European Journal of Neuroscience, Vol. 10, pp. 790–793, 1998 © European Neuroscience Association

SHORT COMMUNICATIONPattern formation in the cerebellum of murine embryonicstem cell chimeras

Richard Hawkes, Beverly Faulkner-Jones,1 Patrick Tam2 and Seong-Seng Tan1

Department of Anatomy and Neuroscience Research Group, Faculty of Medicine, The University of Calgary, Calgary,Alberta T2N 4N1, Canada1Howard Florey Institute of Experimental Physiology and Medicine, The University of Melbourne, Parkville 3052, Victoria,Australia2Embryology Unit, Children’s Medical Research Institute, WNT Wentworthville, NSW 2145, Australia

Abstract

The cerebellar cortex is subdivided into an elaborate, stereotyped array of transverse zones and parasagittalstripes. It has been speculated that (i) all Purkinje cells derive from 10 to 20 precursors allocated early inembryogenesis and (ii) that pattern formation is based on cell lineage restriction in the founder pool. Thesehypotheses have been tested by clonal analysis of embryonic stem cell chimeras. Neither speculation issupported: the analysis suggests that Purkinje cells derive from a founder population of . 102 precursors, andthat neither cerebellar transverse developmental boundaries nor parasagittal stripes have a clonal origin. Weconclude that early lineage restriction plays no role in cerebellar pattern formation.

Purkinje cells in the mouse cerebellum are organized into an elaboratearray of parasagittal bands (e. g. Hawkes & Gravel, 1991; Hawkeset al., 1992; Wassefet al., 1992) and transverse zones (e.g. Hawkes& Eisenman, 1997). Cerebellar compartmentation can be revealed inmany ways. In particular, zebrin II/aldolase C is expressed selectivelyin parasagittal bands of Purkinje cells (Brochuet al., 1990; Eisenman& Hawkes, 1993; Ahnet al., 1994). The zebrin II expressing bandsare numbered from P11 at the midline to P71 laterally. They areseparated by the zebrin II unreactive (P–) bands, numbered accordingto the P1 band immediately medial.

The role that lineage restriction plays in cerebellar developmentand pattern formation is hotly debated (e.g. Herrup, 1988; Jennings,1988). Quantitative analysis of aggregation chimeras between lurcher(Lc) and wild-type mice (1/Lc , – . 1/1) suggested that allPurkinje cells derive from 10 to 20 precursors that are allocated earlyin development (Wetts & Herrup, 1982; Herrup, 1986; Herrup &Suntner, 1986; Vogel & Herrup, 1993). In contrast, the analysis ofclonal sizes in X-inactivation mosaics identified™ 102 precursors inthe founder population, and concluded that the Purkinje cells in eachlineage group are distributed randomly (Baaderet al., 1996). We haveaddressed these issues – the founder population size, and the clonalrelationships between cells in transverse zones and parasagittal bands –by analysing the clonal relationships of Purkinje cells in murinechimeras generated by injecting embryonic stem (ES) cells homo-zygous for anE. coli β-galactosidase (β-gal1) transgene into wild-type blastocysts.

Chimeras were constructed by injection of 1–5 embryonic ES cellsper blastocyst. ES cells carry thelacZ marker gene onboth X chromosomes, and were derived from a transgenic mouse line

Correspondence:Richard Hawkes, as above. E-mail: rhawkes@acs.ucalgary.ca

Received 25 June 1997, revised 22 September 1997, accepted 9 October 1997

with 14 tandem copies of thelacZ gene fused to the HMGCoAreductase promoter (see Tan & Breen, 1993). Both the ES cell lineand the wild type recipient blastocysts, were C57BL/6 J3 DBA/2.Adult chimeras were perfused with 4% paraformaldehyde/0.2% glutar-aldehyde in 0.1M phosphate buffer, pH 7.41 0.15M NaCl, and thebrains immersion fixed for 30 min and stored cold overnight inbuffered 30% sucrose. Serial 150µm transverse frozen sections werereacted with the X-gal substrate to identifylacZ-expressing cells ofES cell-origin, then immunoperoxidase stained for zebrin II/aldolaseC by using mouse monoclonal antizebrin II (Brochuet al., 1990) anddiaminobenzidine (Eisenman & Hawkes, 1993). Eighteen chimeraswere analysed. X-gal stained Purkinje cells were reliably identifiedby their intensely blue nuclei, large size (soma diameter™ 20 µm)and location at the interface of the granular and molecular layers(Fig. 1). In eight chimeras, large numbers of cerebellar neuronesexpressedlacZ, suggesting that they might be polyclones derivedfrom multiple precursors, and these were not considered further. Inthe other 10, fewer Purkinje cells wereβ-gal1 and were likely tocomprise small numbers of clones. These cerebella were seriallyreconstructed, and all labelled Purkinje cells were counted (Table 1).In four cases. 97% of the ES cell-derived Purkinje cells wereconfined to one hemicerebellum. This is consistent with previoussuggestions that the progeny of Purkinje cell precursors rarely crossthe midline (Altman & Bayer, 1985; Oteroet al., 1993), and with thenotion that the labelledβ-gal1 Purkinje cell population in onehemicerebellum is derived from a small number of ES cell derivedneural precursors.

As with X-inactivation transgenic markers and aggregation chi-meras our quantitative estimates rely on the neutrality of ES cells

Cerebellar pattern formation 791

FIG. 1. The cerebellar cortex of embryonic stem (ES) cell chimeras. Sections were X-gal stained (blue-green) to reveal ES-derived neurones, then antizebrin IIimmunoperoxidase counterstained (brown). (A) A glancing section through the Purkinje cell layer from the anterior lobe vermis at the midline. The zebrin II1

band P11 is flanked by the two P1– bands. The X-gal-stained somata include Purkinje cells in both P11 and P1–. (B) X-gal/zebrin II labelling in the vicinityof P21. The ES cell-derived Purkinje cells are stained by X-gal in their somata (Purkinje cell layer – pcl) but their dendrites in the molecular layer (ml) areunreactive. In the granular layer (gl) many granule cell somata are X-gal stained. The white matter (wm) is unreactive. Scale bar5 100µm. (C) Purkinje cellsin no. 584–1. Seventeen examples ofβ-gal1 Purkinje cells are seen here: in P21 (one in VIII dorsal, two in IX dorsal), P2– (one in VIII dorsal), P31 (one inVIII dorsal, two in IX dorsal, three in IX ventral), P3– (two in VIII dorsal), P41 (one in VIII ventral, two in IX), P4– (one in VIII dorsal), and P51 (three inVIII dorsal). Scale bar5 1 mm.

during chimeric development. Evidence from 17 separate lines of EScells suggest that ES cells, without exception, have the capability toproliferate and integrate normally into embryos and can developwithout bias (Evanset al., 1985; Beddington & Robertson, 1989;Nagyet al., 1990). The reporterlacZ transgene has been demonstratedto be developmentally neutral in Purkinje cells (Oberdicket al.,1990), and the participation of this line of ES cells in cortex andretina development is indistinguishable from results obtained usingX-inactivation mosaics (S-S.Tanet al. unpublished observations).

If Purkinje cells are all derived from a small number of precursors,then their genotype ratios in the chimeras will vary in a quantalfashion that reflects the mosaicism of the founder population. Weassume that in four cases (nos. 584–1, 600–1, 605–2, 628–13) the

© 1998 European Neuroscience Association,European Journal of Neuroscience, 10, 790–793

β-gal1 Purkinje cells are a lineage group derived from a singleprecursor and in two cases two precursors contributed equally(nos. 607–1 and 615–1 – one per hemicerebellum). This yields anpreliminary clone size of (12706 84 SEM;n 5 8) which can be usedto estimate the precursor numbers in the other cerebella (Table 1).The goodness of fit of the model in predicting the ES cell-derivedPurkinje cell populations is shown in Figure 2. The data fall close tothe linear regression line (slope5 1223;n 5 24;R2 5 0.97), implyingthat a typical precursor generates 1223 Purkinje cells, and that allPurkinje cells derive from a founder population of 131 (assuming1.663105 Purkinje cells as in the C57BL/6 J cerebellum: Herrup &Suntner, 1986). These estimates are very close to the population sizesidentified independently by analysis of X-inactivation mosaics (129

792 R. Hawkeset al.

FIG. 2. Purkinje cells are found as quantal populations. The number ofembryonic stem (ES)-derived Purkinje cells per hemicerebellum is plottedagainst the estimated number of precursors (see Table 1). The relationship islinear (regression liney 5 20 1 1223x; R2 5 0.97;n 5 15), consistent with aclonal expansion model.

TABLE 1. β-gal1 Purkinje cell counts from the cerebella of 10 embryonic stem(ES) cell chimeras. Counts were made from complete series of serial 150µmcryostat transverse sections and corrected stereologically (physical dissector(Stereo, 1984): this reduced theβ-gal1 Purkinje cell counts by an average ofonly 4/cerebellum). In three cases, the labelled Purkinje cells were confinedentirely to one hemicerebellum (nos. 584–1, 601–1, 628–13); in two others(nos. 600–1, 605–2) the distribution was predominantly unilateral but a smallnumber (, 3%) were located contralateral to the main mass; in the other fivecases (nos. 607–1, 615–1, 618–1, 621–3, 628–14), substantial numbers werecounted each side of the midline. The inferred number of ES derived precursorson each side (based on multiples of 1270: in parentheses) is listed

Animal β-gal1 Purkinje cells Precursors

584–1 1211 (0.95) 11 0600–1 1422 (1.12)1 26 (0.02) 11 0601–1 2122 (1.67) 21 0605–2 1421 (1.12)1 30 (0.02) 11 0607–1 1087 (0.86)1 1073 (0.84) 11 1615–1 887 (0.70)1 1534 (1.21) 11 1618–1 2727 (2.15)1 1270 (1.00) 21 1621–3 3520 (2.77)1 4959 (3.90) 31 4628–13 1467 (1.16) 11 0628–14 1114 (0.88)1 3830 (3.02) 11 3

precursors with a mean clonal expansion of 1265: Baaderet al., 1996)but are inconsistent with models in which the proposed founderpopulation is very small (e.g. Wetts & Herrup, 1982).

Zebrin II/aldolase C is expressed in the cerebellum in alternatingbands of Purkinje cells (e. g. Fig. 1C). Purkinje cells commit to theirzebrin phenotypes early in development and the choice of zebrinphenotype does not appear to involve extracerebellar regulatorypathways (reviewed in Hawkes & Mascher, 1994). We tested thehypothesis that individual zebrin bands are lineage groups by askingif the β-gal1 Purkinje cells in a chimera share a common zebrin II/aldolase C phenotype. In four chimeras theβ-gal1 Purkinje cellsapparently derive from a single ES cell-derived precursor (Table 1).

© 1998 European Neuroscience Association,European Journal of Neuroscience, 10, 790–793

In each case ES cell-derived Purkinje cells are scattered throughoutthe cerebellum and no clonal boundaries, transverse or parasagittalare apparent. Each Purkinje cell clone was partitioned evenly betweenzebrin II1 and zebrin II– phenotypes, with no apparent restriction toindividual bands. For example, in the posterior lobe of no. 584-1 atleast oneβ-gal1 Purkinje cell was identified in every band fromP11/– – P61/– (Fig. 1C).

Based on these data, we propose the following model of earlycerebellar compartmentation. Some™ 102 cerebellar Purkinje cellfounders are allocated at the commencement of Purkinje cell neuro-genesis (, E7.5 from the X-inactivation data: Baaderet al., 1996).They are not committed to a zebrin II lineage. Between allocationand the formation of the cerebellar ventricular zone (. E10) thereis substantial population expansion (™ 102-fold) accompanied byextensive mixing. Commitment to a zebrin II lineage occurs in theventricular zone. Finally, postmitotic Purkinje cells are born betweenE10 and E13 from committed progenitors in the ventricular zone(Miale & Sidman, 1961) and migrate into the cerebellar anlage toform the early clusters that are the genealogical predecessors of themature parasagittal bands.

Acknowlegements

We thank E. Gonza´lez, R, Parkinson and F. Weissenborn for their assistance,Karl Herrup and Dan Goldowitz for their advice and comments on an earlierdraft of this MS, and acknowledge the financial support of the MedicalResearch Council of Canada (R.H.) and the Australian NHMRC (S-S.T., P.T.).

Abbreviations

ES cell embryonic stem cellLc lurcherβ-gal1 β- galactosidase

References

Ahn, A.H., Dziennis, S., Hawkes, R. & Herrup, K. (1994) The cloning ofzebrin II reveals its identity with aldolase C.Development, 120, 2081–2090.

Altman, J. & Bayer, S.A. (1985) Embryonic development of the cerebellum.III. Regional differences in the time of origin, migration, and settling ofPurkinje cells.J. Comp. Neurol., 231, 42–65.

Baader, S.L., Schilling, M.L., Rosengarten, B., Pretch, W., Teutsch, H.F.,Oberdick, J. & Schilling, K. (1996) Purkinje cell lineage and the topographicorganization of the mouse cerebellar cortex: A view from X inactivationmosaics.Dev. Biol., 174, 393–406.

Beddington, R.S.P. & Robertson, E.J. (1989) An assessment of thedevelopmental potental of embryonic stem cells in the midgestational mouseembryo.Development, 105, 733–737.

Brochu, G., Maler, L. & Hawkes, R. (1990) Zebrin II: A polypeptide antigenexpressed selectively by Purkinje cells reveals compartments in rat and fishcerebellum.J. Comp. Neurol., 291, 538–552.

Eisenman, L.M. & Hawkes, R. (1993) Antigenic compartmentation in themouse cerebellar cortex: Zebrin and HNK-1 reveal a complex, overlappingmolecular topography.J. Comp. Neurol., 335, 586–605.

Evans, M., Bradley, A. & Robertson, E.J. (1985) EK contribution to chimaericmice: from tissue culture to sperm. In Jaenisch, R. and Constantini, F. (ed.),Genetic Manipulation of the Mammalian Ovum and Early Embryo. BanburyReport 20, Cold Spring Harbor Laboratory Press, pp. 93–98.

Hawkes, R., Brochu, G., Dore´, L., Gravel, C. & Leclerc, N. (1992) Zebrins:Molecular markers of compartmentation in the cerebellum. In Llina´s, R.and Sotelo, C. (eds),The Cerebellum Revisited. Springer-Verlag, New York,pp. 22–55.

Hawkes, R. & Eisenman, L.M. (1997) Stripes and zones: the origins ofregionalization of the adult cerebellum.Perspect. Dev. Neurobiol., in press.

Hawkes, R. & Gravel, C. (1991) The modular cerebellum.Prog. Neurobiol.,36, 309–327.

Hawkes, R. & Mascher, C. (1995) The development of molecularcompartmentation in the cerebellar cortex.Acta Anatomica, 151, 139–149.

Cerebellar pattern formation 793

Herrup, K. (1986) Cell lineage relationships in the development of themammalian CNS: Role of lineage in the control. of Purkinje cell number.Dev. Biol., 115, 148–154.

Herrup, K. (1988) What do genetic mosaics tell us about cell lineage in themammalian CNS? A reply.Trends in Neurosci., 11, 49–50.

Herrup, K. & Sunter, K. (1986) Cell lineage dependent and independentcontrol. of Purkinje cell number in the mammalian CNS: Further quantitativestudies of lurcher chimeric mice.Dev. Biol., 117, 417–427.

Jennings, C.G.B. (1988) What do genetic mosaics tell us about cell lineagein the mammalian CNS?Trends in Neurosci., 11, 46–49.

Miale, I. & Sidman, R.L. (1961) An autoradiographic analysis of histogenesisin the mouse cerebellum.Exp. Neurol., 4, 277–296.

Nagy, A., Gocza, E., Diaz, E.M., Prideaux, V.R., Iva´nyi, E., Markkula, M. &Rossant, J. (1990) Embryonic stem cells alone are able to support fetaldevelopment in the mouse.Development, 110, 815–827.

Oberdick, J., Smeyne, R.J., Mann, J.R., Zackson, S. & Morgan, J.I. (1990) Apromoter that drives transgene expression in cerebellar Purkinje and retinalbipolar neurons.Science, 248, 223–248.

© 1998 European Neuroscience Association,European Journal of Neuroscience, 10, 790–793

Otero, R.A., Sotelo, C. & Alvarado-Mallart, R.-D. (1993) Chick/quail chimeraswith partial cerebellar grafts: An analysis of the origin and migration ofcerebellar cells.J. Comp. Neurol., 333, 597–615.

Stereo, D.C. (1984) The unbiased estimation of numbers and sizes of arbitraryparticles using the disector.J. Microsc., 134, 127–136.

Tan, S.-S. & Breen, S.J. (1993) Radial mosaicism and tangential cell dispersionboth contribute to mouse neocortical development.Nature, 362, 638–640.

Vogel, M.V. & Herrup, K. (1993) A theoretical and experimental examinationof cell lineage relationships among cerebellar Purkinje cells in the mouse.Dev. Biol., 156, 49–68.

Wassef, M., Angaut, P., Arsenio-Nunes, L., Bourrat, F. & Sotelo, C. (1992)Purkinje cell heterogeneity: Its role in organizing the topography of thecerebellar cortex connections. In Llina´s, R. and Sotelo, C. (eds),TheCerebellum Revisited.Springer-Verlag, New York, pp. 5–21.

Wetts, R. & Herrup, K. (1982) Cerebellar Purkinje cells are descended froma small number of progenitors committed during early development:quantitative analysis oflurcher chimeric mice.J. Neurosci., 2, 1494–1498.