electron tomography study of isolated human centrioles

Electron Tomography Study of Isolated Human CentriolesRANA IBRAHIM,1,2 CEDRIC MESSAOUDI,1,2 FRANCISCO JAVIER CHICHON,3

CLAUDE CELATI,4,5 AND SERGIO MARCO1,2*1Institut Curie, Centre de Recherche, Orsay, F-91405 France2INSERM, U759, Orsay, F-91405 France3Centro Nacional de Biotecnologıa (CNB-CSIC), Department of Macromolecular Structure,Campus de Cantoblanco, 28049 Madrid, Spain4Institut Curie, Centre de Recherche, Paris, F-75248 France5CNRS, UMR144, Paris, F-75248 France

KEY WORDS electron microscopy; centrosome; structure

ABSTRACT Centrioles are components of the centrosome, which is present in most eukaryoticcells (from protozoa to mammals). They organize the microtubule skeleton during interphase and themitotic spindle during cell division. In ciliate cells, centrioles form basal bodies that are involved incellular motility. Despite their important roles in biology, the detailed structure of centrioles remainsobscure. This work contributes to a more complete model of centriole structure. The authors usedelectron tomography of isolated centrosomes from the human lymphoblast KE37 to explore thedetails of subdistal appendages and centriole lumen organization in mother centrioles. Their resultsreveal that each of the nine subdistal appendages is composed of two halves (20 nm diameter each)fused in a 40 nm tip that extends 100 nm from where it anchors to microtubules. The centriole lumenis filled at the distal domain by a 45 nm periodic stack of rings. Each ring has a 30 nm diameter,is 15 nm thick, and appears to be tilted at 538 perpendicular to the centriole axis. The rings areanchored to microtubules by arms. Based on their results, the authors propose a model of the mothercentriole distal structure.Microsc. Res. Tech. 72:42–48, 2009. VVC 2008 Wiley-Liss, Inc.

INTRODUCTION

Centrosomes are ubiquitous organelles present inmost eukaryotic species except land plants. They havebeen described as ‘‘a central body, tightly associatedwith the nucleus, which duplicates once during the cellcycle and which acts as a microtubule organizing cen-ter’’ (Paintrand et al., 1992). The centrosome consistsof a core structure, the centriole pair, that is linked to-gether by the pericentriole matrix (PCM) (Vorobjev andNadezhdina, 1987) composed of large coiled proteins(Dammermann et al., 2004; Schnackenberg and Pal-azzo, 1999). Centrioles are involved in a large numberof essential cell processes, including microtubule nucle-ation and organization during interphase or mitosis,and cilia assembly (Marshall and Rosenbaum, 2003;Rebollo et al., 2007). During cell division cycle, centro-somes reproduce by duplicating their centrioles andmatrix. Pro-centrioles (daughter centrioles) grow atright angles close to the proximal ends of the mothercentrioles. At the onset of mitosis, the two newlyformed centrosomes separate and migrate to each poleof the cell. At the end of mitosis, the mother andappended daughter centrioles disengage from eachother and maintain a link through the PCM. Therefore,depending on the stage of the cell cycle, the centrosomecontains either a pair of centrioles or a pair of duplicat-ing centrioles (Delattre and Gonczy, 2004; Krameret al., 2004; Nigg, 2007).

Despite the major role of centrioles, a detailed struc-tural description is lacking. Recent tomographic stud-ies have provided information on the structural organi-zation of centriole-related structures (basal bodies and

spindle pole bodies) in mammalian (Kenney et al.,1997) and in unicellular organisms (Muller et al., 2005;O’Toole et al., 1999, 2003). More recently, tomographyhas contributed to an understanding of the structureand assembly mechanism of nematode centrioles (Pel-letier et al., 2006). However, knowledge of the 3D struc-ture of human centrioles has mainly depended ontransmission electron microscopy (TEM) studies per-formed on serial sections of resin-embedded material(Nitschke et al., 1995; Paintrand et al., 1992). Humancentrioles are described as barrel-shaped structuresthat are 300–500 nm in length, 200–250 nm in diame-ter, and have ends (proximal and distal) that are notidentical. The proximal domain consists of nine micro-tubule triplets and the distal domain consists of ninemicrotubule doublets. The distal domain of the cen-triole has subdistal and distal appendages that pointtowards the outside of the barrel. Subdistal appen-dages have been described as nine sets of filamentsthat extend out from the nine microtubule doublets atan angle with respect to the centriole axis, and the tips

*Correspondence to: Sergio Marco, Institut Curie, Centre de Recherche, Labo-ratoire Raymond Latarjet, INSERM U759, Imagerie Integrative, Equiped’Imagerie Cellulaire, Centre Universitaire d’Orsay, Bat 112, 91405 OrsayCedex, France. E-mail: [email protected]

Received 24 February 2008; accepted in revised form 31 July 2008

Contract grant sponsor: 3DEM European Network; Contract grant number:LSHG-CT-2004-502828; Contract grant sponsor: Curie Institute (PIC Physicochimie du vivant); Contract grant sponsor: ANR; Contract grant number:PCV06_142771.

DOI 10.1002/jemt.20637

Published online 6 October 2008 in Wiley InterScience (www.interscience.wiley.com).

VVC 2008 WILEY-LISS, INC.

MICROSCOPY RESEARCH AND TECHNIQUE 72:42–48 (2009)

of each set of filaments fuse to a common point. Distalappendages have been described as rigid sticks locatedat the base of the microtubule doublets. In addition, inthe middle of the distal domain, there is an internalstructure described as an amorphous hub (Vorobjevand Chentsov, 1980). Further study of isolated centro-somes in serial sections has proposed a helical organi-zation (Paintrand et al., 1992). Cryo-electron micros-copy studies performed on mammalian centrioles(Chretien et al., 1997) and cryotomography (Kenneyet al., 1997) have confirmed most of the previouslydescribed features. In this work, we used electron to-mography to study isolated resin-embedded humancentrioles. This approach has allowed an improveddescription of the distal lumen and subdistal appen-dages. On the basis of this tomographic data and previ-ous results from serial sections, a model of the centriolestructure is proposed.

MATERIALS AND METHODSCentrosome Isolation

Centrosomes were isolated from lymphoid KE37 cellsaccording to the procedure previously described (Bor-nens et al., 1987). Briefly, KE37 cells were cultured inRPMI 1640 media containing 20% fetal calf serum andmaintained in an atmosphere saturated in 5% CO2.Cultures were treated with nocodazole (0.2 lM) and cy-tochalasin D (1 lg/lL) and lysed with Nonidet (NP40)in the presence of protease inhibitors. Lysates werecentrifuged and the supernatant was applied to a 40–70% sucrose gradient. Centrioles isolated in the 60%sucrose fractions were used for electron tomographypreparations.

Specimen Preparation for Electron Tomography

To enable direct comparisons between electron to-mography and TEM, we used the same protocol for iso-lating centrosomes as that used in previous studies(Paintrand et al., 1992). In brief, fractions containingisolated centrioles were centrifuged on coverslips, andthen fixed with 2.5% glutaraldehyde in 0.1 M cacodyl-ate buffer, pH 7.3, for 30 min. After fixation, centrioleswere washed in 0.1 M cacodylate buffer containing0.15% tannic acid and postfixed in 2% aqueous OsO4

for 10 min at 48C. Next, they were stained in 1% uranylacetate (48C; 60 min), washed in distilled water, dehy-drated in a graded series (50, 70, 90, and 100%) of etha-nol, and flat-embedded in Epon. Semi-thick sections(defined by a deep purple interference that corre-sponded to about 300 nm) were cut from the Epon blocswith a Leica Ultracut E microtome.

Electron Microscopy and TomographicReconstruction of the Centriole

Thirteen tilt series of centrioles were imaged andseven were investigated further. These were acquiredin zero-loss conditions (2508 to 508 with a tilt step of18) in a LEO 912X (Carl Zeiss SMT AG, Oberkochen,Germany) operating at 120 kV. Images were recordedat 16,0003 nominal magnification using plug-insdeveloped in our laboratory for Analysis1 software anda Proscan 1k ssCCD (pixel size 1.6 nm/px). A copy ofthe image stack was filtered (median filter, radius 3)and then a threshold was applied. The resulting stackwas aligned by cross-correlation using the TomoJ align-

ment option (Messaoudi et al., 2007). Shifts from cross-correlated series were applied to the original projectionimages. After the original projections were aligned, 5123 512 pixel subimages were cropped. A second align-ment was performed by cross-correlation on the sub-image stack. After alignment, each tilt series was usedas input for the algebraic reconstruction techniques(ART) reconstruction algorithm (Herman et al., 1973)with four iterations and a relaxation coefficient equalto 0.1. This algorithm has been demonstrated to pro-vide better reconstructions than weighted back projec-tion (WBP) when a missing wedge is important, imagesare noisy, or structures are not clearly delimitated. Allcomputations were carried out with the TomoJ plug-in(Messaoudi et al., 2007) integrated into the ImageJsoftware (Abramoff et al., 2004) on a SUN system (Bi-opteron 250, 16-Gbyte RAM).

For each reconstructed volume, we applied a denois-ing process to enhance characteristic centriole featureswithin the volumetric representation. Thus, Gaussianfiltering (standard deviation set to 0.5) (Sonka et al.,1999) and anisotropic nonlinear diffusion (AND) werecombined using TOMOAND software (Fernandez andLi, 2003, 2005). The AND was run with 10 iterationsusing the following parameters: coherence enhance-ment diffusion constant 5 1; edge-enhancing diffusionconstant 5 1; coherence enhancement diffusion/edge-enhancing ratio diffusion balance 5 1; coherenceenhancement diffusion proportion along 2nd eigenvec-tor 5 1; smoothing proportion based on gray level 5 1;initial standard deviation 5 1; standard deviation foraveraging structural tensor 5 1; time interval (ht) 50.4; and stencil5 5.

Denoised volumes were segmented using AMIRA(TGS Europe, Merignac, France) with semi-automaticmask generation tools. In AMIRA, multiple segmentedregions can be visualized simultaneously in a commonviewer window. Final rendering was performed by Chi-mera (Pettersen et al., 2004) without manual featureenhancement. Thus, different centriole domains thatbelonged to a single reconstructed centriole and thathad been segmented with AMIRA were independentlysaved and imported into Chimera for visualization andrendering. Tomogram resolution was computed withthe Crowther-DeRosier-Klug formula for a flat ex-tended reconstruction volume (Frank, 1992) and withB-soft software (Heymann et al., 2008) using the Fou-rier shell correlation (FSC) criteria as described inCardone et al. (2005).

On the basis of our measures from tomograms of cen-triole components, including the microtubular barrel(total diameter and distal domain length), the internalstructure (thickness, diameter, and period from Fou-rier space), and the distal and subdistal appendages(length, thickness, and angle at which they interactwith the microtubular barrel), a volume was derivedfor each component. To this purpose, four black vol-umes (250 3 250 3 280 voxels, 2 nm/voxel) were builtwith ImageJ. The different components were designedin white, each component in a separate volume, and 9-fold symmetry was imposed. Volumes were combinedinto a single volume and band pass filtered with a co-sine modulation of 0–64 and 16–12 nm. The model wasfinally rendered in Chimera (Pettersen et al., 2004).Validation of this model was performed by cropping

Microscopy Research and Technique

433D STRUCTURE OF HUMAN CENTRIOLES

regions of different height along the centriole axis andprojecting them using the average projection algorithmimplemented in ImageJ (Abramoff et al., 2004). Projec-tions of the model were visually compared with previ-ously published nonsymmetrical and symmetrical ex-perimental data from serial sections (Paintrand et al.,1992). The model was refined by modifying details inthe initial volumes to obtain the best fit with experi-mental data. For example, we adjusted doublet and tri-plet orientations with respect to the centriole axis, rela-tive angles of distal and subdistal appendages, and theheight of the cropped regions. To obtain the best fit ofthe model to the data, we computed cross-correlationcoefficients between the digitized experimental imagesand the model projections.

RESULTS

The Z-loss energy filtered series allowed improve-ment of data quality and provided enough contrast at120 kV on semi-thin sections (Fig. 1) to calculate 3Dreconstructions with 7 and 9 nm resolution (Crowther-DeRosier-Klug formula) or 6 and 7 nm resolution (FSCcriteria) for volumes shown in Figures 2 and 3 respec-tively. These reconstructions displayed the majorknown centriole features. Thus, tomograms showedbarrel-shaped centrioles that comprised microtubules,distal and proximal ends, subdistal and distal appen-dages, and internal structures (Figs. 2 and 3). Thesefeatures were recognizable in the reconstruction even

when the whole centriole was not completely containedwithin the volume (Fig. 2B). Other features (i.e., micro-tubule contacts in doublets and triplets) that wereobserved in serial cross-sectional studies (sections thatwere cut perpendicular to the centriolar axis) were notresolved here due to the centriole orientations on thegrid.

The tomographic slices shown in Figure 2 (with aninterval that spanned 12 planes corresponding to a dis-tance of 14 nm) indicated that microtubules (arrow-head in plane 144) were not parallel to the major cen-triole axis (red line in plane 144). We estimated thatthey were tilted by an angle of 68, as has been previ-ously described with cryotomography (Kenney et al.,1997). The proximal end (curly bracket in plane 96)appeared at 200 nm from where a daughter centriolebudded. A blurred electron-dense material appeared tobe perpendicular to the centriole axis at the proximalend (circled in plane 156, Fig. 2). This may correspondto a budding daughter centriole or to an artifact result-ing from the isolation procedure. The proximal endappeared to have a hollow core; however, the distal endat 225 nm (box in plane 96) was filled with an electron-dense material.

Apart from these major known centriole features,our studies revealed additional structures that pro-vided insight into the organization of centrioles. Forexample, we visualized subdistal appendages (Fig. 2,plane 108; arrows in Fig. 3A) at the distal end of themother centriole. They appeared to be nine, twisted, g-shaped structures that extended perpendicular to thecentriole axis. Each subdistal appendage was composedof two halves fused to a 40 nm tip at 100 nm fromwhere they anchored to the microtubules. The twohalves of the same appendage were attached to adja-cent doublets. Each half formed a hollow 20 nm diame-ter filamentous structure, and the halves were sepa-rated by 68 nm. The most proximal half was tilted at658 with respect to the centriole axis, and the more dis-tal half was tilted at 1008. The distal appendages werethin, straight structures about 100 nm long with a 20-nm diameter (Fig. 2 plane 120, arrowhead in Fig. 3A).These distal appendages were tilted at �408 withrespect to the major axis of the centriole. They joinedthe microtubule barrel at the level of the distal half ofthe subdistal appendage, running almost parallel to it.

The tomograms in Figure 2 also provided insightsinto the structure of the electron-dense materialobserved inside the centriole at the distal domain. Thismaterial formed an internal structure of stacked rings(parallel planes boxed in Figs. 3A and 3B) tilted at 538with respect to the centriole axis. The rings were30 nm in diameter and 15 nm thick. These rings wereassociated with the microtubules by ‘‘spoke-like’’ arms(Fig. 3C arrows) that extended radially to the centriolebarrel walls. Fourier analysis (Fig. 3D) of these stackedrings showed layer lines (45 nm spacing) on the powerspectrum with a periodic organization.

DISCUSSION

Previously published studies on centriole structurehave mostly been performed with TEM on chemicallyfixed, resin-embedded isolated centrioles.

Despite the drawbacks of that protocol, including theloss of centriole proteins due to the isolation procedure,

Fig. 1. Micrograph of a Z-loss filtered centrosome. Lower centriole(arrow) corresponds to that shown in Figure 3. Scale bar, 250 nm.


44 R. IBRAHIM ET AL.

the sample shrinkage due to resin embedding, andmodifications of biological structures due to chemicalfixation, these studies have provided importantinsights on the global structure of centrioles. Althoughelectron tomography does not prevent these potentialartifacts, it offers more opportunity for investigatingthe details of the centriole structures described in pre-vious studies because it allows the observation of 3Darrangements that are unattainable in single projec-tion images.

The first tomographic studies in centrioles were per-formed with frozen mammalian centrioles and imposeda 9-fold symmetry (Kenney et al., 1997). However,because of the bias introduced by symmetry and thelow image contrast, they were only able to confirm fea-tures that were previously described in studies from se-

rial sections. For this reason, we have performed newtomographic studies without symmetry imposition onresin-embedded centrosomes isolated from KE37 cells.Information from these tomograms allowed us to builda 3D model of the centriole (Fig. 4). This modelincluded further details on the subdistal appendagesand the internal structure and was validated by com-paring theoretical projections with images previouslyobtained from centriole sections (Paintrand et al.,1992). The cross-correlation coefficients calculatedbetween the theoretical projections and the experimen-tal images varied between 40% and 45%. This valuewas not very high because experimental images haveless uniform shapes and densities compared to theoret-ical projections. Therefore, at the distal and subdistalappendage levels, projections depicted a ‘‘fan-like’’

Fig. 2. Sections of a mother centriole tomogram. A: Arrow-headspoint to microtubules, arrows to subdistal appendages, and the circleindicates a possible budding of a daughter centriole. Bracket in plane 96highlights the proximal region. Box in the same plane envelops the inter-nal structure in the distal region. The red line in plane 144 corresponds

to the centriole axis. Planes were extracted from the reconstructed vol-umes in an interval that spanned 12 planes (14 nm). Scale bar, 250 nm.B: Volume rendering of a mother centriole showing appendages in pinkand the internal structure in blue.



shape (Fig. 4C, projections 1, 2 and 3) that was previ-ously derived from experimental data from serial sec-tions. Projections from regions at the bottom of the in-ternal structure revealed a ‘‘crescent-like’’ shape (Fig.4C, oval in projection 5). This shape, also observed inexperimental data, has been previously interpreted aspart of a helical structure. However, the ‘‘crescent-like’’shape can also be explained by considering that the in-ternal structure is a stack of rings tilted at 538 withrespect to the centriole-axis, as observed in our tomo-grams (Fig. 3). In addition, the stacked disk organiza-tion can also explain the two stripes observed in projec-tions corresponding to the section between two consec-utive rings in the internal structure (arrows in Fig. 3C,

projection 4 in Fig. 4C). These stripes have not previ-ously been interpreted, although they have also beenobserved in experimental data from sections. Finally,our data showed evidence of ‘‘spoke-like’’ arms (alsoobserved in serial sections from Paintrand et al., 1992)that linked the internal structures to the microtubulesthat formed the walls of the centriole barrel. Thesearms appeared to be ‘‘connection lines’’ in projectionsfrom our model (arrows in Fig. 3C, projection 2 in Fig.4C), visualized when a complete ring was contained ina volume.

In conclusion, this study has led to the re-interpreta-tion of previous observations on the centriole structureby providing details on the internal structure and

Fig. 3. Sections of a mother centriole tomogram mainly focused onthe distal domain. A: Arrows point to subdistal appendages; the inter-nal structure is boxed. Planes are extracted from the reconstructedvolumes in an interval that spanned 12 planes (14 nm). Scale bar, 250nm. B: Zoomed image of the internal structure boxed in A. Scale bar,

30 nm. C: Volume rendering of a mother centriole showing subdistalappendages in pink, one of the distal appendages in red, possible bud-ding of a daughter centriole in green and the internal structure inblue. D: FFT of the boxed region in panels A and B. Arrows point tolayered lines corresponding to a 45.5 nm span.



subdistal appendage organization. The concordancebetween previously published experimental data on se-rial sections and the information revealed in our tomo-graphic reconstructions validates the centriole modelproposed in this work. However, as the centriole is adynamic structure, more detailed studies with syn-chronized cells, cryo-methods, and tomography of cen-trioles inside cells will be required to determine howthe centriole structure is modified during the cell cycle.

ACKNOWLEDGMENTS

We thank M. Paintrand and M. Bornens for provid-ing Epon blocs from KE37-isolated centrioles and Dr.

Jose Lopez Carrascosa for providing access to the FEIelectron microscope.

REFERENCES

Abramoff M, Magelhaes P, Ram S. 2004. Image processing withImageJ. Biophoton Int 11:36–42.

Bornens M, Paintrand M, Berges J, Marty MC, Karsenti E. 1987.Structural and chemical characterization of isolated centrosomes.Cell Motil Cytoskeleton 8:238–249.

Cardone G, Grunewald K, Steven AC. 2005. A resolution criterion forelectron tomography based on cross-validation. J. Struct Biol 151:117–129.

Chretien D, Buendia B, Fuller SD, Karsenti E. 1997. Reconstructionof the centrosome cycle from cryoelectron micrographs. J StructBiol 120:117–133.

Fig. 4. Mother centriole model. A: Rendered volume of the cen-triole model representing mainly the distal part (microtubule dou-blets) and the beginning of the proximal part (microtubule triplets).B: Midsagittal section of the rendered volume shown in A. Numberedboxes correspond to the different sections used in projections shownin panel C. C: Projections of sections extracted at different levels ofthe model shown in panel B. Numbers in projections correspond to

boxes. Arrows in projection 2 point to arms connecting the rings of theinternal structure to the microtubular barrel. Arrows in projection 4point to stripes that correspond to a section between two consecutiverings. Circled region in projection 5 depicts a ‘‘crescent-like’’ shape. A,B: Microtubules are colored yellow, internal structure is blue, subdis-tal appendages are pink, distal-appendages are red, and distal elec-tron-dense material is green.



Dammermann A, Muller-Reichert T, Pelletier L, Habermann B, DesaiA, Oegema K. 2004. Centriole assembly requires both centriole andpericentriole material proteins. Dev Cell 7:815–829.

Delattre M, Gonczy P. 2004. The arithmetic of centrosome biogenesis.J Cell Sci 117:1619–1630.

Fernandez JJ, Li S. 2003. An improved algorithm for anisotropic non-linear diffusion for denoising cryo-tomograms. J Struct Biol 144:152–161.

Fernandez JJ, Li S. 2005. Anisotropic nonlinear filtering of cellularstructures in cryoelectron tomography. Comput Sci Eng 7: 54–61.

Frank J. 1992. Electron tomography. New York: Plenum. pp 107–108.Herman GT, Lent A, Rowland SW. 1973. ART: Mathematics and

applications. A report on the mathematical foundations and on theapplicability to real data of the algebraic reconstruction techniques.J Theor Biol 42:1–32.

Heymann BJ, Cardone G, Winkler DC, Steven AC. 2008. Computa-tional resources for cryo-electron tomography in Bsoft. J Struct Biol161:232–242.

Kenney J, Karsenti E, Gowen B, Fuller SD. 1997. Three-dimensionalreconstruction of the mammalian centriole from cryoelectron micro-graphs: The use of common lines for orientation and alignment. JStruct Biol 120:320–328.

Kramer A, Lukas J, Bartek J. 2004. Checking out the centrosome.Cell Cycle 3:1390–1393.

Marshall WF, Rosenbaum JL. 2003. Tubulin superfamily: Givingbirth to triplets. Curr Biol 13:R55–R56.

Messaoudi C, Boudier T, Sanchez Sorzano CO, Marco S. 2007. TomoJ:Tomography software for three-dimensional reconstruction intransmission electron microscopy. BMC Bioinformatics 8:288.

Muller EG, Snydsman BE, Novik I, Hailey DW, Gestaut DR, NiemannCA, O’Toole ET, Giddings TH, Jr, Sundin BA, Davis TN. 2005. Theorganization of the core proteins of the yeast spindle pole body. MolBiol Cell. 16:3341–3352.

Nigg EA. 2007. Centrosome duplication: Of rules and licenses. TrendsCell Biol 17:215–221.

Nitschke P, Ravise S, Paintrand M, Bornens M, Delacroix H. 1995.The centriole: Twist and pitch. Towards a 3-D reconstruction. JTrace Microprobe Tech 13:383–391.

O’Toole ET, Winey M, McIntosh JR. 1999. High-voltage electron to-mography of spindle pole bodies and early mitotic spindles in theyeast Saccharomyces cerevisiae. Mol Biol Cell 10:2017–2031.

O’Toole ET, Giddings TH, McIntosh JR, Dutcher SK. 2003. Morpho-logically distinct microtubule ends in the mitotic centrosome of Cae-norhabditis elegans. J Cell Biol 163:451–456.

Paintrand M, Moudjou M, Delacroix H, Bornens M. 1992. Centrosomeorganization and centriole architecture: Their sensitivity to diva-lent cations. J Struct Biol 108:107–128.

Pelletier L, O’Toole E, Schwager A, Hyman AA, Muller-Reichert T.2006. Centriole assembly in Caenorhabditis elegans. Nature 444:619–623.

Pettersen EF, Goddard TD, Huang CC, Couch GS, Greenblatt DM,Meng EC, Ferrin TE. 2004. UCSF Chimera—Avisualization systemfor exploratory research and analysis. J Comput Chem 25:1605–1612.

Rebollo E, Sampaio P, Januschke J, Llamazares S, Varmark H, Gon-zalez C. 2007. Functionally unequal centrosomes drive spindle ori-entation in asymmetrically dividing Drosophila neural stem cells.Dev Cell 12:467–474.

Schnackenberg BJ, Palazzo RE. 1999. Identification and function ofthe centrosome centromatrix. Biol Cell 91:429–438.

Sonka M, Hlavac V, Boyle R. 1999. Image processing, analysis, andmachine vision. Pacific Grove, CA: PWS.

Vorobjev IA, Chentsov YS. 1980. The ultrastructure of centriole inmammalian tissue culture cells. Cell Biol Int Rep 4:1037–1044.

Vorobjev IA, Nadezhdina ES. 1987. The centrosome and its role in theorganization of microtubules. Int Rev Cytol 106:227–293.



electron tomography study of isolated human centrioles

Documents