Werner Göbel and Fritjof Helmchen - UZH

22:358-365, 2007. doi:10.1152/physiol.00032.2007 PhysiologyWerner Göbel and Fritjof Helmchen

You might find this additional information useful...

64 articles, 9 of which you can access free at: This article cites http://physiologyonline.physiology.org/cgi/content/full/22/6/358#BIBL

on the following topics: http://highwire.stanford.edu/lists/artbytopic.dtlcan be found at Medline items on this article's topics

Medicine .. Multiphoton Microscopy Physiology .. Neural Circuit

including high-resolution figures, can be found at: Updated information and services http://physiologyonline.physiology.org/cgi/content/full/22/6/358

can be found at: Physiologyabout Additional material and information http://www.the-aps.org/publications/physiol

This information is current as of January 6, 2010 .

http://www.the-aps.org/.Copyright © 2005 by the American Physiological Society. ISSN: 1548-9213, ESSN: 1548-9221. Visit our website at 20814-3991.bimonthly in February, April, June, August, October, and December by the American Physiological Society, 9650 Rockville Pike, Bethesda MD

) publishes brief review articles on major physiological developments. It is publishedNews in Physiological Science (formerly published as Physiology

on January 6, 2010 physiologyonline.physiology.org

Dow

nloaded from

http://physiologyonline.physiology.org/cgi/content/full/22/6/358#BIBL

http://highwire.stanford.edu/lists/artbytopic.dtl

http://physiologyonline.physiology.org/cgi/content/full/22/6/358

http://www.the-aps.org/publications/physiol

http://www.the-aps.org/

http://physiologyonline.physiology.org

excellently suited to study structureand function of cells and networkswithin their native environment. Fordetails on principles of 2PLSM andrecent applications, see Refs. 11, 19,54, 66.

Here, we focus on in vivo two-pho-ton imaging as an emerging tech-nique to reveal activity patterns inlocal networks of the mammalianbrain. Novel methods to stain popula-tions of cells with activity-dependentfluorescent markers, in particular cal-cium-sensitive indicators (17, 18, 33,34, 51), have opened the door forcomprehensive studies of the func-tional organization of microcircuits.We first introduce the basics of popu-lation calcium imaging, includinglabeling and fluorescence read-outmethods.

Calcium Indicator Labeling

Two-photon imaging relies on fluo-rescence excitation and, in general,necessitates staining of cells with fluorescent dyes. Various stainingmethods have been developed for invivo calcium measurements. Singlecells can be filled with membrane-impermeable calcium indicators viaintracellular recording electrodes (22,53) or by single-cell electroporation(34). In addition, several methods areavailable for labeling of cell popula-tions. Injection of dextran-conjugatedforms of organic calcium indicators(e.g., calcium green-1 dextran) intoaxonal pathways leads to local dyeuptake with subsequent anterogradeand retrograde transport over hoursand days (13, 39) (FIGURE 1A). In themammalian brain, this method hasbeen applied to measure calcium

In the nervous system, all sensationsand behaviors are encoded in dynam-ic patterns of activity in cellular net-works. Through a sequence of neuralnetworks, sensory information istransmitted to higher, associativebrain areas. Following integration inthese areas, specific activity patternsare eventually formed in the relevantmotoneuron pools to produce ade-quate behavior. In this chain of events,key processing steps are thought tooccur on the level of local microcir-cuits that contain on the order of1,000–10,000 cells. These local cir-cuits form highly connected three-dimensional networks that, other thanthe projection neurons, typically con-tain various inhibitory and excitatoryinterneuron types. In addition, neu-ronal networks form dense mesh-works with the surrounding glial cells.

Despite the fundamental importanceof microcircuits, their operating prin-ciples remain enigmatic. How localnetworks process incoming signalsand integrate them with ongoingbrain activity is poorly understood.The reason is that experimental toolsto record from large sets of neurons inthe living brain (in vivo) are still limit-ed. Although multi-electrode extra-cellular recordings enable simultane-ous recordings from hundreds of neu-rons in behaving animals (6), suchrecordings sample from widely dis-tributed cell populations with poorlydefined cell types and spatial relation-ships. As an alternative and compli-mentary technique, two-photon laserscanning microscopy (2PLSM) (7)enables high-resolution functionalimaging in living animals. Providing adepth penetration of several hundredmicrons in intact tissue, 2PLSM is

signals in afferent axonal terminals invivo (29, 61) and in retrogradelylabeled neuronal populations in spinalcord and retina in vitro (38, 63).

Groups of 10-100 neurons can beloaded in vivo by using either target-ed electroporation (34) or local elec-troporation, as recently demonstratedfor mouse neocortex, olfactory bulb,and cerebellum (33) (FIGURE 1B). Inthe latter case, a micropipette con-taining a high concentration of calci-um indicator in salt-form or as dextran-conjugate is inserted into thetarget area but is not particularlyaimed at a specific cell. Mild electricalpulses are applied through themicropipette, causing transient mem-brane disruptions in nearby neuronaldendrites and dye uptake. Followingmany repetitions, dye has accumulat-ed intracellularly and diffusedthroughout the affected cells. Anadvantage of the resulting sparselabeling is that subcellular calcium sig-nals can be resolved in dendrites andsynaptic structures because of the lowbackground staining (33).

Large-scale in vivo staining of entirecell populations can be achievedusing membrane-permeable ace-toxymethyl (AM) ester forms of calci-um indicators (60) (FIGURE 1C).Following diffusion through the cellmembrane, the ester groups arecleaved by cytosolic esterases so thatthe dye molecules, now membraneimpermeable, are trapped inside thecells. AM ester loading has been longapplied in cell cultures and in brainslices. Successful in vivo loading even-tually was achieved by bolus injectionof the AM ester dye into the brain(51). This “multi-cell bolus loading”(MCBL) technique at present is the

358

EMERGING TOPICS

In Vivo Calcium Imaging of NeuralNetwork Function

1548-9213/05 8.00 ©2007 Int. Union Physiol. Sci./Am. Physiol. Soc.

PHYSIOLOGY 22: 358–365, 2007; 10.1152/physiol.00032.2007

Werner Göbel and Fritjof Helmchen

Department of Neurophysiology, Brain Research Institute,University of Zurich, Zurich, Switzerland

[email protected] activity patterns in local neural networks are fundamental to brain

function. Network activity can now be measured in vivo using two-photon imaging

of cell populations that are labeled with fluorescent calcium indicators. In this review,

we discuss basic aspects of in vivo calcium imaging and highlight recent develop-

ments that will help to uncover operating principles of neural circuits.


Dow

nloaded from


most commonly used in vivo networklabeling approach and has beenapplied to a number of brain areas invarious animal species (5, 26, 36, 40,52). A detailed protocol for MCBL canbe found in Ref. 12.

An alternative to “traditional”organic dyes are genetically encodedcalcium indicators (GECIs) (forreviews, see Refs. 10, 27, 28, 31).GECIs are calcium-sensitive fluores-cent proteins that can be expressed insubsets of cells in transgenic animalsusing specific promoters (8, 17, 18)(FIGURE 1D). Although traditionalhigh-affinity calcium indicators still aresuperior to GECIs in some respects(42), the sensitivity of GECIs is contin-ually improving. Meanwhile, somaticcalcium signals evoked by only a fewaction potentials (APs), as well as den-dritic calcium signals, can be detectedin vivo (18).

To understand the function of a neu-ral circuit, one would like to discrimi-nate its sub-network components. Tothis end, counterstaining of specificcell types has proven helpful, espe-cially in bulk loaded tissue wheremarkers need not be calcium sensitive(FIGURE 1E). In rodents, cortical astro-cytes are identified with the red fluo-rescent dye sulforhodamine 101 (37).In addition, transgenic mice with fluo-rescent protein expression in specificneuronal subsets, e.g., in GABAergicinterneurons (57), allow separation offunctional signals into different neu-ronal subtypes (50). For the future, wecan expect transgenic animals withcolourful co-expression of differentGECIs in the various cellular sub-net-works.

Network ScanningApproaches

The new options for in vivo popula-tion labeling pose a challenge for2PLSM in terms of reading out net-work dynamics. Simultaneous sam-pling of neuronal calcium signals (typ-ically 0.5-1 s long) in hundreds of cellswith sufficient temporal resolution isnot trivial. In the standard imagingmode, the laser focus is scanned with-in one focal plane in a raster-like

fashion (FIGURE 2A). In this rastermode, the data acquisition rate (forthe individual cells) is determined bythe rate at which the entire scan pat-tern is repeated. Consequently, imag-ing of large fields is slow. Moreover,the choice of cells is somewhat ran-dom, and not all cells can be optimal-ly in focus. More flexibility is achievedby rotation of the image plane usingfocusing devices appropriate for rapidz-scanning (14, 35), for examplepiezo-electric focusing elements thatcan drive microscope objective move-ments at rates of 10 Hz and more(FIGURE 2B).

Entirely new scanning approachesare required for fast network meas-urements in 3D space, taking intoaccount the 3D structure of neural cir-cuits. The simplest option is 2D rasterscanning while continuously movingthe objective (FIGURE 2C) (16).

Acquisition rates in this mode typical-ly are limited to <1 Hz, and valuablescan time is wasted on backgroundstructures. Fast scanning systemsusing resonant scanners or acousto-optical devices (AODs) could furtherincrease 3D acquisition rates.

More easily, rapid scanning in 3D(with rates of 10 Hz and higher) isachieved with non-raster-like scanpatterns (FIGURE 2, D–F). For instance,a closed scan trajectory in 3D spacecan be generated by applying appro-priate signals to the xy-scan mirrorswhile rapidly scanning along the opti-cal axis with a piezo-electric focusingelement (15). Such a 3D line scan canbe defined by analytical functions, butit can also be trimmed to include spe-cific user-defined points of interestsuch as cell bodies (FIGURE 2E). As werecently could demonstrate for neo-cortical networks in vivo, calcium

359PHYSIOLOGY • Volume 22 • December 2007 • www.physiologyonline.org

EMERGING TOPICS

FIGURE 1. In vivo labeling of cellular networks with calcium indicatorA: retrograde labeling using dextran-conjugated calcium-sensitive dye. B: labeling of a subset ofcells using local electroporation. C: multi-cell bolus loading (MCBL) by injection of AM-ester calci-um indicator results in unspecific labeling of cell somata and a diffuse background in the neuropil.D: expression of a genetically encoded calcium indicator (GECI) in a particular cell type. E: multi-color labeling by combining MCBL with specific counterstains. Various fluorescent dyes can helpto distinguish neuronal and glial networks and identify neuronal sub-networks.


Dow

nloaded from

creo


rescence signals. Various signal compo-nents may be present, especially in thecase of MCBL where not only cell bod-ies are stained but also neuropil struc-tures including axonal, dendritic, andglial processes (FIGURE 3, A AND B).Distinct calcium signals are observedin the various tissue compartments aswe illustrate here using simulated arti-ficial data.

Neuronal signals

Due to the presence of voltage-gated calcium channels in neuronalmembranes, each AP causes a brief(around 1 ms) calcium influx (3). Theresulting transient elevations of theintracellular calcium concentrationabove resting level (so-called “calci-um transients”, �[Ca]) typically displaya sharp rise followed by an exponen-tial decay with a time constant ofaround 0.5-1 s for neuronal somata(FIGURE 3C). So far, AP-evoked calci-um transients are the predominantsignals observed in vivo, althoughsubthreshold neuronal activation mayalso lead to either highly localized ormore widespread calcium signals (30,45). Here, we presume AP-evoked cal-cium influx as the only source so thatpopulation spiking activity can be

signals can be sampled from hun-dreds of cells with nearly 100% of cells“hit” in tissue cubes of 0.25-mm sidelength (15).

Finally, novel ways to deflect thelaser beam discontinuously promiseto enable 3D random-access scan-ning, which effectively means to“jump” from cell to cell (FIGURE 2F).Possible implementations are AOD-based random-access scanning in 2D(24, 47) combined with fast z-scanning(15, 16, 35), special arrangements ofmultiple AODs for inertia-free scan-ning in 3D (44), or laser illuminationthrough optical fibers with their tipssuitably positioned to direct the laserfocus to pre-defined locations (46).Assuming point-to-point transitiontimes of around 10 �s, one can imag-ine 3D measurements from 1,000 cellswith 100-Hz resolution. Taken togeth-er, all these developments greatlyextend the possibilities for in vivooptical recordings of the functionalorganization of neural circuits.

Calcium Signaling in VariousCellular Structures

We now take a closer look at the ori-gin, analysis, and optimization of fluo-

indirectly inferred from the somaticcalcium signals (26, 49, 51, 64) (seealso below). Using this approach, themicroanatomical organization of sen-sory-evoked population responses hasbeen mapped, for example, in the pri-mary visual cortex of rodents and cats(32, 40, 41, 50), in rodent barrel cortex(25, 48), and in zebrafish olfactory bulband optic tectum (36, 64, 65). With ahigh-affinity indicator such as OregonGreen BAPTA-1, even single APs canbe optically detected (26).

Calcium signaling in glial networks

Glia cells are electrically mostly silentand, therefore, difficult to investigatein vivo by electrophysiological means.All the more in vivo imaging of calci-um signalling in glial cells has gainedparticular interest during recent years,e.g., to clarify the function of astro-cytes in local neocortical networks(59). Astrocytic calcium signals have amuch slower, wave-like time course(23, 37) (FIGURE 3C). They are linked tothe regulation of cerebral blood flow(56) and show altered properties inmodels of epilepsy (58) and Alzheimer’sdisease (55). Finally, sensory-evokedastrocytic calcium signals (62) suggestparticipation of the astrocytic networkin local network processing.

Neuropil calcium signals

In several studies employing MCBL,calcium signals have been observedin the diffusely stained neuropil, bothduring spontaneous activity and inresponse to sensory stimulation (26,40, 41, 48, 50). The neuropil signalrepresents an average signal frommany densely packed cellularprocesses, and spontaneous signalswere found to strongly correlate withthe electrocorticogram, thus provid-ing an “optical encephalogram”(OEG) (26). Because neuropil calciumsignals are not affected in amplitudeby locally blocking synaptic transmis-sion and since dendrites contributelittle (26), neuropil signals most likelyoriginate primarily from calciuminflux into presynaptic boutons andadjacent axonal segments. Hence,they reflect afferent synaptic input,which makes it possible to study

360 PHYSIOLOGY • Volume 22 • December 2007 • www.physiologyonline.org

EMERGING TOPICS

FIGURE 2. Laser scanning modes for in vivo two-photon microscopyIn each panel, the scan path of the laser focus is indicated by arrows. A: conventional raster scanin a single focal plane. B: with a fast z-scanning device, the image plane can be tilted. C: back-and-forth raster scan in 3D. D: 3D line scan employing an analytical function for the xy-deflection(in this case, an opening and closing spiral pattern). E: user-defined 3D scan trajectory. F: princi-pal idea of 3D random-access scanning by discontinuous displacements of the laser focus in 3D.


Dow

nloaded from

creo


input-output relationships of local cir-cuits in a purely optical manner (26).

In general, the calcium signals in thevarious cellular structures translateinto noisy fluorescence measurements(FIGURE 3D). In addition, some mixingof fluorescence signals between struc-tures may occur in the microscope, aswill be discussed further below. Noisein the best case is limited by shotnoise, which is due to the statisticalfluctuations in the detected photoncounts. Additional noise sourcesinclude signal detection electronicsand tissue movements. Noise is mostconfounding for the detection ofsmall calcium transients like those thatoccur during spiking activity in neu-ronal somata. We therefore considerthis case in more detail.

Optical Detection ofNeuronal Spikes

Assuming APs as the only trigger ofcalcium influx, spike patterns aredirectly reflected in the trains of calci-um transients. Mathematically speak-ing, each fluorescence trace is theconvolution of the spike train with thesingle AP-evoked calcium transientplus added noise (FIGURE 4A). Thechallenge is to reconstruct spike pat-terns from the noisy traces, which canbe done, for example, by using tem-plate-matching routines (26, 50), bycalculating the first derivative of thefluorescence signals and thresholdapplication (43, 49), or with deconvo-lution methods (64). Note that thetemporal resolution will be limited bythe acquisition rate of the networkscanning approach. In addition, thesignal-to-noise ratio (SNR) of fluores-cence signals will be a decisive factorfor the accuracy of the reconstruction.Here, some basic considerations mayhelp to optimize signal conditions.

Single AP-evoked calcium tran-sients depend on a number of factorsand may differ for different celltypes. For example, although singleAP-evoked calcium transients can beresolved in pyramidal cells (26), it isnot clear whether they aredetectable in fast-spiking interneu-rons in vivo. The amplitude �[Ca]

depends on the amount of calciumentry per AP, which is determined bychannel expression and AP width,whereas the time constant � dependson the calcium extrusion mechanismspresent. Both amplitude and timecourse are also affected by the calci-um indicator. Because the dye acts asan additional calcium buffer, calciumtransients are reduced and pro-

longed with increasing dye concen-tration. The magnitude of this effectdepends on the relative amounts ofthe calcium buffering capacity of theindicator dye (�dye) and the calciumbuffering capacity of endogenouscalcium-binding proteins (�S) (21).Buffering capacity is here defined asthe ratio of the change in buffer-bound calcium and the change in


EMERGING TOPICS

FIGURE 3. Signal components in bulk-loaded neocortical networkA: schematic of cellular structures labeled with MCBL. Other than cell bodies of neurons (green)and SR101-counterstained astrocytes (orange), cellular processes are stained in the neuropil,which includes dendrites as well as axons with presynaptic boutons. B: regions-of-interest (ROIs)selected from three neurons (blue), two astrocytes (red), and the neuropil (black). C: illustration oftypical calcium signals in the various cellular structures using artificially generated data. Neuronalcalcium transients reflect underlying spiking activity (top), whereas astrocytes show calcium waveson a longer time scale (middle). In the neuropil, fluctuations of the mean calcium signal areobserved, which during anaesthesia are linked to cortical up and down states (bottom). All struc-tures show spontaneous activity in vivo and may respond to sensory stimulation. D: calcium sig-nals translate into fluorescence signals, which typically are expressed as relative fluorescencechanges �F/F. In practice, fluorescence measurements are noisy, which was simulated here byadding gaussian-shaped noise. Moreover, signals may be contaminated slightly by neighboring structures.

“Distinct calcium signals are observed in the various tissue

compartments.”


Dow

nloaded from

creo


nals? Here, SNR is defined as the ratioof the absolute fluorescence change�F and the standard deviation of thebaseline fluorescence signal �F(FIGURE 4B). In the shot-noise limitedcase, �F equals the square root of theaverage fluorescence value F, whichincreases in proportion to the dyeconcentration (which itself in firstapproximation is proportional to �dye).With increasing dye concentration,however, the AP-evoked fluorescencechange �F follows a saturating func-tion simply because the dye cannotcapture more calcium ions than haveentered the cell. As a result, the noiselevel in the �F/F traces decreases withdye concentration but also so does

free calcium following calcium influx.In mathematical terms

Here, we assumed that �[Ca] is pro-portional to the relative fluorescencechange �F/F, which holds true forsmall calcium elevations far from dyesaturation. Note that �F/F furtherdepends on the baseline fluorescenceand dynamic range of the indicatortype. This discussion leads us to thequestion: Which dye concentrationprovides the highest SNR for themeasurement of somatic calcium sig-

the amplitude of the AP-evoked �F/Fchange (FIGURE 4B). It follows

and for the SNR under shot-noiselimited conditions

This dependence of SNR on the cal-cium indicator concentration is shownin FIGURE 4C. Because of the twoopposing effects of reduced noise butlarger buffering at higher dye concen-tration, the curve shows a maximum.The position of the maximum �max canbe found from the first derivative as (3)

where the simplification is justifiedbecause typical values of �S for corti-cal pyramidal neurons are 100–150(21). Hence, the optimal dye concen-tration is reached when the addedcalcium buffering capacity equals theendogenous one, a situation termed“balanced loading.” For detection ofsingle APs, high-affinity indicatorssuch as Oregon-Green BAPTA-1 (KD =0.2–0.3 �M) are required (which, how-ever, might saturate during trains ofAPs). For such indicators �S values of100–150 for pyramidal neurons corre-spond to an optimal dye concentra-tion in the range of 25–50 �M. Thisestimate is close to the estimates forMCBL (20 �M) (51), local electropora-tion (20 �M) (33), and targeted elec-troporation (50–250 �M) (34). Note,however, that other cell types, e.g.,GABAergic interneurons, may havedifferent requirements because theytypically possess a higher endoge-nous calcium buffering capacity. It isthus worth keeping the simple rule of“balanced loading” in mind and tovary the type of indicator, pipette dyeconcentration, loading times, or GECIexpression levels to optimize the SNRfor reconstruction of neuronal spiking activity.


EMERGING TOPICS

FIGURE 4. Action potential-evoked somatic calcium signalingA: principle of reconstructing spiking activity from calcium indicator fluorescence measurements,exemplified for three simulated neurons. Calcium signals are spike trains convolved by the singleAP-evoked calcium transients. From the resulting noisy fluorescence traces (�F/F), spiking activitycan be reconstructed by estimating either spike times or temporally smoothed instantaneousspike rates. The rate of errors in form of missed spikes (green) or false positives (red) will dependon the noise level. B: fluorescence signals depend on the calcium indicator concentration. Here,we simulated single-AP evoked fluorescence signals for three dye-loading conditions with anincreasing ratio of �dye over �S. Note that the relative fluorescence change �F/F is reduced athigher dye concentration, although �F slightly increases. The SNR is given by the ratio of �F and�F. C: dependency of SNR on calcium indicator concentration. A maximal SNR is achieved under“balanced loading” condition, i.e., when the calcium-binding capacity of the indicator �dye equalsthe calcium-binding capacity of the endogenous buffers �S.

�[Ca]� �

and � � (1 + �S + �dye)

�FF

11 + �S + �dye

�F � �F

1 + �S + �dye

�dye

1 + �S + �dye

SNR = � =

��dye

(1 + �S + �dye)

�F�F

�max = (1+�S) � �S

�dye

(1 + �S + �dye)��dye


Dow

nloaded from

creo


Contamination of SomaticSignals by Neuropil

Another problem one should beaware of, in particular when usingMCBL, is that neuropil calcium signalspossibly contaminate somatic signals, inthe worst case indicating a neuronalspike where none occurred (FIGURE 3D).Although this problem can bereduced by carefully selecting ROIsand restricting the analysis to wellfocused neurons (25, 50), there is afundamental aspect to it that relatesto the image-formation process in themicroscope. Despite “optical section-ing” in 2PLSM, mixing of fluorescencesignals can still occur as illustrated inFIGURE 5.

In a simple model of bolus-stainedneocortical tissue, we simulated a cellbody by a bright sphere embedded ina “sea” of relatively high neuropilbackground fluorescence (FIGURE 5A).The microscope image of any objectis given by the spatial convolution ofthe object with a characteristic func-tion called the “point spread func-tion” (PSF), which critically dependson the numerical aperture (NA) of themicroscope objective. For a two-pho-ton microscope, the two-photon PSF(2P-PSF) is given as the square of thePSF for one-photon excitation (19,66). To determine the relative contri-bution of background fluorescence tothe cell body signal, we simulated thetwo-photon images of neuropil and

cell body separately by convolvingthese objects with a theoretical 2P-PSF (19). Remarkably, cross-sectionalprofiles through the resulting imagedemonstrate a clear contribution ofthe neuropil fluorescence to the signaldetected within the simulated cellbody (FIGURE 5, B AND D). The profilesindicate a strong contamination of thesomatic signal at low NAs; but also forrelatively high NAs, blurred neuropilfluorescence contributes to the signalat the soma center, where back-ground fluorescence should be low-est. Hence, contamination cannot beneglected, especially since the effec-tively used NA typically is lower thanthe NA specified by the manufacturer[which assumes a plane wavefront fill-


EMERGING TOPICS

FIGURE 5. Simulation of neuropil contribution to somatic fluorescence signals.Simulation of neuropil contribution to somatic fluorescence signals. A: cell body (16-�m diameter sphere of intensity 1) in a sea of neuropil fluores-cence (intensity 0.7, mimicking MCBL loading with Oregon-Green BAPTA-1). 3D images of neuropil and cell body are separately processed by con-volving the objects with a theoretical 2P-PSF (shown for NA 0.5). The final image is obtained by a linear superposition of cell body and neuropilimages. B: intensity profiles through objects and images for three different NAs along the lateral (solid lines) and axial (broken lines) dimension (cor-responding to the yellow lines in A). Note that the intensity level in the center of the final image is a combination of fluorescence from cell body andneuropil. C: the effectively used NA for fluorescence excitation depends on how much the laser beam fills the objective's back aperture. D: depend-ency of neuropil contamination of somatic calcium signals on effective NA, taking into account fluorescence within a ROI that corresponds to theoriginal cell body diameter. At NA 0.5, about 30% of fluorescence inside the ROI originates from the neuropil, and even for high NAs (0.8-1.0) thereis a remaining contribution of around 10%.


Dow

nloaded from

creo


Work in the authors’ laboratory is supportedby the Swiss National Science Foundation, theHuman Frontier Science Program Organisation,and the Neuroscience Center Zurich.

References1. Adelsberger H, Garaschuk O, Konnerth A.

Cortical calcium waves in resting newborn mice.Nat Neurosci 8: 988–990, 2005.

2. Arenkiel BR, Peca J, Davison IG, Feliciano C,Deisseroth K, Augustine GJ, Ehlers MD, Feng G.In vivo light-induced activation of neural circuitryin transgenic mice expressing channelrhodopsin-2. Neuron 54: 205–218, 2007.

3. Borst JG, Helmchen F. Calcium influx during anaction potential. Methods Enzymol 293: 352–371,1998.

4. Broome BM, Jayaraman V, Laurent G. Encodingand decoding of overlapping odor sequences.Neuron 51: 467–482, 2006.

5. Brustein E, Marandi N, Kovalchuk Y, Drapeau P,Konnerth A. “In vivo” monitoring of neuronalnetwork activity in zebrafish by two-photonCa2+ imaging. Pflügers Arch 446: 766–773,2003.

6. Buzsaki G. Large-scale recording of neuronalensembles. Nat Neurosci 7: 446–451, 2004.

7. Denk W, Strickler JH, Webb WW. Two-photonlaser scanning fluorescence microscopy. Science248: 73–76, 1990.

8. Diez-Garcia J, Akemann W, Knöpfel T. In vivo cal-cium imaging from genetically specified targetcells in mouse cerebellum. Neuroimage 34:859–869, 2007.

9. Dombeck DA, Khabbaz AN, Collman F, Tank DW.Imaging large scale neural activity with cellularresolution in awake mobile mice. Neuron 56:43–57, 2007.

10. Garaschuk O, Griesbeck O, Konnerth A. TroponinC-based biosensors: a new family of geneticallyencoded indicators for in vivo calcium imaging inthe nervous system. Cell Calcium 42: 351–361,2007.

11. Garaschuk O, Milos RI, Grienberger C, MarandiN, Adelsberger H, Konnerth A. Optical monitor-ing of brain function in vivo: from neurons to net-works. Pflügers Arch 453: 385–396, 2006.

12. Garaschuk O, Milos RI, Konnerth A. Targetedbulk-loading of fluorescent indicators for two-photon brain imaging in vivo. Nat Protoc 1:380–386, 2006.

13. Gelperin A, Flores J. Vital staining from dye-coat-ed microprobes identifies new olfactory interneu-rons for optical and electrical recording. JNeurosci Methods 72: 97–108, 1997.

14. Göbel W, Helmchen F. New angles on neuronaldendrites in vivo. J Neurophysiol. In press.doi:10.1152/jn.00850.02007 2007.

15. Göbel W, Kampa BM, Helmchen F. Imaging cellu-lar network dynamics in three dimensions usingfast 3D laser scanning. Nat Methods 4: 73–79,2007.

16. Hammond AT, Glick BS. Raising the speed limitsfor 4D fluorescence microscopy. Traffic 1:935–940, 2000.

17. Hasan MT, Friedrich RW, Euler T, Larkum ME,Giese G, Both M, Duebel J, Waters J, Bujard H,Griesbeck O, Tsien RY, Nagai T, Miyawaki A, DenkW. Functional fluorescent Ca2+ indicator proteinsin transgenic mice under TET control. PLoS Biol 2:e163, 2004.

18. Heim N, Garaschuk O, Friedrich MW, Mank M,Milos RI, Kovalchuk Y, Konnerth A, Griesbeck O.Improved calcium imaging in transgenic miceexpressing a troponin C-based biosensor. NatMethods 4: 127–129, 2007.

ing the entire objective’s back-aper-ture, whereas for in vivo two-photonmicroscopy the beam profile isGaussian shaped, and the back-aper-ture often is under-filled to maximizetransmission of excitation light (19)].Although the problem of contamina-tion cannot be fully overcome, oneshould therefore aim for an illumina-tion NA as high as possible (tradingoff against power loss). A good way toverify little contamination is to searchfor bright neurons that show nosomatic signals despite large signalsin the surrounding neuropil.

Outlook

In summary, in vivo calcium imagingof network activity creates newopportunities to study the functionalarchitecture of neuronal and glial cir-cuits in the living brain. Furtheradvances in labeling and read-outmethods can be expected, especiallyfor fast 3D measurements from largepopulations. Improved GECIs willfacilitate functional measurementsfrom specific sub-networks, and newlaser scanning approaches will pushcurrent speed limits. Further optimiza-tion of fluorescence signals and appli-cation of sophisticated signal analysistools promise reliable single spikedetection from large numbers of neu-rons in 3D. This will allow examinationof correlations among individual cellsand identification of functionallylinked sub-networks that encode spe-cific stimulus features. Moreover, fromthe reconstructed spiking activity, onecan reveal the temporal evolution ofthe 3D network state, e.g., followingsensory stimulation, using advancedanalysis tools such as principal com-ponent analysis (4). Although mostexperiments at present are carriedout in anesthetized animals, severalapproaches for imaging in awakebehaving animals have been devisedthat ultimately aim at directly correlat-ing neuronal network dynamics withbehavior (1, 9, 20). Finally, throughexpression of light-activated channelproteins, it might become possible inthe future to not only read-out but alsocontrol neuronal networks in vivo (2).

19. Helmchen F, Denk W. Deep tissue two-photonmicroscopy. Nat Methods 2: 932–940, 2005.

20. Helmchen F, Fee MS, Tank DW, Denk W. A minia-ture head-mounted two-photon microscope.High-resolution brain imaging in freely movinganimals. Neuron 31: 903–912, 2001.

21. Helmchen F, Imoto K, Sakmann B. Ca2+ bufferingand action potential-evoked Ca2+ signaling indendrites of pyramidal neurons. Biophys J 70:1069–1081, 1996.

22. Helmchen F, Waters J. Ca2+ imaging in the mam-malian brain in vivo. Eur J Pharmacol 447:119–129, 2002.

23. Hirase H, Qian L, Bartho P, Buzsaki G. Calciumdynamics of cortical astrocytic networks in vivo.PLoS Biol 2: E96, 2004.

24. Iyer V, Hoogland TM, Saggau P. Fast functionalimaging of single neurons using random-accessmultiphoton (RAMP) microscopy. J Neurophysiol95: 535–545, 2006.

25. Kerr JN, de Kock CPJ, Greenberg DS, Bruno RM,Sakmann B, Helmchen F. Spatial organization ofneuronal population responses in layer 2/3 of ratbarrel cortex. J Neurosci. In press.

26. Kerr JN, Greenberg D, Helmchen F. Imaginginput and output of neocortical networks in vivo.Proc Natl Acad Sci USA 102: 14063–14068, 2005.

27. Knöpfel T, Diez-Garcia J, Akemann W. Opticalprobing of neuronal circuit dynamics: geneticallyencoded versus classical fluorescent sensors.Trends Neurosci 29: 160–166, 2006.

28. Kotlikoff MI. Genetically encoded Ca2+ indicators:using genetics and molecular design to under-stand complex physiology. J Physiol 578: 55–67,2007.

29. Kreitzer AC, Gee KR, Archer EA, Regehr WG.Monitoring presynaptic calcium dynamics in pro-jection fibers by in vivo loading of a novel calciumindicator. Neuron 27: 25–32, 2000.

30. Lin BJ, Chen TW, Schild D. Cell type-specific rela-tionships between spiking and [Ca2+]i in neuronsof the Xenopus tadpole olfactory bulb. J Physiol582: 163–175, 2007.

31. Miesenböck G, Kevrekidis IG. Optical imagingand control of genetically designated neurons infunctioning circuits. Annu Rev Neurosci 28:533–563, 2005.

32. Mrsic-Flogel TD, Hofer SB, Ohki K, Reid RC,Bonhoeffer T, Hübener M. Homeostatic regula-tion of eye-specific responses in visual cortex dur-ing ocular dominance plasticity. Neuron 54:961–972, 2007.

33. Nagayama S, Zeng S, Xiong W, Fletcher ML,Masurkar AV, Davis DJ, Pieribone VA, Chen WR. Invivo simultaneous tracing and Ca2+ imaging oflocal neuronal circuits. Neuron 53: 789–803, 2007.

34. Nevian T, Helmchen F. Calcium indicator loadingof neurons using single-cell electroporation.Pflügers Arch 454: 675–688, 2007.

35. Nguyen QT, Callamaras N, Hsieh C, Parker I.Construction of a two-photon microscope forvideo-rate Ca2+ imaging. Cell Calcium 30: 383-393, 2001.

36. Niell CM, Smith SJ. Functional imaging revealsrapid development of visual response propertiesin the zebrafish tectum. Neuron 45: 941-951,2005.

37. Nimmerjahn A, Kirchhoff F, Kerr JN, Helmchen F.Sulforhodamine 101 as a specific marker ofastroglia in the neocortex in vivo. Nat Methods 1:31-37, 2004.

38. O’Donovan MJ, Bonnot A, Wenner P, Mentis GZ.Calcium imaging of network function in the devel-oping spinal cord. Cell Calcium 37: 443-450,2005.


EMERGING TOPICS


Dow

nloaded from



EMERGING TOPICS

39. O’Donovan MJ, Ho S, Sholomenko G, Yee W.Real-time imaging of neurons retrogradely andanterogradely labelled with calcium-sensitivedyes. J Neurosci Methods 46: 91-106, 1993.

40. Ohki K, Chung S, Ch’ng Y, Kara P, Reid R.Functional imaging with cellular resolution revealsprecise micro-architecture in visual cortex. Nature433: 597-603, 2005.

41. Ohki K, Chung S, Kara P, Hubener M, BonhoefferT, Reid RC. Highly ordered arrangement of singleneurons in orientation pinwheels. Nature 442:925-928, 2006.

42. Pologruto TA, Yasuda R, Svoboda K. Monitoringneural activity and [Ca2+] with genetically encodedCa2+ indicators. J Neurosci 24: 9572-9579, 2004.

43. Ramdya P, Reiter B, Engert F. Reverse correlationof rapid calcium signals in the zebrafish optic tec-tum in vivo. J Neurosci Methods 157: 230-237,2006.

44. Reddy GD, Saggau P. Fast three-dimensional laserscanning scheme using acousto-optic deflectors. JBiomed Opt 10: 64038, 2005.

45. Ross WN, Nakamura T, Watanabe S, Larkum M,Lasser-Ross N. Synaptically activated Ca2+ releasefrom internal stores in CNS neurons. Cell MolNeurobiol 25: 283–295, 2005.

46. Rozsa B, Katona G, Vizi ES, Varallyay Z, Saghy A,Valenta L, Maak P, Fekete J, Banyasz A, Szipocs R.Random access three-dimensional two-photonmicroscopy. Appl Opt 46: 1860–1865, 2007.

47. Salome R, Kremer Y, Dieudonne S, Leger JF,Krichevsky O, Wyart C, Chatenay D, Bourdieu L.Ultrafast random-access scanning in two-photonmicroscopy using acousto-optic deflectors. JNeurosci Methods 154: 161–174, 2006.

48. Sato TR, Gray NW, Mainen ZF, Svoboda K. Thefunctional microarchitecture of the mouse barrelcortex. PLoS Biol 5: e189, 2007.

49. Smetters D, Majewska A, Yuste R. Detectingaction potentials in neuronal populations with cal-cium imaging. Methods 18: 215–221, 1999.

50. Sohya K, Kameyama K, Yanagawa Y, Obata K,Tsumoto T. GABAergic neurons are less selectiveto stimulus orientation than excitatory neurons inlayer II/III of visual cortex, as revealed by in vivofunctional Ca2+ imaging in transgenic mice. JNeurosci 27: 2145–2149, 2007.

51. Stosiek C, Garaschuk O, Holthoff K, Konnerth A.In vivo two-photon calcium imaging of neuronalnetworks. Proc Natl Acad Sci USA 100:7319–7324, 2003.

52. Sullivan MR, Nimmerjahn A, Sarkisov DV,Helmchen F, Wang SSH. In vivo calcium imaging ofcircuit activity in cerebellar cortex. J Neurophysiol94: 1636–1644, 2005.

53. Svoboda K, Denk W, Kleinfeld D, Tank DW. In vivodendritic calcium dynamics in neocortical pyrami-dal neurons. Nature 385: 161–165, 1997.

54. Svoboda K, Yasuda R. Principles of two-photonexcitation microscopy and its applications to neu-roscience. Neuron 50: 823–839, 2006.

55. Takano T, Han X, Deane R, Zlokovic B,Nedergaard M. Two-photon imaging of astrocyticCa2+ signaling and the microvasculature in exper-imental mice models of Alzheimer’s disease. AnnNY Acad Sci 1097: 40–50, 2007.

56. Takano T, Tian GF, Peng W, Lou N, Libionka W,Han X, Nedergaard M. Astrocyte-mediated con-trol of cerebral blood flow. Nat Neurosci 9:260–267, 2006.

57. Tamamaki N, Yanagawa Y, Tomioka R, Miyazaki J,Obata K, Kaneko T. Green fluorescent proteinexpression and colocalization with calretinin, par-valbumin, and somatostatin in the GAD67-GFPknock-in mouse. J Comp Neurol 467: 60–79,2003.

58. Tian GF, Azmi H, Takano T, Xu Q, Peng W, Lin J,Oberheim N, Lou N, Wang X, Zielke HR, Kang J,Nedergaard M. An astrocytic basis of epilepsy.Nat Med 11: 973–981, 2005.

59. Tian GF, Takano T, Lin JH, Wang X, Bekar L,Nedergaard M. Imaging of cortical astrocytesusing 2-photon laser scanning microscopy in theintact mouse brain. Adv Drug Delivery Res 58:773–787, 2006.

60. Tsien RY. A non-disruptive technique for loadingcalcium buffers and indicators into cells. Nature290: 527–528, 1981.

61. Wachowiak M, Cohen LB. Representation of odor-ants by receptor neuron input to the mouse olfac-tory bulb. Neuron 32: 723–735, 2001.

62. Wang X, Lou N, Xu Q, Tian GF, Peng WG, Han X,Kang J, Takano T, Nedergaard M. Astrocytic Ca2+

signaling evoked by sensory stimulation in vivo.Nat Neurosci 9: 816–823, 2006.

63. Wong RO. Calcium imaging and multielectroderecordings of global patterns of activity in thedeveloping nervous system. Histochem J 30:217–229, 1998.

64. Yaksi E, Friedrich RW. Reconstruction of firing ratechanges across neuronal populations by tempo-rally deconvolved Ca2+ imaging. Nat Methods 3:377–383, 2006.

65. Yaksi E, Judkewitz B, Friedrich RW. Topologicalreorganization of odor representations in theolfactory bulb. PLoS Biol 5: e178, 2007.

66. Zipfel WR, Williams RM, Webb WW. Nonlinearmagic: multiphoton microscopy in the bio-sciences. Nat Biotechnol 21: 1369–1377, 2003.


Dow

nloaded from


Documents

Werner Göbel and Fritjof Helmchen - UZH