Mouse Genetic Models Reveal Surprising Functions of IκB Kinase Alpha in Skin Development and Skin Carcinogenesis

Gene knockout studies unexpectedly reveal a pivotal role for IκB kinase alpha (IKKα) in mouse embryonic skin development. Skin carcinogenesis experiments show that Ikkα heterozygous mice are highly susceptible to chemical carcinogen or ultraviolet B light (UVB) induced benign and malignant skin tumors in comparison to wild-type mice. IKKα deletion mediated by keratin 5 (K5).Cre or K15.Cre in keratinocytes induces epidermal hyperplasia and spontaneous skin squamous cell carcinomas (SCCs) in Ikkα floxed mice. On the other hand, transgenic mice overexpressing IKKα in the epidermis, under the control of a truncated loricrin promoter or K5 promoter, develop normal skin and show no defects in the formation of the epidermis and other epithelial organs, and the transgenic IKKα represses chemical carcinogen or UVB induced skin carcinogenesis. Moreover, IKKα deletion mediated by a mutation, which generates a stop codon in the Ikkα gene, has been reported in a human autosomal recessive lethal syndrome. Downregulated IKKα and Ikkα mutations and deletions are found in human skin SCCs. The collective evidence not only highlights the importance of IKKα in skin development, maintaining skin homeostasis, and preventing skin carcinogenesis, but also demonstrates that mouse models are extremely valuable tools for revealing the mechanisms underlying these biological events, leading our studies from bench side to bedside.


Introduction
IB kinase alpha (IKK), also known as CHUK, is an 85-kD polypeptide with 745 amino acids (aa). It consists of a putative kinase catalytic domain (KD, aa 15-300) containing 12 regions of homology that are characteristic of a protein serine kinase, a leucine zipper (LZ) motif, and a helix-loop-helix (HLH) motif, and it belongs to a classic zipper protein family, which includes c-myc, Id1, C/EBP, and Jun [1,2]. IKK can form homodimers and heterodimers, which lead to distinct pathways in a cell-specific manner in vivo [3,4]. For example, IKK, IKK, and IKK (NEMO) form an IKK complex, which is essential for activating the canonical NF-B pathway. IKK and IKK are two highly conserved protein kinases, and IKK is a regulatory subunit. On the other hand, IKK homodimers lead to noncanonical NF-B signaling. Unlike IKK and IKK, mouse genetic studies unexpectedly disclose that IKK deletion impairs embryonic skin development, and that IKK homodimers show transcription activity in regulating the expression of many genes encoding proteins that are important for skin and skin tumor development [5][6][7][8]. In this review, we discuss how different mouse models illustrate the physiological functions of IKK in the skin.

Why Is It Surprising that IKK Has a Role in Embryonic Skin Development?
IKK is identified as the first component of the IKK complex, which activates NF-B, following tumor necrosis factor (TNF), interleukin 1 (IL-1), or lipopolysaccharide (LPS) stimulation, through phosphorylating IB, an NF-B inhibitor, thereby inducing IB ubiquitination and degradation in cell lines [4,9,10]. NF-B has an anti-apoptotic activity via the TNF receptor (TNFR) death pathway [4]. Mice lacking p65 (RelA, a major NF-B subunit in the canonical NF-B pathway), IKK, or IKK die at days 15.5, 13.5, and 12.5, respectively, during embryonic development due to liver cell apoptosisinduced hemorrhage [11][12][13]. Depleting TNFR rescues p65 −/− , Ikk −/− , and Ikk −/− mice [14][15][16][17][18][19], indicating that p65, IKK, and IKK antagonize apoptosis downstream of the TNFR-mediated death pathway. In addition, p65 −/− /Tnfr −/− , Ikk −/− /Tnfr −/− , and Ikk −/− /Tnfr −/− newborn mice develop relatively normal skin. No severe skin lesions have been reported in mice lacking one of the other NF-B genes [4]. Most NF-B family members function in the development of lymphoid cells and lymphoid organs [4]. Because IKK has stronger kinase activity than IKK, and IKK and IKK share similar protein motifs and can phosphorylate the same proteins [20], it was speculated that IKK and IKK were functionally redundant in vivo and that IKK knockout might not have phenotypes due to the presence of IKK and IKK.
Surprisingly, Ikk −/− newborn mice exhibit the appearance of a pupa with shiny, wrinkleless skin, and these newborns die soon after birth because of quick loss of body water [15,21,22]. Histological examination and immunostaining show that the Ikk −/− epidermis is markedly thick compared to wild-type (WT), and that the entire Ikk −/− epidermis expresses the basal epidermal cell markers keratin 5 (K5) and K14 and lacks the determinably differentiated cell layers in the epidermis that blocks body water loss. Because of no cornfield layers formed on top of the Ikk −/− skin, the epidermis between the limbs and body begins fuse together after day 13 during Ikk −/− mouse embryonic development. At birth, the invisible limbs are developing beneath the inner skin of Ikk −/− mice [21]. These phenotypes demonstrate that IKK is required for epidermis formation during embryonic development. In the skin, the basal keratinocytes are able to proliferate [23]. Once keratinocytes move to the superficial layer, they stop proliferating and start to differentiate. Those normal differentiating keratinocytes are no longer able to express basal keratinocyte markers, such as K5 and K14. IKK loss cannot turn off the expression of K5 and K14, and K5 and K14 are co-expressed with differentiating keratinocyte markers K1, K10, and involucrin in the superficial epidermal layers of Ikk −/− skin. It is still unknown how IKK loss interrupts the expression patterns of these distinct keratins during the formation of the epidermis. Importantly, NF-B activity is not reduced in Ikk −/− keratinocytes and depleting TNFR does not alter the skin phenotypes in Ikk −/− /Tnfr −/− mice [15,21,24]. Thus, it is clear that Ikk −/− skin phenotypes are not directly caused by a defect in the activation of IKK, IKK, p65, and TNFR pathway.
Interestingly, interferon regulator factor 6 (IRF6) knockout and repeated epilation (Er/Er) newborn mice display a skin phenotype and appearance similar to that of Ikk −/− newborn mice [25][26][27]. IRF6 is a member of the IRF transcription factor family (IRF1-IRF9). Mutations in Irf6 have been reported to cause Van der Woude and popliteal pterygium syndromes, which result in the facial development of cleft lips and palates in humans [28]. IKK and IRF6 knockout mice display similar phenotypes in facial and tooth development [29]. In Er/Er mice, a mutation is found in 14-3-3, which generates an unstable and truncated 14-3-3 protein [26,27]. 14-3-3 functions as a G2/M cell-cycle checkpoint in response to DNA damage [30,31]. Although Ikk −/− , Irf6 −/− , and Er/Er newborn mice exhibit shiny and thick skin with a similar appearance, IKK expression levels are comparable in Irf6 −/− , Er/Er, and WT mice, suggesting that either IKK, IRF6, and 14-3-3 lead to different pathways but crosstalk, or that IKK is an upstream regulator of IRF6 and 14-3-3. We have reported that IKK upregulates 14-3-3 expression at the transcriptional level in an epigenetic manner [31]. Loss of IKK does not completely block 14-3-3 expression and other molecules also regulate the expression of 14-3-3 [32]. Interestingly, our study found that when using the same culture conditions, Ikk −/− and Er/Er keratinocytes do not undergo terminal differentiation, but Ikk −/− keratinocytes can grow into large colonies compared to Er/Er keratinocytes [33], suggesting that IKK and 14-3-3 may lead to different signaling pathways in regulating keratinocyte growth.
It has been speculated for a long time that IKK and p63 crosstalk because Ikk −/− and p63 −/− mice show opposite skin phenotypes [21,34]. p63 −/− mice die at birth and have an extremely thin epidermis compared to WT mice. Several studies have identified IKK as a downstream target of p63 in epidermis formation [35][36][37]. Furthermore, IKK has been shown to regulate the expression of Mad1, Mad2, and Ovol1 in keratinocyte proliferation and differentiation [6,7,38]. Because the defect in the embryonic skin development of Ikk −/− mice has not been rescued by any of those genes described above, the genetic pathways IKK-mediated in regulating embryonic development remain to be disclosed.
Importantly, IKK deficiency is associated with a lethal inherited human disease (called cocoon syndrome) [39]. The human fetuses at 13 to 14 gestational weeks display multiple severe malformations in the cranial and craniofacial area and their appearance is similar to Ikk −/− knockout embryos [21,40]. The defects in multiple organs including the kidney, lungs, skeletal muscles, diaphragm, and limbs and increased vascularization in the paws and omphalocele were also observed.
These lethal fetuses remain normal karyotypes, but contain a mutation from CT at the nucleotide 1,264 in the Ikk gene, which generates a stop codon at position 422 of IKK protein, leading to a truncated polypeptide of 421 amino acids of IKK. Previously, we demonstrated that the truncated IKK proteins were unstable and had no functions in phosphorylating proteins and inducing keratinocyte differentiation [24]. Because these human fetuses die before the keratinization of the epidermis, no defects in the skin development are detected. Consistently, Ikk −/− mouse embryos display epidermal hyperplasia after E16 to E17 days. These results indicate that IKK is critical for human development.

Loss and Gain of IKK Function in Skin Carcinogenesis
Chemical carcinogen and ultraviolet B (UVB) induced skin carcinogenesis approaches have been widely used to study the mechanisms of skin carcinogenesis, to reveal the unknown activity of molecules, and to identify therapeutic reagents to treat and prevent skin tumors [41][42][43][44][45]. These two approaches target different pathways. For example, the chemical carcinogen 7,12-dimethylbenz[a]anthracene (DMBA) induces activating mutations in H-ras, and the tumor promoter 12-O-tetradecanoylphorbol-13-acetate (TPA), an inflammation irritator, expands H-Ras-targeted cells and elicits skin inflammation. On the other hand, UVB causes DNA damage and p53 mutations in the skin. Both skin carcinogenesis models require inflammation to promote skin tumor development [41,46].

Chemical Carcinogen-Induced Skin Carcinogenesis
Because Ikk −/− mice die soon after birth, we studied the activity of IKK in skin tumor development in Ikk +/− mice on a C57/BL6 background using DMBA and TPA treatment [41]. Ikk +/− mice develop two times more benign skin papillomas and 11 times more malignant carcinomas, resembling human squamous cell carcinomas (SCCs), compared to WT mice. The tumor latency is shorter and the tumor sizes are larger in Ikk +/− than in WT mice. Importantly, the WT Ikk allele is lost in half of the Ikk +/− papillomas and in almost all of the Ikk +/− carcinomas. We also observed undifferentiated features in skin carcinomas lacking IKK compared to carcinomas having IKK. These results demonstrate that Ikk is haploid insufficient for suppressing tumor progression, that somatic IKK deletion can facilitate skin carcinoma development, and that the tumor progression may be related to a de-differentiation program in keratinocytes. Furthermore, we found a similar rate of H-ras mutations in both Ikk +/− and WT skin tumors, while Ikk +/− mice developed more skin tumors than WT mice, suggesting that, at the tumor initiation stage, IKK reduction promotes H-Ras-targeted cell expansion following TPA treatment, which accelerates skin tumor formation. In addition, we also identified mutations in the Ikk gene in WT and Ikk +/− skin papillomas and carcinomas. Ikk mutations in Ikk +/− papillomas may trigger Ikk WT allele loss, thus promoting carcinoma development (tumor progression) [41]. These results also suggest that Ikk is a target of mutagenesis during skin carcinogenesis. Following TPA treatment, we observed the increased expression of multiple growth factors, cytokines, and vascular factors, and increased extracellular signal-regulated kinase (ERK) activity in Ikk +/− skin compared to WT skin [41], all of which further support that IKK reduction amplifies the Ras and mitogenic pathways and elevates micro-blood vessel formation. Collectively, these results indicate that IKK is a tumor suppressor of skin SCCs.
On the other hand, IKK overexpression in the epidermis represses tumor progression and metastases in Lori.IKK transgenic mice on a FVB background compared to WT mice in DMBA/TPA-and DMBA/DMBA-induced skin carcinogenesis settings [47]. Transgenic IKK inhibits DMBA/TPA-induced mitogenic and angiogenic activities. The study on gain of IKK function further supports that IKK acts as a tumor suppressor in skin carcinoma development [47].

UVB-Induced Skin Carcinogenesis
UVB is an etiological cause of human skin cancer, and the tumor suppressor p53 is one of its targets [44,48]. UVB induces p53 mutations in the skin at the early stage of skin carcinogenesis and causes skin carcinomas, resembling human SCCs, in hairless mice [49,50]. On the other hand, UVB-induced DNA damage mediates cell death, which serves as a defense mechanism to clear the cells with DNA damage, thereby preventing the expansion of cells carrying DNA mutations [45]. The development of skin tumors requires repeated treatments of UVB. Increased infiltrating inflammatory cells in the skin following UVB treatment are a common phenomenon, and inflammation is pivotal for UVB skin carcinogenesis. UVB treatment induces more p53 mutations and many more skin carcinomas, and shorter tumor latency in Ikk +/− than in WT hairless mice [49]. Cell death numbers are significantly reduced in UVB-treated Ikk +/− skin compared to UVB-treated WT skin. Furthermore, we observed increased cytokines, including TNF, IL-1, IL-6, and monocyte chemoattractant protein-1 (MCP-1), and increased macrophages in the skin of UVB-treated Ikk +/− mice compared to WT mice. Therefore, IKK reduction promotes DNA damage and inflammation but decreases apoptosis. NF-B activity is higher in Ikk +/− than in WT keratinocytes following TNF treatment, which may contribute into increased inflammation and reduced apoptosis. Likely, this increased NF-B is a result of increased IKK activity as IKK replacing IKK in the IKK complex has stronger kinase activity [24]. Although increased p53 mutations and NF-B activity may contribute to decreased cell death following UVB treatment in Ikk +/− skin compared to WT skin, there are likely additional mechanisms to be revealed.
On the other hand, K5.IKK transgenic mice develop normal skin, do not show any histological abnormalities, and rescue the skin phenotype of Ikk −/− mice in an IKK dose-dependent manner [38]. The transgenic IKK has been shown to repress IKK loss-induced the activities of epidermal growth factor receptor (EGFR), ERK, c-Jun N-terminal kinase (JNK), activator protein 1 (AP-1), and signal transducer and activator of transcription 3 (Stat3), and to elevate the expression of Ovol1 and Mad1, which are c-myc antagonists, in the skin of Ikk −/− /K5.IKK mice [38]. Furthermore, the K5.IKK transgene significantly inhibits UVB-induced skin tumor development in K5.IKK mice compared to WT mice (see Figure 1). Lori.IKK also inhibits UVB-induced skin carcinogenesis (data not shown). Together, these results demonstrate that IKK is important for suppressing UVB-induced skin carcinogenesis. Both chemical and UVB skin carcinogenesis models demonstrate that increased IKK has an inhibitory effect on skin carcinogenesis; thus, IKK may be considered a target for preventing skin tumor development.
Although newborn mice lacking IKK, 14-3-3, or IRF6 display similar skin appearance, there is no direct evidence that these molecules function within the same genetic pathway. Interestingly, Er/+ mice are more susceptible to DMBA/TPA-induced skin carcinogenesis than WT mice [51]. Er/+/p63 +/− mice developed fewer skin tumors than Er/+ mice, although the tumor incidence in p63 +/− and WT mice was similar, suggesting that reduced p63 antagonizes 14-3-3 reduction-promoted skin tumor development. In addition, IRF6 is downregulated in human skin SCCs, and has a tumor suppressive function [52,53]. These findings suggest that the developmental-phenotypic-related genes of IKK, 14-3-3, and IRF6 are tumor suppressive. Whether the deletion of 14-3-3 or IRF6 also induces spontaneous skin SCCs in a manner similar to IKK remains to be determined [5].

Figure 1.
Tumor multiplicity in UVB-treated wild-type (WT) and K5.IKK transgenic mice. Each group contains 28 mice in SKH-1 background. Western blot shows IKK levels in the skin of Ikk −/− , WT, and K5.IKK mice. NS, a non-specific band; HA, hemagglutinin-A tag. UVB treatment was described as in reference [49].  [15]. Thus, IKK and IKK are not required for mouse embryonic skin development. It is known that IKK, IKK, and IKK form the IKK complex, but how can we dissect their physiological functions? Conditional gene deletion in specific cell types in mice may help to address this issue. K14.Cre-mediated deletion of IKK or IKK specifically in keratinocytes induces epidermal hyperplasia and skin inflammation, and these Ikk f/f /K14.Cre and Ikk f/f /K14.Cre mice die within three weeks after birth [54,55]. Again, TNFR depletion, but not T-cell depletion, rescues skin phenotypes in Ikk f/f /K14.Cre/Tnfr −/− and Ikk f/f /K14.Cre/Tnfr −/− mice. Inducible K14.CreER-mediated IKK, or IKK deletion in keratinocytes also causes epidermal hyperplasia and skin inflammation, but not skin tumors, and TNFR deletion, indeed, rescues the skin phenotypes in these mice. Collectively, these results suggest that the IKK and IKK deletion-mediated skin phenotypes are associated with a TNFR-mediated pathway. While it is known that Tnfr −/− mice develop fewer skin tumors than WT mice, following DMBA and TPA treatment [56], the effect of IKK or IKK deletion on skin tumor development remains unknown.

Keratinocyte-Specific IKK Deletion Induces Epidermal Hyperplasia and Spontaneous Skin SCCs; EGFR Reduction, but not TNFR Loss, Inhibits IKK Loss-Mediated Epidermal Hyperplasia
We have shown that paternal K5.Cre-or K14.Cre-mediated IKK deletion in keratinocytes induces epidermal hyperplasia; Ikk f/f /K5.Cre and Ikk f/f /K14.Cre mice die within three weeks after birth; and maternal K5.Cre or K14.Cre-mediated IKK deletion induces the same phenotype as that of Ikk −/− mice because K5.Cre and K14.Cre can be activated in oocytes [5,57,58]. Interestingly, TNFR deletion does not rescue these skin phenotypes, but instead slightly enhances the skin phenotypes in Ikk f/f /K5.Cre (paternal Cre) mice. In another study [59], Gareus et al. used K14.Cre to delete IKK in Ikk f/f mice. Since they did not discuss whether they used paternal or maternal K14.Cre to delete IKK in the mutant mice, when IKK was specifically deleted in the epidermis during embryonic development, and whether IKK was indeed expressed in the dermis and other multiple organs, it is premature to evaluate those results before obtaining the additional information. Furthermore, inducible IKK deletion in keratinocytes induces spontaneous skin carcinomas in Ikk f/f /K5.CreER and Ikk f/f /K5.CrePR mice. Cultured IKK-null keratinocytes obtained from both Ikk f/f /K5.Cre and Ikk −/− mice exhibit similar undifferentiated and hyperproliferative features [5,24]. Many pathways, including EGFR/H-Ras/ERK signaling, are altered in IKK-null keratinocytes compared with WT keratinocytes. Inhibiting EGFR, H-Ras, or ERK activity represses keratinocyte proliferation and induces terminal differentiation in IKK-null keratinocytes in culture. Importantly, using genetic and pharmacological approaches, inhibiting EGFR represses epidermal hyperplasia and prevents spontaneous skin tumors in Ikk f/f /K5.CreER mice [5]. In addition, reintroducing IKK inhibits the expression of genes that encode growth factors and growth factor activators, which serve as EGFR ligands, at the gene transcription levels. Thus, IKK loss elevates the EGFR activity, enhancing EGFR/H-Ras/ERK activity and the crosstalk between the EGFR pathway and IKK occurs in the nucleus of keratinocytes [5]. This conclusion may also explain why TNFR depletion does not affect the skin condition of mice lacking IKK in keratinocytes.
Furthermore, we have shown that IKK loss-mediated EGFR activity forms a loop with increased Stat3 and AP-1 activities in the skin [38]. Reducing EGFR or reintroducing IKK can inhibit this autocrine loop, as well as epidermal hyperplasia and skin inflammation including increased infiltrating macrophages and cytokines in Ikk f/f /K5.Cre mice. Because TNFR depletion does not rescue the skin phenotypes in Ikk f/f /K5.Cre/Tnfr −/− mice [5], skin inflammation may not be associated with the TNFR pathway. Increased Stat3 or AP-1 transcription factors also lead to major inflammation pathways [60,61], and, thus, may provide the induction signaling in skin inflammation in Ikk f/f /K5.Cre mice. These targets may be considered therapeutic ones for treating and preventing human skin diseases.

IKK Regulates the Cell Cycle in Keratinocytes
The cell cycle regulates keratinocyte proliferation and differentiation. Increased BrdU incorporation is observed in the Ikk −/− epidermis compared to WT, but overexpression of p21 and p27 cell-cycle inhibitors is not able to drive Ikk −/− keratinocytes to terminal differentiation [21,33], suggesting that the S phase may be an important point that regulates IKK-null keratinocyte proliferation and differentiation. To support this point, we and others have shown that reintroduced domain-negative H-Ras-N17 represses proliferation, induces terminal differentiation, and inhibits EGFR, ERK, and Ras activities in Ikk −/− keratinocytes; increased levels of c-myc antagonists Ovol1 and Mad1 also reverse the status of proliferation and differentiation in Ikk −/− keratinocytes through the transforming growth factor beta (TGF) and Smad pathway [5][6][7]. TGF regulates the c-myc/Ovol1/Mad1 pathway [6]. Thus, both EGFR/H-Ras/ERK and TGF/c-myc pathways are crucial for keratinocyte proliferation and differentiation.
Taken together, we summarize the molecular pathways associated with IKK physiological activities in the development of the skin and skin tumors in Figure 2. The most results were obtained from our published studies from 1999 to now. Thus, this review may not include all the published data and we apologize for this.

Conclusions and Puzzles
It has been puzzling that IKK and IKK display distinct physiological functions although the two protein kinases share similar protein motifs and can phosphorylate the same proteins. For example, K5.IKK transgenic mice develop skin inflammation and epidermal hyperplasia [62], but K5.IKK transgenic mice show no pathological changes in the skin and other organs [38,63]. Depleting TNFR rescues the skin phenotype of Ikk −/− and Ikk f/f /K14.Cre mice, but does not rescue the skin phenotype of Ikk −/− and Ikk f/f /K5.Cre mice. It is known that IKK regulates classical and non-classical NF-B pathways by forming the IKK complex and IKK homodimers [4]. The activation of these pathways largely depends on cell receptors. Noncanonical NF-B is specifically activated in B cells and some stromal cells [64]. We recently showed that in B cells, IKK inactivation causes increased p100, a target of IKK in the noncanonical NF-B pathway [3]. Although IKK and IKK are present in these cells, increased p100 may inactivate canonical NF-B signaling by blocking p65/p50 translocation from the cytoplasm to the nucleus. Thus, IKK inactivation can impair canonical NF-B activity through the noncanonical NF-B pathway in B cells. On the other hand, NF-B and IKK activities are elevated in Ikk −/− keratinocytes compared to WT keratinocytes, following TNF or IL-1 stimulation [24].
Because IKK has stronger kinase activity than IKK, it is speculated that replacing IKK with IKK in the IKK complex in keratinocytes can elevate IKK activity. Interestingly and consistently, we observed increased IKK activity in skin carcinomas lacking IKK, which was induced by DMBA and TPA [41]. It is not clear whether increased IKK activity is involved in promoting the tumor progression. Thus, regulation of the physiological activities of IKK, IKK, and NF-B in different organs remains a mystery.
Although IKK deletion in keratinocytes induces spontaneous skin carcinomas, we observe increased IKK in some human skin SCCs [65,66] (our unpublished data). Immuno-staining reveals that most of the increased IKK is located in the cytoplasm. Also, we have observed that reintroduced IKK induces terminal differentiation in primary cultured IKK-null keratinocytes; however, immortalized IKK-null keratinocytes are not inducible for terminal differentiation by reintroduced IKK, suggesting that the program for regulating cell differentiation in the transformed keratinocytes may not be functional or may be altered [33]. Thus, these transformed cells do not respond to normal IKK signals, whereas increased cytoplasmic IKK may mimic IKK function in tumor cells. Moreover, Ikk mutations have been reported in human skin SCCs [47]. We also isolated IKK mutations from mouse skin carcinomas, which were induced by DMBA/TPA, and demonstrated that the Ikk mutations can inactivate the IKK function in inducing keratinocyte terminal differentiation compared to WT IKK [5]. On the other hand, because we found that overexpressed IKK in the epidermis inhibited UVB-induced skin carcinogenesis, it is also possible that the distinct pathways may pay a predominant role in driving skin tumorigenesis in those human SCCs that express increased IKK. Taken together, the new targets and precise functions of IKK in tumor cells still need to be addressed for clarification of the role of IKK in tumorigenesis.