured plots (left) show occurrenceof detectable spiking in each branch (blue or black) during somatic placefield firing (red) in different co-occurring place fields (A and B, in differentrunning directions) of the same place cell. Right, histograms of branch-spike

prevalence on each traversal for the two place fields. Cartoons (far right) donot represent real data. Note that the running behaviour differed in the tworunning directions causing differences in the number of traversals reachingbehaviour criteria. b, Plot comparing average branch-spike prevalence of twodistinct place fields in the same place cells (black) or of in-place field versusout-of-place field somatic firing in the same cells (green).

RESEARCH LETTER


Extended Data Figure 9 | Place field spatial precision is correlated to placefield stability. a, Schematic showing single-plane imaging of the samepopulation of cell somata over multiple days (1, 2 and 8). b, Population imageswith place cells colour coded by their place field location. Top and bottom rowsshow place cells in the negative and positive track running directions,respectively. Examples of place cell field fate shown at right: transitory fieldsoccurring on day 1 only (indicated with a minus symbol on the right; blackarrows on the left), fields persisting for at least one day (indicated by asterisks

on right, white open arrowheads on left) or seven days (indicated by doubleasterisks on right; white arrows on left); no symbol indicates that the field didnot meet the criteria for inclusion (Methods). c, Mean fields of the samecell measured over different days with COM locations from day 1 (bottom;black circles) indicating a precise place field. d, Spatial precision is significantlyhigher in stable versus transitory place fields (0.127 6 0.016 cm21 versus0.065 6 0.007 cm21, respectively; t-test; P 5 0.0003). In d, individual datapoints are depicted by circles to the right of bars and error bars represent s.e.m.

LETTER RESEARCH


Extended Data Figure 10 | Examples of dissociation between somatic firingand branch spiking in a place cell. a, DF/F traces from the (co-acquired) somaand numbered dendritic branches of a place cell during the same imagingsession demonstrate that somatic firing and branch spiking are oftendissociated. b, Plot of somatic firing intensity versus branch-spike prevalence

for all individual place field traversals from all cells (open circles) and binned(solid circles; error bars represent s.e.m.). Branch spiking did not stronglycorrelate with mean (binned) somatic firing intensity (Spearman’s rankcorrelation coefficient: P 5 0.04; r 5 0.57).

RESEARCH LETTER


calcium transient prevalence across the dendritic arbour...

Documents