Krüppel-like factor5 (Klf5) is a zinc-finger transcription factor normally expressed in the skin. Here, we show that overexpression of Klf5 in the basal layer of the epidermis during embryogenesis affects epidermal development and disrupts epithelial-mesenchymal interactions necessary for skin adnexae formation as well as craniofacial morphogenesis. The transgenic mice exhibited exencephaly, craniofacial defects, persistent abdominal herniation and ectodermal dysplasia. Moreover, the epidermis was hypoplastic and underwent abnormal differentiation with expression of keratin8, a marker for single-layered epithelia, in the stratified epidermis. Correspondingly, we observed a downregulation of ΔNp63 expression in the skin. Overexpression of Klf5 in adult mice led to hyperkeratosis, follicle occlusion and epidermal erosions. Further, we observed decrease and even loss of the stem cell population of bulge keratinocytes, as characterized by the expression pattern of α6 integrin and CD34 markers. Our data suggest a new role of Klf5 as a modulator of p63 expression and the differentiation program of epidermal cells relevant for regenerative potential of the epidermis and epithelial-mesenchymal interactions.
The mammalian epidermis develops from the non-neural embryonic single-layered ectoderm. Through a series of well-orchestrated differentiation and proliferation steps, the ectoderm forms a multilayered, stratified squamous epithelium, which is then continuously renewed during life. Activities of several transcription factors are tightly regulated to achieve a fully functional epidermis. In this regard the Sp1/Klf family of transcription factors, which contains at least 20 identified members in mammals, is emerging as an important player in epidermal biology. Members of this family can act both as transcriptional activators and repressors, and are characterised by a highly conserved C-terminal DNA-binding domain containing three zinc fingers that are similar to those found in the Drosophila protein Krüppel (Kaczynski et al., 2003). In the epidermis, three Klf members are abundantly expressed namely, Klf4/Gklf, Klf5/Iklf and Klf14/Epiprofin (Nakamura et al., 2004; Ohnishi et al., 2000; Sur et al., 2002). Of these, Klf4 is expressed in the differentiating cells of the skin and is required for the generation of the barrier function (Segre et al., 1999). Similar to this, the differentiating cells of the intestine express Klf4 (Shields et al., 1996), whereas the proliferating cells of the intestinal crypts are enriched in Klf5 (Conkright et al., 1999). Klf5 expression in the skin is also predominantly localized to compartments containing proliferating cells. From the published literature, we know that endogenous Klf5 mRNA is not detected in the epidermis at embryonic day (E)10.5 (Ohnishi et al., 2000) and is absent from the epidermis at stages younger that E10.5 (F. Laub, Weill Medical College of Cornell University, NY, personal communication). From E14.5 and onwards, high expression of Klf5 is preferentially seen in the basal layer of the developing epidermis (Ohnishi et al., 2000). High level of expression is also observed in the basal layer of the cells of the inner root sheath and in matrix cells of adult human hair follicles (Sur et al., 2002). Consistent with its expression in proliferating cells of the intestine and skin, Klf5 functions as a positive regulator of proliferation (Sun et al., 2001) in fibroblasts, and partly mediates the transforming activity of oncogenic H-Ras (Nandan et al., 2004) in NIH 3T3 cells. However, Klf5 expression is downregulated in several tumors and overexpression of Klf5 in cell lines derived from these tumors results in growth arrest (Bateman et al., 2004; Chen et al., 2002; Chen et al., 2003). Interestingly, it was recently reported that Klf5 expression is downregulated in quiescent stem cells of hair follicles (Morris et al., 2004).
Mice that lack Klf5 die during embryogenesis, before embryonic day E8.5, whereas heterozygous Klf5-knockout animals survive until adulthood and show diminished levels of arterial-wall thickening, angiogenesis, cardiac hypertrophy and interstitial fibrosis in response to external stress (Shindo et al., 2002). In that study, Klf5 was also shown to interact with the retinoic acid receptor α (RARα) and synthetic RAR ligands affected stress responses in the cardiovascular system in a Klf5-dependent manner. Klf5 thus seems to be a key factor linking external stress and cardiovascular remodeling. Whether or not Klf5 has a stress-related role in the skin remains to be determined.
To address the biological role of Klf5 function in epidermal development and homeostasis, we have generated inducible Klf5-transgenic mice, in which Klf5 overexpression is targeted to the basal layer of the epidermis. Here, we present data showing for the first time that increased Klf5 expression in the epidermis leads to abnormal epidermal development and differentiation, associated with a decrease in ΔNp63 expression and a disruption of the epithelial-mesenchymal interactions required for skin adnexae formation and craniofacial development. In the adult mice, overexpression of Klf5 led to a decrease in the stem cell population of bulge keratinocytes. These results provide support for a new function of Klf5 to regulate the differentiation and fate programs of skin epithelial cells that affects the regenerative potential of the epidermis as well as epithelial-mesenchymal interactions.
Phenotype of K5tTA/TRE-KLF5 bi-transgenic embryos
To study putative roles of Klf5 in epidermis and hair follicles, we generated transgenic mice that expressed the full-length His-tagged human KLF5 cDNA from a tetracycline-responsive promoter. Three founder lines were established. Of these, lines 77 and 17/4 showed high-inducible expression of Klf5, when crossed to two different skin-specific transactivator lines - K14rtTA (Tet-on) and K5tTA (Tet-off) (Diamond et al., 2000; Xie et al., 1999). Subsequently, in all the experiments reported here, we used the transactivator line K5tTA in which the tetracycline-regulated transcriptional transactivator tTA is linked to the bovine keratin 5 promoter.
To determine the effect of Klf5 overexpression on epidermal morphogenesis, TRE-KLF5 transgenic mice from founder lines 17/4 and 77 were crossed with K5tTA mice. In this genetic cross, we did not obtain any bi-transgenic pups among the litters that were born. Caesarian sections were then performed at different time points (E13.5-E18.5) during pregnancy. This revealed the presence of bi-transgenic embryos with severe defects. The overall phenotype observed in the bi-transgenic pups from the two transgenic lines (77 and 17/4) was similar and, subsequently, we used line 77 in all the experiments reported.
The bi-transgenic embryos exhibited exencephaly due to the absence of the cranial vault. Instead, the brain was covered with a richly vascularized connective tissue and limited by a thin epithelial layer; no evidence of intramembranous bone formation was seen (Fig. 1a,b). In addition, the mandible was underdeveloped. Bifurcation of the ribs and partial clefting of the palate was also detected in some embryos (data not shown). We did not observe any limb abnormalities. In the mutant mice the physiological abdominal hernia remained because the ventral midline did not close (Fig. 1a). Thus, these mice exhibit cranioabdominoschisis, a condition also seen in humans (Martinez-Frias, 1997). Overexpression of Klf5 in the epidermis of bi-transgenic embryos was confirmed by reverse transcription (RT)-PCR analysis, immunohistochemistry and western blot analysis (Fig. 1 and supplementary material Fig. S1).
Ectodermal dysplasia upon KLF5 overexpression
Ectodermal dysplasias are developmental syndromes that specifically affect ectodermal organs such as teeth and hair. In the TRE-KLF5 bi-transgenic embryos, a drastic reduction in the number of forming hair follicles on the torso was observed and hair follicles were completely missing in the tail skin. Moreover, tooth development was severely disturbed and we did not detect any incisor development while the first maxillary molar arrested at the early bud stage (Fig. 1c). The embryos had open eyes due to a lack of eyelid formation and vibrissae follicle formation was either aborted at an early stage or was absent (Fig. 1c). The phenotype of the bi-transgenic embryos resembles features of the human syndrome aplasia cutis congenita, which is characterized by defects in the scalp, underlying calvaria and skin (Evers et al., 1995). Interestingly, transgenic mice overexpressing the glucocorticoid receptor (GR) in the epidermis, also show similarities to this syndrome (Perez et al., 2001) and have defects in development of ectodermally derived organs (Cascallana et al., 2005). However, we were unable to detect any change in GR expression. Furthermore, Klf5 did not modulate the transcriptional transactivation capacity of GR in transient transfection assays (data not shown).
Hypoproliferative epidermis and loss of barrier function in bi-transgenics
Bi-transgenic embryos lacked, almost completely, epidermis covering the skull vault, whereas on the rest of the body, the skin was thin and very fragile. Although during embryonal epidermal development there are large regional differences in thickness, the number of epidermal cell layers was drastically reduced, regardless of region, at all stages studied. At E16.5, the epidermis of the bi-transgenic embryos consisted of only one to two cell layers with indication of oedema formation in some regions, compared with the four to five cell-layered epidermis in the dorsal skin of the wild-type embryos (Fig. 2a). It was also evident that the epidermis easily separated from the dermis in a region corresponding to the basement membrane, implicating defective formation of this structure. Staining for the proliferation marker Ki67 confirmed that the epidermis was hypoproliferative (Fig. 2a). In the bi-transgenic epidermis, Ki67 stained only about 25% of the nuclei in the basal layer, in contrast to nearly 80% in the wild type.
Since the epidermis appeared hypoproliferative and Klf4, another member of the Klf family expressed in the epidermis, is required for the barrier function of the epidermis, we tested the barrier formation in the embryos. At E17.5, the bi-transgenic embryos showed a pronounced loss of barrier function (Fig. 2b). No compensatory changes in Klf4 expression were observed at the mRNA level, as determined by RT-PCR (data not shown). The compromised barrier function in embryos overexpressing Klf5 illustrates a striking difference to Klf4, which upon similar overexpression, accelerates barrier formation (Jaubert et al., 2003).
KLF5 overexpression leads to abnormal epidermal differentiation
To determine whether the overall epidermal differentiation program was altered, the expression of a set of markers was analysed in the more mature epidermis at E18.5. Keratin5, expressed by cells of the basal layer, and loricrin, expressed by differentiating cells in the suprabasal layer, were seen in their appropriate locations, indicating that keratinocytes can undergo terminal differentiation (Fig. 3). However, at this developmental stage, the epidermis still expressed keratin8 throughout the basal layer (Fig. 3). Keratin8 is normally synthesised in the surface ectoderm prior to stratification, by peridermal cells and in simple epithelia of the adult. Continued synthesis of keratin8 in this location after stratification of the epidermis in the bi-transgenic embryos was therefore very surprising.
In addition, we found alterations in the expression patterns of keratin1 and keratin17. Keratin17, which is expressed in the basal layer of the epidermis and the developing hair follicles at E16.5 - and for which expression begins to recede towards the developing hair follicles by E18.5 (McGowan and Coulombe, 1998), still showed suprabasal expression in the E18.5 bi-transgenic embryos (Fig. 3). However, keratin1, which is expressed in the keratinocytes as they commit to terminal differentiation and is excluded from the basal layer in the wild type, was expressed in the basal layer of the bi-transgenic skin at E18.5 as well as in the hair follicles (Fig. 3).
There is, thus, abnormal expression of keratin8 (a marker of simple epithelium) as well as keratin1 (a marker of terminal differentiation of stratified epidermis in the basal layer of the bi-transgenic mice) at E18.5. The abnormal morphogenesis of the basal layer in the bi-transgenics was also evident by ultrastructural analysis using transmission electron microscopy in E14.5 embryos (Fig. 4). At this stage, while the wild-type epidermis was well organized with a basal layer of cuboidal cells, the basal cells of the bi-transgenic embryos were elongated and orientated parallel to the basement membrane. Additionally in the bi-transgenic embryos, the suprabasal cells were characterized by thick coalescing bundles of filament resembling the tonofilaments of maturing epidermal cells. Cells of superficial strata also contained discreet granules of varying sizes, showing the semi-crystalline characteristics of keratohyaline granules (Fig. 4).
These data show that the basal layer of the bi-transgenic epidermis is undergoing an abnormal or deregulated differentiation program lacking clear transition from a surface ectoderm to a mature stratified epithelium.
Decrease in ΔNp63 expression
Absence of commitment to epidermal stratification is seen in p63-deficient mice. These mice also show continuous epidermal expression of keratin18, a marker of surface ectoderm that is normally expressed together with keratin8 (Koster et al., 2004). Owing to the use of alternative promoters and transcriptional start sites, the p63 gene encodes proteins that either contain (TA) or lack (ΔN) a transactivation domain. To test the hypothesis that the abnormal differentiation of epidermal cells observed in embryos overexpressing Klf5 is due to interference with p63 expression, we used the p63(4A4) antibody that recognises both the TA and ΔNp63 forms. A marked loss of p63 expression in the interfollicular epidermis of the bi-transgenic embryos was observed (Fig. 5a). Further analysis at the mRNA level by RT-PCR, using primers specific for the two forms, showed that in the bi-transgenic embryos there was a differential downregulation of the ΔNp63 form compared with the p63TA form, the latter even showing a slight increase in expression (Fig. 5b).
The primary defect caused by KLF5 overexpression is intrinsic to the keratinocytes
There are several possibilities as to how Klf5 overexpression in the keratinocytes leads to the observed phenotypes: (1) Klf5 might directly affect the proliferation and/or differentiation of the keratinocytes and, as a consequence, the mesodermal structures could be affected. (2) Klf5 might regulate the expression of a secretory molecule in keratinocytes that affects the mesenchyme and, as a consequence of a defective mesenchymal signaling, epidermal development as well as skeletal development is compromised. To distinguish between these two possibilities, we decided to isolate keratinocytes from the bi-transgenic animals. To prevent transgene expression during development, we administered doxycycline to the pregnant females in their drinking water. We isolated keratinocytes from newborn pups at 0-3 days of age. 5×105 cells from wild-type and bi-transgenic mice were plated per well in 6-well plates and cultured in vitro in the absence of doxycycline to induce the transgene expression. After 48 hours of transgene induction, a decrease in keratinocyte growth in cultures from bi-transgenic pups became apparent and by 7 days these cultures contained very few cells compared with the confluent keratinocyte cultures from the wild-type pups (Fig. 6a,b). Presence of doxycycline in the bi-transgenic cultures effectively inhibited transgene expression and the growth defect (Fig. 6c). The growth defect observed in culture was not associated with increased differentiation as judged from the morphology of the keratinocytes. These results show that the primary defect caused by Klf5 overexpression is intrinsic to the keratinocytes and independent of mesenchymal and/or dermal influence.
To further determine whether the observed growth defect was due to a loss of proliferation or a loss of adhesion, we also performed BrdU-incorporation analysis. We found a marked reduction in the number of BrdU-positive nuclei in bi-transgenic keratinocytes after 24 hours of culture without doxycycline (Fig. 6d), whereas, at this stage, the bi-transgenic cultures were equivalent to the wild-type cultures with regard to growth (Fig. 6b). The induction of transgene expression at 24 hours after doxycycline withdrawal was confirmed using RT-PCR analysis (data not shown). After 48 hours in cultures, we detected a decrease both in the BrdU incorporation and in cell growth. The appearance of a BrdU-incorporation defect prior to the appearance of a growth defect indicates that loss of adhesion is not a causative factor for the observed growth defect in the bi-transgenic keratinocytes.
Decrease or loss of epidermal cells expressing the stem cell marker CD34 in adult bi-transgenics
The thin or missing epidermis in the embryos together with alterations in the p63 levels suggested the possibility of a loss of regenerative potential of the epidermis due to Klf5 overexpression. Since the stem cells and appropriate markers in the embryonic skin are not defined, we investigated the effect of Klf5 overexpression in the adult epidermis in which stem cells residing in the bulge region of the hair follicles have been characterized. K5tTA/Tre-Klf5 adult bi-transgenic animals were obtained by administrating doxycyline to the pregnant females and their progeny. This effectively inhibits the expression of the transgene. The Klf5 overexpression was induced in mice at 4-5 weeks of age by doxycycline withdrawal. In this analysis eight bi-transgenic and eight normal adult mice were used. Initially, the bi-transgenic mice appeared normal but after approximately 6 weeks of transgene induction, all the bi-transgenic animals developed mild hyperkeratosis with an ungroomed appearance and hair loss. Once the phenotype appeared, it rapidly progressed (within one week) to visible skin lesions. Histological analysis showed regions of complete epidermal loss with hyperproliferating epidermis at the edges (Fig. 7a). In addition, hyperkeratosis and follicle occlusion (Fig. 7a) were prominent with formation of utriculae. Development of epidermal erosions in the adult mice would be consistent with a loss of regenerative potential of the epidermis by depleting the stem cell compartment, similar to what is observed upon Myc overexpression in keratinocytes (Gandarillas and Watt, 1997; Waikel et al., 1999).
We further analysed the expression of CD34, an established marker of skin epithelial stem cells in the adult epidermis (Silva-Vargas et al., 2005; Trempus et al., 2003; Tumbar et al., 2004). Keratinocytes were isolated from dorsal skin that showed a mild phenotype (hyperkeratosis, ungroomed hair or hair loss) and that did not present any overly evident epidermal lesions. Isolated cells were analysed for the expression of α6 integrin and CD34 markers using FACS. The α6 integrin is expressed on the basal keratinocytes at the point of contact with the basement membrane, and CD34 expression in mouse keratinocytes of the hair-follicle bulge coincides with label-retaining cells. Together, these markers have been successfully used for physical enrichment of bulge keratinocytes that have stem and progenitor cell characteristics (Silva-Vargas et al., 2005; Trempus et al., 2003; Tumbar et al., 2004). We found a decrease in number (or a complete loss) of α6highCD34high cells in the epidermis of the bi-transgenic compared with wild-type animals (Fig. 7b). The total number of α6highCD34low cells was not altered in the epidermis of the bi-transgenics compared with the wild type, ruling out the possibility of a hyperproliferating epidermis diluting the number of α6highCD34high cells in the FACS analysis. We also found a decrease in the number of α6lowCD34low cells that represent a smaller population of suprabasal stem cells in the bulge (Silva-Vargas et al., 2005; Trempus et al., 2003; Tumbar et al., 2004). The nature of the sorted CD34-positive and CD34-negative populations was verified by keratin5 staining, which showed that more than 98% of the sorted cells expressed keratin5 (supplementary material Fig. S2).
In addition, we also analysed the expression of CD34 by immunohistochemistry. Consistent with the FACS data, there were fewer hair follicles that contained keratinocytes expressing CD34. These hair follicles showed either a completely normal morphology or were mildly affected. No CD34-positive cells were observed within the severely affected hair follicles (supplementary material Fig. S3). The reduction of CD34 expression observed using FACS is, thus, likely to be a consequence of abnormal differentiation and subsequent loss of these cells rather than due to hyperproliferation of epidermal cells.
We report here that, Klf5 overexpression in keratinocytes leads to epidermal and craniofacial defects during embryogenesis associated with alterations in the p63 levels. The epidermis is hypoproliferating and undergoes abnormal differentiation. The observed phenotype suggests a loss of regenerative potential of stem or progenitor cells. This is further substantiated by our finding that the CD34-expressing stem-cell- and/or progenitor-cell-compartments are depleted in the adult epidermis of the bi-transgenics.
Klf5 is known to be a positive regulator of proliferation in fibroblasts as well as in a non-transformed colon cell line (Bateman et al., 2004). The hypoproliferating epidermis in the bi-transgenic embryos and the negative effect of Klf5 on keratinocyte growth in vitro, was therefore surprising. However, overexpression of Klf5 in several cancer cell lines causes growth arrest. Moreover, transcriptional profiling of quiescent hair-follicle stem cells showed that Klf5 was downregulated in such cells (Morris et al., 2004). These results strongly suggest that different cell types respond differently to Klf5 overexpression. Given our data showing missing or thinner epidermis with altered cytokeratin expression during embryogenesis, it is conceivable that overexpression of Klf5 in cells capable of fate change, for instance the stem and progenitor cells of the epidermis, can lead to a loss of these cells due to disruption of their fate-determination programs and abnormal differentiation. The decrease or loss of the CD34-positive bulge-stem-cell population and the development of epidermal erosions in the adult bi-transgenic mice are consistent with such a model. Interestingly, the bulge-stem-cell pool is distinct from the epidermal stem cell pool and does not contribute to the interfollicular epidermis in the absence of injury or trauma (Ito et al., 2005; Levy et al., 2005). In this regard, development of epidermal erosions in the adult bi-transgenic mice suggests that their skin loses both the bulge and the epidermal stem cell pool upon Klf5 overexpression.
How does Klf5 overexpression lead to abnormal differentiation? The abnormal differentiation observed in the bi-transgenic embryos is associated with alterations in p63 levels. The transcription factor p63 is required for commitment of the surface ectoderm to stratification and maintenance of the proliferative potential of the keratinocytes (Koster et al., 2004). Owing to the use of alternative promoters and transcriptional start sites, the p63 gene encodes proteins that either contain (TA) or lack (ΔN) a transactivation domain. Although, p63 was initially described as a factor required for the regenerative potential of epidermal stem cells, its role in stem cell maintenance is still a matter of debate. However, it is now established that expression of the p63TA form in the surface ectoderm is necessary for commitment to stratification. The ΔNp63 form is considered to be essential for further differentiation by counteracting the effect of the p63TA form (Koster and Roop, 2004). In line with this view, we propose that Klf5-mediated downregulation of ΔNp63 expression underlies the abnormal and incomplete epidermal differentiation caused by Klf5 overexpression. Additionally, the alteration in the p63 level is likely to be the key factor in the loss of proliferation observed in the bi-transgenics. A role of ΔNp63α protein in maintenance of the proliferative potential of keratinocytes has been previously inferred (King et al., 2003; Westfall et al., 2003). Thus, the decrease in p63 levels due to Klf5 overexpression could explain both the loss of proliferative potential of bi-transgenic keratinocytes through the effect of p63, for instance on p21 (Westfall et al., 2003), and the expression of keratin8 in the basal layer of the bi-transgenic epidermis due to a disruption of commitment to stratification. Furthermore, the expression of keratin1 in the basal layer might be due to activation of a default differentiation pathway, perhaps as an indirect consequence of loss of proliferation or due to an effect of p63 on adhesion molecules, e.g. Perp (Ihrie et al., 2005). Although in our experiments the proliferation block appeared independently of any adhesion defect, subtle changes in adhesion properties of keratinocytes cannot be ruled out.
Whether or not all the defects observed in the bi-transgenic embryos are due to a decrease in ΔNp63 still needs to be determined. However, the similarities observed in the phenotypic changes of embryos presented here, and some of the phenotypic features of patients with Hay-Wells syndrome (e.g. cleft palate, ectodermal dysplasia and epidermal erosions in the scalp region) (Mancini and Paller, 1997) further suggest that p63 plays a major role. In Hay-Wells syndrome, mutations in the p63 gene are associated with a decrease in the ΔNp63 repressor activity (Westfall et al., 2003).
Klf5 overexpression might regulate p63 levels either by directly interacting with the p63 promoter or by interacting with signal transduction pathways targeting p63. Furthermore, the more severe phenotype of the Klf5-transgenic mice (exencephaly and midline-closure defect) compared with the p63-knockout animals suggests that Klf5 overexpression also targets genes and/or pathways in addition to p63. In this regard, proteins that directly interact with Klf5, e.g. NF-κB and retinoic acid receptorα, or genes that are direct targets of Klf5, e.g. PDGFA, are potential candidates in the skin. Interestingly, the craniofacial and midline-closure defects observed in the Klf5-transgenic mice are reminiscent of the phenotype of the AP-2α chimeric knockout mice (Nottoli et al., 1998). However, in these mice the defect appears to be due to a loss of AP-2α function in the non-epidermal cells (mesenchymal and neural crest cells) because K14-cre-AP2α mice do not share this phenotype (Wang et al., 2006). Since signals originating from epidermis are necessary for normal skeletal and craniofacial morphogenesis (Sil et al., 2004), Klf5 overexpression in the keratinocytes might target AP-2α indirectly by regulating signaling molecules required for epithelial and mesenchymal cross talk. In this regard, we have observed an increase in mRNA levels of one such signaling molecule, Fgf8, in the bi-transgenic epidermis (our unpublished data).
Given our results on Klf5 overexpression and the decrease in the p63 levels, it is intriguing that, at least in some developmental stages for which data is available, high levels of Klf5 mRNA are expressed in the normal basal cells that also express ΔNp63. We have analysed the expression of Klf5 using immunohistochemistry and found that the endogenous mouse Klf5 protein levels are below detection level in the embryonal skin, even at E16.5 where high levels of Klf5 mRNA are reported. The transgene, however, is readily detected. Similar results were obtained using western blotting (supplementary material Fig. S1). Probably, the endogenous Klf5 protein in these cells is unstable and is prevented from targeting ΔNp63. Recently, it was shown that the Klf5 protein levels are tightly regulated by the ubiquitin-proteasome pathway and that Klf5 has a short half-life that is further decreased in tumor cells (Chen et al., 2005). A similar mechanism might regulate Klf5 levels in the basal cells of the embryonic skin. Our data also suggests that most keratinocytes in the basal layer of the embryonal epidermis are susceptible to Klf5-mediated changes in differentiation, based on the uniform expression of keratin8 in the basal layer. In the adult epidermis, however, only a certain pool of the basal keratinocytes is likely to be susceptible to such changes, explaining the requirement of at least 6 weeks for the development of phenotype in the adult mice after induction of the transgene.
Potential impact on carcinogenesis
Klf5 is downregulated in breast, prostate and colon tumors, and overexpression of Klf5 in cell lines derived from these tumors results in growth arrest. Based on our results, we hypothesize that this happens because the tumor stem cells or cells in the tumor mass with a more immature cell fate are susceptible to Klf5-mediated changes in fate and/or differentiation programs and hence undergo growth arrest. The growth-promoting ability of Klf5, however, might be restricted to cells committed to specific fates. In this regard, the hyperproliferation observed in the adult bi-transgenic epidermis is of significance, although it might be a secondary reaction to the epidermal erosion but, more interestingly, it could also be a direct response to Klf5 overexpression in the adult lineage-committed cells.
A context-dependent role for Klf4 (another krüppel-family member expressed in the skin) has been demonstrated (Rowland et al., 2005). In normal cells, Klf4 represses p53 expression and induces the expression of p21cip1, leading to growth arrest. However, in Ras-transformed cells in which p21cip1 is blocked, Klf4-mediated suppression of p53 contributes to carcinogenesis. Thus, Klf4 acts as a tumor suppressor or as an oncogene in a context-dependent manner. Moreover, in a recent study, in the basal cells of adult mouse epidermis overexpression of Klf4 elicited partially overlapping changes to the ones we observe in the adult Klf5 bi-transgenic animals (Foster et al., 2005), namely development of hyperkeratosis, hyperproliferation and cystic degeneration of hair follicles. However, Foster et al. found that these changes subsequently progressed to dysplasia and squamous cell carcinomas. The absence of dysplastic changes in the adult Klf5 bi-transgenic animals could be due to the loss of regenerative potential, leading to the development of epidermal erosions, which are not observed in Klf4 bi-transgenic mice. Nevertheless, this suggests that Klf5 and Klf4 share some target genes in the basal cell compartment of the epidermis and that the balance between Klf5 and Klf4 expression in the epidermis is important. The possibility that Klf5 has a context-dependent role similar to the one reported for Klf4, and that cells with a progenitor potential in the epidermis are particularly sensitive to Klf5-mediated ΔNp63 reduction are two interesting questions originating from our study.
In summary, we have described a new function of Klf5 in the regulation of the keratinocyte-differentiation program associated with a specific downregulation of the expression of the ΔNp63 isoform. Our data suggests that Klf5 can regulate the regenerative potential of stem cells in the epidermis and epithelial-mesenchymal interactions. Identification of proteins that interact with Klf5, and the targets of Klf5 in the epidermis and specifically in stem cells will be essential to identify the signaling pathways relevant for Klf5 function in the skin. These pathways are probably also important for carcinogenesis because the proposed ability of Klf5 to regulate the cellular differentiation program and proliferative potential might be the reason why, in several types of tumors, downregulation of Klf5 expression is observed.
Materials and Methods
Generation of transgenic mice
We cloned a H3-XhoI fragment from the pcDNA3his vector that contained the full-length human KLF5 cDNA fused in-frame with a His-tag at the N-terminus into the pB/pA TRE vector (M. Nilsson, M. Hagerlund and Å.B., unpublished) downstream of a promoter with a tetracycline-responsive element. The Not1-Sal1 fragment was purified and used for microinjection to generate transgenic animals at the Karolinska Center for Transgene Technologies. Microinjections were done in F2 (CBA×C57Bl/6) oocytes. The founders were crossed to C57Bl/6 mice to check for germ-line transmission of the transgene. Three founder lines were established (55, 77, 17/4). Lines 77 and 17/4 showed good inducible expression of Klf5 transgene when crossed to the transactivator lines K5-tTA (Diamond et al., 2000) and K14rtTA (Xie et al., 1999). Experiments described here were performed using the transactivator line K5-tTA. Tail biopsies were used to genotype the mice using the following transgene specific primers: TRE-F1: 5′-ACCCGGGTCGAGTAGGCGTGTA-3′ and TRE-R1: 5′-CCCGGTGTCTTCTATGGAGGTCAA-3′. Pregnant females were inspected for the presence of mating plugs, and the morning when the mating plug was observed was taken as embryonic day 0.5 (E0.5). All animal experiments were done in accordance with the Swedish national requirements.
Skin-permeability assay and skeletal staining
Four-micrometer-thick sections were prepared from formalin-fixed, paraffin-embedded tissue blocks. The slides were heated in an oven at 60°C for 1 hour and subsequently hydrated with H2O through xylene and decreasing concentrations of ethanol. The immunohistochemistry analysis was performed using the Zymed Histostain SP kit. The following primary antibodies and dilutions were used: keratin1 (1:6000; Covance), keratin5 (1:5000; Covance), keratin8 (1:50; Troma1 and kindly provided by Igor Mikaelian, The Jackson Laboratory, Bar Harbor, ME), keratin17 (1:1000; kind gift of Pierre Coulombe, The Johns Hopkins University School of Medicine, Baltimore, MD), loricrin (1:1000; BAbCO), p63(4A4) (1:200; Santa Cruz Biotechnology, Inc.), Ki67 (1:2000; Novocastra), Klf5 (1:1000; Santa Cruz), CD34 (1:50; Abcam) Positive staining was indicated by brown coloration. The slides were counterstained with hematoxylin to show histological details.
E14.5 embryos were fixed in 2% glutaraldehyde, 0.5% paraformaldehyde, 3 mM CaCl2, 0.1 M sodium cacodylate buffer pH 7.2. Small pieces of skin were postfixed in 2% OsO4 in 0.07 M sodium cacodylate, 1.5 mM CaCl2 pH 7.2. These were then dehydrated and embedded in LX-112. The sections were examined with a Leo 906 microscope at accelerating voltage of 80 kV. In each section, the entire area of the epidermis was examined and representative images were photographed for analysis.
To determine the expression profiles of different transcripts in the skin, we extracted total RNA from dorsal skins of E16.5 and E18.5 embryos using the RNA-Bee extraction solvent (BioSite). Reverse transcription (RT) of 1 μg of RNA was performed using random primers (Promega) and Superscript™ RNase H reverse transcriptase (Invitrogen). The expression of different transcripts was detected using the following primers p63TA: forward 5′-TCGCAGAGCACCCAGACA-3′; reverse 5′-GCATCGTTTCACAACCTCG-3′, ΔNp63: forward 5′-TTGTACCTGGAAAACAATG-3′; reverse 5′-GCATCGTTTCACAACCTCG-3′, KLF5-tg: forward 5′-TGACCTCCATAGAAGACACC-3′ reverse 5′-TTGCTGTCCACCAGTCATGCTAGCCAT-3′, Transgene-specific primers crossed the intron boundary between β-globin intron and the Klf5 transgene. β-actin: forward 5′-GACAGGATGCAGAAGGAGAT-3′, reverse 5′-TTGCTGATCCACATCTGCTG-3′. For β-actin, serial dilutions were used to ensure the linear range of amplification.
Primary keratinocyte culture
Cells were isolated from newborn pups (0-3 days old). To obtain viable bi-transgenic pups, dams were given 2 mg/ml doxycycline in 5% sucrose solution throughout the pregnancy. Skins were floated overnight on DispaseII (Roche) at 4°C and primary keratinocytes were isolated after a brief trypsin treatment. Cells were cultured in Ca2+-free and Mg2+-free Eagle's minimum essential medium (EMEM) supplemented with 9% Chelex-treated (BioRad) fetal bovine serum (FBS), 0.05 mM CaCl2, 4 ng/ml epidermal growth factor (EGF) and 1% antibiotic mixture (Invitrogen) on CollagenIV (R&D Systems, Inc.) -coated plates.
A colorimetric assay, based on the cleavage of the tetrazolium salt WST-1 (Roche) by mitochondrial dehydrogenases in viable cells was used for growth analysis. We added 100 μl of WST-1 reagent to cells cultured in 6-well plates. These were then incubated at 37°C under 5% CO2 atmosphere for 4 hours. Absorbance of the sample was measured using a spectrophotometer at a wavelength of 450 nM.
Cells were labeled with 10 μM BrdU for 6 hours. Detection of BrdU incorporation was performed using the in situ cell proliferation kit, FLUOS (Roche) according to the manufacturer's protocol. Slides were evaluated with fluorescence microscopy. Nuclei were counterstained with propidium iodide.
Keratinocytes from the adult mice were isolated as previously described (Wu and Morris, 2005). Single-cell suspensions in 1% BSA-SMEM (GibCo) were incubated for 1 hour with PE-Cy5-conjugated anti-α6 integrin (CD49f from BD Pharmingen) and FITC-conjugated anti-CD34 (BD Pharmingen) antibodies. After staining, cells were analysed on FACS caliber flow cytometer and 50,000 events were counted per sample.
The authors wish to thank J. E. Kudlow and A. Glick for the kind gifts of K14rtTA and K5tTA mouse strains, respectively, Pierre Coulombe for anti-keratin17 and Igor Mikaelian for the anti-keratin8 (Troma1) antibodies, Carin Lundmark and Pinelopi Vlachos for help with sectioning and immunohistochemistry respectively, Peter Zaphiropoulos for comments on the manuscript. V.J. is an IARC fellow. This study was supported by grants from the Swedish Cancer Fund and National Institutes of Health, grant POIAR47898-02. B.R. was supported by a grant from the Wallenberg Consortium North.