Supporting InformationHill et al. 10.1073/pnas.1312599110SI Materials and MethodsAll experimental procedures were performed in accordance withinstitutional animal welfare guidelines and were approved by thestate government of Bavaria, Germany.

Animal Preparation. The animal preparation procedures were thesame as described previously (1). C57BL/6 mice (P28–P35, n = 34)were used for in vivo dendrite and spine imaging experiments.All procedures were performed under isoflurane anesthesia (0.5–1.0%) while monitoring breathing rate and body temperature.Craniotomies were centered at 1.0 mm anterior and 1.0 mmlateral to bregma to target the whisker representation of whiskermotor cortex where layer 5 (L5) is most superficial (2). In a subsetof experiments, this location was confirmed by intracorticalstimulation that elicited whisker movements. The cranial openingextended 1 mm in the anterior–posterior direction and 2 mm inthe medial–lateral direction to allow travel of the pipette mediallyto the deep cortical layers.

Whole-Cell Recordings and Electroporation. In the majority of ourexperiments (n = 25), we used the whole-cell patch technique toload L5 pyramidal neurons with calcium dye and record theirsomatic membrane potentials. Due to the long diffusion time fordye to reach distal tuft dendrites, imaging began 40 min afterloading began. Identity as a L5 pyramidal neuron was confirmedby the depth (>400 μm below the pia), the presence of an apicaldendrite that terminated as a tuft in L1, and the presence ofspines. Patch pipettes with a resistance of 5–8 MΩ were pulledfrom borosilicate glass capillaries on a vertical puller (PC-10;Narishige). The pipette solution contained 135 mM K-gluconate,10 mM Hepes, 4 mM KCl, 10 mM Na-phosphocreatine, 4 mMMg-ATP, 0.3 mM Na2-GTP, 100 μM Oregon Green 488 1,2-bis(o-aminophenoxy)ethane-N,N,N,N-tetraacetic acid (BAPTA–1)hexapotassium salt (OGB1), and 12.5 μM Alexa Fluor 594 hy-drazide sodium salt (Alexa). Current clamp recordings wereacquired with an EPC-9 HEKA amplifier and saved on a com-puter using the software tool Pulse. Signals were low-pass-fill-tered at 3 kHz and digitized at 10 kHz. Series resistances rangedfrom 10 to 40 MΩ.In a subset of experiments (n = 6), neurons were filled with

calcium dye via the electroporation technique as previously de-scribed (3). Pipettes and internal solution were the same as usedfor whole-cell recordings, except that OGB1 and Alexa con-centrations were increased 100-fold due to the lower efficiency ofelectroporation filling. Dye was injected with 2–3 pulses of 25 msduration at 500 nA current until the cell was judged sufficientlybright. The pipette was then retracted from the brain. Althoughthis procedure did not allow membrane potential recordings, theimaging quality was superior to whole-cell patch experiments andallowed for deep imaging of spine activity on the basal dendrites.In a separate set of pharmacological experiments designed to

block glutamatergic synaptic transmission, multiple L5 cells werefilled using the electroporation technique (n = 8 cells from threeanimals). After recording baseline data, a new pipette was in-troduced containing 100 μM 2,3-dioxo-6-nitro-1,2,3,4-tetrahy-drobenzo[f]quinoxaline-7-sulfonamide disodium salt (NBQX),500 μM D-(-)-2-amino-5-phosphonopentanoic acid (APV), and12.5 μM Alexa in Hepes Ringer. The pipette was guided towithin 50 μm of the imaged dendrites, and then short puffs wereapplied at 100 millibars until the Alexa was visibly surroundingthe dendrites. New imaging data were collected ∼1 min afterdrug application.

Two-Photon Calcium Imaging of Single Branches. Two-photon cal-cium imaging was performed on a custom-built setup similar towhat has been described previously (4). A resonant scanner unit(GSI Lumonics), including one fast axis with a 12 kHz resonantmirror and one slow axis with a standard galvanometric mirror,was mounted on an upright microscope chassis (BX51; Olympus)with a long working distance water-immersion objective (40×/0.8; Nikon). Fluorescent excitation light was delivered by apulsing infrared laser (wavelength, 800 nm; pulse width, 100 fs;repetition rate, 80 MHz) equipped with a prechirper (modelDeepSee; Spectra-Physics). Laser power under the objective wastypically 20–120 mW (depending on imaging depth). Emittedfluorescent photons were separated by a beamsplitter mirror at570 nm (F33-572, AHF; Analysentechnik) and detected by high-sensitivity photomultiplier tubes (H-7224; Hamamatsu) and an-alog current amplifiers (DHPCA-100; FEMTO Messtechnik).The laser scanning, image acquisition, and stimulation triggeringwere controlled by a real-time PXI computer (PXIe-1082; Na-tional Instruments) equipped with FPGA I/O board (PXI-7831R)and a high-speed digitizer (PXI-5122). The user interface soft-ware was developed in LabVIEW (version 2009) on an eight-corePC connected to the PXI computer. Full-frame images consistedof 512 lines, and they were scanned at a 40 Hz repetition rate.Each line had 1,041 raw pixels that were digitized at 25 MHz (100MHz internal sampling rate) and reconfigured into 512 pixelsaccording to the sinusoidal trajectory of the resonant mirror.Neurons were imaged at several focal planes distributed across

the tuft, apical trunk, and basal dendrites. The width of the field ofview was controlled by the amplitude of scanning mirror rotationand ranged between 100 and 300 μm to fit the layout of dendritesin the focal plane. Spontaneous activity was monitored at eachfocal plane for 5–10 trials of 30 s each. Z-stacks were taken at theend of each experiment using the largest possible field of view(300 × 300 μm) with a step size of 0.5 μm along the z-axis.

Two-Photon Calcium Imaging of Spines. The same two-photon im-aging setup was used for the imaging of single spines. For theimplementation of the low-power temporal oversampling pro-cedure (5), the scanner was configured for the following mode:the number of lines was reduced to 64, the number of pixels ineach line was resampled to 256, and the imaging frame rate wasincreased to 200 Hz. The width of the field of view was ac-cordingly reduced to the range of ∼40 μm. Laser power wasreduced to 10–60 mW (depending on imaging depth) duringspine imaging. The experimental protocol for the spine imagingwas the same as for dendritic imaging.

Data Analysis. Calcium imaging data were analyzed by custom-written programs in MATLAB (version 2010b; Mathworks).Whole-cell recordings were extracted to text files using Igor (IgorPro v5.0.1.0; Wavemetrics) and then processed in MATLAB.Filtering of calcium imaging data and transformation into ΔF/F signals. Aregion of interest (ROI) was manually drawn around each con-tinuous dendritic branch in a focal plane. The average pixel in-tensity within the ROI was extracted on each frame to form a rawintensity signal. The baseline value at a particular time sample wastaken as theminimumvalueof thefiltered signal over an8 swindowcentered at the time point. The baseline signal was then subtractedfromanddivided into the rawsignal to form the fractional change influorescence, ΔF/F. This signal was then smoothed using a causalexponential filter with a time constant of 200 ms.

Hill et al. www.pnas.org/cgi/content/short/1312599110 1 of 5

Identification and quantification of calcium transients. Significant cal-cium transients were found automatically with a threshold crossingprocedure. First, the SD of the noise was estimated by high-passfiltering the unsmoothed ΔF/F signal at 15 Hz to remove anyphysiological calcium signal. The filtered signal was assumed torepresent shot-noise, which is characterized by equal noise varianceat all frequencies. The noise variance in the smoothed signal wastaken as the variance per Hz in the high-pass-filtered signal multi-plied by the range of frequencies spared by smoothing. A thresholdof 3 SDswas used, with a typical value of∼10%ΔF/F. For the specialcase of a spine signal, the threshold was set to 3 SDs above thecalcium signal in the neighboring dendritic shaft. To identify isolatedcalcium events, a transient was used only if it remained below

threshold for 1 s before threshold crossing and then stayed abovethreshold for at least 500 ms. The amplitude of the event was takenas the difference between the peak value within 500 ms afterthreshold crossing and the minimum value 500 ms before.For the analysis of the correlation between calcium amplitude

and spike count, the amplitude was defined relative to spiking.Here an isolated spike event was defined as one or more spikesoccurring in a window of less than 500 ms, with no spikes for 500ms before or after the event. The amplitude of the accompanyingcalcium signal was defined as the maximum value of the signaltaken up to 500 ms after the last spike minus the minimum valuewithin 500 ms before the first spike.

1. Jia H, Rochefort NL, Chen X, Konnerth A (2011) In vivo two-photon imaging of sensory-evoked dendritic calcium signals in cortical neurons. Nat Protoc 6(1):28–35.

2. Hooks BM, et al. (2011) Laminar analysis of excitatory local circuits in vibrissal motorand sensory cortical areas. PLoS Biol 9(1):e1000572.

3. Judkewitz B, Rizzi M, Kitamura K, Häusser M (2009) Targeted single-cell electroporationof mammalian neurons in vivo. Nat Protoc 4(6):862–869.

4. Varga Z, Jia H, Sakmann B, Konnerth A (2011) Dendritic coding of multiple sensoryinputs in single cortical neurons in vivo. Proc Natl Acad Sci USA 108(37):15420–15425.

5. Chen X, Leischner U, Rochefort NL, Nelken I, Konnerth A (2011) Functional mapping ofsingle spines in cortical neurons in vivo. Nature 475(7357):501–505.

30

20

10

0

Cal

cium

am

plitu

de (

F/

F)

Time relative to onset of up-state (s)

-1 0 1

Fig. S1. Subthreshold dendritic calcium activity is mostly locked to the onset of up-states. Example traces of subthreshold calcium transients from a repre-sentative tuft dendrite (n = 25 events). Traces are aligned on the transition of the membrane potential signal from down-state to up-state. Thick black trace isensemble average.

Hill et al. www.pnas.org/cgi/content/short/1312599110 2 of 5

Time

Spine #2

Spine #1

Tuft dendritic shaft

100% F/F

10 s

* * * *

*

10

m10

m

Time

Basal dendritic shaft

50% F/F

Spine #4

Spine #3*

***

*

10 s

A

B

0 5 10

0.2

0.1

00 50 100

0.5

0.25

0

Amplitude (%[Fspine - Fshaft]/Fspine) Events per minute

Pro

porti

on o

f eve

nts

Pro

porti

on o

f spi

nesBasal spines

Tuft spines

C DBasal spinesTuft spines

Fig. S2. Spines show active and passive calcium dynamics. (A) Imaging plane, blow-up of spines, and example traces for spine and shaft calcium signals froma tuft dendrite. Colored lines on images represent ROIs used to extract calcium signals shown at right. Each spine signal is overlaid in gray with a copy of theshaft signal. Asterisks denote spine calcium transients that rise significantly above the shaft signal, suggesting that the transient is not due to merely passivediffusion of calcium from the shaft. (B) Same as in A but for a basal dendrite from the same neuron. (C) Differential amplitude of calcium transients in spines.Amplitude in shaft was subtracted from amplitude measured in spine. Histogram was smoothed with a boxcar filter of width 5% ΔF/F. (D) Frequency of calciumtransients in spines. Histogram was smoothed with a boxcar filter of width two events per minute.

Hill et al. www.pnas.org/cgi/content/short/1312599110 3 of 5

Number of spikes

Eve

nt a

mpl

itude

%F

/F

0

20

Eve

nt a

mpl

itude

%F

/F

10 % F/F

Tuft dendrite

40

A

D

Number of spikes

0

100

50

Down-stateUp-state

21 2

21 2

50 mV

1 s

Calcium

Vm

Calcium

Vm

Basal dendrite

Up-stateDown-state

B CTuft dendrites

Spi

ke in

vasi

on %

Tuft dendrites Basal dendritesor apical trunk

Stimulation

Stimulation

Time

0

20

40

Basal dendrites

0 642 0 642

Fig. S3. Evoked spikes do not propagate efficiently into the tuft. (A) Example traces from a tuft and basal dendrite illustrating relation between evokedspikes, calcium activity, and cortical state. Branches came from the same L5 pyramidal neuron with soma located 480 μm below the cortical surface. Number ofevoked spikes is shown beside membrane potential. For these experiments, neurons were kept at resting potential (about –65 mV during down-state) andspontaneous action potentials (APs) were discarded from the analysis. (B) Scatterplot of calcium amplitude versus number of events for all trials from tuftdendrites (n = 6 branches). Blue data points indicate trials where APs were evoked during the up-state, and red data points indicate the down-state. (C) Sameas B but for data from basal dendrites and apical trunks (n = 11 branches). (D) Bar graph of success rate for observing calcium transient evoked by spikes in theup-state and down-state. Success was defined as whether a calcium transient could be detected in the 500 ms following current injection. Vertical black linesrepresent 95% confidence intervals.

Hill et al. www.pnas.org/cgi/content/short/1312599110 4 of 5

10 m50 mV

1 s

Loca

l cal

cium

Vm

50% F/F

Loca

l cal

cium

Vm10 m

Tuft dendrite (depth = 73 m )A

50 mV

50% F/F

Basal dendrite (depth = 490 m )B

Event amplitudes

Time

1 s

Event amplitudes

Time

Fig. S4. Calcium signals during AP activity are largely uniform throughout the dendritic branch. (A) Example local calcium signals in tuft dendritic branch froma L5 pyramidal cell. Soma was located at 520 μm below the pia. (Left) Two-photon fluorescence image of a tuft dendritic branch. Colored lines show ROIssegmented at 10 μm intervals along the length of the branch. At right are traces for five example AP events. Left-most example illustrates a calcium transientthat was not spatially uniform while others are more uniform. Colored traces are calcium signals recorded from the ROI of the same color as in the fluorescenceimage. Black trace is the membrane potential recorded in the whole-cell patch configuration. Bottom row shows amplitude for each calcium trace as given bythe height of a bar of the same color. (B) Same as A but for a basal dendritic branch from a L5 pyramidal cell with soma located at 480 μm below the pia.

Hill et al. www.pnas.org/cgi/content/short/1312599110 5 of 5