4.2 Characterising wound healing in murine epidermis
4.2.5 Quantifying keratinocyte hyperproliferation during epidermal healing
Tissue samples were collected from 3 female wild type mice of between 6 and 9 weeks old at each of 2 dpw and 4 dpw, and from 4 mice at 5 dpw. Three sections from each wound were stained with ki67, a cell cycle marker; this enables visualisation of keratinocytes that were undergoing mitosis at the time that the tissue was collected. CellProfiler was used to quantify ki67+ cells in 5 images, and the results compared to the manual counts of 2 operators. As shown in Figure 4.8, no significant difference between the 3 counts were observed (rank ANOVA: F=17.3; d.f.=2,8 ; p= 0.302), and the manual counts showed a strong positive correlation with values output from the CellProfiler pipeline. This may be attributed to the relative sparsity of ki67+ cells compared to DAPI+ keratinocytes, and their more consitent level of brightness (an example of the staining is shown in Figure 4.9c).
Figure 4.7: Wound space morphology at 5 dpw. Variations in keratinocyte infiltration of the wound space at 5 dpw. (a) Keratinocytes at the wound edge only, (b) a single continuous layer of keratinocytes across the wound space, (c) a multi-layered structure of keratinocytes. Sectioned stained for DAPI (blue) and integrin-α6 (green). Scale bars 200 µm.
dpw; at 4 dpw there were two distinct subsets – half of the wound edges had only 10-20 ki67+ keratinocytes whereas the others had between 50 and 100; a similar spread of counts was observed at 5 dpw, but with a higher mean. It is possible that the variation could be attributed to the wide range of total keratinocyte cell numbers in the wound space (as illustrated in Figure 4.7). To compensate for this, the number of ki67+ cells is represented as a percentage of DAPI+ cells in Figure 4.9b. This shows a much wider spread of percentages at 4 dpw, ranging from 5 to 31%, whereas at 5 dpw ki67+ keratinocytes make up 5-18% of epidermal cells in the wound space.
Figure 4.8: Comparison of automated and manual ki67+ cell counts. (a) Com- parison of ki67+ cell counts in 5 images, collected by CellProfiler and two operators. (b) Automated and manual DAPI+ cell counts are well correlated. Error bars – mean and range.
A representative example of the location of ki67+ cells at 5 dpw is shown in Figure 4.9c: keratinocytes in cell cycle are clustered in a small region of epidermis adjacent to the wound space, and are also located in a wide band immediately within the wound space. Where a multi-layered population of keratinocytes has formed, some keratinocytes in the lower 2-3 layers (though predominantly in the lowest layer) are also ki67+, suggesting that after migrating into the wound space they take on a proiferative phenotype to contribute further to re-epithelialisation.
Staining with ki67 identifies keratinocytes that were in cell cycle at the time the tissue was collected, which can be used to give estimates of the probability that a cell is proliferating at any given timepoint. However, it does not provide reliable estimates of the overall pro- liferation rate, as no information is provided on how many rounds of division each cell has undergone. To investigate this we used TetO-Hist GFP mice, which have previously been used to quantify turnover of haematopoietic stem cells [223]. Doxycycline was first used to induce H2B-GFP expression in haematopoietic stem cells; then, during a chase period when doxycycline was no longer administered, cells lost GFP expression with each round of cell division they underwent. It was hoped that a similar system could be used to label epidermal keratinocytes, which would allow the number of divisions undergone by cells in various spatial locations to be estimated. However, the period of 3-6 weeks between pups
age-associated differences in skin structure described previously, their putative effects on healing rates, and the effects of hair cycle stage on the healing response, it is important to standardise the age at which mice are used. Therefore it was not feasible to extend the doxycycline induction period to achieve the initial level of labelling required.
One possible reason for the labelling inefficiency is that GFP expression is thought to be induced in cells that divide during the induction period. Therefore we used the induction period to label cells that had divided at least once during the timeframe. 8 female mice between 6 and 9 weeks old were moved to a diet including 0.2 mg/g of doxycycline the day before wounding, and tissue was collected from each of two wounds on 4 animals at 2 dpw and a further 4 animals at 5 dpw. At 2 dpw the percentages of GFP+ cells were very variable, with between 47 and 93% of cells seeming to have undergone at least one round of division; at 5 dpw the percentages were more consistent, with a mean of 91% (Figure 4.10a).
To investigate whether variation in the percentage of GFP+ cells at 2 dpw could be at- tributed to varying numbers of keratinocytes in the wound space, the two variables were compared for each wound (Figure 4.10b). Wounds with higher DAPI+ cell counts showed a relatively low percentage of GFP+ cells, whereas those with a keratinocyte count that suggested significant wound space infiltration had not taken place showed a greater variety of expression levels. This could be a biological effect, where proliferation is less consistent during the early stages of keratinocyte migration into a wound, but may also be a technical issue: it is more difficult to determine where the wound space ends and intact epidermis begins when keratinocytes have not migrated as far as the wound bed. Nonetheless, the finding that the majority of keratinocytes have proliferated at least once during the 5 dpw period is novel, as ki67 staining suggests that only cells in certain spatial regions undergo hyperproliferation. Figure 4.10c shows that keratinocytes in all layers of the intact epidermis adjacent to the wound, and the majority of keratinocytes in the wound space (regardless of their proximity to the initial wound edge) had divided at least once within 5 days of a wound event. Whilst it is disappointing that the TetO-Hist GFP system is not well suited to continuous monitoring of keratinocyte proliferation rates in young mice, information on the spatial locations of previously-dividing cells may be valuable in the development of PDE models and ABMs.
Figure 4.9: Automated ki67+ cell counts and localisation within the wound space. (a) Counts from half the wound space at various timepoints post healing, (b) counts from the full wound space at various timepoints post healing. (a-b) Bars – median. (c) ki67+ cells
Figure 4.10: Automated quantification of GFP+ cells within the wound space in TetO-Hist GFP mice. (a) GFP+ cells as a percentage of DAPI+ cells within the wound space. Bars – mean. (b) Correlation between %GFP+ cells and total number of DAPI+ cells within the wound. (c) Cells in intact epidermis adjacent to the wound are also GFP+. Sectioned stained for DAPI (blue), integrin-α6 (red) and GFP (green). Scale bar 200µm.