Female mice, hemizygous for the H253 X-linked nLacZ transgene (here termed XLacZ+/-), are X-inactivation mosaics and show variegated patterns of β-galactosidase (β-gal) reporter expression in all of their tissues . These mosaics have been widely used to study lineage relationships during development [2,3] but they can also be used to analyse maintenance of adult tissues by stem cells [4,5].
Maintenance of the corneal epithelium by stem cells
The corneal epithelium is an excellent model system for the study of
tissue maintenance and repair because it is a discrete 5–6 cell thick
epithelium replenished by a regionalised stem cell population, which is
confined to the basal layer of the limbus at the periphery of the
These limbal stem cells (LSCs) produce transient (or transit)
amplifying cells (TACs), which proliferate rapidly and migrate
centripetally in the basal epithelial layer until their final division
when both daughter cells move into the superficial layers,
differentiate and are eventually lost from the epithelial surface by
desquamation [8-11]. Previous studies with XLacZ+/- mosaics from our group have shown that LSCs become active after birth  and identified possible genetic influences on LSC function .
Our previous mosaic analysis, suggesting that LSCs become active
after birth, was based on a transition from a pattern of patches to one
of radial stripes. The sequence of events shown in Fig. 1A–D
proposes that some of the basal limbal epithelial cells are specified
as LSCs after the limbal epithelium has been determined. Subsequently
LSCs become activated and the cornea is maintained by centripetal
migration of TACs. Thus, it is predicted that the initial mosaic
pattern of patches is established during fetal and early postnatal
development and the emergence of stripes indicates when stem cell
function begins. Stem cells are required to maintain tissues throughout
life and the idea that stem cell function may decline with age and so
contribute to age-related changes in tissue homeostasis is currently of
great interest [12-14]. However, this possibility has not yet been investigated systematically for LSCs maintaining the corneal epithelium.
Wound healing in the corneal epithelium
LSCs are involved in wound healing, as well as normal tissue homeostasis, and they are up-regulated to replace the lost cells .
Corneal epithelial wound healing proceeds through three stages: (i) an
initial migratory stage to cover the wound with a cell monolayer, (ii)
a proliferative stage to restore the epithelial thickness and (iii) a
period of differentiation to restore the complex epithelial structure .
Several possible mechanisms could drive cell movement during the
initial migratory stage, including population pressure of streams of
cells moving centripetally from the limbus, population pressure from
the wound-margin or other forces, such as chemotaxis or electric fields
Cell proliferation is stimulated in the peripheral limbal epithelium
and to a lesser extent in the corneal epithelium, within 12 hours of
wounding, to replace lost cells . However, proliferation at the wound margin may be suppressed to maintain tissue integrity .
Thus, many new cells will arise in the periphery and migrate
centripetally, as in normal tissue homeostasis, to restore cell numbers
and tissue morphology.
The source of the cells and the extent of cell mixing during the
early phase of wound healing can be investigated experimentally by a
combination of mosaic analysis and organ culture. Wound healing during
24-hour organ cultures reproduces the initial rapid movement of
surrounding corneal epithelial cells to cover the exposed stroma but
after 24-hours the experimental wound is only covered by a single layer
of cells and the epithelium does not stratify during this initial ex vivo healing response [19-21].
The alternative experimental approach of wounding cultured corneal
epithelial cell monolayers is thought to cause a loss of spatial
constraints and induce motility of sheets of epithelial cells rather
than individual cells . However, this may not reflect the situation in vivo or
in organ culture where wounding of a multi-layered stratified
epithelium allows more scope for cell mixing during wound healing. This
is because cells from the upper epithelial layers could contribute to
the cell monolayer that forms during the initial movement phase. It has
also been suggested that cells at the wound margin become less adhesive
and may detach from the epithelial sheet , so promoting cell mixing. Distributions of retrovirus-labelled skin epithelial cells during ex-vivo wound healing  have been interpreted to suggest cell mixing is quite extensive  but it has yet to be determined to what extent cell mixing also occurs during corneal epithelial wound healing.
Quantitative mosaic analysis of limbal stem cell function
Quantitative analysis of distributions of the two cell populations
in mosaic tissues (e.g. analysis of the relative numbers, size, shape
and distribution of patches) can provide more information than
qualitative mosaic analysis [25-27]
but this has not yet been widely exploited. The striping patterns in
the adult cornea are produced by LSC function and cell movement in the
epithelium. LSC function can be compared in different experimental
groups by quantitative analysis of stripe numbers.
In an adult corneal epithelium, showing mosaic expression of a LacZ transgene, the stripes of β-gal-positive cells are elongated patches formed from one or more β-gal-positive
coherent clones whose descendents have migrated centripetally. (The
terms 'patch' and 'coherent clone' are defined in the Methods section.)
A stripe spans the corneal radius, so its length is not affected by the
number of LSCs and is not relevant to the analysis. The stripe width,
however, is variable and depends in part on the number of adjacent
corneal epithelial coherent clones belonging to the same cell
population (either β-gal-positive or β-gal-negative). Clearly, an individual stripe is more likely to be made up of multiple adjacent β-gal-positive corneal epithelial coherent clones when the proportion of β-gal-positive
cells in the corneal epithelium is higher. This source of variation in
stripe width can be factored out by dividing the observed mean width of
β-gal-positive stripes by the function 1/(1-p), where p is the proportion of β-gal-positive cells around the circumference [25,28].
The resultant 'corrected mean stripe width' can be used to derive a
'corrected stripe number' (see Methods section). This is proportional
to the number of corneal epithelial coherent clones and can, therefore,
be used to compare LSC function in different groups. A coherent clone
of β-gal-positive limbal stem cells will
produce a coherent clone of cells in the basal layer of the corneal
epithelium that extends to the centre as cells move centripetally and
extends to the suprabasal and outer epithelial layers as cells leave
the basal layer. Each corneal epithelial coherent clone is assumed to
be formed from a single active coherent clone of LSCs. Thus, the number
of active LSC coherent clones can be compared in different groups of
mosaic eyes by comparing the corrected stripe numbers.
Although, the corrected stripe number is related to the number of
active LSC coherent clones it does not provide a direct estimate of LSC
numbers. This is partly because the proportion of LSCs that are active
may vary and also because the number of LSCs per LSC coherent clone may
vary. For example, variation in the number of LSCs per LSC coherent
clone may occur because of differences in the extent of cell mixing
during development of the surface ectoderm, from which the corneal and
limbal epithelia develop (compare Fig. 1A–D, showing the consequences of extensive cell mixing during development, with Fig. 1E–H, showing the consequences of less cell mixing).
The aims of this study were to better characterise LSC function and
the streaming and mixing behaviour of the cells they produce during
maintenance, repair and ageing of the mouse corneal epithelium.
Analysis of mosaic patterns in intact and wounded corneas demonstrated
that (i) LSC function declines with age, (ii) little cell mixing occurs
either during normal maintenance of the corneal epithelium or during
wound healing, (iii) the main driving force during wound closure is not
population pressure from centripetally streaming cells produced by LSCs
and (iv) quantitative and temporal mosaic analyses provide new
possibilities for studying stem cell function in tissue maintenance and