Impaired NF-κB Activation and Increased Production of Tumor Necrosis Factor α in Transgenic Mice Expressing Keratin K10 in the Basal Layer of the Epidermis

Both the diversity and the precisely regulated tissue- and differentiation-specific expression patterns of keratins suggest that these proteins have specific functions in epithelia besides their well known maintenance of cell integrity. In the search for these specific functions, our previous results have demonstrated that the expression of K10, a keratin expressed in postmitotic suprabasal cells of the epidermis, prevents cell proliferation through the inhibition of Akt kinase activity. Given the roles of Akt in NF-κB signaling and the importance of these processes in the epidermis, a study was made into the possible alterations of the NF-κB pathway in transgenic mice expressing K10 in the proliferative basal layer. It was found that the inhibition of Akt, mediated by K10 expression, leads to impaired NF-κB activity. This appears to occur through the decreased expression of IKKβ and IKKγ. Remarkably, increased production of tumor necrosis factor α and concomitant JNK activation was observed in the epidermis of these transgenic mice. These results confirm that keratin K10 functions in vivo include the control of many aspects of epithelial physiology, which affect the cells not only in a cell autonomous manner but also influence tissue homeostasis.

The keratins are the main components of the intermediate filament cytoskeleton in epithelial cells. The functions of these proteins were clarified when human epithelial fragility syndromes were attributed to mutations of epidermal keratin genes (for reviews see Refs. [1][2][3][4]. However, this shared function does not clearly explain the great diversity of these proteins, which suggests they may have additional family member-specific functions. In the search for possible specific keratin functions, a study was made of keratin K10. This protein is expressed in postmitotic differentiating keratinocytes in epidermis in vivo (5), and its expression is severely reduced in hyperproliferative situations, including skin tumors (6,7). It has been shown that the expression of this keratin inhibits cell proliferation in cultured cells and in transgenic mice (8 -10). The modulation of cell growth by keratin K10 is linked to the retinoblastoma (pRB) protein and the molecular machinery controlling cell cycle progression during G 1 , in particular cyclin D1 expression (8,9). This activity appears to be promoted by the sequestration of Akt to the keratin cytoskeleton, mediated by keratin K10 through its amino terminus. This leads to decreased Akt kinase activity (9). More recently these results have been amplified to include in vivo situations (10). Transgenic mice were generated in which human keratin K10 gene expression was targeted to the basal layer of the epidermis by using the bovine keratin K5 promoter (11). These animals displayed severe alterations in the epidermis, including decreased proliferation, which results in hypoplastic epidermis, associated with impaired activation of Akt kinase activity in the skin. Finally, by using chemical skin carcinogenesis protocols, it was demonstrated that K10 expression reduces the formation of tumors in vivo (10). These results are in agreement with the recently described fundamental roles of Akt kinase in mouse skin carcinogenesis (12).
Akt is a serine/threonine kinase that plays an important role not only in tumorigenesis but also in many aspects of cell physiology (reviewed in Refs. [13][14][15][16]. This kinase phosphorylates many cellular substrates involved in the control of cell proliferation, apoptosis, and metabolism. Its role in activating Akt in NF-B signaling has recently been demonstrated, which may be an essential antiapoptotic event. However, the mechanism responsible for the activation of NF-B by Akt remains uncertain and may depend on the cell type analyzed (17)(18)(19).
The canonical activation of NF-B requires the phosphorylation of the inhibitory IB protein by a high molecular weight complex that includes IKK␣, IKK␤, and IKK␥ 1 proteins. This allows the NF-B to move to the nucleus and activate target genes. In this context, it has been shown that Akt may increase the transactivation of the p65 subunit of NF-B, without interfering with its nuclear localization (20 -22). Similarly, the increased transactivation of p50 subunit by Akt phosphorylation has been reported (23). Furthermore, the phosphorylation of IKK proteins by Akt, leading to increased phosphorylation and thus degradation of IB proteins, has been widely reported (24 -27). Finally, in breast cancer lines, it has been suggested that Akt mediates the calpain degradation of IB protein instead of the consensus ubiquitin pathway (28). Collectively, all these studies demonstrate the importance of Akt in activat-ing NF-B signaling, but also emphasize the controversy in identifying the molecular mechanisms responsible for such activation.
The importance of NF-B signaling in epidermis has been highlighted in recent years (reviewed in Ref. 29). This stems from findings obtained in transgenic mice ectopically expressing a non-degradable IB␣ protein (IB␣M), or which overexpress p50 and/or p65 subunits in the epidermis (30). In these animals, increased NFB activity leads to epidermal hypoplasia and growth inhibition, probably in association with p21 waf/cip1 expression (30,31). The repression of NF-B activity, on the other hand, produces epidermal hyperplasia (30) and may cause spontaneous squamous cell carcinomas in transgenic mice (32). Nevertheless, we have observed increased endogenous NF-B activity during chemically induced mouse skin tumorigenesis or upon UV treatment of mouse skin (33,34). This apparent discrepancy might be explained in that the overexpression of regulatory molecules used in earlier transgenic studies may differ from those involved in endogenous activity. In this regard, one has to take into account that dominant negative IB␣ expression does not generate a true NF-B null phenotype and that nuclear accumulation of Bcl-3 and its interaction with other NF-B dimers is independent of inhibition by other IB proteins and IKK stimulation. In agreement with this, neither tumors nor phenotypic alterations were observed in the absence of IBM overexpression in epidermis (32). This suggests that a threshold in the level of expression of this protein is required to produce the phenotype. More recently, using a knock in approach, which does not produce massive overexpression, it has been shown that the repression of NF-B promotes severe epidermal defects affecting hair follicle development and increased apoptosis but not increased proliferation or abnormalities in differentiation (35). Finally, the epidermal-specific ablation of IKK␤, which is accompanied by a deficiency in NF-B activation, results in a TNF-mediated inflammatory skin disease but does not led to hyperproliferation or impaired keratinocyte differentiation (36).
Another line of information on the roles of NF-B signaling in keratinocytes comes from the strong phenotypes observed in the epidermis of mice lacking the IKK␥ or IKK␣ subunits of the IKK complex (37)(38)(39)(40)(41). In the case of IKK␣, the abnormalities are not only due to the possible alterations of NF-B signaling but also to the decreased production of a yet-unidentified soluble factor capable of inducing keratinocyte differentiation (42).
Given the above importance of Akt modulation of NF-B activity in a range of systems, and the importance of NF-B in epidermal physiology, an investigation was made into the possible alterations in this pathway that might arise as a consequence of keratin K10 ectopic expression in the basal layer of the epidermis. A dramatic decrease was seen in NF-B activity in both transgenic skin and cultured mouse keratinocytes. This decrease is attributed to the inhibition of IKK activity associated with decreased expression of IKK␤ and IKK␥. Remarkably, the transgenic animals showed aberrant overproduction of TNF␣, and concomitant increased JNK activity, which may account for some of the phenotypic characteristics observed in bK5hK10 transgenic mice.

MATERIALS AND METHODS
Transgene Construction and Generation of Transgenic Mice-The plasmid bK5hK10 was used to generate transgenic mice in a (C57 BL/10 ϫ BALB/c) F 1 and (C57BL/10 ϫ DBA/2) F 1 genetic background as previously described (10). The presence of the transgene was analyzed by Southern blots. Homozygous and heterozygous mice were identified using a PhosphorImager scanner (Bio-Rad) as reported (10). Primary keratinocyte cultures were established isolating keratinocytes from newborn mice cultured in Eagle's minimal medium containing 8% Chelex-treated serum and 0.03 mM Ca 2ϩ as previously described (43).
Histological Analysis-Dorsal skin samples and tumors were fixed either in formalin or ethanol and embedded in paraffin prior to sectioning. 4-m sections were cut and stained with hematoxylin-eosin. Immunodetection of dermal inflammatory cells including T lymphocytes (anti-CD3e), granulocytes (anti-Ly-6G/Gr-1), and macrophages (anti-CD11b/Mac1) was performed in frozen sections using fluorescein isothiocyanate-labeled specific rat anti-mouse monoclonal antibodies (145-2C11, RB6 -8C5, and M1/70 clones, respectively; Amersham Biosciences). Sections were also stained for the expression of K10 as previously described (10). The localization of TNF␣ and IL-6 was carried out in formalin-fixed, paraffin-embedded sections from transgenic and non-transgenic littermates using specific rabbit polyclonal antibodies purchased from Calbiochem and diluted 1/100, followed by horseradish-peroxidase-labeled anti-rabbit antibody (Jackson ImmunoResearch Laboratory, diluted 1:2000). Positive staining was determined using diaminobenzidine as a substrate (diaminobenzidine kit, Vector, Burlingame, CA) following the manufacturer's recommendations. Sections were then counterstained with hematoxylin and mounted. For ultrastructural analysis, skin samples were fixed in 2.5% glutaraldehyde in PBS and postfixed in 1% osmium tetroxide prior to dehydration and embedding in Epon 812 resin. Ultrathin sections were stained with uranyl acetate and lead citrate.
Plasmids and Transfection-The 8xB-Tk-CAT plasmid reporter gene used for transfection assays contained eight copies of the Ig enhancer NF-B binding site. The expression vectors for IKK␣, IKK␤, and IKK␥ containing the open reading frame of the indicated genes, and cloned in pCDNA3 under the cytomegalovirus promoter/enhancer were a generous gift of Dr. A. Israël. The plasmid coding for hK10 has been described previously (8,9). Plasmids coding for wt and dominant-negative Akt were a generous gift of Dr. J. S. Gutkind. Chloramphenicol acetyltransferase assays were performed using an ELISA kit (Roche Molecular Biochemicals) following the manufacturer's instructions.
Retroviral Constructs and Infection-Retrovirus coding for wt Akt or dominant negative Akt were generated by subcloning the corresponding cDNAs into pStMCS vector (kindly provided by Dr. J. C. Segovia, CIEMAT). These plasmids were transfected into Nxe cells and the supernatants were collected 48 h after transfection. For infections, exponentially growing PB keratinocytes were cultured in the presence of retroviral supernatants for 3 h, and afterward fresh medium was added and the cultures were further incubated overnight. Protein extracts were collected after 48 h and used in Western blotting analyses as commented below.
Western Blots-Protein extracts (20 g) were boiled in Laemmli buffer and separated on 10% SDS-PAGE and transferred to nitrocellulose filters (Hybond ECL, Amersham Biosciences, Aylesbury, UK). Filters were blocked with 5% nonfat dry milk in PBS/0.1% Tween 20 at 4°C overnight, washed three times in PBS/0.1% Tween 20, and incubated with the indicated antibodies. After washing, membranes were incubated with a peroxidase-conjugated secondary antibody, washed again, and analyzed using the enhanced chemiluminescence method (West Picosignal, Pierce), according to the manufacturer's instructions. Membranes were stripped by incubation with 62.5 mM Tris-HCl (pH 6.7)/2% SDS/100 mM ␤-mercaptoethanol at 55°C for 30 min and reprobed with different antibodies. No cross-reactivity with other NF-B/ IB/IKK proteins was detected.
Band Shift Analysis-Electrophoretic mobility shift assays (EMSA) were performed by incubating whole cell extracts from mouse skin with a labeled oligonucleotide corresponding to a palindromic B site as previously described (44). The sequence of the B oligonucleotide coding strand was: 5Ј-GATCCAACGGCAGGGGAATTCCCCTCTCCTTA-3Ј (44).
Complexes were separated on 5.5% native polyacrylamide gels in 0.25ϫ Tris borate-EDTA buffer, dried, and exposed to Hyperfilm-MP (Amersham Biosciences) at Ϫ70°C. The composition of the B complexes in newborn mouse skin has been previously described (34).
In Vitro Kinase Assays-Whole skin extracts from newborn mice were obtained in buffer A (1% Triton X-100, 10% glycerol, 137 mM NaCl, 20 mM Tris-HCl, pH 7.5, 1 g/ml aprotinin and leupeptin, 1 mM phenylmethylsulfonyl fluoride, 20 mM NaF, 1 mM disodium pyrophosphate, and 1 mM Na 3 VO 4 ). The kinase activity for Akt and IKK complexes was determined by immunoprecipitation of the endogenous kinase proteins using anti Akt1/2 antibody or a mixture of different IKK␣ antibodies (Santa Cruz Biotechnology). Histone H2B was used as a substrate for Akt, and full-length IB␣ (Santa Cruz Biotechnology) for IKK in in vitro kinase assays (essentially as described in Refs. 12 and 45). Jun kinase assay was performed essentially as described (46) upon immunoprecipitation of JNK using GST-c-Jun (kindly provided by Dr. J. S. Gutkind) as substrate.
Northern Blots-Total RNA from freshly harvested mouse epidermis and frozen tumors was isolated by guanidine isothiocyanate-phenolchloroform extraction. Northern blots containing total RNA (15 g/lane) were probed for expression of TNF␣, IL-6, and IL-1 employing DNA probes prepared by random primed reactions using the complete sequences. The membranes were also hybridized with a keratin K14 cDNA probe to verify that equal amounts of mRNA were loaded and transferred.
TNF Determination-The quantification of TNF␣ in mouse serum and in the culture supernatant from primary keratinocytes was performed using an ELISA kit (R&D) following the manufacturer's recommendations.

Decreased NF-B Activity in bK5hK10 Transgenic Mice-
Transgenic mice ectopically expressing K10 in the basal layer of the epidermis display severe epidermal abnormalities associated with decreased proliferation in close conjunction with Akt kinase activity inhibition (10). Given the importance of Akt in the activation of NF-B (see the introduction), the present study was designed to investigate NF-B binding activity in the epidermis of transgenic mice through electrophoretic mobility shift analysis using a B-labeled oligonucleotide. In non-transgenic samples, two retarded complexes were observed (Fig. 1A, lanes 1 and 2) that were identified as p50-p50 homodimers and p50-p65 heterodimers by supershift experiments using specific antibodies (not shown; see also Ref. 34). Severely decreased DNA binding activity was observed in heterozygous transgenic mice (Fig. 1A, lanes 3 and 4). The B binding activity in homozygous animals was barely detectable (Fig. 1A, lanes 5 and  6). Interestingly, the inhibition of Akt kinase activity runs in parallel with the increased K10 expression observed in heterozygous and homozygous transgenic mice (Fig. 1AЈ) (not shown, see also Ref. 10). Given that we used whole skin extracts for these experiments, and to rule out the possibility that other cell types were actually responsible for the observed effect, similar analysis were performed using primary keratinocytes derived from wt and bK5hK10 transgenic mice. In addition, we also tested whether stimulation with IL-1␣ could induce increased DNA binding in these cells. IL-1␣ treatment led to increased NF-B activity in non-transgenic keratinocytes but not in keratinocytes derived from transgenic animals (Fig.  1B). Western blot analysis against keratin K14 (Fig. 1BЈ) also demonstrated that this effect was not due to different amounts of keratinocyte protein in the assays. These data show that the expression of K10 leads to a dramatic reduction in NF-B activity in keratinocytes and impedes the activation of this complex on stimulation.
Experiments were then performed to see whether this decreased NF-B activity was due to K10 expression and K10mediated Akt inhibition. PB mouse keratinocytes were transfected with a NFB reporter element (8xBCAT) plus empty vector, K10, or K10 plus an expression vector for wt Akt (Fig.  1C). It has previously been shown that co-expression of wt Akt is sufficient to override K10-induced cell cycle arrest in keratinocytes (10). The expression of K10 produced a severe inhibition of NF-B transcriptional activity induced either by IL-1␤ or TNF␣ (Fig. 1, C and CЈ, respectively), whereas co-expression of Akt almost completely abolished such inhibition (Fig. 1, C   FIG. 1. Expression of human keratin K10 in the epidermis of transgenic mice results in impaired NF-B signaling. A, whole cell extracts from non-transgenic, heterozygous, and homozygous bK5hK10 transgenic mice were analyzed by EMSA using a labeled B oligonucleotide. The composition of the complexes is indicated on the right. AЈ, in vitro Akt kinase activity of the same protein extracts. B, cultured primary keratinocytes obtained from non-transgenic and homozygous transgenic mice were analyzed by EMSA as in A. Where indicated, the keratinocytes were also stimulated by incubation (10 min) in the presence of IL-1␣ (10 ng/ml). Note that in transgenic epidermis and primary keratinocytes there is a dramatic inhibition of NF-B binding activity. No stimulation was observed. BЈ, the same protein extracts shown in B were analyzed by Western blot against K14 to rule out the possibility that the observed differences were due to different protein concentrations. C and CЈ, the K10-mediated inhibition of Akt is responsible for impaired NF-B signaling. PB keratinocytes were transfected with the quoted plasmids. 48 h after transfection cells were stimulated as indicated with TNF␣ or IL-1␤ (both 10 ng/ml for 10 min). Chloramphenicol acetyltransferase assays were performed using ELISA (Roche Molecular Biochemicals), and the activities were normalized to ␤-galactosidase activity. Data were taken from duplicate experiments and are shown as mean Ϯ S.D. and CЈ). Finally, to determine whether the inhibition of Akt activity might be sufficient to account for the decreased NF-B activity, the reporter plasmid was co-transfected with a dominant negative form of Akt, resulting in almost complete inhibition of NF-B activity in response to TNF␣ or IL-1␤ (Fig. 1, C  and CЈ). Together, these results demonstrate that K10-mediated Akt inhibition leads to impaired NF-B basal activity and NF-B activation in cultured keratinocytes and in epidermis in vivo.
Ectopic Expression of K10 Leads to Decreased IKK␤ and IKK␥ Expression in Skin-A study was then undertaken to determine whether the decreased NF-B activity found in the transgenic mouse epidermis was due to altered expression of different NF-B family members ( Fig. 2A). No major alterations in p65 (RelA) or IB␣ protein levels were detected by Western blotting (Fig. 2A), and a decrease in only p50 was observed in whole skin extracts from homozygous transgenic mice ( Fig. 2A). Because the NF-B activity is dependent on the activity of the IKK complex, the expression of IKK␣, IKK␤, and IKK␥ was analyzed. Surprisingly, in homozygous transgenic mice there was a significant decrease in IKK␤ and IKK␥ levels (Fig. 2AЈ). Because these two components of the IKK complex are essential for IB phosphorylation and for NF-B activation (47,48), IB kinase activity was evaluated in the transgenic mouse extracts (45). Almost complete inhibition of the IB kinase activity was seen in homozygous mouse extracts (Fig.  2AЉ), and, surprisingly, a significant inhibition of this activity in extracts from heterozygous animals despite the fact these show no decrease in IKK␤ or IKK␥ levels. These data clearly suggest that the decrease in Akt activity elicited by K10 also results in a dramatic inhibition of IB kinase activity.
The decreased protein levels of the IKK components and the concomitant inhibition of IKK activity are very striking. It was therefore, through experiments similar to those shown in Fig. 1  (C and CЈ), determined whether expression vectors for IKK␤ or IKK␥ were able to rescue the K10-induced inhibition of NF-B activity. IKK␤ co-expression partially restored the NF-B inhibition elicited by K10, whereas IKK␥ co-expression totally rescued it (Fig. 2, B and BЈ). These results strongly indicate that the reduction of IKK␤ and IKK␥ in transgenic epidermis is probably responsible for the K10-induced impaired activation of NF-B activity. In addition, the partial rescue observed with IKK␤ also indicates that a residual amount of IKK␥ must be FIG. 2. Decreased expression of IKK␤ and IKK␥ in the epidermis of bK5hK10 transgenic mice. A, protein extracts from nontransgenic, heterozygous, and homozygous bK5hK10 transgenic mice epidermis were probed by Western blot for the expression of p65, p50, and IB␣. AЈ, the same protein extracts shown in A were probed by Western blot for the expression of IKK␣, IKK␤, and IKK␥. Note than in homozygous mice there is a dramatic reduction of IKK␤ and IKK␥ levels. AЉ, in vitro kinase assay following immunoprecipitation of IKK␣ from the same protein extracts shown in A and AЈ and using IB␣ as substrate. B and BЈ, co-expression of IKK␤ and IKK␥ restores the NFB activity. Chloramphenicol acetyltransferase assays were performed essentially as in Fig. 1 (C and CЈ) in PB keratinocytes transfected with the quoted plasmids. Data were taken from duplicate experiments and are shown as mean Ϯ S.D.
FIG. 3. Decreased Akt activity leads to decreased IKK␤ and IKK␥ protein levels. PB keratinocytes were infected with retrovirus coding for wt or dominant negative Akt or empty backbone (mock). Protein extracts were obtained and used to analyze the expression of Akt, Akt phosphorylated in Ser-473, IKK␣, IKK␤, and IKK␥ by Western blotting as in Fig. 3. Note that the expression of dominant negative Akt leads to a decreased expression of IKK␤ and IKK␥, whereas IKK␣ remains unaffected. present upon K10 expression, because this protein is essential for IB kinase activity (48).
Inhibition of Akt Leads to Decreased IKK␤ and IKK␥ Expression-The above commented results prompted us to analyze if the observed decrease in Akt activity might be responsible of the decrease in IKK␤ and IKK␥ proteins. To this, PB keratinocytes were infected with retrovirus coding for dominant negative Akt, wt Akt, or empty retroviral backbone. Forty-eight hours after infection protein extracts were obtained and used for Western blot determination of IKK␣, IKK␤, IKK␥, and Akt as well as Akt phosphorylated in Ser-473. The results (Fig. 3) demonstrate that the inhibition of Akt leads to a significant decrease in IKK␤ and IKK␥ proteins, whereas IKK␣ remained unaffected. This indicates that the observed decrease in these two proteins in skin extracts from bK5hK10 transgenic mice can be attributed to the inhibition of Akt mediated by the expression of K10. In addition, these data are of great importance, because, to our knowledge, this is the first evidence suggesting that Akt activity might regulate the protein level of these two components of the IKK complex.
Phenotypic Alterations in Epidermis of Transgenic Mice Are Different to Those of IKK␥-deficient Mice-IKK␥ is an X-linked gene both in humans and mice. Male mice lacking IKK␥ are embryonic lethal (37,38). Early after birth, heterozygous females show a strong phenotype in skin closely related to that seen in patients with the hereditary disorder retinitis pigmen-tosa (37,38). In particular, the epidermis is characterized by a generalized edema with increased intercellular spaces, severe alterations in differentiation, and increased proliferation and apoptosis (37,38). Given the decreased IKK␥ protein levels observed in the epidermis of bK5hK10 transgenic mice, one might expect that they would correlate with phenotypic alterations similar to those of IKK␥-deficient mice. However, as reported previously (10), heterozygous animals show no overt alterations, and homozygous mice exhibited severe, although different, abnormalities to those reported for IKK␥-deficient females. By day 21 after birth, homozygous bK5hK10 transgenic mice show an epidermal phenotype characterized by a clear decrease in epidermal thickness caused by reduced proliferation as a consequence of K10 expression in proliferative keratinocytes, along with an increased stratum corneum (data not shown, see Ref. 10). Electron microscopy analysis also shows flattened and degenerative basal cells but no intercellular edema in transgenic mice (Fig. 4AЈ, see also Ref. 10). In the present work, prominent irregularities of the basal membrane of the epidermal cells were observed (Fig. 4AЉ). Another characteristic feature of these animals is their progressive generalized phenotype. Transgenic animals were markedly underweight, became very frail, and showed signs of impaired movement and wasting 3-5 weeks after birth (Fig. 4B). In addition, the skin of these transgenic mice was characterized by a severe reduction of the proliferation of the epidermal cells, as demonstrated by BrdUrd incorporation experiments (Fig.  4C, see also Ref. 10). In contrast, we detected increased proliferation of the dermal cells in these animals compared with non-transgenic littermates (Fig. 4C). Finally, a severe reduction in the amount of dermal adipose tissue was also observed (Fig. 4, compare D with E). This causes the bending of hair follicles (arrow in Fig. 4D) due to the 2-to 4-fold decrease in the distance between the epidermis and the muscle layer. None of these alterations are present in animals lacking the IKK␥ subunit of the IB kinase complex (37,38). On the other hand, most of these features are characteristics of animals overexpressing TNF␣ in epidermal cells (49) or as a consequence of IB␣ inactivation (50), which also show increased production of this cytokine (50). Consequently, the expressions of TNF␣ and related cytokines were studied by Northern blotting. Homozygous mice showed a dramatic increase in the mRNA levels of TNF␣, IL-6, and IL-1 (Fig. 5A). To further confirm this, we measured the levels of TNF␣ in the serum of these animals and found a significant increase in circulating levels (Fig. 5B). Finally, experiments were performed to determine whether the keratinocytes were the source of these increased levels of TNF␣. For this, the production and release of TNF-␣ by primary keratinocytes derived from non-transgenic and transgenic mice were analyzed. Elevated levels of TNF-␣ were found in the supernatant of cultured primary transgenic keratinocytes (Fig. 5B).
To further confirm that keratinocytes are the source of TNF␣ in vivo, two experiments were performed. First, we monitored the localization of TNF␣ and IL-6 in skin samples from transgenic and non-transgenic mice using specific antibodies. We observed a positive TNF␣ staining in the non-transgenic samples located deep in the dermis, close to the muscle layer (Fig.  6A), and a few scattered areas more close to the epidermaldermal border (Fig. 6AЈ). On the contrary, in transgenic samples TNF␣ was located primarily at the epidermal-dermal border (Fig. 6B) and in some cases positive staining was also observed in basal keratinocytes (Fig. 6BЈ, arrows). Finally, although no IL-6 was detected in non-transgenic mice samples (Fig. 6C), a clear expression was detected in the epidermis of homozygous littermate samples (Fig. 6D). Interestingly, in this case, a stronger reaction was observed in the hair follicles (Fig. 6D).
As a second approach, the possible presence of inflammatory cells in the dermis of transgenic and non-transgenic skin samples was monitored using antibodies against T lymphocytes (CD-3), granulocytes (GR-1), and macrophages (Mac-1) in frozen skin samples. No increase in any of these cell populations was observed in the transgenic samples (Fig. 7, compare A, B, and C with AЈ, BЈ, and CЈ, respectively). In fact, the dermis of transgenic mice displayed lower number of these cell types as compared with non-transgenic littermates. Given that the increased production of cytokines in the transgenic skin (Fig. 6) would allow for the recruitment of inflammatory cells, the observed decrease would reflect a possible alteration in the production and/or maturation of these cells in transgenic mice. In agreement, we have observed defective thymus development FIG. 5. Increased production of TNF ␣ in bK5hK10 transgenic mice epidermis. A, the expression of TNF␣, IL-1, IL-6 genes in the quoted mouse skin RNA extracts were analyzed by Northern blot using the corresponding cDNA probes. Note the increased expression in the homozygous bK5hK10 transgenic mice. The loading was normalized by hybridization with a K14 cDNA probe. B, the levels of secreted TNF␣ in serum and primary keratinocyte culture supernatants were quantified by ELISA. in bK5hK10 homozygous transgenic mice (not shown). 2 Together, these results demonstrate that the expression of K10 in the basal layer of the epidermis leads to the production of higher levels of TNF-␣, which might explain some of the features of the phenotype of these animals. However, it is worth mentioning that such increased levels were not sufficient to produce liver apoptosis in vivo (not shown).
Activation of the JNK Pathway in Transgenic Mouse Epidermis-The pro-inflammatory cytokine TNF-␣ plays an important role in several cellular events such as septic shock, induction of other cytokines, cell proliferation, differentiation, and apoptosis. In response to TNF treatment, the transcription factor NF-B and c-Jun terminal kinase (JNK) are activated in most cells. Because it was observed that bK5hK10 transgenic mice keratinocytes do not respond to TNF␣ either in vivo or in vitro (Figs. 1 and 2), an investigation was made into the possible stimulation of JNK as a consequence of increased production of TNF␣ in these keratinocytes. JNK activity was analyzed by an in vitro kinase assay using GST-c-Jun as a substrate. The results showed increased JNK activity in the homozygous mice (Fig. 8A). In agreement, increased phosphorylation of ATF2 and phosphorylated JNK-1 was also observed (Fig. 8A).
The binding of TNF-␣ to its receptors TNFR-1 or TNFR2 induces receptor aggregation and results in the recruitment of a number of cytoplasmic proteins to these complexes. TRAF2 is one of these signaling complexes. It interacts directly with TNFR2 and indirectly through TRADD with TNFR1. TRADD also recruits the death domain kinase RIP to these complexes.
Interestingly, TRAF2 and RIP are differentially involved in the modulation of NF-B and JNK signals upon TNF␣ stimulation (51)(52)(53)(54). Consequently, we studied the expression of TRADD, TRAF2, and RIP in the epidermal extracts of nontransgenic, heterozygous, and homozygous bK5hK10 transgenic mice. No differences were seen in the expression of these molecules among the different genotypes (Fig. 8B), therefore, ruling out the possibility that the different NF-B and JNK activities were due to altered expression of these signaling intermediates. DISCUSSION The functional diversity of the keratins is a matter of debate. This report focuses on the possible functions of keratin K10, a protein expressed in non-proliferative suprabasal skin keratinocytes, which in addition is down-regulated during hyperproliferative situations. Using cultured cells and transgenic mouse models we have previously demonstrated that the expression of keratin K10 inhibits cell cycle progression through its ability to bind and inhibit the activation of Akt and PKC (8 -10). More remarkably, transgenic mice expressing hK10 in the basal proliferative keratinocytes are almost completely resistant to skin tumorigenesis (10). This is in agreement with our recent observations indicating a fundamental role for Akt in the process of mouse skin carcinogenesis (12). Akt kinase is involved in many aspects of cell physiology, including the regulation of NF-B signaling. Because this pathway is also very important in the regulation of epidermal functions, we sought to analyze the possible alterations in NF-B in the epidermis of bK5hK10 transgenic mice.
Decreased NF-B activity was seen in the skin and primary keratinocytes derived from bK5hK10 transgenic mice (Fig. 1). Moreover, such inhibition was dependent on K10-induced re- pression of Akt activity (Fig. 1). One of the main phenotypic alterations in these mice is the decreased proliferation in vivo and in vitro (10). This seems to be in disagreement with the results shown by other authors who indicate that NF-B acts in epidermis to arrest cell proliferation (reviewed in Ref. 29). However, it is important to notice that this antiproliferative function has been inferred from transgenic mouse models overexpressing different molecules that interfere with NF-B activity. It is therefore possible that this effect is owed to such overexpression. Recent knock-in and epidermal-specific null animal models have provided data to support this hypothesis (35). Moreover, increased endogenous NF-B activity during mouse skin carcinogenesis has been observed (33).
The present data clearly indicate that decreased NF-B activity can be reversed by Akt co-expression, suggesting that the inhibition of this kinase is responsible for the observed impaired NF-B signaling observed in vitro and in vivo. The experimental evidence suggesting such functional relationship is very ample and still growing (see the introduction). However, the molecular mechanism responsible for such a connection has not been fully demonstrated. Here we show that the inhibition of the IB kinase complex is involved. In fact, the co-expression of either IKK␤ or IKK␥ was, respectively, able to rescue NF-B inhibition partially or totally (Fig. 2). More strikingly, it was observed that in animals expressing higher amounts of K10, and thus with the strongest inhibition of Akt kinase and NF-B activities, there was a dramatic reduction IKK␤ and IKK␥ levels. These data are further confirmed by the use of retroviral constructs coding for wt or dominant negative Akt. We have found that the inhibition of Akt activity leads to a decreased expression of IKK␤ and IKK␥ proteins (Fig. 3). To our knowledge, this is the first evidence suggesting the involvement of Akt in the modulation of IKK expression. The molecular mechanism underlying this process will be studied in the future. Finally, the fact that IKK␤ co-expression can partially rescue NF-B activity clearly points to the possibility that IKK␥ levels were reduced but not completely absent, because this protein is absolutely necessary for this process (37,38).
The phenotype of the high expressing bK5hK10 transgenic mice was, however, clearly different to that reported for IKK␥deficient mice. These animals die in utero while heterozygous IKK␥ females display severe dermatopathy characterized by keratinocyte hyperproliferation, skin inflammation, hyperkeratosis, intercellular edema, and increased apoptosis (37,38). Of all these characteristics, however, only hyperkeratosis was observed in bK5hK10 mice along with some alterations suggestive of increased apoptosis. In addition, the reduced proliferation observed in keratinocytes, in parallel with increased BrdUrd incorporation in the dermal cells, the flattened appearance of the keratinocytes, the reduction in the subepidermal adipose tissue, and the characteristic progressive phenotype (Fig. 4), were all similar to reported alterations in transgenic mice expressing TNF␣ in the epidermis (49). Similar alterations have also been described in the epidermis of mice lacking IB (50), which also have increased levels of circulating TNF␣. In agreement with this, the results show that bK5hK10 keratinocytes produce increased amounts of this cytokine (Figs. 5 and 6). This increased production may be responsible for some of the phenotypic alterations observed in bK5hK10 transgenic mice.
The increased production of TNF␣ is particularly striking. In fact, it is well established that TNF␣, IL-1, and IL-6 genes are predominantly regulated by NF-B elements (55). In this regard it is worth mentioning that similar increases in TNF␣, as well as the increased expression of several genes normally controlled by NF-B factor, have been reported in mice lacking IKK␥ (37), and in epidermal-specific IKK␤-null mice (36). However, in these cases, the increased production of these cytokines has been attributed to the inflammatory cells that invade the tissue (36,37). In bK5hK10 this is not the main cause, as confirmed by immunofluorescence analysis (Fig. 7). On the other hand, we observed a decreased number of inflammatory cells in the dermis of bK5hK10 homozygous transgenic mice. This might be in agreement with our findings indicating that these animals display severe immunodeficiency due to altered thymus development. 2 The decreased expression of IKK␥ observed in bK5hK10 transgenic epidermis clearly indicates the existence of alternative mechanisms for the production of TNF␣, given that this subunit is essential for NF-B activation (48). TNF synthesis and secretion are regulated at several points, including the transcriptional and post-transcriptional levels. Among the elements that may control TNF␣ gene expression, besides NF-B sites, are Ets, ATF-2/c-jun, Sp1, Elk1, CBP, and p300 (56 -59). This clearly points to a central role of JNK activity in the positive modulation of TNF␣ gene expression. Interestingly, JNK activity is inhibited by Akt kinase through direct binding and phosphorylation of SEK1/MKK4 (46,60). Consequently, although we do not know at present how TNF␣ is produced in the absence of normal NF-B signaling in transgenic keratinocytes, a possible explanation might be that Akt inhibition promoted by K10 expression can lead to the activation of JNK. This would allow the production of TNF␣, which, upon binding to TNFR1, might induce increase binding of ATF-2/c-jun to the TNF␣ promoter, therefore, inducing increased transcription (57). Finally, such increased secretion of TNF␣ may account for the expression of IL-1 and Il-6 genes. Alternatively, it is tempting to speculate that differentiation-specific keratins, such as K10, could directly modulate TNF␣. In this regard, it has been demonstrated that simple epithelial keratins modulate different aspects of TNF␣ signaling through direct binding with several components of the TNF-dependent network (61)(62)(63). Therefore, K10 and K8 and/or K18 could act in opposite manners. These aspects will be the subject of future experiments.
This report provides evidence that the specific expression of keratins in epithelia may affect the transcriptional program executed by these cells, probably through the modulation of signaling molecules. This may lead to abnormal production of cytokines, which results, not only in a cell autonomous effect but in a complete disturbance of tissue homeostasis. In this regard, it is worth mentioning that the recently described effect of K10 loss in adult mice results in hyperproliferation in basal layer of epidermis (64). This observation points to a paracrine effect of keratin expression. Whether this might be attributed to altered expression of cytokines as those described here is an attractive possibility that merits future investigation.