Dear Author

The proof of your article, to be published in Neuroscience, is attached to this e-mail as a ‘PDF’ file. Also attached are instructions on the annotation of PDF files. You may also have a ‘Query Form’ as an attachment if we have any questions regarding your article.

Please note that proof corrections can now be annotatedon-screen, which allows you to mark directly in the PDF file, and return the marked file as an e-mail attachment. To take advantage of this method, please download the free Adobe Reader version 7 (or higher; see the attached instructions).

Alternative methods of returning proof corrections: If you do not wish to use the PDF annotations function, you may list the corrections (including replies to the Query Form) in an e-mail and return them to us using the ‘reply’ button to this e-mail. Please list your corrections quoting line number.

If, for any reason, this is not possible, mark the corrections and any other comments (including replies to the Query Form) on a printout of your proof and fax this to the number given below, or scan the pages and return them by e-mail, or post them to the address given below.

Please use this proof for checking the typesetting, editing, completeness and correctness of the text, tables and figures. Please ensure that you answer any questions (see the Query Form) that have arisen during the preparation of your proof. Significant changes to the article as accepted for publication will only be considered at this stage with permission from the Editor. We will do everything possible to get your article published quickly and accurately. Therefore, it is important to ensure that all of your corrections are sent back to us in one communication: please check carefully before replying, as inclusion of any subsequent corrections cannot be guaranteed.

If you submitted usable colour figures with your article they will appear in colour on the web, at no extra charge, as you can see in the attached PDF proof of your article. In the printed issue, colour reproduction depends on journal policy and whether or not you agree to bear any costs (not applicable to journals which appear only online). Any ‘supplementary’ material to your article (i.e., not appearing in print) will be accessible after your corrected article is placed online; such material is not part of the proofing procedure and is therefore not attached here.

Please respond promptly (even if you have no corrections): the sooner we hear from you, the sooner your corrected article will appear online. Note that we may proceed with publication of your article if no response is received.

Kind regards,

Elsevier E-mail corrections to: corrections.essd@elsevier.cadmus.com Fax: +1 (619) 699 6859

Contact Editorial Production Customer Support: +1 888 834 7287 (toll-free for US & Canadian customers)

UNCORREC

TED P

ROO

F

JOBNAME: AUTHOR QUERIES PAGE: 1 SESS: 3 OUTPUT: Sat Aug 21 21:25:50 2010/tapraid3/zpn�nsc/zpn�nsc/zpn99907/zpn4979d07z

AQ1— Please ensure that affiliations, affiliation addresses, present addresses, and correspondingauthor’s information are complete and accurate (street address, city, state if applicable, ZIPcode if applicable, province if applicable, country, e-mail address, and telephone and faxnumbers with country codes) and that all abbreviations are spelled out.

AQ2— Out of seven Keywords: please choose six keywords as per journal style.

AQ3— Please provide the manufacturer/donor’s location (city, state, province, country) and searchthe text for other instances of missing manufacturer/donor information.

AQ4— Please expand all abbreviations at first use or create the abbreviation at first use of theterm (except those that are never expanded or are expanded only in the “Abbreviations”section), and add them to the “Abbreviations” section if used more than once. Treat the title,abstract, and the text separately. Please check throughout the text for other instances ofinitially unexpanded abbreviations and ensure that any that have been expanded for you arecorrect.

AQ5— The following references were noted in the text but appear to be missing from yourreference list. Bower and Sultan (1998); Duch and Mentel, 2004; Klintsova et al., 2002;Gundersen et al., 1988; Heckroth, 1994. Please complete the list or remove the references fromthe text.

AQ6— Please provide an explanation for the superscript indicator or remove it from the table.

AUTHOR QUERIES

AUTHOR PLEASE ANSWER ALL QUERIES 1

UNCORREC

TED P

ROO

F

POPULATION BASED QUANTIFICATION OF DENDRITES: EVIDENCEFOR THE LACK OF MICROTUBULE-ASSOCIATE PROTEIN 2A,B INPurkinje CELL SPINY DENDRITES

S. HAMODEH,a D. EICKE,b R. M. A. NAPPER,c

R. J. HARVEYc AND F. SULTANa*aDepartment of Cognitive Neurology, HIH for Clinical Brain Research,Otfried-Müller-Street 21, 72076 Tuebingen, GermanybDepartment of Cellular Neurology, HIH for Clinical Brain Research,Otfried-Müller-Street 21, 72076 Tuebingen, GermanycDepartment of Anatomy and Structural Biology, University of OtagoMedical School, Dunedin, New Zealand

Abstract—The high molecular weight isoforms (a and b) ofmicrotubule-associate protein 2 (MAP2a,b) are widely be-lieved to be specific markers for neuronal somata and den-drites. We analyzed and quantified MAP2a,b stained den-drites of the cerebellar molecular layer using a novel ap-proach that segmented and 3D reconstructed them, and theresults have been compared with those obtained by othermethods, including single-cell reconstruction and analysis ofelectron micrographs. Our results show that the molecularlayer dendritic volume fraction is lower than in the neocortex(10% compared to neocortical 29%). The low total volumefraction of dendrites in the molecular layer is best explainedby the majority of the afferents to the dendrites being fromthe very densely packed parallel fibers, which allows thedendritic fields of individual neurons to be smaller and morecompact than in the cerebral cortex. However, the MAP2a,bdendritic volume fraction is even lower (5.2%) than the totalvolume fraction of dendrites in the molecular layer (10%).Analysis of the material shows that this difference betweenthe two results is due to the unexpected finding that therewere few MAP2a,b stained Purkinje cell spiny dendrites.© 2010 Published by Elsevier Ltd on behalf of IBRO.

Key words: molecular layer, dendritic molecular architecture,Purkinje cells, molecular interneurons, basket- and stellatecells, 3D reconstruction, electron microscopy.

Dendrites are an important feature of the neurons of thebrain since the vast majority of synapses terminate onthem (Fiala and Harris, 1999). The percentage of volumethat they occupy in the cerebral cortex varies—dependingon region—between 20 and 30% of the total brain volume(Braitenberg and Schüz, 1991; Chklovskii et al., 2002).Neurons come in a wealth of different types, the greatmajority originally described by Ramón y Cajal (1909) onthe basis of his studies using the rapid-Golgi staining tech-nique. Together with the cells’ axons and somata, the formof the dendrites is often used to distinguish the differentclasses of neurons (Ramón y Cajal, 1909). During devel-

opment the morphogenesis of dendritic trees is regulatedby innate genetic factors (Miller and Kaplan, 2003), exter-nal molecular cues (Kim and Chiba, 2004), and neuronalactivity (Duch and Mentel, 2004). In addition changes inneuronal morphology may occur during aging and neuro-degenerative diseases (de Brabander et al., 1998). Hence,the quantification of dendrites may help towards under-standing their role in developmental, physiological andcomputational, as well as pathological processes.

Currently the determination of dendritic morphology ismostly achieved by intracellular filling and 3D manual orsemi-automatic reconstruction. This single cell approachhas yielded detailed anatomical information for compart-mental models (De Schutter and Bower, 1994a,b). How-ever, this approach is time-consuming and hence only asmall number of cells have been studied. For instance,currently there are a handful of well-reconstructed Purkinjecells from only two species (Rapp et al., 1994; Roth andHäusser, 2001), which is not surprising if one considers theoverall length of a Purkinje cell dendritic tree (�10 mm inthe rat). We therefore opted for a population based anal-ysis by attempting to stain the dendrites of all neurons andrecovering some quantitative information (e.g. volumefraction, dendritic thickness and length) from semi-auto-mated 3D reconstructions.

A good candidate to specifically stain dendrites are thehigh molecular weight isoforms (a and b) of the microtubule-associated protein 2 (MAP2a,b). These isoforms are cy-toskeletal proteins confined to the neuronal dendritic com-partments and somata (Matus et al., 1981, 1983; Bernhardtand Matus, 1984). It is further considered a marker of struc-tural integrity because it is involved in morphological stabili-zation of dendritic processes (Farah and Leclerc, 2008;Conde and Caceres, 2009). Moreover, MAP2 expressioncoincides with dendritic outgrowth, branching, and post-le-sion dendritic remodeling, suggesting that this protein plays acrucial role in plasticity (Conde and Caceres, 2009).

In this study we quantified MAP2a,b distribution in themolecular layer of the cerebellar cortex after 3D recon-struction. We have implemented a set of previously estab-lished reconstruction algorithms to extract parameters thatdescribe the dendrite characteristics. In order to evaluatethe method as to its potential to quantify total dendriticvolume of a brain region, we have applied it to the molec-ular layer of the rat cerebellar cortex because this regionhas been well studied (Harvey and Napper, 1988, 1991;Napper and Harvey, 1988a,b) and its elements are welldefined (Sultan and Bower, 1998).

*Corresponding author. Tel: �49-7071-2980464; fax: �49-7071-295724.E-mail address: fahad.sultan@uni-tuebingen.de (F. Sultan).Abbreviations: MAP, microtubule associated protein 2; PB, phosphatebuffer.

Please cite this article in press as: Hamodeh S, et al., Population based quantification of dendrites: Evidence for the lack of microtu-bule-associate protein 2a,b in Purkinje cell spiny dendrites, Neuroscience (2010), doi: 10.1016/j.neuroscience.2010.08.021

Neuroscience xx (2010) xxx

0306-4522/10 $ - see front matter © 2010 Published by Elsevier Ltd on behalf of IBRO.doi:10.1016/j.neuroscience.2010.08.021

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455

AQ: 1

AQ: 2

123456789

10111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455

tapraid3/zpn-nsc/zpn-nsc/zpn99907/zpn4979d07z routa S�1 8/21/10 21:24 Art: 12338

1

UNCORREC

TED P

ROO

F

EXPERIMENTAL PROCEDURES

Adult rats (Sprague–Dawley) with body weights 175–325 g (2–3.5months of age) were used in this study. All experiments werecarried out in accordance with the institutional, national, and NIHguidelines on the use of animals in research. Rats were raised andkept in a room with a 12-h light/dark rhythm and with free accessto food and water. Animals were deeply anesthetized with amixture of ketamine (60 mg/kg body weight) and xylazine (6.5mg/kg body weight) plus atropine (1 mg/kg) and perfused throughthe ascending aorta with 4% paraformaldehyde prepared in 0.1 Mphosphate buffer (PB) pH 7.4. The cerebellum or forebrain wasremoved and postfixed in 4% paraformaldehyde (1–2 h) andcryoprotected in an ascending series of sucrose solutions (10%,15% and 30% sucrose). The cerebellum was then mounted forsectioning with a freezing microtome and serial sagittal or coronalsections were cut at 40 �m. The forebrain including hippocampuswas sectioned coronally at 40 �m.

Sections of the cerebellar cortex or hippocampus werewashed (3�5 min) in 0.1 M PB. Free-floating sections were incu-bated in a blocking buffer containing 10% horse serum (PAALaboratories, Coelbe, Germany) and 0.3% Triton X-100 in 0.1 MPB for 1 h at room temperature (RT). The sections were thenincubated in the primary monoclonal antibody (MAP2a �2b) pro-duced in mouse cells (clone AP-20, Sigma-Aldrich, Steinheim,Germany) for 36–48 h at 4 °C. The antibody was diluted at 1:1000in 0.1 M PB containing 2% horse serum and 0.1% Triton X-100.This antibody recognizes an epitope common to both a and bisoforms of high molecular weight MAP2 (Kalcheva et al., 1994).Both of these isoforms have been well characterized and shown tobe specifically associated with dendrites and somata (Binder etal., 1986; Hoskison et al., 2007; Huang et al., 2007; Lopez andWagner, 2009). Sections were then washed (3�5 min) in 0.1 MPB and incubated in the secondary antibody Alexafluor 488 goatanti-mouse IgG (H�L) antibody (Invitrogen, Karlsruhe, Germany)for 2 h at RT at a dilution of 1:500. Sections were then washed(3�5 min) in 0.1 M PB and finally mounted with mowiol 4–88(Merck, Darmstadt, Germany) in glycerol on glass slides. Fordouble-immunolabeling for MAP2a,b calbindin D-28K the sec-tions were incubated in both the MAP2 (2a�2b) and the rabbitpolyclonal anti-calbindin D-28K antibody (Millpore Chemicon,#AB1778) for 56 h at 4 °C. Both antibodies were diluted at 1:1000in 0.1 M phosphate buffer containing 2% horse serum and 0.1%Triton X-100. Sections were then washed (3�5 min) in 0.1 M PBand incubated in the secondary antibodies Alexafluor 488 goatanti-mouse and Alexafluor 568 goat anti-rabbit (Invitrogen,Karlsruhe, Germany) for 2 h at RT at a dilution of 1:500.

Sampling and fluorescence image acquisition

Sections were imaged with a confocal laser scanning microscope(LSM 510, Carl Zeiss, Jena, Germany) using the Argon laser(wavelength 488 nm) to acquire image stacks. Confocal imageswere taken at a distance below the slice surface to minimizecontribution of neurons damaged at the slice surface by the slicingprocedure. Samples were taken systematically at regular spacing(every 50 �m in x and y-direction, see Fig. 1A, B) from within

Fig. 1. Location of sample probes in the molecular layer are shown forsagittal (A) and coronal (B) sections. Squares indicate the 256�256pixel frame of the LSM z-stack probes. Square side length is 37.9 �m.(C, D) Dependence of antibody penetration on section depth. (C1–3)

shows projections views for the three planes for a probe with 91 slicescorresponding to 28.9 �m. (C3) is the projection view along the depthof the section. (C1, C2) show projection views perpendicular to (C3)that is parallel to section plane. Evident is the region of weaker stainingin the middle of projection views (C1, C2). Quantification of antibodypenetration is shown in (D). The maximal intensity values found withina z-section of the z-stack is plotted against its z-location and normal-ized to the overall maximal intensity found within the total volume dataset (number of probes�176; Error bars: standard deviation units;interrupted line: 95.4%).

Please cite this article in press as: Hamodeh S, et al., Population based quantification of dendrites: Evidence for the lack of microtu-bule-associate protein 2a,b in Purkinje cell spiny dendrites, Neuroscience (2010), doi: 10.1016/j.neuroscience.2010.08.021

S. Hamodeh et al. / Neuroscience xx (2010) xxx2

5657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114

AQ: 3

F1

5657585960616263646566676869707172737475767778798081828384858687888990919293949596979899

100101102103104105106107108109110111112113114

tapraid3/zpn-nsc/zpn-nsc/zpn99907/zpn4979d07z routa S�1 8/21/10 21:24 Art: 12338

UNCORREC

TED P

ROO

F

lobulus IV of the cerebellum using a 40� oil objective (Plan-Neofluar 40�/1.3 Oil DIC). Hippocampal samples were taken withthe same parameters from the CA1 stratum radiatum, except thatsample spacing was every 100 �m in x and y-direction. For eachsample, a stack of images were taken using the following imageacquisition parameters: pinhole diameter of 0.6 �m, z-stack inter-val of 0.31 �m (z-stack on average with 90 (sd�18) optical slices)and voxel x, y-size of 0.15�0.15 �m2 (matrix size 256�256pixels). Combined MAP2a,b and calbindin double immunofloures-cence imaging was performed using the Argon laser (wavelength488 nm) and Helium-Neon laser (wavelength 543 nm) in a multi(sequential)-tract configuration with a dichroic beam splitter at 545nm and a band pass filter 505–550 nm for the MAP2/488 nmchannel and a long pass filter at 550 nm for the calbindin/543 nmchannel. Image parameters were: 0.11�0.11 �m2 pixel size and amatrix of 1024�1024 pixels.

Segmentation and thickness reconstruction ofMAP2-labeled dendrites

The MAP2-labeled dendrites were reconstructed in 3D with amethod previously used to reconstruct blood vessels (Fouard etal., 2006). In this approach dendrites are represented by theircenterlines with local estimates of their diameters. Centerlines areachieved by iterative shrinkage, that is peeling off the boundarylayer-by-layer. Over shrinkage is prevented, in that the points aredeleted if they are not end points and if their deletion does notchange the object’s topology (Fouard et al., 2006), that is it doesnot lead to the splitting of a segment. Analysis was performed withdifferent routines supplied by the Amira software package (Amira4.1.1, Mercury Computer Systems, Chelmsford, MA, USA). Thebasic steps of segmentation and reconstruction algorithm con-sisted of the following: deconvolution, filtering, segmentation bythresholding, chamfer distance map calculation, a thinning pro-cess and finally conversion to a skeleton structure (see Fig. 2).

LSM images were deconvoluted using AutoQuant X2 software(Media Cybernetics, Bethesda, MD, USA) and its iterative-blindpoint spread function estimation routine (10 iterations) using max-imum likelihood estimation and constrained iteration. The pointspread function was further approximated by the imaging settings

such as wavelength, lens specifications, and voxel size (Fig. 2B).The images were then filtered in 3D with a Gaussian filter and a5�5�5 kernel and ��1 (Fig. 2C). Segmentation of the 8-bitgray-level images into a binary image was performed by settingthe threshold value to 8 (Fig. 2E). Next we calculated a distancemap using Amira’s chamfer map routine. This corresponds to agray level image where the value of each foreground voxel cor-responds to its shortest distance to the next background voxel.Voxels with relatively high values lie in the dendrite’s core andlower values lie nearer to the dendrite’s surface (Fig. 2F). Next weapplied the thinning procedure to the distance maps (Fouard et al.,2006) that preserves the topology and gives us the centerline ofthe dendrite (Fig. 2G). In the following step a skeleton structurewas produced in which the dendritic segments were representedas multiple cylinders (called nodes) with the dendritic thicknesscorresponding to the cylinder’s diameter (see Fig. 2H). The den-dritic radius is calculated between the center line and the bound-ary. The cylinder length corresponds to 0.5 �m, equal to thesampling rate along the dendrite. In the final step we manuallyremoved the stained somata of the neurons.

Dendritic diameter distribution—single cell approach

The Purkinje cell dendrite was obtained from the study of Roth andHäusser (2001) and is from a 19 postnatal day old rat Purkinje cellfilled intracellularly with biocytin in the version available fromNeuroMorpho (http://neuromorpho.org). The dendritic parametersof the interneurons of the molecular layer (IML, the so-calledbasket and stellate cells) were obtained from rapid Golgi stainedcells of the rat that we have described previously (Sultan andBower, 1998). The density of molecular interneurons and Purkinjecells were taken from Korbo et al. (1993). We then estimated thetotal dendritic volume fraction as the product of the average neu-ron volume and its density.

Electron microscopic analysis

Cerebellar sections from nine young adult rats that formed thecontrol series for a study of the effects of alcohol consumption oncerebellar development (Klintsova et al., 2002) were employed.

Fig. 2. Basic steps of segmentation and dendritic 3D reconstruction. (A) Projection view of LSM z-stack across the depth of the section of themolecular layer. (B, C) after deconvolution and Gaussian spatial smoothing, respectively. (D) dendrite in inlet of (C) shown at larger magnification. (E)Segmentation into fore- and background by thresholding. (F) Result of Chamfer map calculation. Voxels with high gray values lie within the dendrite’scenter whereas lower gray values lie nearer to the dendrite’s surface. (G) Extraction of centerline (dotted black line in dendritic center) by applying athinning process (see material and methods for further details). (H) Calculating dendritic radii by taking the distance from the centerline to thesegmentation border. This is done every 0.5 �m along the dendrite’s length. (I) Visualization of dendritic surface.

Please cite this article in press as: Hamodeh S, et al., Population based quantification of dendrites: Evidence for the lack of microtu-bule-associate protein 2a,b in Purkinje cell spiny dendrites, Neuroscience (2010), doi: 10.1016/j.neuroscience.2010.08.021

S. Hamodeh et al. / Neuroscience xx (2010) xxx 3

115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173

F2

115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173

tapraid3/zpn-nsc/zpn-nsc/zpn99907/zpn4979d07z routa S�1 8/21/10 21:24 Art: 12338

UNCORREC

TED P

ROO

F

The micrographs (magnification�21,600) were of pairs of serialsections (thickness 80 nm) at intervals of approximately 100 �mthrough the middle third of the molecular layer. To estimate thevolume fraction of Purkinje cell dendrites and dendritic spines andof interneuron somata and dendrites, the point-counting method(Gundersen et al., 1988; Heckroth, 1994) was used on one of thesections from each pair.

RESULTS

Description of MAP-2 immunofluorescent staining

Image stacks of 176 probes from 10 different locationsfrom three rats located within vermal and paravermal lob-ulus IV were acquired. Two locations are shown in Fig. 1A,B at low magnification. We also obtained 52 probes fromthe stratum radiatum of the hippocampal CA1 region ascontrols.

We first quantified the dependence of immunofluores-cence staining on section depth penetration (Fig. 1C, D).The maximal intensity value found within a section of thez-stack is plotted against its z-location. Intensity values arenormalized to the maximal intensity found within the totalvolume dataset. Both sagittal and coronal tissues showeda good (�94.5%) antibody penetration in the region of 2–7�m depth from the section surface. Dendrites within thisregion were taken as fully labeled.

Examples of MAP2 staining in the molecular layer areshown in Fig. 3. The antibody reaction was weak in the somataof the Purkinje cells (Fig. 3D). In comparison the somata of themolecular layer interneurons were more strongly stained(Fig. 3C3). The thick dendrites of the Purkinje cells were themost noticeable structures in the lower two-thirds of the mo-lecular layer (Fig. 3B, C1–3). Most of these thicker dendritescorrespond to the smooth dendrites of the Purkinje cells andhence not surprisingly were not observed close to the pialsurface. Staining in these smooth dendrites could be vari-able, possibly due to their location in different depths of thesection (Fig. 3B, C1–3). Strong MAP2 staining was observedin the interneuron dendrites and these structures constitutedthe most numerous structure in most of our samples. Inter-neuron dendrites located in the lower and middle part ofmolecular layer (Figs. 3C and 4C, above the Purkinje celllayer) often showed the most intense MAP2 staining.

Reconstruction algorithm

3D Reconstructions of MAP2-stained dendrites from differ-ent levels within the rat cerebellar molecular layer areshown in Fig. 4. Our reconstruction algorithm generallyextracts the 3D features of thick and thin dendrites remark-ably well. Comparing the 3D visualization of dendritic sur-face to the deconvoluted images shows that the majority ofvery thin dendrites are well reconstructed and the algo-rithm did not show any errors in which close-by dendritesmerged (Fig. 4A, B3). Some of thick dendrites exhibitedsmall holes in the deconvoluted images (Fig. 4A1-2; C1–2)that did not lead to a bifurcation in the reconstruction, thatis splitting of the single dendrite into two. Not surprisingly,in a few cases, where very thin dendrites were scantilylabelled, the reconstruction algorithm failed (Fig. 4A3).

Dendritic diameter distribution in the molecular layer

In Fig. 5A the diameter distribution of the MAP2 dendritesis shown. We binned and counted the diameters of thenodes (i.e. individual dendrite segments—see Material andMethods) as reconstructed from our MAP2-stained confo-cal images. The mean of the distribution was 0.54 �m(sd��0.4 �m). The smallest diameter found was 0.15 �m.This is the value we expected from the smallest recon-structable diameters determined by the minimal distancefrom one voxel center to the next voxel center.

The diameters obtained by the MAP2 reconstruction ap-proach were compared to published data previously obtainedby single cell staining (see Material and Methods). The majorneuronal elements of the molecular layer—apart from theparallel fibers—are the Purkinje cells and the molecularlayer interneurons, the basket and stellate cells. The Pur-kinje cell dendrites can be divided into the thicker smoothdendrites and the thinner more numerous spiny dendrites(Palay and Chan-Palay, 1974). The dendrites of the inter-neurons are thinner and shorter (1189 �m; Sultan andBower, 1998) but total length of their dendrites is compa-rable to that of the Purkinje cells. The dendritic length percell for Purkinje cells is about 10,000 �m (Roth andHäusser, 2001), however, the interneurons outnumber thePurkinje cells by a factor of 11.3 according (Korbo et al.,1993). Based on the different diameters obtained fromprevious studies (Sultan and Bower, 1998; Roth andHäusser, 2001) we can classify the dendrites into threeclasses, interneuron dendrites, Purkinje cell smooth den-drites and Purkinje cell spiny dendrites. The histogramdistribution of the dendritic diameters is shown in Fig. 5B.The mean diameter of the interneuron dendrites is 0.5 �m.The distribution of the Purkinje cell spiny dendrites is be-tween 0.5 and 2.2 �m (mean�1.09 �m). Finally the diam-eter of the Purkinje cell smooth dendrites is on average3.45 �m. Comparing our MAP2ab dendritic diameter dis-tribution (Fig. 5A, C) to that which we obtained from singlecell reconstruction, it is evident that most of the spinydendrites are absent in our MAP2-stained samples. On theother hand the thinner dendrites corresponding to the mo-lecular layer interneurons are stained and reconstructed.Also the thick smooth Purkinje cell dendrites were wellstained by MAP2. Most of the Purkinje cell spiny dendritesseem to lack MAP2 immunoreactivity (Figs. 3 and 4).

Quantification of dendritic volume fraction

The dendritic volume fraction that we obtained by quanti-fying the reconstructed MAP2 dendrites was 5.2%(sd��0.04; Fig. 6). This result was significantly lower thanthat which we obtained in the hippocampal CA1 stratumradiatum (22.4%, Fig. 6, see insert).

In Table 1 we compared the results of dendritic volumefraction of the cerebellar molecular layer estimation withthose obtained using two other approaches.

The first of these approaches was based on data ob-tained from single cell reconstructions. The volume of therat Purkinje cell dendrite (11824 �m3; n�1) was obtainedby Roth and Häusser (2001). The dendritic volume of

Please cite this article in press as: Hamodeh S, et al., Population based quantification of dendrites: Evidence for the lack of microtu-bule-associate protein 2a,b in Purkinje cell spiny dendrites, Neuroscience (2010), doi: 10.1016/j.neuroscience.2010.08.021

S. Hamodeh et al. / Neuroscience xx (2010) xxx4

174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232

F3

F4

174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232

F5

F6

T1

tapraid3/zpn-nsc/zpn-nsc/zpn99907/zpn4979d07z routa S�1 8/21/10 21:24 Art: 12338

UNCORREC

TED P

ROO

F

interneurons (231 �m3; n�29) was obtained from rapidGolgi-stained rat cerebella. The dendritic volume fractionof the cerebellar molecular layer can then be derived bymultiplying the individual dendritic volumes by their respec-tive neuronal density as obtained from the study of Korboet al. (1993). The Purkinje cell density is 6778/mm3 andthat of the interneurons is 76,667/mm3 of molecular

layer volume. This yields together a dendritic volumefraction of about 10%, with the interneuron fractionamounting to 1.8%. The Purkinje cell spiny dendriteswould account for 6% of the total dendritic volume frac-tion based on the data by Roth and Häusser (2001) onthe total surface area of the spiny dendrites excludingthe spine area (see Table 1).

Fig. 3. Examples of MAP2 immunofluorescence stained fibers in the cerebellar molecular layer. Shown are projection views across the LSM z-stackfrom different heights of the molecular layer (sagittal sections): (A1–3) upper molecular layer; (B1–3, C1–3) middle and lower molecular layer,respectively. (D1–3) Purkinje cell layer. Generally large and very thin diameter dendrites were well stained with the thinner dendrites exhibiting moreintensive staining. Purkinje cell layer (D1–3) was dominated by the faintly stained Purkinje cell somata. Many of the finer fibers can be followed overa considerable distance (up to 30 �m, see A2,3 and B3 marked by arrowheads). The morphological appearance of the thinner fibers fits well with thoseof the interneurons of the molecular layer and their previous description as long and thin dendrites (Palay and Chan-Palay, 1974; Sultan and Bower,1998). The thicker fibers that we observed were in the range of 1–5 �m diameter. Abbreviations: Pso, Purkinje cell soma; gl, granular layer; Id,Interneuron dendrite; Psd, Purkinje cell smooth dendrites; Iso, Interneuron soma; Scale bar�5 �m.

Please cite this article in press as: Hamodeh S, et al., Population based quantification of dendrites: Evidence for the lack of microtu-bule-associate protein 2a,b in Purkinje cell spiny dendrites, Neuroscience (2010), doi: 10.1016/j.neuroscience.2010.08.021

S. Hamodeh et al. / Neuroscience xx (2010) xxx 5

233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291

233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291

tapraid3/zpn-nsc/zpn-nsc/zpn99907/zpn4979d07z routa S�1 8/21/10 21:24 Art: 12338

UNCORREC

TED P

ROO

F

In the second approach, we estimated volume frac-tions from electron micrographs. A square grid was over-laid on the electron micrograph of one of each serial pair ofsections, as indicated in Fig. 7. At each grid crossing, theunderlying profile was identified, and counted if it was aPurkinje cell dendrite, or the soma or dendrite of a molec-ular layer interneuron. In a few cases the identificationcould be made more secure by examining the correspond-ing profile on the adjacent section. More than 10,000 in-

tersections on 89 micrographs were examined, corre-sponding to a total area of 23,500 �m2 of molecular layer.Purkinje cell dendrites were identified by their high con-centration of mitochondria, with usually obvious sub-plas-malemmal cisternae, and, often by their continuity withspines or spine necks. The majority of the Purkinje celldendrite profiles were of spiny dendrites, but a number oflarge (smooth) dendrite profiles were also seen. Purkinjecell spines were identified by their clear profiles, sur-

Fig. 4. Reconstructions of MAP2-stained dendrites from sagittal sections at different heights within the rat cerebellar molecular layer. (A1–3) uppermolecular layer; (B1–3, C1–3) upper and lower portions of middle molecular layer; (D1–3) lower molecular layer. Left column (A1–D1) shows the projectionviews of LSM z-stack within molecular layer after deconvolution. Middle column (A2–D2) shows the filtered images after Gaussian filtering. Rightcolumn (A3–D3) presents the visualization of the dendritic surface. Arrows indicate sites of holes in the dendrite. Arrow heads indicate site of dendriticloss. Asterisk indicates soma of interneuron. Scale bar�5 �m.

Please cite this article in press as: Hamodeh S, et al., Population based quantification of dendrites: Evidence for the lack of microtu-bule-associate protein 2a,b in Purkinje cell spiny dendrites, Neuroscience (2010), doi: 10.1016/j.neuroscience.2010.08.021

S. Hamodeh et al. / Neuroscience xx (2010) xxx6

292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350

F7

292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350

tapraid3/zpn-nsc/zpn-nsc/zpn99907/zpn4979d07z routa S�1 8/21/10 21:24 Art: 12338

UNCORREC

TED P

ROO

F

rounded by glia, except where they are in contact with apre-synaptic element, and their typical cisternae. Interneu-ron dendrites have less densely stained cytoplasm thanPurkinje cell dendrites, a lower concentration of mitochon-dria, no sub-plasmalemmal cisternae and are less com-pletely surrounded by glia (Palay and Chan-Palay, 1974).The counts showed that about 7.6% of the molecular layerwas occupied by Purkinje cell dendrites, 5.7% by Purkinjecell spines, 3.4% by interneuron somata and 2.7% byinterneuron dendrites (see Table 1 for comparison withother results).

Comparison of MAP-2a,b and calbindin D-28kimmunofluorescent staining

We compared molecular layer dendrites labeled withMAP2a,b with those labeled with calbindin D-28K. Theresults are shown in Fig. 8. As predicted from our resultsdescribed above, different patterns of staining were ob-served. MAP2a,b (Fig. 8A) stains the smooth dendrites ofthe Purkinje cells and the dendrites of the interneuronsvery well (in addition to the neurons’ somata). CalbindinD-28k also stains the smooth Purkinje cell dendrites welland, in addition, also stains the Purkinje cell spiny den-

Fig. 5. Comparison of MAP-2 dendritic diameter distribution in themolecular layer with previous results. Included are MAP-2 resultsobtained from sagittal and coronal sections. (A) Diameter distributionof the MAP-2 dendritic nodes. (B) Diameter distribution of dendriticsegments of interneurons (black) and Purkinje cells (white). Countsnormalized to correspond to the numeric relationship of interneuronsto Purkinje cells (Korbo et al., 1993). (C) Probability distribution ofMAP-2 results (gray) and the sum of distribution in (B) (interrupted)normalized to maximal counts. It is evident that the MAP-2 distributionlacks dendrite diameters corresponding to the spiny dendrites of thePurkinje cells (0.5–2 �m).

Fig. 6. Dendritic volume fractions of the cerebellar molecular layerobtained from MAP-2 reconstructed dendrites (176 probes from 3rats). The mean value was a dendritic volume fraction of 5.2%. Insert:comparison of MAP-2 volume fractions obtained from the cerebellarmolecular layer with volume fractions obtained from the hippocampalCA1 stratum radiatum. The latter had 22.4% (52 probes from one rat),a highly significant larger dendritic volume fraction than the former(t-test P�0.001).

Table 1. Summary and comparison of dendritic volume fraction ob-tained with different methods

Method Single cell MAP2a,b EM

Interneuron dendrites 1.8 5.2 2.7Purkinje cell smooth 2.4 7.6Purkinje cell spiny 6.0 n.s�

Total 10.2 5.2 10.3

n.s�not stained; single cell: data for interneuron dendrite fromBower and Sultan (1998); data for Purkinje cell dendrite from Roth andHäusser (2001).

Please cite this article in press as: Hamodeh S, et al., Population based quantification of dendrites: Evidence for the lack of microtu-bule-associate protein 2a,b in Purkinje cell spiny dendrites, Neuroscience (2010), doi: 10.1016/j.neuroscience.2010.08.021

S. Hamodeh et al. / Neuroscience xx (2010) xxx 7

351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409

F8

351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409

AQ: 6

tapraid3/zpn-nsc/zpn-nsc/zpn99907/zpn4979d07z routa S�1 8/21/10 21:24 Art: 12338

UNCORREC

TED P

ROO

F

drites, while the interneuron dendrites are weakly stained(Fig. 8B). These differences in staining also show up wellin doubly stained sections (Fig. 8C—MAP2a, b green,calbindin D-28k red).

DISCUSSION

Methodological considerations in relation toMAP2a,b studies

A number of potential factors could have influenced ourMAP2a,b quantification approach. These include for in-stance antibody specificity, low signal-to-noise (i.e. fore-ground-to-background) ratio and discontinuity in stainedstructures such as those visible in Fig. 4A1–2 and C1–2. Themonoclonal MAP2a,b antibody we have used is well char-

acterized and has been repeatedly used with consistentresults showing specific labeling of neuronal somata anddendrites in adult brains. The antibody recognizes bothisoforms (a and b) of the heavy molecular weight MAP2(Kalcheva et al., 1994; Chung et al., 1996). The otherremaining MAP2 isoforms are of low molecular weight andare only present during early development (Tucker et al.,1988; Matus, 1988). MAP2a,b is dendrite specific through-out development. According to Bernhardt and Matus(1984) its expression shows a distal to proximal gradient,with distal dendrites showing stronger labeling throughoutadulthood (older than 70 days). Our rats are of similar ageand although we may tentatively detect a distal to proximalgradient (see Fig. 3), this is present within the smoothdendrites of the Purkinje cells and does not extend into the

Fig. 7. Electron micgrograph of middle third of molecular layer overlaid with a square grid as used for volume fraction estimations. A selection of theprofiles counted are indicated in color. Red: Purkinje cell dendrites. Yellow: Purkinje cell dendritic spines. Blue: Interneuron soma. Green: Interneurondendrites. For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.

Fig. 8. Laser scanning micrograph of a sagittal section of the rat cerebellar cortex. (A) MAP2a,b staining showing strong labeling of in Purkinje cellsmooth dendrites and of interneuron dendrites. (B) Calbindin D-28k showing strong staining of Purkinje cells spiny dendrite. (C) Double immunoflu-orescence staining using MAP2 (green color) and calbindin D-28k (red) showing overlap of both labeling (yellow) in the Purkinje cells smooth dendritesand their somata. Purkinje cell spiny dendrites are largely stained for calbindin D-28k with only weak expression of MAP2a,b. In contrast, interneurondendrites show a preferential staining of MAP2a,b and a lack of calbindin D-28k expression. Abbreviations: Pso, Purkinje cell soma; Psd, Purkinje cellsmooth dendrites; spd, spiny dendrites of the Purkinje cells. Arrowheads: interneuron dendrite. Scale bar�10 �m. For interpretation of the referencesto color in this figure legend, the reader is referred to the Web version of this article.

Please cite this article in press as: Hamodeh S, et al., Population based quantification of dendrites: Evidence for the lack of microtu-bule-associate protein 2a,b in Purkinje cell spiny dendrites, Neuroscience (2010), doi: 10.1016/j.neuroscience.2010.08.021

S. Hamodeh et al. / Neuroscience xx (2010) xxx8

410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468

410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468

tapraid3/zpn-nsc/zpn-nsc/zpn99907/zpn4979d07z routa S�1 8/21/10 21:24 Art: 12338

UNCORREC

TED P

ROO

F

spiny dendrites. This result is confirmed by our doubleimmunoflourescence staining for MAP2a,b and calbindinD-28K which shows strong staining of the spiny dendritesfor the latter and very little for the former.

In contrast to most other neurons, Purkinje cell den-drites show a weaker staining of their main dendrite (Huberand Matus, 1984). This was also observed in our study,and could lead our reconstruction algorithm to split thedendrite into two. However careful inspection showed thatthis affected our reconstruction algorithm in only 5% (2/40)of cases. More distal dendrites and the dendrites of mo-lecular basket and stellate cells have generally been re-ported to be well-stained (Bernhardt and Matus, 1984).

Immunostaining is critically dependent on the antibodybeing able to penetrate the tissue and reach its targetantigen. To achieve full penetration and staining the tissuehas to be sectioned as thinly as possible (�2–6 �m).However, this would then lead to difficulties in reconstruc-tion and quantification. In this study we have limited ouranalysis to the regions within the tissue that show goodstaining (see Results). Our sections show that good stain-ing is on average present in depths between 2 and 7 �mfrom the section surface. These values are comparable toanother study showing good antibody penetration up to 12�m (Calhoun et al., 1996) in paraffin and ethanol perme-ated tissue.

Additional benefits were obtained by deconvolution ofthe raw images; the confocal image data can be correctedfor the microscope and tissue-specific point spread func-tion by applying deconvolution algorithms. The deconvo-luted filtered images yielded good foreground/backgroundseparation. A final critical factor in reconstruction was thelevel of thresholding. Our threshold was chosen to mini-mize linking of separate dendrites (out of 30 critical regionsnone showed linking at the threshold chosen here) whilefew very fine dendrites were lost.

MAP2 in Purkinje cell dendrites

A major finding of our study is the low volume fraction ofMAP2a,b in the cerebellar molecular layer. Our results indi-cate that this is both due to a lower dendritic volume fractionoverall than in cerebral cortex and to a negligible amount ofMAP2a,b staining in the spiny dendrites. The lower values ob-tained by our MAP2 approach could be due to severalmethodological problems (see beginning of Discussion).However, the values for dendrites other than Purkinje cellspiny dendrites are comparable to values we obtained bysingle-cell reconstruction estimates and also by electronmicroscopic analysis. In addition calbindin D-28k stainingrevealed well stained spiny dendrites.

MAP2a,b are used in dendrites to crosslink the micro-tubules, a role maintained largely by tau and MAP1b inaxons (Conde and Caceres, 2009). In the adult brainMAP2a,b suppression or degradation leads to dendriticloss or alterations of morphology (Faddis et al., 1997;Sanchez et al., 2000). MAP2a,b deficient mice show areduction in dendritic length (Harada et al., 2002). How-ever, it has also been shown that other proteins (MAP2c ortau) can help maintain some of the functions of MAP2a,b

(for review see Conde and Caceres, 2009). MAP2a,b arenot present in the dendritic spines of Purkinje cells (Bern-hardt and Matus, 1984). This is in agreement with theabsence of microtubules in Purkinje cell spines. The spinybranchlets are also known to express few microtubules(Palay and Chan-Palay, 1974). Hence our observations ofweak MAP2a,b expression in spiny dendrites is best ex-plained by the fewer microtubules cross-linked withMAP2a,b. The fewer number of microtubules is likely re-lated to the special arrangement of the Purkinje cell den-drite (see next paragraph).

Dendroarchitecture of the molecular layer

The dendritic volume fraction has been estimated to ac-count for 29% in the neocortex of mice based on electronmicroscopic and single cell approaches (Braitenberg andSchüz, 1991). This result was similar to values obtained inthe visual cortex of cats (Foh et al., 1973) and has alsobeen confirmed in the piriform cortex and hippocampuswith a dendritic volume fraction ranging from 20 to 30%(Chklovskii et al., 2002). The dendritic volume fraction ofthe hippocampal CA1 stratum radiatum that we obtainedhere with our MAP2a,b 3D reconstruction algorithm was22.4% and hence very similar to the previous studies (25%as deduced from Fig. 3 in Chklovskii et al., 2002). Thisindicates that our algorithm indeed works well if the den-drites express MAP2a,b in all their sub-compartments.Furthermore, it is becoming evident that the variabilityfound in the volume contributions by dendrites, axons,vessels and glia in different regions of the brain is less dueto methodological uncertainties and more due to real dif-ferences between regions. In a number of recent studies(Chklovskii et al., 2002, 2004; Stepanyants et al., 2004;Wen et al., 2009) it has been proposed that dendrites andaxons have different volume contributions reflecting prin-ciples of optimal wiring that reflect the presence of differentneuron classes and the computations they perform.

The present results are in good agreement with thesuggestion that brain architecture influences features suchas dendritic volume fraction. Our results using single cellreconstruction and electron microscopy indicate that vol-ume fraction of dendrites in the molecular layer is 10.2 or10.3% (see Table 1). These two values agree very well,although there are some differences in the contributionsfrom the dendrites of different types of cell. The valuesfrom single cell reconstruction will have a degree of uncer-tainty because some of the values included in the total arebased on figures obtained from a small number of individ-ual neurons. The EM studies used a sufficient area ofmicrograph so that the values given are close to the truevolume fractions for the middle third of the molecular layer.The values obtained using MAP2a,b material are consis-tent with the others if none or only a small number ofPurkinje cell spiny dendrites were stained and quantified.

The cerebellar cortex’s architectural/computational fin-ger print is feed-forward short-term excitation (in the milli-second range) mediated by the granule cells (Braitenberget al., 1997). The number of granule cells is very large,making them one of the most numerous neurons in the

Please cite this article in press as: Hamodeh S, et al., Population based quantification of dendrites: Evidence for the lack of microtu-bule-associate protein 2a,b in Purkinje cell spiny dendrites, Neuroscience (2010), doi: 10.1016/j.neuroscience.2010.08.021

S. Hamodeh et al. / Neuroscience xx (2010) xxx 9

469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527

469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527

AQ: 4

tapraid3/zpn-nsc/zpn-nsc/zpn99907/zpn4979d07z routa S�1 8/21/10 21:24 Art: 12338

UNCORREC

TED P

ROO

F

mammalian brain (Andersen et al., 2003) and allowing fora vast number of parallel computations (Nelson andBower, 1990). However the effect of a single granule cellupon its main postsynaptic target—the Purkinje cell––isweak (Isope and Barbour, 2002). The low efficacy of theseindividual synaptic connections is perhaps related to thefact that a parallel fiber normally contacts each Purkinjecell only once. Even so, the large number of granule cellsleads to a large number of granule cell/Purkinje cell syn-apses. These synapses are located on Purkinje cell spinesand have been estimated to number 150,000–175,000 perPurkinje cell, corresponding to 17.2 spines per linear mi-cron of a spiny branchlet (Napper and Harvey, 1988b),although for many of them, presynaptic activity may nothave any postsynaptic effect (Isope and Barbour, 2002).Because of the very dense packing of the parallel fibersand the large spine density, in combination with the planeof the Purkinje cell dendritic tree being perpendicular to theparallel fibers, the large number of fibers can connect to aPurkinje cell with much less dendritic length (and volume)than for a cerebral cortical neuron (as indicated diagram-matically in Fig. 9A). Purkinje cell dendrites are remarkablefor their dense branching, so that very little of the area

within the spread of the dendritic tree is more than about 2�m (the maximum spine length, Palay and Chan-Palay,1974; Napper and Harvey, 1988a) from a dendrite, and is,therefore, within the “spine-reach zone” of the dendrites(Wen et al., 2009 see Fig. 9B1, B2). Although a Purkinjecell dendritic tree is of similar length to that of a neocorticalpyramidal cell it is much more tightly packed and occupiesa much smaller space overall. Such an arrangement mayalso allow the Purkinje cell to require a much small numberof microtubules, which could be a metabolic advantage.Searching for appropriate presynaptic axons requires amuch larger volume to be covered in the neocortex than inthe cerebellar molecular layer, where the Purkinje celldendritic tree is contacted by the majority of the potentialafferent axons (the parallel fibers) passing through itsplane. It would not be surprising that such a dense ar-rangement of dendrites and spines may require specialmaintenance as indicated by the elaborate glia of themolecular layer the Bergmann glia is estimated to occupyup to 18% of the volume of the molecular layer (Andersonet al., 1994; Grosche et al., 2002; a value of about 20%was found in preliminary counts on the electron micro-graphs studied here), considerably greater than found inthe neocortex, where only 11% is occupied by glia (Brait-enberg and Schüz, 1991).

REFERENCES

Andersen BB, Gundersen HJ, Pakkenberg B (2003) Aging of thehuman cerebellum: a stereological study. J Comp Neurol466:356–365.

Anderson BJ, Li X, Alcantara AA, Isaacs KR, Black JE, Greenough WT(1994) Glial hypertrophy is associated with synaptogenesis follow-ing motor-skill learning, but not with angiogenesis following exer-cise. Glia 11:73–80.

Bernhardt R, Matus A (1984) Light and electron microscopic studies ofthe distribution of microtubule-associated protein 2 in rat brain: adifference between dendritic and axonal cytoskeletons. J CompNeurol 226:203–221.

Binder LI, Frankfurter A, Rebhun LI (1986) Differential localization ofMAP-2 and tau in mammalian neurons in situ. Ann N Y Acad Sci466:145–166.

Braitenberg V, Heck D, Sultan F (1997) The detection and generationof sequences as a key to cerebellar function: experiments andtheory. Behav Brain Sci 20:229–245.

Braitenberg V, Schüz A (1991) Anatomy of the cortex. Berlin:Springer-Verlag.

Calhoun ME, Jucker M, Martin LJ, Thinakaran G, Price DL, Mouton PR(1996) Comparative evaluation of synaptophysin-based methodsfor quantification of synapses. J Neurocytol 25:821–828.

Chklovskii DB, Mel BW, Svoboda K (2004) Cortical rewiring andinformation storage. Nature 431:782–788.

Chklovskii DB, Schikorski T, Stevens CF (2002) Wiring optimization incortical circuits. Neuron 34:341–347.

Chung WJ, Kindler S, Seidenbecher C, Garner CC (1996) MAP2a, analternatively spliced variant of microtubule-associated protein 2.J Neurochem 66:1273–1281.

Conde C, Caceres A (2009) Microtubule assembly, organization anddynamics in axons and dendrites. Nat Rev Neurosci 10:319–332.

de Brabander JM, Kramers RJ, Uylings HB (1998) Layer-specificdendritic regression of pyramidal cells with ageing in the humanprefrontal cortex. Eur J Neurosci 10:1261–1269.

De Schutter E, Bower JM (1994a) An active membrane model of thecerebellar Purkinje cell. I. Simulation of current clamps in slice.J Neurophysiol 71:375–400.

Fig. 9. Diagrammatic model relating cortical architecture to dendriticvolume fraction. (A) Low dendritic volume fraction in molecular layerand high length density of spines in Purkinje cells compared to otherneurons. Less dendritic length is required in the case of Purkinje cells.(A1: Pyramidal cells, A2: Purkinje cells.) (B) Comparison of dendriticmorphologies and spatial extent of a Purkinje cell (B2, from Roth andHäusser, 2001) and a Pyramidal cell (B1: cell g0703P1 from neuro-morpho.org/Svoboda). These two cell types have on average similarlylong dendrites (approximately 10 mm) but have radically differentspatial extensions. The Purkinje cell dendrite samples its space verydensely, establishing contacts with about 50% of the axons that tra-verse its tree (6�106/mm2 gives 31,2500 axons crossing the 0.052mm2 of the PC dendritic tree, and these provide 150,000� synapses).Scale bar 100 �m.

Please cite this article in press as: Hamodeh S, et al., Population based quantification of dendrites: Evidence for the lack of microtu-bule-associate protein 2a,b in Purkinje cell spiny dendrites, Neuroscience (2010), doi: 10.1016/j.neuroscience.2010.08.021

S. Hamodeh et al. / Neuroscience xx (2010) xxx10

528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586

F9

528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586

AQ: 5

tapraid3/zpn-nsc/zpn-nsc/zpn99907/zpn4979d07z routa S�1 8/21/10 21:24 Art: 12338

UNCORREC

TED P

ROO

F

De Schutter E, Bower JM (1994b) An active membrane model of thecerebellar Purkinje cell II. Simulation of synaptic responses. J Neu-rophysiol 71:401–419.

Faddis BT, Hasbani MJ, Goldberg MP (1997) Calpain activation con-tributes to dendritic remodeling after brief excitotoxic injury in vitro.J Neurosci 17:951–959.

Farah CA, Leclerc N (2008) HMWMAP2: new perspectives on apathway to dendritic identity. Cell Motil Cytoskeleton 65:515–527.

Fiala JC, Harris KM (1999) Dendrite structure. In: Dendrites, (Stuart G,Spruston N, Häusser M, eds). Oxford, UK: Oxford University Press.

Foh E, Haug H, Konig M, Rast A (1973) Determination of quantitativeparameters of the fine structure in the visual cortex of the cat, alsoa methodological contribution on measuring the neuropil. MicroscActa 75:148–168.

Fouard C, Malandain G, Prohaska S, Westerhoff M (2006) Blockwiseprocessing applied to brain microvascular network study. IEEETrans Med Imaging 25:1319–1328.

Grosche J, Kettenmann H, Reichenbach A (2002) Bergmann glial cellsform distinct morphological structures to interact with cerebellarneurons. J Neurosci Res 68:138–149.

Harada A, Teng J, Takei Y, Oguchi K, Hirokawa N (2002) MAP2 isrequired for dendrite elongation, PKA anchoring in dendrites, andproper PKA signal transduction. J Cell Biol 158:541–549.

Harvey RJ, Napper RMA (1988) Quantitative study of granular andPurkinje cells in the cerebellar cortex of the rat. J Comp Neurol274:151–157.

Harvey RJ, Napper RMA (1991) Quantitative studies on the mamma-lian cerebellum. Prog Neurobiol 36:437–463.

Hoskison MM, Yanagawa Y, Obata K, Shuttleworth CW (2007) Calcium-dependent NMDA-induced dendritic injury and MAP2 loss in acutehippocampal slices. Neuroscience 145:66–79.

Huang J, Furuya A, Furuichi T (2007) Very-KIND, a KIND domaincontaining RasGEF, controls dendrite growth by linking Ras smallGTPases and MAP2. J Cell Biol 179:539–552.

Huber G, Matus A (1984) Differences in the cellular distributions of twomicrotubule-associated proteins, MAP1 and MAP2, in rat brain.J Neurosci 4:151–160.

Isope P, Barbour B (2002) Properties of unitary granule cell—�Pur-kinje cell synapses in adult rat cerebellar slices. J Neurosci22:9668–9678.

Kalcheva N, Albala JS, Binder LI, Shafit-Zagardo B (1994) Localizationof specific epitopes on human microtubule-associated protein 2.J Neurochem 63:2336–2341.

Kim S, Chiba A (2004) Dendritic guidance. Trends Neurosci27:194–202.

Korbo L, Andersen BB, Ladefoged O, Moller A (1993) Total numbersof various cell types in rat cerebellar cortex estimated using anunbiased stereological method. Brain Res 609:262–268.

Lopez V, Wagner CK (2009) Progestin receptor is transiently ex-pressed perinatally in neurons of the rat isocortex. J Comp Neurol512:124–139.

Matus A (1988) Microtubule-associated proteins: their potential role indetermining neuronal morphology. Annu Rev Neurosci 11:29–44.

Matus A, Bernhardt R, Hugh-Jones T (1981) High molecular weightmicrotubule-associated proteins are preferentially associated withdendritic microtubules in brain. Proc Natl Acad Sci U S A78:3010–3014.

Matus A, Huber G, Bernhardt R (1983) Neuronal microdifferentiation.Cold Spring Harb Symp Quant Biol 48 (Pt 2):775–782.

Miller FD, Kaplan DR (2003) Signaling mechanisms underlying den-drite formation. Curr Opin Neurobiol 13:391–398.

Napper RMA, Harvey RJ (1988a) Quantitative study of the Purkinjecell dentritic spines in the rat cerebellum. J Comp Neurol274:158–167.

Napper RMA, Harvey RJ (1988b) Number of parallel fiber synapses onan individual Purkinje cell in the cerebellum of the rat. J CompNeurol 274:168–177.

Nelson ME, Bower JM (1990) Brain maps and parallel computers.Trends Neurosci 13:403–408.

Palay SL, Chan-Palay V (1974) Cerebellar cortex. Cytology and orga-nization. Berlin: Springer-Verlag.

Ramón y Cajal S (1909) Histolgie du système nerveux de l’homme etles vertébrés. Paris: Maloine.

Rapp M, Segev I, Yarom Y (1994) Physiology, morphology and de-tailed passive models of guinea-pig cerebellar Purkinje cells.J Physiol (Lond) 474:101–118.

Roth A, Häusser M (2001) Compartmental models of rat cerebellarPurkinje cells based on simultaneous somatic and dendritic patch-clamp recordings. J Physiol 535:445–472.

Sanchez C, Diaz-Nido J, Avila J (2000) Phosphorylation of microtu-bule-associated protein 2 (MAP2) and its relevance for the regu-lation of the neuronal cytoskeleton function. Prog Neurobiol61:133–168.

Stepanyants A, Tamas G, Chklovskii DB (2004) Class-specific fea-tures of neuronal wiring. Neuron 43:251–259.

Sultan F, Bower JM (1998) Quantitative Golgi study of the rat cere-bellar molecular layer interneurons using principal componentanalysis. J Comp Neurol 393:353–373.

Tucker RP, Binder LI, Viereck C, Hemmings BA, Matus AI (1988) Thesequential appearance of low- and high-molecular-weight forms ofMAP2 in the developing cerebellum. J Neurosci 8:4503–4512.

Wen Q, Stepenyants A, Elston GN, Grosberg AY, Chklovskii DB(2009) Maximisation of the connectivity repertoire as a statisticalprinciple governing the shapes of dendritic arbors. Proc Natl AcadSci U S A 30:12536–12541.

(Accepted 11 August 2010)

Please cite this article in press as: Hamodeh S, et al., Population based quantification of dendrites: Evidence for the lack of microtu-bule-associate protein 2a,b in Purkinje cell spiny dendrites, Neuroscience (2010), doi: 10.1016/j.neuroscience.2010.08.021

S. Hamodeh et al. / Neuroscience xx (2010) xxx 11

587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645

587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645

tapraid3/zpn-nsc/zpn-nsc/zpn99907/zpn4979d07z routa S�1 8/21/10 21:24 Art: 12338