Epidermal Stem Cells Are Defined by Global Histone Modifications that Are Altered by Myc-Induced Differentiation

Histone marks in human epidermis

We began by investigating whether stem cells in human interfollicular epidermis were characterised by specific histone modifications. We prepared epidermal whole mounts [9], [20] and labelled them with antibodies specific for histone H3 methylation at lysines 4 (H3diK4) or 9 (H3diK9, H3triK9) and an antibody that detects acetylation of H4 (H4Ac) (Figure 1).

Human interfollicular epidermal stem cells express high levels of ß1 integrins and are arranged in clusters in the epidermal basal layer, surrounded by their progeny, transit amplifying (TA) cells, and cells that have initiated terminal differentiation. Double labelling for β1 integrins and antibodies to modified histones revealed that although levels of H3diK4 or H3di, triK9 methylation varied, there was no correlation with high or low expression of ß1 integrins (Fig. 1A–I). In contrast, cells that expressed high levels of ß1 integrins had low levels of H4 acetylation (Fig. 1J–L). Therefore low levels of H4 acetylation are a marker of human interfollicular epidermal stem cells.

Histone marks in the mouse hair follicle bulge

In mouse epidermis the best characterised pool of stem cells is not in the interfollicular epidermis but in a region of the hair follicle known as the bulge [8]. One characteristic of bulge stem cells is that if they incorporate BrdU in neonatal epidermis they retain the label in adulthood; such cells are known as label retaining cells (LRC) [21], [22]. We therefore examined the state of histone modifications in label retaining cells (LRC) in whole mounts of mouse tail epidermis (Fig. 2A–F) [22]. Antibodies to H3diK4, H3diK9 and H3triK9 gave uniform labelling along the outer root sheath (ORS) of the hair follicle, and most LRC, like the rest of the bulge cells, were strongly labelled (Fig. 2A–C; line; insert). To quantify these results we counted the total number of LRC per bulge and scored each cell as strongly (high) or weakly (low) labelled (Fig. 2D–F, Fig. S1).

Whereas levels of histone methylation appeared to be uniform in bulge cells, we detected heterogeneity amongst cells of the interfollicular epidermis (IFE) (Figure 2G–I). The majority of BrdU positive LRC in wild-type interfollicular epidermis were low in H3diK4 or H3diK9 methylation (Fig. 2 G,H and M,N), but high in H3triK9 (Fig. 2I,O).

We conclude that LRC in the bulge and the IFE represent distinct cell populations when characterised by the level of H3 methylation.

Effect of Myc activation on epidermal histone marks

Epidermal stem cells can be stimulated to exit the stem cell compartment by activation of the transcription factor Myc [23]. In response to Myc, stem cells move into the TA compartment, where they undergo several rounds of division and then differentiate along the lineages of the interfollicular epidermis and sebaceous gland [11], [12], [15], [17]. To investigate the consequences of overexpressing Myc on histone modifications in epidermal cells, we prepared whole mounts of tail epidermis from K14MycER mice that had been treated with 4-hydroxy-tamoxifen (4OHT). The keratin 14 promoter drives transgene expression in stem and transit amplifying cells in all regions of the epidermis. MycER is a chimeric protein in which the C-terminus of Myc is fused to a mutant oestrogen receptor; thus Myc is only activated when cells are treated with 4OHT [11], [12].

As reported previously [22], activation of Myc did not alter the number or location of LRC in the bulge or IFE (Fig. 2D–F, M–O and data not shown). In the bulge, the level of methylation of H3diK4 and H3di, triK9 was similar in 4OHT treated K14MycER mice and wild-type littermates (Fig. 2D–F). However, the IFE differed from the bulge in its response to Myc. We still detected LRC with low levels of H3diK4 and H3diK9 methylation (Fig. 2M,N), but the proportion of cells with high levels of these modifications was increased compared to wild-type control LRC (Fig. 2M,N).

The most dramatic effect of Myc was on H3triK9 methylation of interfollicular epidermal LRC. The majority of LRC in Myc activated epidermis were low in H3triK9 methylation, whereas in controls most of the LRC had high levels (Fig. 2O). This suggests that Myc increased H3diK9 methylation at expense of H3triK9 methylation.

Low levels of H4 acetylation are a mark of label retaining cells in mouse epidermis that is regulated by Myc

Since low H4 acetylation was common in stem cells in human interfollicular epidermis (Fig. 1), we examined whether this was also the case in mouse epidermis (Fig. 3A–D). The entire bulge had much lower levels of histone acetylation than the rest of the hair follicle (Fig. 3A,B; Bg). Double labelling for BrdU and H4Ac demonstrated that LRC in both the bulge and interfollicular epidermis of wild-type mice had low acetylation of H4 (Fig. 3C,D; insert). We conclude that in human and mouse interfollicular epidermis and in the mouse hair follicle bulge, stem cells are characterised by low levels of histone H4 acetylation.

Acetylation of histone H3 and H4 is a chromatin modification known to be mechanistically associated to Myc-induced transactivation. Myc regulates gene transcription by recruiting the histone acetyltransferases GCN5 and TIP60, and by directly up-regulating GCN5 [19], [24], [25]. Therefore it was not surprising that in 4OHT treated K14MycER epidermis there was increased H4 acetylation of bulge LRC (Fig. 3D). Indeed, on activation of Myc the bulge could no longer be distinguished on the basis of low H4 acetylation (Fig. 3A,E; Bg), and acetylation was also increased in the interfollicular epidermis (Fig. 3A,E; IFE).

We next asked whether H4 acetylation was correlated with methylation at lysine 20 (K20). Mono-methylation of histone H4 at lysine 20 (H4monoK20) is increased in the promoter and coding regions of active genes and correlates with histone hyperacetylation [26]. In the epidermis of wild-type mice, levels of H4monoK20 correlated with acetylation of H4, being low in the bulge (Bg) and high in other regions of the epidermis, particularly the hair follicle bulb (Bb) (Fig. 3 B,F).

The increase in H4 acetylation resulting from activation of Myc correlated with increased levels of H4monoK20 in bulge and interfollicular epidermal LRC (Fig. 3G–I). However, the increase was modest and did not occur throughout the whole epidermis (Fig. 3J), leading us to speculate that H4monoK20 might accumulate only transiently and be replaced with another modification, such as H4di- or triK20. Since the number of LRC with high levels of H4triK20 was decreased in 4OHT treated K14MycER epidermis (Fig. 3K,L), we speculated that H4monoK20 might be replaced by H4diK20.

Myc-induced switch from H4monoK20 to H4diK20 depends on HDAC activity

To demonstrate that H4diK20 was up-regulated in vivo, we treated K14MycER and wild-type control mice for 6 days with 4OHT. H4diK20 was dramatically increased in K14MycER mice compared to the controls (Fig. 3N,Q). Whereas nuclei high in H4monoK20 were localised to the basal layer of the IFE in control and transgenic mice (Fig. 3M,P), cells with high levels of H4diK20 were found in the differentiating, keratin 10 positive, suprabasal layers of K14MycER mice (Fig. 3N–R).

The re-organisation of constitutive heterochromatin from H4monoK20 to H4diK20 requires HDAC activity [27]. If Myc replaces H4monoK20 with H4diK20, the process should be blocked with the HDAC inhibitor, trichostatin A (TSA) [28]. When we treated wild-type mouse epidermis with TSA, increased numbers of cells with H4monoK20 were found in the IFE, hair follicles and sebaceous glands (Fig. 4A,B). The effect was even more pronounced when TSA was combined with Myc activation (Fig. 4C,D).

To examine the mechanism by which Myc induced the switch from H4monoK20 to H4diK20, cultured primary human keratinocytes transduced with a MycER retroviral vector or an empty vector control (pBabe) were labelled with antibodies to the different states of H4 methylation (Fig. 4E–P). The number of cells with high levels of H4diK20 increased by approximately 25% on activation of Myc (Fig. 4I,K,R), an effect that was inhibited by TSA (Fig. 4K,L,R). In both pBabe and Myc expressing keratinocytes, H4monoK20 increased on treatment with TSA (Fig. 4M–Q).

Replacement of H4monoK20 with H4diK20 in response to Myc activation should be accompanied by activation of specific histone methyltransferases. Set8 is an H4K20 specific mono-methylase [29], [30], while H4K20 di-methylation is dependent on the histone methyltransferase Ash-1 [31]. Western blotting showed that expression of Set8 was transiently up-regulated within 12 hours of activating Myc in keratinocytes (Fig. 5A; left hand panel). Induction of the RNA methyltransferase Misu (NSun2), a direct down-stream target of Myc [16], served as a positive control for Myc activation in the Western blots (Fig. 5A; left hand panel).

TSA treatment had different effects on control keratinocytes and keratinocytes with activated Myc. In control cells transduced with pBabe, TSA caused a transient upregulation of Set8 (Fig. 5A; right hand panel), which may reflect an initial increase in S phase cells (detected as elevated Misu expression), followed by arrest in G2+M of the cell cycle [16]. In TSA treated cells with activated Myc, Set8 was highly expressed at 12 and 24 h (Fig. 5A; right hand panel). The effect of activation of Myc on Ash-1 was to stimulate nuclear accumulation, a process that was inhibited when the cells were treated with TSA (Fig. 5B).

Inhibition of HDACs has similar effects on the epidermal stem cell compartment to activation of Myc

Our results show that acetylation of H4 and mono-methylation of H4 at lysine 20 are low in epidermal stem cells, and that activation of Myc or treatment with TSA increases both modifications. If the modifications contribute to the mechanism of Myc-induced exit from the epidermal stem cell compartment [11], [23], then TSA treatment of wild-type mouse epidermis should result in a similar phenotype to Myc activation. Indeed we found that both TSA treatment of wild-type epidermis and 4OHT treatment of K14MycER epidermis led to a marked increase in the number of cell layers in the interfollicular epidermis (Fig. 6A–C and E–G; arrows), enlargement of the sebaceous glands (Fig. 6E–G; arrowheads and Q–S) and expansion of the hair follicle bulb (Fig. 6Q,R). TSA treatment, like Myc activation, led to an increase in proliferation, as measured by Ki67 labelling and incorporation of a 1 hour pulse of BrDU into S phase cells (Fig. 6I–K, M–O; arrowheads).

Activation of Myc in the presence of TSA exacerbated the phenotype of K14MycER epidermis (Fig. 6D,H,L,P,T). There was massive accumulation of cornified cells in the IFE (Fig. 6D); in some areas no viable cell layers remained and the epidermis began to detach from the underlying dermis (Fig. 6D; arrow). Both TSA and Myc cause arrest of cultured keratinocytes in G2/M-phase of the cell cycle, prior to initiation of terminal differentiation [16], [23]. Consistent with these findings, the number of proliferative cells was lower in K14MycER epidermis treated with 4OHT and TSA (Fig. 6L,P) than with 4OHT alone (Fig. 6K,O). In addition to the effects on the interfollicular epidermis, the combination of TSA treatment and Myc activation resulted in further increases in the number of differentiated sebocytes (Fig. 6H; arrowhead and 6T) and aberrant hair follicle morphology (Fig. 6T).

We conclude that Myc-induced chromatin modifications play a major role in Myc-induced exit from the stem cell compartment and differentiation into interfollicular epidermis and sebocytes.

Transgenic mice and treatment

K14MycER transgenic mice (2184C.1) [11] and nontransgenic littermates (wild-type controls) received topical applications of 4OHT (Sigma, 1 mg per mouse per day) or trichostatin A (TSA) (Sigma, 1mg per kg per day) to a shaved area of dorsal skin. 4OHT and TSA were dissolved in ethanol. When mice were treated with both 4OHT and TSA, the animals received 4OHT or TSA on alternate days over the treatment period of 6 days. Mouse breeding and experimental protocols were subject to Institutional ethical approval and were performed under the terms of a U.K. Government Home Office licence.

BrdU labelling

To generate label-retaining cells (LRC), 10-day-old mice were injected with 50 mg/kg (body weight) 5-bromo-2′-deoxyuridine (BrdU; 20 µl of 12.5 mg/ml BrdU) every 12 hours for a total of four injections, as described previously [22]. In some experiments mice were injected with BrDU one hour prior to sacrifice, in order to label S phase cells.

Tissue preparation

Frozen sections (5–7 µm) of mouse dorsal and tail skin were fixed for 10 minutes in 4% paraformaldehyde prior to labelling. Wholemounts of mouse tail skin [22] and adult human skin [9] were prepared as described previously. Human skin was obtained as surgical waste from mastectomy operations with appropriate ethical approval.

Immunostaining and image analysis

Immunostaining was performed as described previously [15], [16]. Antibodies used were: H3diK4 (clone RR302; Upstate), H3diK9 (kindly provided by T. Jenuwein), H3triK9 (kindly provided by T. Jenuwein), H4Ac (anti-acetyl-Histone H4; Upstate), ß1 integrin (P5D2) [43], H4monoK20 (ab9051; Abcam), H4diK20 (ab14092, Abcam), H4triK20 (ab9053; Abcam), Set8 (PR/SET07; Abgent), Ash-1 (ab4477; Abcam), tubulin (Sigma), Misu (NSun2) (EF1) [16], keratin 14 (MK14, Babco), keratin 10 (MK10, Covance), Ki67 (Novacastra), BrDU (Becton Dickinson) and Myc (N262, Santa Cruz).

Images were acquired and expression levels were estimated using a Zeiss 510 confocal microscope. Approximately 30 optical sections of each epidermal sheet were captured with an increment of 1 µm. Scans are presented as z-projections, scanned from the dermal side towards the epidermal surface. The number of LRC was counted per hair follicle bulge and per optical field in the interfollicular epidermis (Zeiss 20/NA 0.75).

Human keratinocyte culture, retroviral infection and Western blotting

Primary human keratinocytes were isolated from neonatal foreskin and cultured in the presence of a feeder layer of J2-3T3 cells in FAD medium (1 part Ham's F12 medium, 3 parts Dulbecco's modified Eagle's medium (DMEM), 1.8×10⁻⁴ M adenine) supplemented with 10% fetal calf serum (FCS) and a cocktail of 0.5 µg / ml hydrocortisone, 5 µg/ml insulin, 10⁻¹⁰ M cholera toxin and 10 ng/ml epidermal growth factor (EGF) as described previously [23]. J2-3T3 cells were cultured in DMEM containing 10% donor calf serum.

Keratinocytes were infected with the following retroviral vectors: pBabe puro (empty vector) [44] and pBabeMycER [23]. Keratinocytes were infected by co-culture with retroviral producer cells as described previously and used within one or two passages after infection [23]. Activation of the steroid-inducible constructs was performed by adding 200 nM 4OHT (Sigma) to the culture medium.

Keratinocytes were solubilised in RIPA buffer, resolved by SDS-polyacrylamide gel electrophoresis and subjected to Western blotting as described previously [15].