Download Cell dispersion patterns in different cortical regions

Development 121, 1029-1039 (1995)
Printed in Great Britain © The Company of Biologists Limited 1995
1029
Cell dispersion patterns in different cortical regions studied with an
X-inactivated transgenic marker
S.-S. Tan1,*, B. Faulkner-Jones1, S. J. Breen1, M. Walsh2, J. F. Bertram1 and B. E. Reese3
1Embryology Laboratory, Department of Anatomy and Cell Biology, The University of Melbourne, Parkville 3052, Victoria, Australia
2Department of Genetics, LaTrobe University, Bundoora, Australia
3Neuroscience Research Institute and Department of Psychology, University of California at Santa Barbara, California 93106,
USA
*Author for correspondence
SUMMARY
Inactivation of the X-linked lacZ transgene provides a
novel and powerful way of distinguishing between clonally
related cellular populations in X inactivation mosaics. This
ability to distinguish between clonal populations of cells in
the mature cortex permits inferences to be made about
cellular dispersion patterns during cortical development.
The present study addresses the extent to which radial and
tangential dispersion patterns contribute to different
regions of the cerebral cortex by quantifying the extent of
cellular mixing between clonally distinct cells in separate
domains of the medial, dorsolateral and lateral cortices. We
show that stripes running perpendicular to the cortical
layers are more likely to be seen in the medial and dorsolateral regions, and that the appearance of a stripe is attributed to about two-thirds of the cells being of the same
colour. Both neurons and glia appeared to exhibit the same
ratio of cell mixing. In the lateral regions of the cortex,
INTRODUCTION
During early development, neurons in the mammalian cerebral
cortex are generated in a proliferative layer known as the ventricular zone (Boulder Committee, 1970). From this germinative zone, recently generated neuroblasts ascend into the
cortical plate, guided by radial glial fibers which extend from
ventricular to pial surfaces (Rakic, 1972). It has been suggested
that progenitor cells give rise to radially organized clones of
cells, which subsequently become represented as repeating
units of functional columns in the mature cortex (the ‘radial
unit hypothesis’; Rakic, 1988). Associated with this hypothesis is the further idea that the adult cytoarchitectonic map is
anticipated by a topologic representation within the ventricular zone (the ‘protomap hypothesis’, Rakic, 1988). According
to this view, the positional information of different cortical
areas is encoded by migrating postmitotic neurons as they
move into the emerging cortical plate. Others have argued
against the idea of a ‘protomap’, preferring instead to regard
the developing cortex as a naive structure upon which the
organization is imposed by extrinsic influences that operate
stripes were not apparent, and cell mixing was roughly
equal. In the barrel-field region of the somatosensory
cortex we looked for a correspondence between cytoarchitectural features and clonal borders but found no correlation. These results demonstrate, first, that although there
is widespread radial dispersion, no cortical region is
composed of radially arrayed stripes of cells in which all
members of a stripe are derived from a single progenitor.
Second, they demonstrate that, within regions containing a
sizeable fraction of cells that do migrate radially, the
boundaries of individual stripes do not always coincide
with single anatomical units of cortical specialization, such
as individual barrels.
Key words: cerebral cortex, mouse, clonal analysis, genetic mosaics,
X chromosome inactivation
after the cessation of neurogenesis (see O’Leary, 1989, for
review). The crucial element in this debate is the way that the
newly born cells move from the germinal zone into the cortical
plate. If there is appreciable tangential dispersion of cells in
the intermediate zone and cortical plate, then clearly the
relative positions of migrating cells would be scrambled, preventing the formation of clonally related radial columns and
any associated protomap information dependent on them.
To address this problem, different approaches to studying
cortical cell dispersion have been employed, including construction of mouse chimaeras, in situ application of fluorescent
dyes, and retroviral lineage labelling. Experiments using
mouse chimaeras, whose cells are composed of two different
genotypes, indicate widespread mixing within the somatosensory barrel fields (Goldowitz, 1987), but elsewhere in the
cortex there appears to be a radial bias to the migration of
neurons (Crandall and Herrup, 1990; Fishell et al., 1990;
Nakatsuji et al., 1991). Application of vital fluorescent dyes to
precursor cells in the neuroepithelium allows video image
analysis of cell migration: such studies indicate that substantial horizontal dispersion occurs in the ventricular zone (Fishell
1030 S.-S. Tan and others
et al., 1993) but in the overlying intermediate zone, over 80%
of imaged cells at any one period of observation appeared to
migrate radially (O’Rourke et al., 1992).
Retroviral lineage studies so far provide the strongest
argument for significant tangential dispersion with the implication that clonally related cells tend to be spread far and wide
(Price and Thurlow, 1988; Walsh and Cepko, 1988; Austin and
Cepko, 1990). However, other studies using this approach also
provide contrary evidence, suggesting that clonally related
cells may be found tightly clustered together in the mature
cortex, implying that they have dispersed radially (Luskin et
al., 1988, 1993; Grove et al., 1993). Some of the discrepancies
between retroviral experiments may be due to inaccurate determination of clonal boundaries, a weakness inherent in the
technique. Thus, two descendants of a marked clone could be
widely separated and therefore perceived as members of two
different clones – so called ‘splitting’ errors. Alternatively,
‘lumping’ errors can occur where labelled cells in a cluster are
wrongly classified as belonging to one clone when in fact
infection of two precursors has taken place. It is possible to
minimise ‘lumping’ errors by using very low retroviral titers
that attempt to infect an average of less than one progenitor
cell per hemisphere (Acklin and van der Kooy, 1993).
‘Lumping’ errors may also be reasonably ruled out where
clones are found to exist in clusters by the reasoning that since
clones are often composed of a single cell type, they are
unlikely to be the progeny of more than one labelled progenitor since this should generate clones with mixed cell types
(Luskin et al., 1988, 1993; Grove et al., 1993). A more secure
method of identifying clonal relationships is to infect progenitor cells with retroviruses carrying different lengths of DNA
inserts. Subsequent amplification of cellular DNA by polymerase chain reaction will permit cells with similar size inserts
to be classified as belonging to a clone (Walsh and Cepko,
1992, 1993). Such studies suggest the widespread dispersion
of neuronal siblings across several functional areas of the
cortex, although it may be pointed out that interpretation of
such data can be complicated by the possibility that apparently
split clones may in fact be multiple infections by one or more
members of the infecting cocktail (Kirkwood et al., 1992).
Besides uncertainties inherent in the various cell marking
techniques, there are a number of other factors that may
obscure radial dispersion after labelling of progenitor cells.
The most important one is that horizontal mixing of progenitor cells in the ventricular zone can be quite extensive (Fishell
et al., 1993) and sibling cells belonging to a clone may subsequently appear in divergent positions even if they all obey
radial migratory pathways. Position in the cortex may also be
important; tangential dispersion may be perceived to be more
widespread in the lateral cortical areas because neuroblasts
have to travel tortuous distances along curved glial fibers in the
intermediate zone (Austin and Cepko, 1990; Misson et al.,
1991). Finally, neurons that are normally stacked together in
the radial axis may be derived from more than one progenitor
in the ventricular zone (e.g. pyramidal versus non-pyramidal
cells, Parnavelas et al., 1991; deep versus superficial layers,
Crandall and Herrup, 1990; Fishell et al., 1990); therefore,
labelling strategies that mark only one progenitor at a time are
limited to studying dispersion patterns produced by one class
of cells.
We previously described a novel method of labelling cortical
progenitors without the need for surgical intervention but
which instead relies upon the natural phenomenon of X-chromosome inactivation (Tan and Breen, 1993). Differential
marking was accomplished by expression of β-galactosidase
that is encoded by the lacZ transgene in approximately half of
the cells of the embryonic telencephalon in females. This was
achieved by insertion of the transgene into one of the two X
chromosomes of the founder H253, and inactivation of the
transgene in all female descendants from this line (Tam and
Tan, 1992). X-chromosome inactivation in mouse embryos
begins at embryonic day 5.5 (E5.5) and is a random process
with respect to whether the paternal or maternal X is silenced
in a given cell. Once this decision is taken, every descendant
of a cell has the same inactive X-chromosome throughout all
successive cell divisions (for review, see Grant and Chapman,
1988). Heterozygous female embryos from line H253 have the
lacZ transgene on only one of the two X chromosomes,
therefore random inactivation would produce approximately
50% lacZ-expressing cells, including cortical progenitor cells
of the forebrain (Tan et al., 1993). The descendants of these
progenitor cells form clones which, after histochemical processing, are either blue (transgene expressed on active X) or
white (transgene silent on inactive X), thereby providing a retrospective method of visualizing cellular dispersion patterns in
the cortex of these transgenic mosaics. While this technique is
unable to indicate which cells belong to a clone, its novel application here is its ability to distinguish borders between clonal
populations in the presence of limited cell mixing. Because X
inactivation (and hence the marking process) of the neuroepithelium is completed at E9.5 before the formation of the telencephalon, every cortical cell is labelled blue or white
depending upon the status of the lacZ-bearing X chromosome
of the progenitor, permitting the entire population of cortical
cells to be studied. By contrast, retroviral and dye application
studies typically label only small numbers of cells for analysis
and at relatively later stages of cortical development. We have
previously reported the use of these X inactivation transgenic
mosaics to label the grey matter of the cortex and found that
in the medial cortex, the pattern of cortical cell dispersion did
not produce a random mixture of marked and unmarked cells;
instead, the neocortex was divided into mixed bands (or
stripes) of variable widths (Tan and Breen, 1993). A large proportion of cells within each stripe was found to have the same
expression phenotype (i.e. mostly blue or mostly white), and
the radial orientation of the stripes suggest that the majority of
cells found within a stripe have dispersed radially. Superimposed upon this was a significant contribution by oppositecoloured cells which did not respect borders between the
stripes, with the implication that some of these cells may have
dispersed tangentially across the borders. The object of the
present study was to examine variation within different cortical
loci by quantification of samples at varying locations across
the cortex. Our results suggest that cell dispersion patterns are
similar in different areas of the dorsal and medial cortices, but
contrast greatly with dispersion in the lateral areas.
MATERIALS AND METHODS
Transgenic animals and tissue preparation
Details on the derivation of transgenic mouse line H253 (Tam and
Cortex cell dispersion patterns in transgenic mosaics 1031
Tan, 1992) and its suitability for marking 50% of cortical progenitors
by X-chromosome inactivation of the lacZ transgene have previously
been described (Tan et al., 1993). For the present study, female adult
mice (7-12 months in age) heterozygous for the transgene were used.
Anaesthetized animals were perfused intracardially using 4%
paraformaldehyde/0.2% glutaraldehyde in 0.1 M phosphate buffer.
Brains were removed and post-fixed in the same fixative for 1hour,
followed by cryoprotection in 30% buffered sucrose overnight.
Frozen sections were cut at 100 µm in the parasagittal, coronal and
horizontal planes. Additional frozen brains were sectioned at 200-300
µm in the tangential plane (parallel to the pial surface). These mice
were perfused with 2% paraformaldehyde in 0.1 M phosphate buffer,
after which the cortices were dissected and then flattened in the
presence of 4% paraformaldehyde/0.2% glutaraldehyde. After rinsing
in phosphate buffer (3× 20 minutes), sections were reacted with Xgal in 24-well tissue culture plates overnight at 37°C. The reaction
solution contained 0.1% 4-chloro-5-bromo-3-indolyl-α-D-galactopyranoside (X-gal, Sigma), 2 mM MgCl2, 5 mM EGTA, 0.01% (w/v)
sodium desoxycholate, 0.02% (w/v) Nonidet P-40, 5 mM potassium
ferricyanide, and 5 mM potassium ferrocyanide. On the following
day, sections were rinsed with buffer and photographed wet. Blue and
white stripes in parasagittal and coronal sections were identified from
defined locations (see below) and removed using a scalpel blade; each
stripe consisted of all cortical layers plus a small segment of the white
matter for orientation. These were dehydrated in alcohols, and
embedded in glycolmethacrylate (Polaron Embedding Medium, Biorad). En face sections were cut at 20 µm; on average each stripe
yielded 5 sections. These were mounted on gelatinized glass slides
and counterstained using 0.1% cresyl violet in 1% acetic acid (pH 3.3)
for 50 minutes at 40°C before analysis using the optical dissector (see
below).
For immunohistochemical detection of β-galactosidase and the
calcium-binding protein, calbindin, anaesthetised animals were
perfused via the left ventricle with 0.5% w/v sodium nitrate in halfstrength phosphate-buffered saline (PBS), followed by 4%
paraformaldehyde in PBS. The brains were dissected free, post-fixed
for 2 hours in the same fixative, and then sectioned with a vibratome
in the parasagittal plane at 50 µm intervals. Sections were recovered
into PBS containing 0.02% sodium azide and stored at 4°C until
required for immunohistochemistry.
Sections chosen for the simultaneous detection of β-galactosidase
and calbindin, were washed in PBS containing 10% normal horse
serum and 0.02% Triton X-100 for 12 hours at room temperature.
These were then incubated for 48 hours at room temperature in PBS
containing 0.02% Triton X-100, and rabbit anti-β-galactosidase
(Cappel) and sheep anti-calbindin at dilutions of 1:1000 and 1:800,
respectively. After washing in three changes of PBS, the sections were
re-incubated in biotinylated donkey anti-sheep immunoglobulins
(Jackson Immuno Research), diluted to 1:100 with 0.02% Triton X100 in PBS, for a further 2 hours. The unbound secondary antibody
was removed by washing in PBS. The final incubation was for 2 hours
in PBS containing 0.02% Triton X-100, Texas Red-conjugated
donkey anti-rabbit immunoglobulins (Jackson Immuno Research) and
Streptavidin-FITC (Amersham) diluted to 1:100 and 1:50 respectively. After washing in PBS for 30 minutes, sections were recovered
onto glass slides and viewed by fluorescence microscopy.
Control sections from the same brains were incubated without the
primary antibodies, to act as negative controls. No specific immunostaining was observed.
Location of blue and white stripes
We compared the ratios of blue and white cells (neurons and
macroglia) in stripes located at different cortical regions. In the first
comparison, we examined stripes from the anterior, middle, and
posterior thirds of the medial cortex from brains that had been
sectioned in the parasagittal plane. Two blue and two white stripes
were obtained from each of these locations per animal, each pair of
blue and white stripes coming from parasagittal sections at two of
three distances from the midline (e.g. 1.2-1.3 mm from the midline;
1.8-2.3 mm from the midline; or 2.6-2.9 mm from the midline). To
confirm that stripes were radially sectioned, stripes embedded in
methacrylate were focussed throughout their entire thicknesses to
ensure that they were uniformly white or blue in three dimensions. A
total of 16 parasagittal sections were used, 12 of these were 2.3 mm
or less, and 4 sections came from 2.6-2.9 mm from the midline. A
total of 4 adult animals were sampled, yielding 8 blue and 8 white
stripes per anteroposterior locus (×3) giving a total of 24 blue and 24
white stripes in this part of the study.
In the second comparison, we examined stripes that were situated
at larger variable distances from the midline in the coronally sectioned
brains by dividing the cortex into thirds and classifying those subdivisions as medial, dorsolateral and lateral domains. Two blue and two
white stripes were obtained from each domain per animal (n=4) in
near-adjacent coronal sections (from the middle half of the cortex)
except in the lateral domain where no stripes appeared to be visible
(see below). For this region, four equivalent pieces of cortical tissue
per animal were used for comparison with medial and dorsolateral
domains, two of which came from the pyriform cortex ventral to the
rhinal fissure. Thus a total of 16 blue and 16 white identifiable stripes
were obtained from medial and dorsolateral areas, the remaining 16
pieces of tissue without clear stripes came from the lateralmost third
of the cortex.
To study the barrel-field region of the somatosensory cortex, tangential slabs of posteriomedial cortices from the right hemisphere of
3 transgenic animals were cut with a scalpel blade and flattened. Serial
100 µm sections cut parallel to the pial surface were obtained and
reacted for β-galactosidase. Blue and white patches within the barrelfield region (see Fig. 1J) were dissected out, embedded in glycolmethacrylate and 20 µm serial sections obtained. These were counterstained and examined under the optical disector.
The optical disector method and statistical analysis
The optical disector provides unbiased estimates of nuclear, and
thereby cell numbers. It does not require assumptions or knowledge
of nuclear shape or size (Gundersen et al., 1988). The nuclei of
neurons and macroglia in the cortex differ considerably in size but the
disector ensures equal treatment of all cells, regardless of size, shape,
colour, and so on. Individual stripes measured 300-400 µm in width,
these 100 µm sections were embedded in glycolmethacrylate and five
20 µm sections were obtained. The glycolmethacrylate resin is transparent and hence allows microscopic identification of all cells
embedded within a 20 µm section; transgene-active (blue) cells could
be identified by the X-gal reaction product, and transgene-inactive
(white) cells by the cresyl violet counterstain. Nuclei were counted
using an unbiased counting frame of 1600 µm2 (40×40) taken from
the central 10 µm of the section thickness. Fields on the stripe were
sampled using a systematic uniform sampling procedure with a
random starting point. Movement from one field to the next was
achieved using a motorized microscope stage with predetermined
movement settings in the X and Y axes. On average, 35-40 fields were
sampled from each stripe. A total of 13,710 cells were counted from
96 stripes, with an average of 142 cells per stripe. Cells were further
classified as neurons or glia based on previously established criteria,
which used electron microscopy to validate light microscopic identification of cortical cells (e.g. Vaughan, 1984). Cells were considered
to be glia if they were more darkly stained with cresyl violet, and
possessed nuclear somata with smaller, irregular shapes and contained
dense chromatin around the nuclear periphery. Neurons were identified on the basis of their abundant cytoplasm, and large, round, pale
nuclei, which frequently contained visible nucleoli.
The method of logistic regression was used to analyse whether
anteroposterior position, in the first study, or mediolateral position, in
the second study, showed a significant effect, after adjusting for
mouse-to-mouse variation and possible differences across sections
1032 S.-S. Tan and others
(Collett, 1991). Specifically, the test asks whether there is a changing
trend in the proportion of blue (or white) cells in blue (or white)
stripes, respectively, across either the anteroposterior or mediolateral
axes.
Chromosomal localization of the lacZ transgene by
fluorescent in situ hybridization (FISH)
Primary cell cultures were established from body walls of transgenic
female embryos a few days before birth. Tissues were embedded in
chick plasma clots and after primary outgrowth, cells were trypsinized
and maintained as confluent monolayers in DMEM with 10% fetal
calf serum. For metaphase chromosome spreads, cells were first
incubated with colcemid (8 µg/ml) for 60 minutes, harvested with
trypsin and centrifuged (1500 rpm, 10 minutes). Cells were resuspended in 0.75 M KCl, and fixed in 3:1 methanol/acetic acid. After
further centrifugation, the pellet was resuspended in fresh fixative and
the procedure repeated twice. The final pellet was resuspended in 0.5
ml fixative and individual drops of cell suspension spread onto clean
glass slides. RNase treatment of chromosome spreads was performed
using 200 µl (100 µg/ml) for 1 hour at 37°C followed by 3× washes
in SSC and then dried in alcohol. The cell spreads were then treated
with Proteinase K (2 µg/ml) for 10 minutes at 37°C. Chromosomal
DNA in metaphase spreads was denatured by dipping into deionised
formamide for 2 minutes and washed with absolute alcohol. Localization of the transgene was performed with FISH using a modified
protocol of Pinkel et al. (1986). Cell spreads were hybridized with a
nick-translated probe containing biotin-labelled dATP using 1 µg of
plasmid transgenic cDNA as template. In supplementary experiments,
a cocktail of probes against the lacZ transgene and human TIMP
DNA, a known X-linked gene with homology to mouse, was used in
a double-labelling experiment. Hybridization was performed
overnight at 37°C and slides were then washed in 2× SSC and 0.1 M
phosphate buffer containing 0.05% Nonidet P-40. Binding of biotin
to FITC-conjugated avidin was carried out in a solution containing 5
µg/ml of FITC-avidin in phosphate buffer containing 5% skimmed
milk powder. Slides were washed after 30 minutes in phosphate buffer
and further incubated with fluorescent anti-FITC antiserum for 30
minutes to amplify the signal. Finally, cell spreads were stained with
propidium iodide (2 µg/ml) and photographed using an epifluorescence microscope.
RESULTS
Chromosomal position of the lacZ transgene
Breeding data from over 7 generations have previously
confirmed that the lacZ transgene is inserted into one of the X
chromosomes of the female founder (Tan and Tam, 1993).
FISH experiments produced intense labelling of lacZ on both
chromatids of a large chromosome, identified as the X by
double labelling with TIMP (known to be on the mouse X),
distinguished by a weaker signal right next to the lacZ
(compare large and small arrows, Fig. 1A). The signal from
the Timp locus mapped to the proximal region of the chromosome whereas the stronger signal from the transgene was seen
to be more distal (Fig. 1A).
Localization of the transgene on the X chromosome was
performed by measuring midpoint positions of fluorescent
signals relative to a reference cytogenetic map where the X
chromosome had been divided into bands and converted to percentage lengths (Lyon and Kirby, 1992). Hybridization signals
for the transgene on 30 different chromatids were performed
and the mean distance from the centromere was found to be
39.5%, which localizes the transgene to approximately band
A6 (Fig. 1B), close to the Hprt locus.
Fig. 1. Chromosomal localization of the lacZ transgene.
(A) Flourescent in situ hybridization with biotinylated probes for the
transgene and TIMP shows a strong signal for multiple copies of the
transgene (small arrow), and a faint signal for Timp (large arrow).
These chromosomal preparations are tetraploid, hence the extra set
of labelled chromatids. (B) Hybridization signals for the transgene
on 30 different chromatids were performed and the mean distance
from the centromere was found to be 39.5%, which corresponds
roughly to band A6 and close to the Hprt locus. (C) Parasagittal, (D)
coronal, and (E) horizontal sections of 100 µm thickness showing
blue and white stripes in the cerebral cortex after reacting for the
lacZ encoded enzyme, β-galactosidase. The parasagittal section has
been obtained from the medial region of the cortex (rostral is to the
left) and this particularly striking example shows radial stripes
running from the white matter to the pial surface (top of the picture).
The stripes are also apparent in coronal (D) and horizontal (E)
sections, shown here at the level of the lateral geniculate nucleus. In
coronal sections (D), stripes are seen in the medial and dorsolateral
parts of the cortex (corresponding to boxes M and DL in Figs 4, 5)
but not in the lateralmost parts (ventral to arrows). In the horizontal
section (E), rostral is to the right. (F) Higher magnification of a
parasagittal section showing the presence of blue cells within a white
stripe (arrows). (G) Thin-section (20 µm) of a white stripe after
embedding in glycolmethacrylate. Staining with cresyl violet permits
both blue and white cells to be visualized. In this section, blue
neurons (large arrows) and blue glia (small arrows) have dispersed
tangentially into a white stripe. Cells were considered blue when the
nucleus is stained for β-galactosidase, or has a blue rim around the
nuclear membrane. Stained nuclei frequently have large blue dots on
their periphery. Smaller blue dots represent cytoplasmic spots that
are possibly associated with dendrites (e.g. see Groves et al., 1993).
(H) Immunohistochemical verification of cells identified as neurons
on morphological criteria. Double-labelled immunohistochemical
detection for neuron-specific calbindin and β-galactosidase in
cortical stripes. Neurons that stain for calbindin alone are fluorescent
green (large single arrow), cells that stain for β-galactosidase only
are red (small single arrow) and neurons that are immunoreactive for
both calbindin and β-galactosidase appear orange (double arrows).
(I) Higher power view of double-labelled neuron positive for both
calbindin and β-galactosidase. Neuronal processes may be seen from
calbindin-positive cell (fluorescent green, large single arrow) and
from a calbindin plus β-galactosidase-positive cell (orange, double
arrows). (J) En face section (200 µm) of a flattened portion of
cerebral cortex showing the mosaic pattern of blue and white stripes
in layer 4. Notice the somatosensory barrel field (arrow) in the
parietal cortex, and that the entire barrel field as well as the
individual barrels are unrelated to the clonal mosaicism. Medial is to
the right, and rostral to the top of the figure. Scale bars, 900 µm in C;
600 µm in D,E,J; 200 µm in F; 40 µm in G; and 50 µm in H and I.
Qualitative analysis of stripes
Stripes were seen in the cortex of all hemizygous females from
line H253 while stripes were never seen in brain sections
obtained from homozygous females where the lacZ transgene
is present on both Xs and therefore expressed by every cell
irrespective of which X became inactivated (Tan and Breen,
1993). Similarly, males with the lacZ on the constitutively
active single X also showed completely blue cortices, implying
that the stripes seen in females have not arisen from chromosomal position effects on the transgene. Further confirmation
of this was provided by line H185 where the transgene had
inserted into an autosome; both male and female animals from
this line showed completely blue cortices (unpublished observations). For the present study, hemizygous females from line
H253 only were used. Staining the cortices of these transgenic
Cortex cell dispersion patterns in transgenic mosaics 1033
1034 S.-S. Tan and others
mosaics for β-galactosidase allowed us to analyze the extent
to which clonally distinct populations remain segregated.
The cortex of these transgenic mosaics is striking in its
columnar organization, even with the unaided eye. At low
power magnification, parasagittal sections from the medial
cortex displayed blue and white stripes of variable size and
frequency; a particularly extreme example is displayed in Fig.
1C. Stripes of similar widths, although less clearly defined,
were also seen in coronal and horizontal sections (Fig. 1D,E).
The stripes appeared to be radial in orientation, running from
white matter to the pial surface with the long axis of each stripe
being roughly perpendicular to the cortical layers. The
exception to this was found at the frontal and occipital poles
where the stripes were occasionally broader in the superficial
relative to the deeper cortical layers, coincident with the
dorsoventral curvature of the cortex in these regions. The width
of stripes in parasagittal sections varied greatly, from 100 to
650 µm.
At higher magnification, blue and white stripes were
observed to contain cells of the opposite colour, indicating
mixing by clonally unrelated cells within stripes (Fig. 1F).
Both neurons and glia contributed to the cellular mixing. For
example, in white stripes, blue nuclei with either neuronal or
glial morphologies were seen (Fig. 1G). The degree of mixing
appeared to be uniform across cortical layers II to superficial
VI. Stripes were not obvious in layer I, although this may be
partially a consequence of its relatively cell-sparse nature.
Deep layer VI, by contrast, is cell-rich, but did not always show
stripes continuing through it. For this reason, counts of cells
described below were always taken superficial to the deeper
half of layer VI and deep to layer I.
Cells that were classified as neurons, based on morphological criteria, were verified using immunohistochemical
detection of calbindin, a calcium-binding protein specific to
neurons in the nervous system (Andressen et al., 1993). This
antibody labels both the cell soma and axon (Fig. 1H,I).
Calbindin immunoreactivity using conjugated fluorescein was
noted in both β-gal-positive (orange) and β-gal-negative
(green) neurons (cells which do not stain with calbindin but
nevertheless expressed β-galactosidase were labelled with
Texas Red).
In parasagittal sections, the presence of stripes was most
conspicuous in medial sections. Further laterally, where the
parasagittal plane is progressively less likely to section the
cortex orthogonal to its layers, stripes were less apparent. Horizontal sections showed the reverse trend, with more ventral
sections that cut the cortex orthogonal to the layers (for
example at mid-height) showing the presence of stripes (Fig.
1E). Curiously, at more ventral levels approaching the rhinal
fissue, the stripes were less readily detected. This difference
between the dorsal and ventralmost parts of the cortex is
apparent in coronal sections (Fig. 1D) where stripes can be
most readily detected medially and dorsolaterally, and are least
apparent ventrolaterally (beneath arrows, Fig. 1D).
Quantitative comparisons of stripes (anterior versus
posterior)
To assess whether the degree of cell mixing was uniform
across different cortical locations, the ratio of blue to white
cells (both neurons and glia) was quantified at different anteroposterior and mediolateral locations. In the first comparison,
the ratio of blue and white cells in both a blue and a white
stripe were determined in the anterior (A), middle (M), and
posterior (P) regions of the medial cortex, expressed as the percentage of blue cells in blue stripes (Fig. 2A) and the percentage of white cells in white stripes (Fig. 3A). This analysis was
conducted on samples from parasagittal sections taken between
1.2 and 2.9 mm from the midline. The results from both blue
and white stripes of four animals confirmed our previous
results (Tan and Breen, 1993): in a blue stripe, approximately
two-thirds of the cells appear blue (giving the stripe its
dominant blue colour) with the remaining one-third being
white (Figs 2A, 3A). These results indicate that a stripe is
formed by a mixture of descendants from at least two clonally
unrelated progenitor cells and that on average, cells of the predominant colour make up about a two-thirds majority.
The degree of mixing by clonally unrelated cells in stripes
appeared to be constant regardless of whether stripes were
obtained from the anterior, middle, or posterior regions of
medial cortex. When neurons and glia were separately
analysed, an average of two neurons to one glial cell were
found to be of the opposite colour in a given stripe. Thus, both
neurons and glia exhibited a similar cell-mixing tendency
within stripes (Figs 2B,C, 3B,C). A logistic regression analysis
was performed to determine whether the ratio of blue to white
cells varied across the anteroposterior axis, revealing no effect
of position (P>0.05). The same analysis was performed separately on neurons and glia yielding the same result (P>0.05),
even though opposite-coloured neurons outnumber oppositecoloured glia by two to one.
Quantitative comparisons of stripes (medial versus
lateral)
In the second comparison, ratios of blue to white cells were
similarly scored in coronal sections, samples being taken from
the medial, dorsolateral and lateral cortices. The latter samples
came from regions of cortex that did not show distinct stripes,
there appeared to be no distinction between samples sourced
from either neo- or paleocortical regions. As found in the above
analysis of parasagittal sections, the medial and dorsolateral
samples taken from coronal sections revealed ratios of roughly
two-thirds blue to one-third white cells in a blue stripe (Fig.
4A), and vice versa for white stripes (Fig. 5A). This trend was
also seen when neurons and glia were considered independently (Figs 4B,C, 5B,C). By contrast, samples from the lateral
cortex showed equal mixing of blue and white cells (thus precluding the detection of stripes), 50% of the cells being blue
and 50% being white (Figs 4A, 5A), and this was also true for
neurons and glia when they were considered separately (Figs
4B,C, 5B,C). A logistic regression analysis, progressing from
medial to lateral cortices, revealed a significant effect of
position on the ratio of blue to white cells (P<0.001). This
decrease in the proportion of blue cells in blue stripes (and
white cells in white stripes) was also significant when the cells
were separately classified as either neurons (P<0.001) or glia
(P<0.01).
Taken together, these results indicate that stripes are most
apparent in medial and dorsolateral areas of the cortex, but are
virtually non-existent in the lateralmost one-third of the
cerebral cortex. Where stripes are present, they invariably
showed mixing with cells of the opposite colour, indicating
that no individual cortical stripe has arisen by clonal expansion
Fig. 2. Schematic representation of a parasagittal section showing the sampled locations
taken from anterior (A), middle (M) and posterior (P) regions of the medial cortex. The
proportion of blue cells found within blue stripes obtained from each sampled area is
shown in the histograms representing pooled data from four animals. Error bars represent
standard error of the mean. A shows the data when neuronal and glial cells are combined,
Fig. 3. Proportion of white cells found within white stripes, sampled from the same mice
as in Fig. 2. Conventions are identical to those in Fig. 2.
while data for neurons (B) and glia (C) are separately represented. Notice that the
proportion does not change appreciably across the cortical loci.
Cortex cell dispersion patterns in transgenic mosaics 1035
Fig. 4. Schematic representation of a coronal section showing the sampled locations taken
from medial (M), dorsolateral (DL) and lateral (L) regions of the cortex. The proportion of
blue cells found within blue stripes obtained from each of the four sampled areas is shown
in the histograms, which represent pooled data from four mice. A shows the data when
neurons and glia are combined, while neurons (B) and glia (C) are separately shown. Note
Fig. 5. Proportion of white cells found within white stripes, sampled from the same mice
as in Fig. 4. Conventions are identical to those in Fig. 4.
that the proportion of blue cells in blue stripes declines in the lateral cortex, approaching
50%. The same trend can also be observed for neurons and glia. V, ventricle.
1036 S.-S. Tan and others
Cortex cell dispersion patterns in transgenic mosaics 1037
Fig. 6. Proportion of blue cells found in blue barrel fields (A) and
white cells in white barrel fields taken from the right hemispheres of
three adult transgenic female animals. The blue/white ratios of cells
(roughly two-thirds in a given barrel field) appear similar to the
ratios found elsewhere in the non-barrel dorsomedial cortex (Figs 2,
3).
of a single progenitor cell. Nevertheless, the existence of
stripes points toward radial mosaicism of cellular subpopulations during corticogenesis. This mosaicism may be directly
viewed in the adult cortex by taking en face sections that cut
the cortex parallel to the cortical layers. Such sections showed
blue and white patches of irregular size and shape that are particularly striking in the medial and dorsolateral regions of the
cortex (Fig. 1J). When such sections pass through layer 4 of
the parietal cortex, the barrel field of the primary somatosensory representation (Woolsey and Van der Loos, 1970) is
readily observed (arrow in Fig. 1J), even in tissue that has not
been counterstained. In the region of the barrel field, the individual patches are not sufficiently large to encompass the entire
barrel field, and individual barrels may lie within either a blue
or a white patch, and in some instances, straddle across the
blue/white borders. To ascertain whether the stripes passing
through the barrel field contained similar ratios of blue and
white cells found elsewhere in the dorsomedial cortex, we
performed the same analysis using the optical dissector. The
results show that the ratio of blue to white cells in blue barrels,
and vice versa for white barrels, is similar to the ratios obtained
for blue and white stripes elsewhere in the dorsomedial cortex
(compare Fig. 6A,B with Figs 2 and 3). These results confirm
that there is nothing unique about the blue/white composition
of cells in the barrel field and imply that during development,
the barrel field draws upon the same pool of dispersing cells,
whether radial or tangential. As with other areas of the cortex,
it would mean that specification of either the barrel field or
individual barrel units is not directly linked to clonal expansion
of predefined cortical progenitors (Goldowitz, 1987).
DISCUSSION
We have examined the issue of mixing between clonally
distinct cells in the cortex of transgenic mosaics. In this
approach, every cortical progenitor cell and its descendants in
the developing forebrain are marked non-invasively by either
the expression, or non-expression, of a lacZ transgene. There
are two major differences between the present technique and
other contemporary methods of marking cortical cells. Retroviral infection and DiI cell labelling both mark small numbers
of cells at a time, revealing the behaviour of only a small
fraction of cortical progenitors. In contrast, the technique
described here provides a global view of cortical development
by labelling 50% of the progenitor cells of the forebrain at the
same time. A second difference relates to timing; in the
present study, forebrain progenitors were labelled during early
stages, before embryonic day 9.5 (X inactivation is completed
in the neural tube at this age), prior to the establishment of
cerebral vesicles and before postmitotic neurons are born. By
contrast, labelling by retroviruses, requiring surgery to the
embryo, has not been feasible before E12.5 in the mouse, and
E14.5 in the rat (Austin and Cepko, 1990; Walsh and Cepko,
1993). By these stages, cortical neurogenesis is well
underway. Finally, there is an important conceptual difference
in interpreting the data generated by the two techniques.
Whilst retroviral studies may be useful in revealing the distribution and clonal relationships of labelled cells, they are
less effective for estimations of mixing by clonally distinct
cells. For example, widespread dispersion between sibling
cells of a retrovirally marked clone may have resulted from
tangential dispersion of both cells, or from radial dispersion
of one cell and tangential dispersion of the other. By contrast,
X-inactivation transgenic mosaics mark every cell as being
either transgene active or inactive, and so labelled cells that
are grouped together define boundaries. By determining the
ratio of unlabelled to labelled cells in such regions, one may
estimate the degree of cell mixing within different cortical
regions. For instance, severe random tangential dispersion of
all cells in the cortex would produce a fine-grained mosaic of
blue and white cells, whereas total radial dispersion would
produce blue and white stripes in which all cells were of the
same colour (which we have failed to detect in any part of the
cortex). That we might reasonably expect to detect such a
degree of clonal restriction with this transgenic mouse is
evidenced in the retina, wherein an analysis of stripes in the
outer nuclear layer reveals virtually no cell mixing (i.e. nearly
every cell in a stripe is of the same colour; Reese et al., unpublished observations).
Our experimental model closely resembles mouse chimaeras
with the important difference that mosaicism is induced in our
system by the naturally occurring phenomenon of X chromosome inactivation, whereas chimaeras rely on the composition
of genotypically dissimilar cellular populations with the
attendant risks of chimeric drift (Warner et al., 1977). Nevertheless, chimeric mice derived from either morula aggregation
or blastocyst injection have been successfully used to detect
lineage relationships in deep versus superficial layers of the
cortex and also in the anteroposterior plane (Crandall and
Herrup, 1990; Fishell et al., 1990). Many of the discoveries in
chimeric mice have since been confirmed using other cell
marking techniques, for example, retroviral markers have
confirmed that different classes of cells normally found in a
radial column can arise from separate progenitors (Parnavelas
et al., 1991; Luskin et al., 1993); whereas at the same time,
other retroviral studies have described clones that span across
1038 S.-S. Tan and others
the entire cortical depth (Luskin et al., 1988; Walsh and Cepko,
1988; Price and Thurlow, 1988).
The present data from transgenic mosaics allow two related
questions to be asked concerning the patterns of cortical cell
dispersion during development, given that we have previously
suggested that both radial and tangential cell dispersion
patterns exist in the mouse neocortex (Tan and Breen, 1993).
First, what is the extent of cell mixing within stripes and is
there variation from one cortical location to another? Second,
does the mixing pattern for neurons versus glia differ? To
answer these questions, we have quantified the relative contributions of blue and white cells in individual stripes, and further
subdivided them into neurons or glia using morphological
criteria. Stripes were further classified according to their
position in the cortex to compare effects of changing cortical
position. In the medial and dorsolateral cortex, we found that
two-thirds of cortical cells within a given stripe are of the same
colour, suggesting that the majority of the cells in these regions
have dispersed radially. This assessment would be an underestimate if earlier horizontal mixing of progenitor cells in the
developing neuroepithelium is taken into consideration (Fishell
et al., 1993). In the extreme case, complete mixing of blue and
white progenitor cells in the neuroepithelium followed by only
radial migration of all progeny could provide results such as
these even where ratios deviate appreciably from 100%. Such
an interpretation of radial dispersion by the majority is
supported by direct imaging of dye-labelled cells. In ferrets,
over 80% of cells in the intermediate zone observed at any one
time appeared to migrate radially (O’Rourke et al., 1992).
Evidence for predominantly radial dispersion is also available
from retroviral studies, which showed that labelled cells tend
to clump together in clusters, and that the majority of such
clusters tend to consist of a single cell type, either neurons,
astrocytes, or oligodendrocytes (Luskin et al., 1988, 1993;
Grove et al., 1993). This would imply that the progeny of
marked precursors have not only followed a common differentiation pathway, but also tend to remain together during cell
dispersion. These studies strongly suggest that cell dispersion
in the cortex operates at the level of cell clusters, not as individual cells. Thus, one way of interpreting split clusters would
be to consider that retrovirally marked progenitors have given
rise to sibling progenitors that later move apart in the ventricular zone; subsequent radial dispersion by their respective
progeny migrating together as a group could account for the
split clusters.
However, our results cannot rule out the alternative interpretation: that the majority of cells in these cortical regions
have obeyed tangential routes of cell dispersion. For instance,
in a white stripe, we note that two-thirds of the cells appear
white (giving the stripe its dominant white colour) with the
remaining one-third being blue. These latter cells may have
entered a stripe by tangential dispersion from others parts of
the brain but for every blue cells that enters a stripe, a white
cell is equally likely to follow. Thus, it could be argued that in
a stripe with two-thirds white cells and one-third blue cells,
only one third white would represent radially dispersed cells,
while two-thirds (comprising one-third white and one-third
blue) would represent random tangential dispersion. This interpretation is reliant upon the assumption that tangential dispersion is both diffuse and long range, capable of traversing
multiple stripe borders. Indeed, retroviral lineage studies
suggest that tangential dispersion may be as great as one centimetre in the rat brain (Walsh and Cepko, 1992).
When the ratio of blue to white cells was estimated in
different dorsal cortical locations, we found a constant 70/30
ratio, irrespective of whether the stripe was obtained from
anterior, middle or posterior locations in the cortex. When the
same analysis was performed in the mediolateral plane, we
found a significant departure from this ratio in the lateral cortex
where the ratio was 50/50, suggesting complete random cell
mixing. These results suggest a place-dependent variation in
the degree of cell mixing across the cortex. Retroviral studies
have also commented on this difference, finding that presumptive clones tended to show more dispersion in the mediolateral than in the antero-posterior axes (Austin and Cepko,
1990). It has been suggested that the geometry of the radial
glial fiber system in the lateral cortex, especially in the intermediate zone, would tend to encourage neurons to shift from
one fiber to another, leading to widely separated termination
points in the cortical plate (Misson et al., 1991). Thus the
neighbour relations of neurons here would not be the same as
their progenitors.
When neurons and glia were separately analysed, they both
showed the same ratio of cell mixing, i.e. one-third of all
neurons and one-third glia in a stripe were of the opposite
colour. Furthermore, when cells of the opposite colour in a
given stripe were sorted into either neurons or glia, it was
found that neurons outnumber glia by about two to one. This
is not surprising given that the mature rodent cortex has a 2:1
ratio of neurons to glia (Parnavelas et al., 1983) and would
suggest that factors that control the direction of dispersion may
be common for both neurons and glia. Such an idea has also
been proposed by others who found that glial progenitors from
the subventricular zone tend to follow the same pathways that
were pioneered by earlier migrating neuroblasts (Levison et al.,
1993). It has been further suggested that tangential dispersion
of neurons may occur by the radial migration of neuroblasts
along radial glial fibers, during which the neuroblast may
switch from one glial fiber to the next (Rakic et al., 1974), or
alternatively, by so called ‘neurophilic’ attachment to axons
running parallel to the surface of the brain (Rakic, 1990). It
would be interesting to see if the tangential mode of neuronal
dispersion in the cortex is related to its ultimate differentiation,
as has been reported in the chick optic tectum (Gray and Sanes,
1991).
What are the implications of our results for the radial unit
and protomap hypotheses of Rakic (1988)? At the heart of the
protomap hypothesis is the radial dispersion of clonally related
cells. The present results indicate that although there is widespread radial dispersion, no cortical area is composed of radial
columns of cells in which every cell in a column is derived
exclusively from a single progenitor cell. However, the
presence of clear cut radial stripes indicate that there must be
a subpopulation of cortical cells, and their progenitors, which
dispersed very little or none at all in the tangential plane. If
there is a protomap, our results would suggest that the imprint
for the protomap would be carried by this influential subpopulation of radially dispersing cells. We suggest that these
radially dispersing cells may then be responsible for parcelling
the cortex into different functional areas by recruitment of
uncommitted (and tangentially dispersed) cells. In this context,
radial dispersion is not only an intrinsic organizational feature
Cortex cell dispersion patterns in transgenic mosaics 1039
of the developing cortex, but also an important mechanism for
imparting positional information from the germinal layer to the
adult structure.
We would like to thank Frank Weissenborn, Suzanne Pearson and
David Francis for excellent technical assistance, Jenny Graves and
Malgorzata Schmidt for advice, and Richard Nowakowski for constructive criticism. The anti-calbindin antibody was a kind gift from
Dr P. Emson. Research in our laboratories was funded by the NHMRC
and NIH.
REFERENCES
Acklin, S. E. and van der Kooy, D. (1993). Clonal heterogeneity in the
germinal zone of the developing rat telencephalon. Development 118, 175192.
Andressen, C., Blumcke, I. and Celio, M. R. (1993) Calcium-binding
proteins : selective markers of nerve cells. Cell Tissue Res. 271, 181-208.
Austin, C. P. and Cepko, C. L. (1990). Cellular migration patterns in the
developing mouse cerebral cortex. Development 110, 713-732.
Boulder Committee (1970). Embryonic vertebrate central nervous system:
revised terminology. Anat. Rec. 166, 257-262.
Collett, D. (1991). Modelling Binary Data. London: Chapman and Hall.
Crandall, J. E. and Herrup, K. (1990). Patterns of cell lineage in the cerebral
cortex reveal evidence for developmental boundaries. Exp. Neurol. 109, 131139.
Fishell, G., Rossant, J. and van der Kooy, D. (1990). Neuronal lineages in
chimeric mouse forebrain are segregated between compartments and in the
rostrocaudal and radial planes. Dev. Biol. 141, 70-83.
Fishell, G., Mason, C. A. and Hatten, M. E. (1993). Dispersion of neural
progenitors within the germinal zones of the forebrain. Nature 362, 636-638.
Goldowitz, D. (1987). Cell partitioning and mixing in the formation of the
CNS: analysis of the cortical somatosensory barrels in chimeric mice. Dev.
Brain Res. 35, 1-9.
Grant, S. G. and Chapman, V. M. (1988). Mechanisms of X-chromosome
regulation. Annu. Rev. Genet. 22, 199-233.
Gray, G. E. and Sanes, J. R. (1991). Migratory paths and phenotypic choices
of clonally related cells in the avian optic tectum. Neuron 6, 211-225.
Grove, E. A., Williams, B. P., Li, D-Q., Hajihosseini, M., Friedrich, A. and
Price, J. (1993). Multiples restricted lineages in the embryonic rat cerebral
cortex. Development 117, 553-561.
Gundersen, H. J. G., Bagger, P., Bendtsen, T. F., Evans, S. M., Korbo, L.,
Marcussen, N., Moller, A., Nielsen, K., Nyengaard, J. R., Pakkenberg,
B., Sorensen, F. B., Vesterby, A. and West, M. J. (1988). The new
stereological tools: disector, fractionator, nucleator and point sampled
intercepts and their use in pathological research and diagnosis. Acta Pathol.
Microbiol. Immunol. Scand. 96, 857-881.
Kirkwood, T. B. L., Price, J. and Grove, E. A. (1992). The dispersion of
neuronal clones across the cerebral cortex. Science 258, 317.
Levison, S. W., Chuang, C., Abramson, B. J. and Goldman, J. E. (1993).
The migrational patterns and developmental fates of glial precursors in the
rat subventricular zone are temporally regulated. Development 119, 611-622.
Luskin, M. B., Pearlman, A. L. and Sanes, J. R. (1988). Cell lineage in the
cerebral cortex of the mouse studied in vivo and in vitro with a recombinant
retrovirus. Neuron 1, 635-647.
Luskin, M. B., Parnavelas, J. G. and Barfield, J. A. (1993). Neurons,
astrocytes, and oligodendrocytes of the rat cerebral cortex originate from
separate progenitor cells: an ultrastructural analysis of clonally related cells.
J. Neurosci. 13, 1730- 1750.
Lyon, M. F. and Kirby, M. C. (1992). Mouse chromosome atlas. Mouse
Genome 90, 43.
Misson, J. -P., Austin, C. P., Takahashi, T., Cepko, C. L. and Caviness, V.
S. Jr. (1991). The alignment of migrating neural cells in relation to the
murine neopallial radial glia fiber system. Cereb. Cortex 1, 221-229.
Nakatsuji, N., Kadokawa, Y. and Suemori, H. (1991). Radial columnar
patches in the chimeric cerebral cortex visualized by use of mouse embryonic
stem cells expressing β-galactosidase. Dev. Growth Differ. 33, 571-578.
O’Leary, D. D. M. (1989). Do cortical areas emerge from a protocortex?
Trends Neurosci. 12, 400-406.
O’Rourke, N. A., Dailey, M. E., Smith, S. J. and McConnell, S. K. (1992).
Diverse migratory pathways in the developing cerebral cortex. Science 258,
299-302.
Parnavelas, J. G., Luder, R., Pollard, S. G., Sullivan, K. and Lieberman, A.
R. (1983). A qualitative and quantitative ultrastructural study of glial cells in
the developing visual cortex of the rat. Phil. Trans. R. Soc. Lond. B301, 5584.
Parnavelas, J. G., Barfield, J. A., Franke, E. and Luskin, M. B. (1991).
Separate progenitor cells give rise to pyramidal and nonpyramidal neurons in
the rat telencephalon. Cereb. Cortex 1, 463-468.
Pinkel, D., Straume, T. and Gray, J. W. (1986). Cytogenetic analysis using
quantitative, high-sensitivity, fluorescence hybridisation. Proc. Natl. Acad.
Sci. USA 83, 2934-2938.
Price, J. and Thurlow, L. (1988). Cell lineage in the rat cerebral cortex: a
study using retroviral-mediated gene transfer. Development 104, 473-482.
Rakic, P. (1972). Mode of cell migration to the superficial layers of fetal
monkey neocortex. J. Comp. Neurol. 145, 61-84.
Rakic, P. (1988). Specification of cerebral cortical areas. Science 241, 170-176.
Rakic, P. (1990). Principles of neural cell migration. Experientia 46, 882-891.
Rakic, P., Stensas, L. J., Sayre, E. P. and Sidman, R. L. (1974). Computeraided three-dimensional reconstruction and quantitative analysis of cells
from serial electron microscopic montages of foetal monkey brain. Nature
250, 31-34.
Tam, P. P. L. and Tan, S.-S. (1992). The somitogenetic potential of cells in the
primitive streak and the tail bud of the organogenesis-stage mouse embryo.
Development 115, 703-715.
Tan, S.-S. and Breen, S. J. (1993). Radial mosaicism and tangential dispersion
both contribute to mouse neocortex development. Nature 362, 638-640.
Tan, S.-S. and Tam, P. P. L. (1993). X-chromosome inactivation in transgenic
mice – studies at the single cell level. In Sex Chromosomes and SexDetermining Genes. (eds. K. C. Reed and J. A. M. Graves), pp. 265-278.
Melbourne:Harwood Academic Publishers.
Tan, S.-S., Williams, E. A. and Tam, P. P. L. (1993). X-chromosome
inactivation occurs at different times in different tissues of the postimplantation mouse embryo. Nature Genet. 3, 170-174.
Vaughan, D. W. (1984). The structure of neuroglia cells. In Cerebral Cortex
vol. 2 (eds. E. G. Jones and A. Peters), pp. 285-329. New York: Plenum
Press.
Walsh, C. and Cepko, C. L. (1988). Clonally related cortical cells show
several migration patterns. Science 241, 1342-1345.
Walsh, C. and Cepko, C. L. (1992). Wide dispersion of neuronal clones across
functional regions of the cerebral cortex. Science 255, 434-440.
Walsh, C. and Cepko, C. L. (1993). Clonal dispersion in proliferative layers of
developing cerebral cortex. Nature 362, 632-635.
Warner, C. M., McIvor, J. L. and Stephens, T. J. (1977). Chimeric drift in
allophenic mice. Transplantation 24, 183-193.
Woolsey, T. A. and Van der Loos, H. (1970). The structural organization of
layer IV in the somatosensory region (S1) of the mouse cerebral cortex: The
description of a cortical field composed of discrete cytoarchitectonic units.
Brain Res. 17, 205- 242.
(Accepted 12 December 1994)