Key changes in chromatin mark mammalian epidermal differentiation and ageing

ABSTRACT

Dynamic shifts in chromatin states occur during embryonic epidermal development to support diverse epigenetic pathways that regulate skin formation and differentiation. However, it is not known whether the epigenomes established during embryonic development are maintained into adulthood or how these epigenetic mechanisms may be altered upon physiological ageing of the tissue. Here, we systematically profiled the nuclear enrichment of five key histone modifications in young and aged mouse epidermis and identified distinct chromatin states that are tightly correlated with cellular differentiation, as well as chromatin alterations that accompanied epidermal ageing. Our data showed that histone modifications, which become differentially enriched in undifferentiated basal or differentiated suprabasal cells during embryonic development, retained their distinct cell-type specific enrichment patterns in both young and aged adult tissues. Specifically, high levels of H3K4me3, H4K20me1 and H4K16ac marked the proliferative basal cells, while differentiated suprabasal cells accumulated H3K27me3 and H4K20me3 heterochromatin with a concomitant deacetylation of H4K16. We further identified shifts in the chromatin in the aged basal epidermis, which exhibited markedly reduced levels of H4K16ac, absence of high H4K20me1 staining and increased cell-to-cell variability in total histone H3 and H4 content. Changes in the chromatin profiles in aged tissues paralleled the altered expression of their corresponding histone modifiers in the basal keratinocytes. These results thus reveal the key histone signatures of epidermal differentiation that are conserved from embryonic development to adult homoeostasis, and provide insights into the epigenetic pathways underlying physiological skin ageing.

Introduction

Post-translational modifications (PTMs) to histone proteins play a significant role in regulating gene expression mechanisms associated with cell fate determination, cell cycle progression, DNA replication, and repair during development and in adulthood [1–3]. Post-translational marks such as acetylation, methylation, or phosphorylation at specific residues of core histone proteins are known to alter chromatin architecture, modulate DNA accessibility and facilitate the locus-specific recruitment of transcription factors [4–6]. Histone acetylation mediated by histone acetyltransferases is often associated with activated transcription, while hypoacetylation by histone deacetylases leads to gene repression [7]. For example, H3K27ac and H4K16ac marks are enriched at active enhancers and transcription start sites but depleted on inactive enhancer regions in embryonic stem cells [8]. Methylation of lysines 4, 36 and 79 on histone H3 at promoter and enhancer loci is well-associated with transcriptional activation while tri-methylation of H3K9, H3K27 and H4K20 elicits gene repressive and silencing effects [9]. In addition, histone modifications such as phosphorylation of H3 at Ser10 and Ser28, mono-methylation of H4K20 and deacetylation of H4K12 and H4K16 facilitate chromosome condensation and impact cell cycle progression [10]. The effects of histone PTMs are often synergistic and dynamic, as multiple marks are put on or erased in combination to regulate distinct chromatin states and gene expression as cells respond to external and internal signals to alter cellular processes or adopt new functions during development and growth.

The skin epidermis is characterized by stratified layers of proliferative stem-like keratinocytes in the stratum basale and terminally differentiated cells in the suprabasal strata spinosum, granulosum and corneum. The stem-like basal keratinocytes can divide to give daughter cells that either re-populate the basal cell layer or undergo stepwise terminal differentiation to replace cells within the superficial epidermal layers [11]. A tightly regulated balance between progenitor self-renewal and daughter cell terminal differentiation is critical to ensure epidermal homoeostasis [12]. The control of epidermal cell proliferation, differentiation, and maintenance of cell identity during development and adulthood has been shown to be regulated via gene regulatory mechanisms conferred by chromatin modifying factors [13–15]. Epidermal stratification during embryonic development is perturbed upon the deletion of histone deacetylases Hdac1, Hdac2, and methyltransferases Setd8 and Jarid2, while knockout of the histone deubiquitinase, Mysm1, led to aberrant epidermal differentiation and homoeostasis in the adult mouse [13,16–18]. Hence, the epidermis is a good model for investigating the complex chromatin dynamics associated with shifts in gene expression programmes that reflect the different proliferative and differentiated cell states within the tissue. Indeed, specific histone modifications have been shown to distinguish basal progenitor cells from suprabasal differentiated keratinocytes during epidermal development. Notably, elevated levels of H3K4me3, H4K20me1, and H4K16ac were present in the proliferative basal keratinocytes compared to the differentiated cells in the suprabasal layers of the mouse embryonic epidermis. By contrast, high levels of H3K27me3 and H4K20me3 were distinctly associated with the differentiated epidermal cell identity [19]. Nuclear H3K4me3, panH3ac and panH4ac were also significantly altered as keratinocytes undergo differentiation in vitro, pointing to a strong correlation of these PTMs with the self-renewing and differentiating epidermal cell states [20]. While a number of studies have shown that progenitor and differentiated cells in the developing epidermis exhibit distinct histone modification profiles [19,21,22], how these epigenetic landscapes and gene regulatory mechanisms are altered with age in the skin epidermis has not yet been fully explored.

Natural ageing is often related to the loss of tissue-specific stem cells that maintain regenerative homoeostasis in the various organs of the body. Studies have demonstrated that aged cells in the heart, brain, bone marrow and liver acquire significant epigenetic changes, including global and locus-specific alterations in DNA methylation and histone methylation and acetylation, which may be correlated with the loss of regenerative potential [23–28]. In addition, epigenome alterations associated with genomic instability and deterioration of nuclear and chromatin architecture are commonly linked to age-related loss of cellular functions [29–32]. The aged skin epidermis is marked by reduced keratinocyte turnover, slower basal cell proliferation and reduced regenerative capacity [33]. The age-associated changes to the structure and function of the skin in both mouse and humans, such as epidermal thinning and delayed re-epithelialization during injury, are indicative of the altered proliferative and differentiation potential of the basal keratinocytes in the skin. However, epigenetic changes that correlate with the regenerative states of these cells in the aged epidermis have not been well-characterized. As specific chromatin signatures are linked to stemness and differentiation of keratinocytes in normal embryonic and neonatal skin [19], systematic analysis of these histone marks in adult young and aged skin will provide important insights into the epigenetic mechanisms that underlie the decline of epidermal regeneration upon ageing.

In this study, we explored whether histone modifications that are distinctly associated with different proliferative and differentiation states of epidermal keratinocytes during embryonic development are maintained in adulthood, and how these chromatin landscapes may be affected by ageing-related mechanisms. Our findings revealed that the global enrichment profiles of H3K4me3, H3K27me3, H4K20me3, H4K16ac and H4K20me1, which become distinctly defined in proliferative basal or differentiated suprabasal keratinocytes by late embryonic development, are stably maintained throughout adulthood in mice. We further identified ageing-related declines in the global H4K16ac and H4K20me1 levels, concomitant with decreased expression of the associated enzymatic modifiers in the basal keratinocytes. Our results highlighted a strong correlation of specific epigenomic changes with the proliferative and regenerative potential of young and aged epidermal tissues and provided key insights into the epigenetic mechanisms underlying skin ageing.

Results

Higher variability of core histone H3 and H4 in the aged tail epidermis compared to young tissue

To examine whether epidermal histone modification patterns observed during embryonic development were conserved in adulthood as well as assess how they were affected by ageing-related mechanisms, we carried out a systematic characterization to analyse the distribution of five specific histone marks, H3K4me3, H3K27me3, H4K16ac, H4K20me1 and H4K20me3, in young (3 months) and aged (24–26 months) mouse skin tissues. The five histone modifications were selected on the basis that they were found to be differentially enriched and associated with either the proliferative stem-like or the terminally differentiated cell states of basal and suprabasal keratinocytes in the developing embryonic epidermis [19]. For this assay, tail skin sections collected from young and aged mice were simultaneously processed and co-immunostained for a selected histone modification and the basal cell marker Keratin 14 (K14) to allow for the quantification and comparison of the global levels of each chromatin mark between basal and suprabasal cells in the young and aged epidermis.

Using this assay, we first examined the levels of total core histones H3 and H4 in young and aged epidermal tissues. Immunofluorescence analyses revealed no differences in the mean nuclear intensities of panH3 and panH4 in the basal and suprabasal cells of the young tail epidermis (Figure 1(a–d)). This observation is congruent with a consistent total chromatin content in the keratinocytes present in the different layers of the epidermis, which was similarly noted in the embryonic epidermis. Unlike in the young tissues, differences in total H3 and H4 content were observed in the aged skin. While the distribution of panH3 or panH4 staining in the basal and suprabasal nuclei of the aged epidermis were similar, the mean fluorescence intensities of both panH3 and panH4 were found to be reduced in the terminally differentiated cells, reflecting differences in the core histone content between cell types in the aged tissue (Figure 1(c,d)). When compared with young epidermal cells, we found that the mean levels of core histone H3 and H4 levels were highest in the basal cells of the aged (Figure 1(c,d)). We further noted that these cells also exhibited the widest spread of nuclear intensities compared to the aged suprabasal and the young basal and suprabasal cells. These results indicated a greater variability of total histones H3 and H4 content in the aged basal epidermal cells.

Figure 1.

Total chromatin content differs in the basal cells of young versus aged tail epidermis. (a, b) Representative micrographs of young (3 months) and aged (24 months) WT mouse tail epidermis immunostained with DAPI, K14, panH3, or panH4. Scale bar = 25 µm. (c) Mean fluorescence intensity per nucleus of panH3 in young and aged basal (B) versus suprabasal (SB) cells. (d) Mean fluorescence intensity per nucleus of panH4 in young and aged basal (B) versus suprabasal (SB) cells. 200–240 nuclei were analysed per group (10–12 FOVs), n = 5–6 mice. Data shown with mean ± SD, and analysed by two-way ANOVA and Tukey’s multiple comparisons test. n.s: not significant, ***p < 0.001.

H3K4me3 and H3K27me3 are differentially associated with epidermal differentiation and are not altered in the ageing epidermis

Histone modifications, H3K4me3, and H3K27me3, associated with transcriptional activation and repression, respectively, were found to be differentially correlated with epidermal differentiation in embryonic and neonatal skin [19]. To investigate whether these chromatin marks are altered in adult skin or upon ageing, we profiled the global enrichment of H3K4me3 and H3K27me3 in basal and suprabasal cells of young and aged tail epidermis. We found the mean global levels of H3K4me3 to be significantly lower in the suprabasal cells compared to the undifferentiated basal keratinocytes in both young and aged tissues (Figure 2(a,b)). By contrast, markedly higher enrichment in the mean fluorescence intensity of the repressive mark H3K27me3, was observed in the differentiated suprabasal cells compared to the basal cells (Figure 2(c,d)). These findings parallel the distinct distribution of H3K4me3 and H3K27me3 observed in embryonic epidermal development, indicating that the correlation of high H3K4me3 with the undifferentiated, and increased H3K27me3 with the differentiated, cell states is conserved in both the embryonic and adult epidermis.

Figure 2.

Enrichment of histone modifications differentially associated with epidermal differentiation is consistent in young and aged skin. (a, c) Representative images of young and aged mouse tail epidermis immunostained with DAPI, K14, histone H3K4me3 or H3K27me3. Scale bar = 25 µm. (b, d) Mean fluorescence intensity per nucleus in young and aged basal (B) versus suprabasal (SB) cells for histone H3K4me3 and H3K27me3. Data shown with mean ± SD, and analysed by two-way ANOVA and Tukey’s multiple comparisons test. n.s: not significant, **p = 0.002, ***p < 0.001. 150–360 nuclei were analysed per group (10–12 FOVs), n = 5–6 mice. (e-f) Cumulative frequency distribution of histone H3K27me3⁺ nuclei in young and aged basal or suprabasal epidermal tail skin. n.s: not significant, ***p < 0.001, Kolmogorov-Smirnov test (Basal KSD = 0.09722; Suprabasal KSD = 0.1694). 360 nuclei were analysed per group, n = 6 mice.

No notable differences in the mean intensities of H3K4me3 and H3K27me3 were found between young and aged tissues (Figure 2(b,d)). However, the distribution of H3K27me3-stained suprabasal, but not basal, nuclei was significantly different between young and aged tissues (Figure 2(e,f)). We applied the Kolmogorov-Smirnov test to compare the cumulative distribution between samples and found that the KS-distance of H3K27me3 intensities between young and aged suprabasal cells to be 0.1694, compared to 0.0972 for basal cells. The age-associated difference in the distribution of H3K27me3 across suprabasal keratinocytes suggest that while H3K27me3 is acquired upon differentiation in both the young and aged epidermis, the extent of enrichment is more variable in individual aged cells. Overall, we found that the enrichment of these two histone modifications to be clearly correlated with changes in epidermal cell states but not with ageing.

Acquisition of H4K20me3-marked heterochromatin correlates to epidermal cell differentiation in young and aged tissues and is altered in aged suprabasal cells

The H4K20me3 histone mark is concentrated in punctate constitutive heterochromatin foci in the nucleus and is associated with transcriptional silencing. Immunostaining revealed significant enrichment of H4K20me3 heterochromatin foci in the suprabasal cells in both young and aged adult tail epidermal tissues (Figure 3(a,b)). In the young tissues, the mean intensity per nucleus in the suprabasal layer was 29% higher than in the basal cells, while in the aged skin, H4K20me3 level was 16% higher in the differentiated cells. These results are consistent with previously reported accumulation of H4K20me3-rich heterochromatin foci in the suprabasal cells of embryonic dorsal skin tissues [19].

Figure 3.

Heterochromatic H4K20me3 foci are elevated in suprabasal cells and are reduced upon ageing. (a) Immunofluorescence analyses of histone H4K20me3⁺ heterochromatin foci in the young and aged tail epidermis. Scale bar = 25 µm. (b) Mean H4K20me3 fluorescence intensity per nucleus in young and aged basal (B) versus suprabasal (SB) cells. 200–300 nuclei were analysed from 10 FOVs per group, n = 5 mice. (c) Normalized number of H4K20me3⁺ nuclear foci in the young and aged basal versus suprabasal epidermal cells. The number of foci per nucleus was normalized to the mean nuclear area of all cells analysed (47.8µm²). 40 nuclei were analysed per group. (d) Mean fluorescence intensity per H4K20me3⁺ foci in young and aged basal and suprabasal epidermal skin cells. 360 foci were analysed per group. All data shown with mean ± SD, and analysed by two-way ANOVA and Tukey’s multiple comparisons test. **p = 0.009, ***p < 0.001. n.s: not significant.

To account for the increase in mean nuclear signal intensities in the suprabasal cells, we quantified the number of H4K20me3 foci in each nucleus, as well as the mean fluorescence intensity of each foci. Average counts of 5–6 foci per nuclear plane (normalized to average area of 48μm²) were obtained for both basal and suprabasal cells of the young epidermis, while the aged basal and suprabasal cells had an average of 6–7 foci per nuclear plane (Figure 3(c)). Hence, the number of H4K20me3-marked heterochromatin foci was largely unchanged upon epidermal differentiation and ageing. By contrast, we found that the mean fluorescence intensities of the H4K20me3 foci were significantly increased in the differentiated suprabasal cells in both young and aged epidermis (Figure 3(d)). The difference in H4K20me3 foci enrichment was much more distinct in the young epidermis, with both the spread and the mean signal intensities of H4K20me3 foci in the suprabasal cells being greater in the young compared to the aged tissues. The distribution of mean intensity of H4K20me3 foci in the young suprabasal cells was much broader, with 71% of these foci having signal intensities greater than the mean intensity (87.823) of the aged suprabasal cells. No major age-related differences were observed in the basal cells. Taken together, these findings provided further support that enrichment of H4K20me3-marked heterochromatin foci is a hallmark of epidermal cell differentiation. The results also suggest that aged cells, in comparison with young keratinocytes, may acquire fewer H4K20me3-modified nucleosomes per foci upon differentiation. Hence, unlike the histone modifications H3K4me3 and H3K27me3, mechanisms involved in the formation of H4K20me3 heterochromatin foci appear to be altered in aged tissues.

Enrichment of cell cycle-related histone modifications, H4K20me1 and H4K16ac, in basal keratinocytes decreases with age

We also sought to investigate whether chromatin modifications are associated with changes in the regenerative potential of skin epidermis. We thus examined two histone marks, H4K20me1 and H4K16ac, which oscillate with different phases of the cell cycle and had been linked to the control of cell division and proliferation [34–36]. The total amount of H4K20me1 was reported to be markedly low in the G0/G1 phase but peak in late G2/M phase of cycling cells, and high levels of H4K20me1 correlated with Ki67- and phosphoH3- positive proliferative cells [19,35,36]. As had been previously reported in the embryonic epidermis, two distinct H4K20me1-positive populations in the basal epidermal layer can be distinguished based on the level of mean fluorescence intensity in the young adult tail skin, while H4K20me1 was barely detectable in the suprabasal cells (Figure 4(a)). We observed a subset of intensely labelled H4K20me1⁺ nuclei with mean fluorescence intensity above 100, which we defined as H4K20me1^hi, while cells with mean intensity <100 were termed H4K20me1^lo (Figure 4(a,b)). Given the correlation of fluctuating H4K20me1 levels with different cell cycle phases, we proposed that H4K20me1^hi nuclei represented the subset of cells that were cycling through G2/M. To correlate high H4K20me1 nuclear enrichment with cell proliferation, we carried out co-staining of H4K20me1 with proliferative cell marker, Ki67 (Figure 4(c), Fig S1). We observed that all H4K20me1^hi nuclei were Ki67-positive, indicating that high H4K20me1 enrichment indeed marks a subset of proliferative cells (Figure 4(c), Fig S1A). In young skin, 20.2% of all basal epidermal cells were Ki67-positive, of which 12.5% were also H4K20me1^hi. By contrast, 10.9% of basal cells in the aged tissues were Ki67-positive and only 5.4% of these cells had high H4K20me1 levels (Figure 4(c), Fig S1A). These observations demonstrated that 50% of proliferative cells in both young and aged skin are characterized by high H4K20me1 nuclear enrichment and aged epidermis had significantly fewer such proliferative cells compared to young tissues.

Figure 4.

Histone modifications associated with cell cycle and proliferation are decreased in the aged epidermis. (a) Representative images showing the distribution of histone modification H4K20me1, DAPI and K14 in young and aged tail skin. Arrows indicate intensely stained H4K20me1^hi nuclei. (b) Mean H4K20me1 fluorescence intensity per nucleus in young and aged basal (B) versus suprabasal (SB) cells. Dashed line demarcates mean nuclear intensity of 100 for segregation of H4K20me1^hi and H4K20me1^lo groups. Data shown with mean ± SD, and analysed by two-way ANOVA and Tukey’s multiple comparisons test. *p = 0.024, ***p < 0.001. 240–360 nuclei were analysed from 12 FOVs per group, n = 6 mice per group. (c) Percentage of Ki67⁺ only versus Ki67⁺ and H4K20me1^hi cells in the basal layer of young and aged epidermis. Error bars = SD. *p < 0.05, two-tailed unpaired t-test. 630–740 cells were analysed from 9 FOVs per group, n = 3 mice per group. (d) Micrographs showing the enrichment of H4K16ac in the basal layer of young and aged epidermis. Arrows indicate K14-negative suprabasal cells with very low or undetectable H4K16ac. (e) Mean H4K16ac fluorescence intensity per nucleus in young and aged basal (B) versus suprabasal (SB) cells. Data shown with mean ± SD, and analysed by two-way ANOVA and Tukey’s multiple comparisons test. ***p < 0.001, n.s: not significant. 180–300 nuclei were analysed per group, n = 6 mice. (f) Cumulative frequency distribution of H4K16ac⁺ nuclei in young and aged basal cells. ***p < 0.001, Kolmogorov-Smirnov test (KSD = 0.2467). 300 nuclei were analysed per group. Dashed lines demarcate the proportion of basal cells with intensities below 100 in both young and aged. (g-h) Maximum intensity z-projection images of H4K20me1 and H4K16ac in young and aged dorsal epidermis. Scale bars = 25 µm.

H4K16ac had been shown to be rapidly lost upon epidermal cell differentiation in embryonic skin [19]. We observed similar distributions of H4K16ac in the adult epidermis from both young and aged mice, which exhibited high H4K16ac staining of all cells in the basal layer while the outermost suprabasal layers had almost undetectable levels of H4K16ac (Figure 4 (d,e)). Hence, H4K16 hyperacetylation appears to be a distinctive feature of the progenitor cells in the basal layer and the swift erasure of H4K16ac is a hallmark of epidermal cell differentiation. To investigate whether H4K16 hyperacetylation is affected by ageing, we compared the levels of H4K16ac in the basal cells of the young and aged epidermis and found that H4K16c was significantly lower in the aged epidermis, despite these cells having higher total H4 content compared to young basal keratinocytes (Figure 4 e and f)). Cumulative distribution analyses further showed that the spread of H4K16ac-positive basal cells was different between the young and aged groups (KS distance = 0.2467), with a greater proportion of aged basal cells (72.3%), compared to 53.3% of young basal cells, having lower H4K16ac intensities of <100 (Figure 4(f)). These findings showed that H4K16 hyperacetylation is reduced in ageing basal cells. Co-staining of H4K16ac and Ki67 also revealed that the aged skin contained ~50% less double H4K16ac-, Ki67- positive cells than the young epidermis (Fig S1B,C).

To corroborate differences in the levels of H4K16ac as well as H4K20me1 in young and aged basal cells, we also examined the enrichment of these modifications in the dorsal skin. We found that, similar to the tail epidermis, the basal layers of young dorsal skin also contained higher global H4K16ac levels and more H4K20me1^hi nuclei compared to the aged dorsal skin (Figure 4(g,h)). Taken together, high H4K20me1 and H4K16ac content appear to be characteristic of proliferative basal epidermal cells and alterations in the levels of these chromatin modifications are correlative with age-associated reduction in the regenerative potential of the tissues.

Writers and erasers of the ageing-related histone modifications are differentially expressed in cells derived from young and aged epidermal tissues

We then sought to explore whether the age-related changes in the chromatin marks were caused by altered regulation in the expression of the corresponding histone modifying enzymes. To address this, we analysed the expression of some of the known writers and erasers of the five mapped histone modifications in basal epidermal cells which were either directly purified from dorsal skin (dorsal KCs), or cultured from the tail epidermis (tail MPEKs). We found that the expression of modifiers of H4K20me1, H4K20me3 and H4K16ac, histone marks that exhibited significant age-related changes in young and aged epidermal tissues, were distinctly different in the young and old purified dorsal KCs (Figure 5(a-c)). The expression of Kmt5a, the methyltransferase that mono-methylates H4K20, was found to be markedly downregulated by 69% in aged dorsal KC (Figure 5(a)). Expression of Kmt5c, the H4K20me3 methyltransferase, was greatly reduced by 90% in the aged dorsal KCs compared to young keratinocytes, though expression of the related Kmt5b methyltransferase was not different in the young and aged cells (Figure 5(b)). We also compared the expression levels of H4K16ac acetyltransferases, KAT8 and KAT5, and deacetylases, SIRT2 and HDAC2, in young and aged keratinocytes. We found that Kat8 and Kat5 were both significantly downregulated, by 36% and 80% respectively, in aged KCs, while the expression of Sirt2 and Hdac2 were reduced at a lesser extent (Figure 5(c)).

Figure 5.

Expression of specific histone modifiers in isolated dorsal epidermal cells correlates with histone modification levels in skin tissues. (a-e) Relative gene expression of histone modifiers (writers – blue; erasers – red) associated with H4K20me1 (a), H4K20me3 (b), H4K16ac (c), H3K4me3 (d) and H3K27me3 (e) examined by RT-qPCR in either purified dorsal basal keratinocytes or tail-derived primary epidermal keratinocytes from young (Y) and aged (A) mice. Data are shown as mean normalized expression ± SD. Data were normalized to Actb, and fold change calculated relative to the average expression in young keratinocytes. **p < 0.01, ***p < 0.001, two-tailed unpaired t-test with multiple comparisons. n = 4–5 mice for dorsal basal KCs, n = 9 for mouse tail primary epidermal KCs.

Though we did not detect age-associated differences in the enrichment of H3K4me3 and H3K27me3, we went ahead to examine whether the expression of the associated methyltransferases and demethylases were altered in the basal keratinocytes. We found that the expression of Kmt2a was 36% lower in aged dorsal KCs but transcript levels of Kmt2b was unchanged in young and aged cells (Figure 5(d)). No differences in Kdm2b expression was detected. The transcript levels of Ezh2, Kdm6a and Kdm6b were also not significantly different between the young and aged basal keratinocytes (Figure 5(e)).

Together, these results demonstrated a strong correlation between the enrichment of the histone modifications with the changes in the expression of their associated enzymatic modifiers in the young and aged epidermal cells. In vivo reductions in the global enrichment of H4K20me1, H4K20me3 and H4K16ac in the aged epidermis corresponded to an age-associated loss in the expression of their enzymatic writers. Notably, while major gene expression changes were detected in the in vivo purified young and aged basal keratinocytes for Kmt5a, Kmt5c, Kat5 and Kat8, we did not detect significant transcript changes in any of the genes analysed in the cultured MPEKs (Figure 5(a-e)). We postulate that age-related differences present in vivo may have been obscured by the selection and expansion of highly proliferative cell populations in vitro, highlighting a major potential drawback in using cultured cells for studies of ageing mechanisms.

Discussion

The present study addressed two main questions. First, we investigated whether chromatin changes which are established upon epidermal formation during embryonic development are maintained into adulthood. Secondly, we examined the in vivo shifts in histone marks that are acquired by epidermal cells upon natural ageing. Our data showed a clear conservation of the chromatin-based mechanisms that follow epidermal differentiation in embryonic, young and aged adult skin. While adult tail skin was analysed in this study, the chromatin landscapes of basal and suprabasal cells were highly comparable to that observed in late embryonic and postnatal dorsal skin epidermis [19]. High levels of H3K4me3, H4K20me1 and H4K16ac clearly mark proliferative undifferentiated basal keratinocytes, while differentiation of these cells leads to the accumulation of H3K27me3, increased H4K20me3-marked constitutive heterochromatin foci, and the rapid deacetylation of H4K16. Hence, we propose that these dynamic histone profiles represent the epigenetic signatures of epidermal differentiation. Our results also highlighted key correlations of the H4K20me1, H4K20me3 and H4K16ac epigenomes with epidermal ageing.

The systematic parallel characterization of the young and aged skin tissues in this study afforded the analysis of chromatin modification changes at the single cell level, allowing us to map age-related epigenomic differences to the different groups of cells present in the epidermis. The basal keratinocytes exhibited the most age-associated chromatin changes such as decrease in global H4K16 acetylation, absence of H4K20me^hi staining, as well as greater cell-to-cell variability in total histone H3 and H4, and H4K16ac content. This is not surprising, given that basal cells are the self-renewing stem-like cells of the epidermis that reside in the adult tissues over a lifetime, while the suprabasal cells are short-lived cells that transit through the epidermal layers and are constantly shed [37]. While sub-populations of ‘ageing-resistant’ epidermal stem cells that do not exhibit age-associated changes in gene expression and multipotency had been reported [38], we did not identify distinct subsets of cells in the aged epidermis that retained the chromatin landscapes of young basal cells. This could be due to the rarity of such ageing-resistant cells in the tissues, which would impede their identification within this study.

Of all the age-associated changes observed, the presence of aged basal cells that contained higher total H3 and H4 than young cells was the most unexpected. Increased total histone content in these aged cells was counter-intuitive to reports of large-scale H3 and H4 reduction, resulting from increased degradation or decreased biosynthesis, in ageing organisms such as yeast and worm, and senescing lung fibroblasts [36,39–41]. However, it has to be noted that there have been very few direct demonstrations of global histone depletion in physiologically aged mammalian tissues. While the expression of histone genes was reportedly reduced in aged quiescent muscle satellite cells, total H3 levels in the liver and cerebellum of young and aged mice were shown, using western blot and ELISA assays, to be largely unchanged [26,42]. Instead, both genomic loci- specific increases and decreases in H3 nucleosome occupancy were observed in a cell-type or context- dependent manner in the aged tissues. Thus, rather than a general mechanism of ageing, changes in total histone content in response to physiological ageing is likely to be highly regulated in a tissue-specific manner in multicellular organisms.

The complexity and tissue specificity of epigenetic alterations associated with ageing mechanisms extend from total histone content to highly diverse remodelling of specific histone modifications in different organisms, tissues and cells. While alteration in H3K4me3 levels was found to impact lifespans of worms, no major changes in global H3K4me3 levels were found in the ageing mouse brain, muscles, and haematopoietic stem cells (HSCs) [23–26,28,43]. Gains of H3K27me3 were reported in aged muscle satellite cells and ageing killifish, while reductions in the same modification were observed in aged worms and mouse brains [23,26,44,45]. Likewise, both hyper- and hypo- acetylation of H4K16 had been linked to longevity in organisms and cells such as yeast, mouse HSCs and human fibroblasts [28,36,46–48]. In our study of the ageing epidermis, we find that significant changes in H3K4me3 and H3K27me3 levels corresponded more strongly with cell identity than with age when the same cell types were compared. It is possible that H3K4me3- and H3K27me3- driven ageing mechanisms in these cells are mediated through locus-specific remodelling and genomic re-distribution instead of changes in global enrichment. Indeed, the increased cell-to-cell variability observed in aged epidermal cells may be reflective of age-associated genomic redistribution of H3K27me3 and H4K16ac modifications. In support of this, cells derived from the cerebellum and blood of aged, but not young, mice were found with high sample-to-sample variations in H3K27me3 and H3K4me3 genomic profiles that corresponded to increased transcriptional noise in the aged cells [24,27]. It is thus important to recognize that physiological organismal ageing is highly complex and likely regulated by diverse epigenetic mechanisms in different tissues and cells.

The two chromatin marks, H4K20me1 and H4K16ac, which are reduced at a global level in basal epidermal cells in association with ageing, are both known to oscillate throughout the cell cycle. While these two modifications had been found to be enriched around transcription start sites of actively transcribed genes [8,9,49], both marks were thought to mediate gene repression and cell cycle by altering higher order chromatin compaction and chromosomal condensation [49–54]. Notably, decreases in H4K16ac and H4K20me1 levels had been independently associated with cell cycle exit and replicative senescence in cultured cells [36,47,52,54,55]. Loss of H4K20 mono-methylation and H4K16 hypoacetylation were also linked to ageing mechanisms in the neuronal and blood cells of aged mice [48,56]. Similarly, both H4K20me1 and H4K16ac levels were markedly reduced in the aged basal epidermal cells in this study. Collectively, these studies support the theory of stem cell ageing, which postulates that ageing is the consequence of a decline in the number of stem cells that can be triggered to undergo active cell division to support homoeostasis and replenishment of the tissue [57]. Interestingly, H4K16ac levels are increased in the newly formed epidermis following injury, providing further correlation of this histone modification with activated proliferative epidermal cells [58]. It will be interesting to examine whether more cells with high H4K20me1 enrichment are also detected in the re-epithelializing wounds. The analysis of H4K20me1 and H4K16ac dynamics could be useful for assessing the regenerative capacity and healing potential in wounds.

This study was a systematic characterization of how the epigenomes of embryonic epidermal cells may be dynamically altered under normal homoeostatic conditions and upon physiological ageing. The assessment of global enrichment revealed key cell intrinsic changes that are correlated with differentiation and ageing mechanisms. High resolution ChIP-seq mapping of these histone marks will next be necessary to delineate the chromatin changes at the genomic level and elucidate the transcriptomic outputs that are regulated by these epigenetic networks in the ageing skin. In conclusion, this study contributes to our understanding of how epigenetic modifications associated with maintaining skin homoeostasis are altered in aged skin. The findings of major age-associated expression changes in the related histone modifiers give insights into key factors that may be targeted for improving cell proliferation and epidermal turnover, thereby improving skin health and reducing susceptibility to ageing-related skin conditions.

Materials and methods

Tissue preparation, immunofluorescence staining and microscopy

Female and male C57BL/6 J wild type young (12 weeks) and aged (24–26 months) mice were internally bred and housed at the Biological Resource Centre at the A*STAR Institute (Singapore). All animal work was carried out according to approved IACUC protocols at the Biological Resource Centre, A*STAR. Tail and shaved dorsal skin tissues were excised and embedded in Frozen Section Media (Leica Biosystems, Germany). Tail skin cryoblocks were generated from 6 mice in each group and stored at −80^°C. 7 µm frozen serial sections were cut and air-dried for 10 minutes. Sections were fixed in 0.5% PFA in PBS for 1 hour at room temperature (RT). Antigen retrieval was performed using 0.5% SDS/PBS for 5 minutes at RT, and tissues were thoroughly washed with 1X PBS. Non-specific binding was blocked by incubation in 1% bovine serum albumin (BSA) and 5% normal goat serum in PBS for 60 minutes at RT. Each section was then incubated overnight at 4°C with a primary antibody cocktail containing antibodies against a specific histone modification and Keratin14. Secondary antibody incubation was carried out at room temperature for 2 hours in the dark. Tissue sections were washed in 1X PBS and counterstained with 0.3 µM DAPI for 10 minutes. Slides were mounted using Prolong Gold antifade mountant (Life Technologies, USA) and dried for 24 hours. Slides were imaged using the Olympus FV3000 inverted confocal laser scanning microscope and software (Olympus, Japan). Z-stacks were collected for all images and merged using FIJI software. The suprabasal layer was distinguished from the basal layer based on K14 distribution, and 20–30 nuclei from each layer were demarcated per field of view (FOV). Altogether, 10–12 FOVs were collected per group of 5–6 mice, and average fluorescence intensities of 150–360 nuclei were quantified on FIJI and plotted. Additional graphs with colour-coded datasets were plotted to examine the contribution from each animal and the extent of animal to animal variation in each experiment (Fig S2).

For sequential triple staining using at least two primary antibodies raised form the same host, sections were blocked as previously described and incubated with rabbit anti-mouse Ki67 and mouse anti-mouse Keratin14 overnight at 4°C. Secondary antibodies were added for 2 hours at RT. Sections were then blocked with 5% normal rabbit serum and 1% BSA in PBS for 2 hours at RT to saturate the first anti-rabbit secondary antibody. Tissues were washed and incubated with either conjugated rabbit anti-mouse H4K20me1 or unconjugated H4K16ac for 2 hours at 37 degrees. For H4K16ac co-staining, secondary antibody detection was carried out for 2 hours at room temperature. Tissues were counterstained, mounted and imaged as previously described. 9 FOVs were imaged from 3 mice per group and the total number of K14⁺, Ki67⁺, H4K20me1⁺/H4K16ac⁺ cells were quantified using FIJI.

Purification of dorsal basal keratinocytes and RNA extraction

To isolate single cells for sorting, shaved dorsal skin excised from young or aged mice were cut into small pieces and digested in C-tubes (Miltenyi, Germany) containing 3 ml of complete RPMI media supplemented with 0.25 mg/ml Liberase TM (Roche, Switzerland) and 75 units/ml DNase 1 (New England Biolabs, USA) for 1 hour and 45 minutes in a 37°C shaking incubator. Tissues were dissociated twice on a gentleMACS dissociator (Miltenyi, Germany). EDTA was added to the suspension to stop the digestion. Cells were passed through a 100 µm cell strainer and washed twice with 1% BSA/PBS. Cells were then stained at 4°C in the dark for 20 minutes with fluorophore-conjugated anti-mouse antibodies diluted in 1% BSA/PBS. 3 µM DAPI was used for live-dead discrimination. Fluorescence-activated cells sorting (FACS) was carried out using a BD FACSAria II cell sorter (BD Biosciences, USA). Live, single CD45⁻ CD31⁻, TER119⁻, TIE2⁻, EPCAM⁺ cells (30 000 to 100 000 cells per sample) were directly sorted into 350 µl of lysis buffer supplemented with 1% β-Mercaptoethanol (Sigma Aldrich, USA). Lysates were homogenized using QIAshredders (Qiagen, Germany) and RNA was extracted using the RNeasy Plus Micro Kit (Qiagen, Germany) according to the manufacturer’s instructions.

Isolation and culture of mouse tail epidermal cells

To derive epidermal keratinocytes, freshly dissected tail skin tissues were first washed in 3 ml of PBS with 1% pen-strep for 15 minutes and then in CnTprime medium (CellnTec, Switzerland) supplemented with 1% pen-strep for another 30 minutes on a rotating platform. The tail tissues were digested in CnTprime medium supplemented with 1% pen-strep and 4 mg/ml dispase (Gibco, USA) for 8hrs at 4°C to separate the epidermal and dermal layers. The epidermis was further incubated in 1X trypsin-EDTA (Lonza, Switzerland) at 37°C for 1 hour and single cells were scraped into 5 ml of DMEM with 10% FBS (Gibco, USA). Cells were passed through a 70 µm filter, centrifuged and resuspended in KGM-Gold media (Lonza, Switzerland) containing 1% pen-strep. Cells were seeded onto collagen-coated plates at a high density and cultured in KGM-Gold media (supplemented with bovine pituitary extract, human epidermal growth factor, bovine insulin, Gentamicin, hydrocortisone, Epinephrine and Transferrin) at 37°C with 5% CO₂. To extract RNA, all media was removed, and cells were briefly rinsed with 1X PBS. 500 µl of TRIzol reagent (Invitrogen, USA) was directly added to the plate for 5 minutes and lysates were transferred to 1.5 ml tubes. 0.2 ml of chloroform was added to each tube for phase separation. Samples were centrifuged at 12000xg for 15 minutes at 4°C, and the aqueous phase was transferred to a new microcentrifuge tube. An equal volume of 70% ethanol was added to each sample, and total RNA was purified using the RNeasy Mini Kit (Qiagen, Germany) according to the manufacturer’s instructions. Eluted RNA was quantified and assessed for purity using a Nanodrop 8000 spectrophotometer (Thermo Fisher Scientific, USA).

Real-time quantitative polymerase chain reaction (RT-qPCR)

Total RNA harvested from sorted dorsal cells and mouse tail epidermal cells were reverse transcribed into cDNA using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems, USA) according to the manufacturer’s instructions. cDNA samples were diluted 10x and used for RT-qPCR analysis using PowerUp SYBR Green Master Mix (Applied Biosystems, USA). RT-qPCR was performed on a QuantStudio5 Real-Time PCR System (Applied Biosystems, USA). Primer sequences used in the qPCR analyses are provided in Supplemental Table 1.

Statistical analyses

Statistical analyses were performed on GraphPad Prism 8 (GraphPad Software, USA). Two-way ANOVA and the Tukey’s multiple comparison tests were used to compare significant differences between young and aged mice as well as changes between basal and suprabasal cells. The cumulative distribution differences were analysed using the Kolmogorov Smirnov test at a confidence level of 95%.

Antibodies

Primary histone antibodies used were against panH3 (ab1791, Abcam); panH4 (08–858; Millipore); H4K20me1 (ab9052, Abcam); H4K16ac (07–329; Millipore); H3K4me3 (ab8580, Abcam); H3K27me3 (07–449, Millipore); H4K20me3 (C15410207, Diagenode). Other antibodies used were anti-K14 (LL001, gift from Professor Birgitte Lane), anti-Ki67 (MA5-14520, Thermo Scientific) and secondary antibodies chicken anti-rabbit Alexa Fluor-594 and chicken anti-mouse AlexaFluor-488 (A21442; A21200 Invitrogen, USA). Antibodies used for FACS purification were EpCAM (#47-5791-80) and TIE2 (# 12–5987-82) from eBiosciences and CD11B (#101249), CD45 (#103138), PDGRFα (#135922), CD31 (#102408), TER119 (#116223) from Biolegend.

Funding Statement

This work was supported by the Agency for Science, Technology and Research, Singapore under IAF-PP Program H17/01/a0/004. C.T.D. was supported by the A*STAR Research Attachment Programme (ARAP) in collaboration with The University of Manchester, Faculty of Biology, Medicine and Health.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary material

Supplemental data for this article can be accessed here.