A tale of terminal differentiation

Abstract

Keratinocyte differentiation is the process of cellular maturation from a mitotic state to a terminally differentiated state during which skin builds up a tough yet soft skin barrier to protect the body. Its irreversibility also allows the shedding of excessive keratinocytes, thereby maintaining skin homeostasis and preventing skin diseases. Although the entire journey of keratinocyte differentiation is intricate and not well understood, it is known that Ras is able to block keratinocyte terminal differentiation and instead induce keratinocyte proliferation and transformation. It appears that uncontrolled proliferation actually interrupts differentiation.

However, it has been unclear whether there are any innate surveillants that would be able to induce terminal differentiation by antagonizing excessive mitotic activities. Inhibitor of nuclear factor κB kinase-α (IKKα, previously known as Chuk) emerges as a master regulator in the coordinative control of keratinocyte differentiation and proliferation and as a major tumor suppressor in human and mouse skin squamous cell carcinomas. IKKα does so largely by integrating into the epidermal growth factor receptor (EGFR)/Ras/extracellular signal-regulated kinase (Erk)/EGFR ligand pathways during mitosis and differentiation. We discuss these findings herein to extend our understanding of how IKKα-mediated terminal differentiation serves as an innate surveillant in skin.

Introduction

Skin is the largest organ in the body and consists of the epidermis and the dermis. Keratinocytes constitute the stratified epidermis. Cells in the basal epidermis are mitotic. After moving to the suprabasal layers, the cells start to gradually differentiate and die when they reach the skin surface. The terminal differentiation process specializes the cells in the formation of the skin barrier. During the journey of keratinocyte maturation, keratinocyte proliferation and differentiation are tightly regulated. Keratinocytes produce different markers at different stages of differentiation.¹ Based on these markers, keratinocytes can be classified as proliferating, intermediately differentiating, and terminally differentiating keratinocytes in vitro and in vivo.

The inhibitor of nuclear factor κB kinase (IKK) complex containing IKKα, IKKβ and IKKγ [NFκB essential modulator (NEMO)] is essential for nuclear factor κB (NFκB) activation.² NFκB is a group of protein-structure-related transcriptional factors that regulate the expression of many genes encoding the proteins involved in inflammation, immunity and apoptosis. Inhibitors of NFκB (IκBs) interact with NFκB in the cytoplasm, thus blocking NFκB nuclear translocation and preventing its transcriptional activation. IKKα and IKKβ phosphorylate IκBs, which leads to IκBs protein degradation and frees NFκB, resulting in NFκB activation.² IKKα and IKKβ were believed to be functionally redundant because they both phosphorylate IκBs. However, genetic studies revealed, surprisingly, that IKKα and IKKβ each have distinct physiological functions in the development of embryonic murine skin.^3–6

IKKα is required for the formation of the embryonic epidermis.^3,4 Loss of IKKα prevents keratinocytes from terminal differentiation and promotes keratinocyte proliferation in vivo and in vitro.^7,8 Reintroduction of IKKα and kinase inactive IKKα rescues these phenotypes in Ikkα^−/− keratinocytes and mice. Furthermore, downregulation, altered localization, mutations and/or loss of heterozygosity (LOH) of Ikkα have been reported in squamous cell carcinoma (SCC) of the skin, lungs, esophagus and head and neck in humans, highlighting the importance of IKKα in human cancers.^9–13 Experiments in animals have shown that reduction in IKKα expression promotes skin carcinogenesis.¹⁴ Moreover, IKKα deletion in keratinocytes causes spontaneous skin SCC in mice.¹⁵ Thus, IKKα is a bona fide tumor suppressor in SCC. Given the biological significance of IKKα in the development of both skin and skin tumors, we focus here on the process of IKKα-mediated keratinocyte terminal differentiation and its role in these critical events.

IKKα, Ras and Calcium in Terminal Keratinocyte Differentiation

In the stratified epidermis, the basal keratinocytes are mitotic and small. Suprabasal keratinocytes terminate proliferation, increase levels of keratins and filaments, and grow large (Fig. 1A).⁷ Finally, terminally differentiated keratinocytes lose nuclei and eventually shed from the skin surface. The same process applies when keratinocytes form colonies in primary cultures, wherein terminally differentiated cells can be 10 times the size of mitotic keratinocytes during detachment from the colony when they eventually die (Fig. 1B).^7,16 Thus, the cellular events in in vitro and in vivo systems appear to share similarities.

Figure 1.

Differentiation Status of Keratinocytes in Murine Skin and in Cultures. (A) Hematoxylin and Eosin (H&E)-stained skin of a 1-day-old mouse. Epi, Epidermis; Derm, Dermis; Differ, Differentiation; SC, suprabasal keratinocytes; BC, basal keratinocytes; HF, hair follicle. The arrow indicates the direction of differentiation. The dash line separates the epidermis and dermis. Scale bar, 30 μm. (B) Morphology of primary keratinocytes in cultures. Terminal differentiated cells are detached. Undifferentiated cells are attached. Scale bar, 30 μm.

Calcium (Ca⁺⁺) is an important cell sensor involved in many cellular events, such as the induction of terminal differentiation in keratinocytes.¹⁷ Thus, in order to culture keratinocytes, the Ca⁺⁺ concentration in the medium must be minimal because it induces terminal differentiation in keratinocytes. Alterations in Ca⁺⁺ levels have been observed in different epidermal layers.¹⁸ The association of the Ca⁺⁺ gradient and keratinocyte differentiation suggests that Ca⁺⁺ may play a role in inducing and maintaining the terminal differentiation status in the epidermis.¹⁹

In theory, intermediately differentiating keratinocytes do not proliferate. Interestingly, oncogenic H-Ras is able to block Ca⁺⁺-mediated terminal differentiation and induce proliferation and transformation in keratinocytes.²⁰ Overexpression of Ras in the suprabasal epidermis, where such partially differentiated cells are located, has been reported to induce cell proliferation and skin tumors in mice.^21,22 Thus, excessive mitotic activity can redirect the differentiation program toward proliferation.

Ca⁺⁺ does not induce terminal differentiation in primary cultured Ikkα^−/− keratinocytes.⁷ Instead, keratinocytes in media supplemented with the elevated concentrations of Ca⁺⁺ formed even larger colonies (Fig. 2A). IKKα loss over-activates the autocrine loop of EGFR/Ras/Erk/EGFR ligands/ligand activators.¹⁵ A dominant-negative form of Ras (RasN17), re-expressed IKKα as well as either EGFR or Erk inhibitors, are able to induce terminal differentiation and repress hyperproliferation in Ikkα^−/− keratinocytes.¹⁵ Thus, IKKα regulates the cellular events utilizing the EGFR/Ras/Erk signaling cascade. Our findings further revealed that IKKα represses the cascade activity by downregulating expression of EGFR ligands and their ligand activators at the transcription level.¹⁵ Dlugosz et al.²³ also reported that inactivation of EGFR elevated differentiation in keratinocytes overexpressing Ras. Thus, IKKα services as an innate surveillant that monitors the intrinsic EGFR/Ras cascade in regulating keratinocyte proliferation and differentiation. However, the mechanism by which Ca⁺⁺ triggers the entire process of keratinocyte terminal differentiation remains to be determined.

Figure 2.

IKKα Null Keratinocytes are Resistant to Induction of Terminal Differentiation. (A) Keratinocyte colonies in cultures for 16 days. Cell colonies were stained with 0.5% crystal violet solution. WT, wild-type keratinocytes; control, keratinocytes cultured in serum-free keratinocyte media (10785; GIBCO), −/−, Ikkα^−/− keratinocytes; serum, keratinocytes cultured in medium supplemented with 5% bovine serum (in which the calcium concentration is higher than that in the serum-free keratinocyte media); Ca⁺⁺, keratinocytes cultured in medium supplemented with 0.5 mM Calcium. Keratinocyte culture conditions were described previously.⁷ (B) Detection of filaggrin in keratinocytes infected by adenoviruses expressing p21 (p21^CIP), p27 (p27^kip1) and 14-3-3σ or treated with Ca⁺⁺ (0.5 mM) by western blotting. β-Actin used as a protein loading control.

Cell Cycle Arrest is Not Enough to Induce Terminal Differentiation in Keratinocytes Lacking IKKα

Activated Ras increases the S phase of the cell cycle in the epidermal keratinocytes.²³ Increased bromodeoxyuridine (BrdU) signals, an S-phase marker, were also detected in the epidermis of Ikkα^−/− mice and mice with IKKα deletion in keratinocytes (Ikkα^F/F/K5.Cre).^3,15 Primary cultured IKKα-null keratinocytes showed elevated S phase and reduced G₂/M phase.²⁴ Thus, increased cell cycle progression may block the pathway to terminal differentiation in keratinocytes.

14-3-3σ is a G₂/M phase cell cycle checkpoint regulatory molecule that is highly expressed in response to DNA damage.²⁵ Downregulation of 14-3-3σ expression was found to be associated with hypermethylation of 14-3-3σ CpG islands in many human SCCs and breast cancers.^26,27 As in Ikkα^−/− mice, repeated epilation (Er) mice with a mutation in the 14-3-3σ gene lack terminally differentiated keratinocytes in the epidermis.^28,29 IKKα was found to prevent the CpG methylation of 14-3-3σ by blocking K9-H3 trimethylation, a repressive transcription mechanism.^24,30 We found that 14-3-3σ was downregulated in Ikkα^−/− keratinocytes and that loss of IKKα and downregulation of 14-3-3σ caused decreases in the G₂/M phase in keratinocytes.²⁴ Importantly, expression of 14-3-3σ was downregulated in chemical carcinogen-induced skin carcinomas lacking IKKα in mice.³¹ Thus, cell cycle regulation involving IKKα/14-3-3σ may be critical for tumorigenesis. Furthermore, Er mice express normal levels of IKKα and still display a skin phenotype similar to mice in which IKKα is deleted.²⁹ It remains to be tested whether 14-3-3σ is a downstream target of IKKα in embryonic development.

It has been thought that differentiation requires the withdrawal of keratinocytes from the cell cycle.³² Yet we found that overexpression of cell cycle checkpoints or cyclin-dependent kinase inhibitors p21^CIP1, 14-3-3σ, p27^kip1, alone or in combinations with Ca⁺⁺ did not induce terminal differentiation in primary cultured Ikkα^−/− keratinocytes (Fig. 2B), although the inhibitors did stop cell proliferation (data not shown). Thus, cell cycle arrest may not be enough to initiate keratinocyte terminal differentiation on its own.

Furthermore, it is known that the terminally differentiating keratinocytes express increased levels of keratins and filaments and change their morphology dramatically, completing the journey of keratinocyte maturation. Many molecular and cellular events are involved in this process. We found that there are significant differences in the expression of many genes between wild-type and Ikkα^−/− keratinocytes. For example, the expression levels of claudins that participate in the skin barrier formation were significantly reduced in Ikkα^−/− compared with wild-type keratinocytes.¹⁵ In addition, although we found that inactivation of EGFR, Ras and/or Erk was able to induce Ikkα^−/− keratinocytes to terminal differentiation, the efficiency of the inhibitors in inducing terminal differentiation was lower than that of re-expression of IKKα. Taken together, these results suggest that the terminal differentiation process requires two steps: termination of cell proliferation and initiation of maturation. Because IKKα is fully functional for inducing keratinocyte terminal differentiation, it may have many targets present at various stages throughout these events.

In Vivo Evidence: Impact of IKKα-Mediated Differentiation on Skin Homeostasis and Skin Tumor Development

Mice overexpressing the active form of heparin-binding (HB)-EGF or H-Ras develop epidermal hyperplasia and die after birth, as do Ikkα^F/F/K5.Cre mice with IKKα deleted in keratinocytes.^15,33,34 In contrast, reintroduction of IKKα induces terminal differentiation and represses uncontrolled proliferation in the epidermis of mice and rescues the IKKα mutant mice.¹⁵ The epidermis of newborn mice contains 4 to 6 cell layers, which reduces to 2 to 3 layers in a few days. EGFR levels are reported to be higher in the thick epidermis of newborns than in the thin epidermis of older mice.³⁵ We found that most IKKα localized in the cytoplasm of keratinocytes in mouse embryos and newborns, however, the IKKα was found in the nuclei of keratinocytes in older mice (data not shown), where IKKα has a repressive transcriptional activity for the genes encoding growth factors and growth factor activators by binding to their promoters. Mitogenic signals were found to reduce the binding of IKKα to these promoters, elevating cell proliferation and reducing terminal differentiation.¹⁵ Thus, IKKα is physically and functionally associated with alterations in the epidermal cell proliferation and differentiation.

Treatment with 12-O-tetradecanoylphorbol-13-acetate (TPA), a tumor promoter, induces proliferation as well as terminal differentiation in the skin of mice.⁹ TPA-induced mitotic activity is lower but the levels of the TPA-induced terminal differentiation marker filaggrin is higher in the epidermis of the transgenic mice overexpressing IKKα than in the epidermis of wild-type mice.⁹ TPA elevates the expression of IKKα in the skin of mice,³⁶ and the IKKα transgenic mice express elevated levels of filaggrin as well.⁹ Moreover, the IKKα transgenic mice develop normally, yet develop fewer malignant carcinomas and metastases than do wild-type mice after exposure to chemical carcinogens. Therefore, IKKα-mediated terminal differentiation may allow excessive cells to shed out, maintaining skin homeostasis and inhibiting skin tumor development.

Noticeably, excessive mitotic activity in the epidermis not only causes epidermal hyperplasia, but also generates “hard” and “scar-type” skin in mice. The hard/scarred skin contains many terminally differentiated keratinocytes in the process of dying and would therefore express elevated levels of terminal differentiation markers. For example, transgenic mice overexpressing H-Ras show epidermal hyperplasia with hard/scarred skin, elevated keratin products and/or developed skin tumors.^21,33 Thus, elevated levels of terminal differentiation markers in the hard/scarred skin may not reflect a real consequence of keratinocyte proliferation and differentiation.

Transformation of IKKα Deficient Keratinocytes

Oncogenic Ras induces keratinocyte transformation.²⁰ Mutations in the H-Ras gene and the elevated H-Ras cascade signaling have been reported in human skin SCC.^37,38 Chemical carcinogens targeting the H-Ras gene induce skin papillomas and SCC in mice.³⁹ We reported that chemical carcinogens induce 10 times more carcinomas and two times more papillomas in Ikkα^+/− mice than in Ikkα^+/+ mice.¹⁴ These tumors carried the carcinogen-induced Ras mutations and LOH of Ikkα, a classical tumor suppressor feature, was found in most carcinomas from Ikkα^+/− mice. These results suggest that IKKα loss cooperates with Ras to promote malignant keratinocyte transformation. Furthermore, we found that IKKα-null keratinocytes were more easily transformed in cultures in which wild-type keratinocytes underwent spontaneous terminal differentiation (Fig. 3A). These transformed keratinocytes formed colonies in soft agar and tumors in nude mice (Fig. 3B and C). Thus, keratinocytes unable to achieve terminal differentiation may be more susceptible to transformation, although other intrinsic deficiencies also contribute to the transformation of IKKα null keratinocytes.³¹

Figure 3.

Transformation of IKKα null keratinocytes in vitro and in vivo. (A) Ikkα^−/− keratinocytes were transformed in cultures for 3 months. Scale bar, 60 μm. (B) Transformed Ikkα^−/− keratinocytes form colonies in soft agar medium for 2 weeks. Untransformed keratinocytes did not form colonies in soft agar medium (data not shown). Soft agar culture was described previously.⁴¹ Scale bar, 60 μm. (C) Transformed Ikkα^−/− keratinocytes (1×10⁶ cells/site) developed tumors in nude mice (3 months) but untransformed Ikkα^−/− keratinocytes did not form tumors (data not shown). H&E-stained sections showed tumor histology. Red arrowheads show mitotic cells in the tumor. Scale bar, 60 μm.

Although the intrinsic defects described above are necessary for tumorigenesis in mice with IKKα deletion in keratinocytes, we found that mice with IKKα deletion induced by K15.Cre in hair follicle keratinocytes developed skin tumors earlier and developed more of them than did mice with IKKα deletion in keratinocytes induced by K5.Cre or MMTV.Cre.¹⁵ We also noticed that more tumors developed on the upper part of the body in the mutant mice. These results suggest that different subgroups of keratinocytes may have different susceptibilities to IKKα deletion-related tumorigenesis or that the microenvironment plays a critical role in regulating tumorigenesis.

Previously, we showed that wild-type keratinocytes secreted keratinocyte differentiation inducible factor (kDIF), which induced terminal differentiation in Ikkα^−/− keratinocytes and mice.⁷ IKKα also regulates kDIF. Descargues et al. reported that IKKα was involved in Smad-mediated keratinocyte differentiation, which was related to kDIF.¹¹ Keratinocytes expressing K15 are further from the suprabasal epidermis than are keratinocytes expressing K5. Thus, the differentiating keratinoytes in the suprabasal epidermis may secrete kDIF into the skin microenvironment, which influences more keratinocytes in the suprabasal epidermis than keratinocytes in the hair follicles. The differentiation-inducible microenvironment may also be important for antagonizing hyperproliferation and reducing the potential susceptibility of keratinocytes to proliferation and transformation in vivo. Moreover, IKKα loss was reported to deregulate multiple important molecules, such as VEGF-A, p63 and Smad,^9,11,13,40 which may contribute to the skin tumor development in vivo as well.

In conclusion, IKKα and IKKα-mediated terminal differentiation act as an innate surveillant in maintaining skin homeostasis and preventing skin cancer through inducing intrinsic and extrinsic mechanisms by utilizing its many targets (Fig. 4), although the further mechanisms remain to be revealed. Thus, IKKα may be a potential therapeutic target, which can be used to prevent skin diseases.

Figure 4.

Scheme of IKKα activity in maintaining skin homeostasis and preventing skin tumors. Arrows indicate that IKKα exerts its functions at different hierarchies.

Abbreviations:

IKKα

inhibitor of nuclear factor κB kinase-α

NFκB

nuclear factor κB

EGFR

epidermal growth factor receptor

Erk

extracellular signal-regulated kinase

TPA

12-O-tetradecanoylphorbol-13-acetate

BrdU

bromodeoxyuridine