Defining a region of the human keratin 6a gene that confers inducible expression in stratified epithelia of transgenic mice.

Injury to the epidermis and other stratified epithelia triggers a repair response involving the rapid induction of several genes, including keratin 6 (K6). The signaling pathways and mechanisms presiding over this induction in keratinocytes at the wound edge remain to be defined. We reported previously that of the multiple genes encoding K6 isoforms in human, K6a is dominant in skin epithelia (Takahashi, K., Paladini, R., Coulombe, P. A. (1995) J. Biol. Chem. 270, 18581-18592). Using bacterial LacZ as a reporter gene in transgenic mice, we show that the proximal 5.2 kilobases of 5'-upstream sequence from the K6a gene fails to direct sustained expression in any adult tissue, including those where K6 is constitutively expressed (e.g. hair follicle, nail, oral mucosa, tongue, esophagus, forestomach). In contrast, the proximal 960 base pairs of 5'-upstream sequence suffice to mediate an induction of beta-galactosidase expression in a near-correct spatial and temporal fashion after injury to epidermis and other stratified epithelia. Transgene expression also occurs following topical application of phorbol esters, all-trans-retinoic acid, or 2-4-dinitro-1-fluorobenzene, all known to induce K6 expression in skin. Our data show that critical regulatory sequences for this inducibility are located between -960 and -550 bp in the 5'-upstream sequence of K6a and that their activity is influenced by enhancer element(s) located between -2500 and -5200 base pairs. These findings have important implications for the control of gene expression after injury to stratified epithelia.

Injury to the epidermis and other stratified epithelia triggers a repair response involving the rapid induction of several genes, including keratin 6 (K6). The signaling pathways and mechanisms presiding over this induction in keratinocytes at the wound edge remain to be defined. We reported previously that of the multiple genes encoding K6 isoforms in human, K6a is dominant in skin epithelia (Takahashi, K., Paladini, R., Coulombe, P. A. (1995) J. Biol. Chem. 270, 18581-18592). Using bacterial LacZ as a reporter gene in transgenic mice, we show that the proximal 5.2 kilobases of 5-upstream sequence from the K6a gene fails to direct sustained expression in any adult tissue, including those where K6 is constitutively expressed (e.g. hair follicle, nail, oral mucosa, tongue, esophagus, forestomach). In contrast, the proximal 960 base pairs of 5-upstream sequence suffice to mediate an induction of ␤-galactosidase expression in a near-correct spatial and temporal fashion after injury to epidermis and other stratified epithelia. Transgene expression also occurs following topical application of phorbol esters, all-trans-retinoic acid, or 2-4-dinitro-1fluorobenzene, all known to induce K6 expression in skin. Our data show that critical regulatory sequences for this inducibility are located between ؊960 and ؊550 bp in the 5-upstream sequence of K6a and that their activity is influenced by enhancer element(s) located between ؊2500 and ؊5200 base pairs. These findings have important implications for the control of gene expression after injury to stratified epithelia.
Injury to skin triggers a repair response aimed at restoring epithelial continuity and barrier function. The activity of several genes encoding intracellular, cell surface, and secreted proteins is rapidly modulated in the epithelial and mesenchymal cells involved in this response (1). Keratins 6,16, and 17, the gap junction protein connexin 26, the receptor for the urokinase-type plasminogen activator, and various proteases are induced in wound edge keratinocytes within hours after injury, and their subsequent accumulation correlates with major changes in keratinocyte cytoarchitecture that precede the onset of migration toward the wound site (2). While the significance of these changes remains to be elucidated, the study of the regulation of the corresponding genes offers an opportunity to decipher the molecular mechanisms underlying the onset of the wound repair response in stratified epithelia. Indeed, even though the levels of several potent growth factors are greatly elevated in the wound site early after injury to the skin (reviewed in Ref. 3), those playing a critical role in this vital homeostatic response remain to be identified.
We recently cloned several human genes and cDNAs predicted to encode highly related keratin 6 (K6) isoforms (4). K6 (56 kDa) is a type II keratin that belongs to the superfamily of intermediate filament proteins and is ususally co-expressed with one or two type I keratins, K16 and K17 (5,6). The K6 isoforms show a complex pattern of expression in epithelia, with constitutive and inducible components (7). They are normally found in the outer root sheath (ORS) 1 of hair follicles, in glandular tissues, in tongue, gingiva and oral mucosa, esophagus, forestomach, and certain reproductive tract epithelia (e.g. Ref. 8). With the exception of palm and sole, K6 is not expressed in normal interfollicular epidermis (5,7,8). The K6 isoforms are better known for their much enhanced expression during hyperproliferation and abnormal differentiation in stratified epithelia (4,9,10). Thus, K6 and K16 are induced in wound edge keratinocytes as early as 4 -6 h after injury to human skin and disappear after closure (11,12). K6 expression is induced as well in a variety of diseases affecting complex epithelia, such as infections, squamous metaplasia, carcinoma, and chronic hyperproliferative disorders, including psoriasis (9,10). In these conditions, K6 expression may be very abundant, but is usually restricted to the suprabasal compartment of the epithelium (10). In mouse skin, K6 expression is induced after topical application of a variety of chemicals (e.g. phorbol esters, retinoic acid; see Ref. 13). K6 induction also occurs in primary cultures of mitotically active keratinocytes from epidermis, esophagus, trachea, and cornea (7,14). Understanding the regulation of K6 gene expression is thus of great interest at various levels, one being the control of gene expression in contexts such as wound repair, psoriasis, and carcinoma. Using a transgenic mouse approach, we report here on the identification of a segment of 5Ј-upstream sequence in the human K6a gene that is both necessary and sufficient for the inducible expression of an heterologous reporter gene in adult mouse epithelia.

EXPERIMENTAL PROCEDURES
DNA Constructs and Production of Transgenic Mice-Our starting template was the human K6a gene, the dominant K6 isoform in hair follicle outer root sheath, foot sole epidermis, and skin squamous carcinoma samples (4). Segments containing 5.2 kb (SmaI-NcoI), 2.56 kb (HindIII-NcoI), 0.96 kb (EcoRI-NcoI), and 0.55 kb (SacI-NcoI) of 5Ј-upstream sequence from the translation initiation codon were isolated from the human K6a gene (GenBank accession numbers L42575-L42583; see Ref. 4) by restriction digestion and subcloned into a LacZ expression cassette. We used a LacZ coding sequence (plasmid pCH110) modified to contain a nuclear localization sequence at its 5Ј-coding end in addition to the SV40 poly(A) sequence at its 3Ј end (15 Transgenic mice were produced by pronuclear injection of DNA constructs in single cell C57B6/BalbC3 embryos (16). Founders were identified by Southern blotting of genomic DNAs using a probe to the coding portion of LacZ. Transgenic lines were established by matings in the C57B6/BalbC3 mixed background (Jackson Laboratories). Doubletransgenic mice were produced by mating the previously described PC5-7-K16 transgenic mice, which contain 8 -10 copies of the fulllength human K16 gene (17), with KT2-2m transgenic mice (Table I).
Experimental Injury and Chemical Treatment of Mouse Tissues-All studies involving animals were reviewed by the Johns Hopkins University Animal Use and Care Committee. For studies involving skin, adult mice (3-6 months old) were anesthetized with avertin and their backs epilated with Nair cream. For injury, the surgical area was disinfected, and full thickness skin wounds were made with a 4-mm punch (Acu-Punch; Acuderm Inc., Ft. Lauderdale, FL). For studies involving other stratified epithelia, adult mice were anesthetized, and short superficial incisions were made with a sterile scalpel to either foot pad epidermis, cornea, oral mucosa, or tongue. Tissues were harvested after 24 h and processed for ␤-galactosidase histochemistry as described above. For studies involving chemical treatment of skin, solutions of PMA (phorbol-12-myristate-13-acetate, 150 l of a 50 M stock in acetone; Sigma) and all-trans-retinoic acid (150 l of a 100 g/ml stock in ethanol; Sigma) were applied topically on Nair-epilated skin every 3rd day for three times. To induce a delayed-type skin hypersensitivity reaction, mice were sensitized with an application of 25 l of 0.25% 2-4-dinitro-1-fluorobenzene (DNFB) at the base of the tail and challenged 5 days later by application of 10 l of the same solution onto the dorsal neck area as described (20). The mice were sacrificed and the skin processed for ␤-galactosidase histochemistry on the next day. Skin papillomas were induced using the two-step chemical carcinogenesis procedure (21), involving initiation with 7,12-dimethylbenz[␣]anthracene and promotion with PMA for 11 weeks. Papillomas were harvested 1 month after cessation of treatment and processed for analysis. In all experiments involving chemical inducers, controls consisted in application of the vehicle only. Table I lists the transgenic lines produced for each of the constructs and reports on transgene expression assessed by ␤-galactosidase histochemistry in situ. None of the constructs, including KT1 (5.2 kb of K6a 5Ј-upstream sequence), shows consistent expression in the ear, trunk, tail, or paw skin of adult transgenic mice (Fig.  1A). In KT1, KT2, and KT3 lines, occasional ORS keratinocytes display ␤-galactosidase activity in a subset of hair follicles (Table I; Fig. 1A). Generally, fewer that three to five follicles show sporadic X-gal staining in the ORS in a typical section (1-2 cm wide) of adult skin tissue (Fig. 1B). In contrast, endogenous K6 is easily detected in mouse hair follicles (Fig. 1C). These findings are supported by immunostainings using antibodies against the ␤-galactosidase protein, by Northern blot-

The Human K6a 5Ј-Upstream Sequence Fails to Direct Sustained Expression in Adult Mouse Tissues-
Does not apply K6 expression in human skin Ϫ ϩϩϩ ϩϩ ϩϩ Does not apply a Expression was assessed using adult mouse tissues incubated with X-gal staining solution, paraffin-embedded, and sectioned for light microscopy. Key: Ϫ, no expression; Ϯ, very sproradic expression; ϩ, modest but consistent expression; ϩϩ, moderately strong expression; ϩϩϩ, very strong expression.
b Wounded tissues (skin and oral mucosa) were examined at 24 h following full-thickness injury (see "Experimental Procedures"). c Expression is very sporadic and restricted to a very small number of outer root sheath keratinocytes in occasional hair follicles (fewer than 1:100).
d Expression in these other tissues is sporadic as well, with only a small subset of cells positive for ␤-galactosidase activity. Abbreviations: esoph, esophagus; kid, kidney; trach, trachea; pericard, pericardium; sm intest, small intestine.
ting (data not shown), as well as by enzymatic assays performed in soluble extracts prepared from intact skin of adult transgenic mice (Table II). A notable exception occurs in the vibrissae follicles of whisker pads in several KT1 transgenic lines, which show weak LacZ activity in a greater number of ORS keratinocytes (Fig. 1, compare D and D-inset). Expression remains patchy, however, and is not seen in transgenic lines made with shorter K6a promoter-based constructs (KT3, KT4; not shown). A survey of other stratified epithelia known to express K6, such as nail, cornea (limbus), tongue, oral mucosa, esophagus, and forestomach fails to reveal ␤-galactosidase activity in the majority of transgenic lines produced (Fig. 1, E-H), albeit with a few exceptions (Table I). Among such exceptions are lines KT3-2m and KT3-4m, which show ␤-galactosidase activity in a very small subset of epithelial cells in tongue, esophagus, and/or cornea (typically, only 1-3 cells/entire histological section; Table I). The three other lines made with the KT3 construct do not show expression in these epithelia (Table  I). Expression of the LacZ transgene is occasionally detected in other types of tissues as well (Table I). Thus, for each K6a promoter construct tested, one or two lines show sporadic expression in the retina, where K6 is not detectable by immunohistochemistry (Fig. 1, I and IЈ). Since this "ectopic" expression does not appear to correlate with transgene copy number (Table I), it appears likely that the transcriptional activity of the LacZ transgenes is somewhat sensitive to their site of integration in the mouse genome.
Induction of Transgene Expression by Injury in Stratified Epithelia of Transgenic Mice-K6 expression is induced following injury to the skin and other stratified epithelia ( Fig. 2A), and this prompted us to examine transgene expression under such conditions. Full-thickness injury to the skin of adult mice induces lacZ expression in epidermis and hair follicles at the wound edge in most KT1, KT2, and KT3 transgenic lines, but in none of the KT4 lines (Table I). In the responsive lines, ␤-galactosidase activity occurs in keratinocytes proximal to the wound edge as early as 2.5 h after injury (Fig. 2, B and C) and extends further away from the wound site at later time points, in a pattern analogous to mouse endogenous K6 (Fig. 2D). Immunostaining for the ␤-galactosidase protein indicates that it is restricted to suprabasal keratinocytes in wounded skin tissue (Fig. 2E). In contrast, mouse endogenous K6 typically extends down to the basal layer in epidermal tissue at the proximal edge of the wound ( Fig. 2A), underscoring a potential difference in the regulation of mouse endogenous K6 and our human K6a promoter-based transgenes (see Ref. 8 for similar observations when using the bovine K6␤ promoter in transgenic mice). Induction of LacZ expression also occurs after injury to other stratified epithelia, including oral mucosa ( Fig.  2F; Table I), cornea (Fig. 2G), tongue (not shown), and foot pad epidermis (Fig. 2H), with all but the shortest transgene (KT4). Thus, these observations establish that the critical information required to mediate rapid induction following injury is contained within the proximal 5Ј-upstream sequence of the human K6a gene. They further suggest that the signaling pathways involved in K6 activation after injury are likely to be related, if not the same, in these four different stratified epithelia.
The histochemistry findings are supported by ␤-galactosidase enzymatic assays performed on soluble extracts prepared from wounded skin tissue, which also reveals differences in the extent of transgene induction depending upon the amount of K6a 5Ј-upstream sequence involved. At 48 h after skin injury, KT1 transgenic mice show a greater induction of ␤-galactosidase activity compared with KT2 and KT3 mice (Table II). We selected transgenic lines showing strong expression (KT1-3m, KT2-2m, KT3-3m) to compare the extent of ␤-galactosidase induction as a function of time after skin injury. The data obtained (Fig. 3) confirmed the rapid induction of the three types of transgene in wound edge tissue. However, peak enzymatic activities were reached at a later time and were three to five times as large in KT1 mice compared with KT2 and KT3 mice (Fig. 3). These data suggest the presence of enhancer elements sensitive to injury and located upstream from Ϫ2500 bp in the 5Ј-upstream sequence of K6a.

Induction of Transgene Expression by Other Acute Stimuli in the Skin of Adult Transgenic
Mice-Topical application of the phorbol ester PMA and of all-trans-retinoic acid (RA), both known to induce K6 expression in mouse epidermis (e.g. Refs. 8 and 13), also result in LacZ induction in [hK6a 5Ј]-LacZ transgenic mice (Fig. 2, I and IЈ, J, and JЈ). As following injury, the extent of ␤-galactosidase enzymatic activity in extracts prepared from skin treated with PMA or RA is clearly greater in KT1 transgenic lines than in KT2 and KT3 lines (Table II). Under the treatment regimens tested, RA appears stronger than PMA in its ability to induce LacZ expression (note that unlike PMA, treatment with RA significantly alters terminal differentiation of skin keratinocytes; see Ref. 13). We also tested whether DNFB, a potent contact allergen that triggers a delayed-type hypersensitivity reaction (20), could induce transgene expression. We found that challenging presensitized skin with a second application of DNFB to a distinct body site causes a modest LacZ induction in many KT1 transgenic lines (Fig. 2,  K and KЈ). Collectively, our data demonstrate that the 5Јupstream region of the human K6a gene contains sufficient regulatory information for its chemical induction in adult transgenic mouse skin using agents that produce enhanced proliferation (via PMA), altered differentiation (via RA), or a contact dermatitis-like reaction (via DNFB).
We next examined the activity of the [hK6a 5Ј]-LacZ transgenes in contexts featuring chronic hyperproliferation and altered differentiation in adult mouse skin. First, we applied the two-step 7,12-dimethylbenz[␣]anthracene-12-O-tetradecanoylphorbol-13-acetate skin carcinogenesis protocol (21) to produce skin papillomas in the various lines of transgenic mice. As expected (22), abundant expression of K6 occurs in premalignant papilloma lesions produced in our various lines of transgenic mice (data not shown). Somewhat surprisingly, a rela-tively small number of keratinocytes express the transgene in fully developed papillomas isolated from KT1 and KT2 transgenic mice, and the LacZ-positive keratinocytes tend to be located in the uppermost portion of the much thickened epidermis (Fig. 2, L and LЈ). Second, we took advantage of K16overexpressing transgenic mice available in our laboratory to produce double-transgenic animals via matings with KT2-2m transgenic mice (Table I). A particular line of transgenic mice containing 8 -10 copies of the full-length human K16 gene (5-7-K16) develops striking lesions in hair follicle ORS and epidermis in the first week after birth, coinciding with the emergence of fur (17). As expected, double-transgenic mice developed similar skin lesions affecting the hair follicle ORS and adjacent epidermis in the first week after birth. However, only patchy LacZ transgene expression could be evidenced in the skin of various body sites in these mice, even though mouse endogenous K6 was present at high levels (Fig. 2, M and MЈ). We verified that the transgene retained its ability to respond to acute skin injury in the hK16-LacZ double transgenic mice (data not shown). Together with the data gathered on chemically induced skin papillomas, these findings suggest up to 5.2 kb of proximal 5Ј-upstream sequence from the human K6a gene may not contain sufficient information for its sustained expression in contexts akin to chronic hyperproliferative diseases.

The Organization of Regulatory Elements in the K6a Gene Appears Unique among Skin Keratin
Genes-Several of the keratin genes are expressed in a stable and predictable fashion in well defined epithelial contexts (5,7). In normal interfollicular epidermis and a few other cornifying epithelia, for instance, the K5-K14 and K1-K10 genes are expressed in a pairwise and constitutive fashion in the progenitor and differentiating layers, respectively (23)(24)(25)(26). In striking contrast, the K6 isoform genes show a complex regulation with constitutive and inducible components in various stratified epithelia, such that there is no obvious relationship between K6 expression and a defined program of terminal differentiation (see Refs. 7 and 14). Yet, the predicted genomic structure and amino acid sequence of the human K6 isoform genes are very related to K5 (a type II keratin as well), and accordingly these have been postulated to originate from a common ancestral gene (4,27). Since the relevant gene duplication event, however, the regulation of these genes has diverged significantly more than their coding sequences (28). Byrne and Fuchs (18) showed that 6 kb of 5Ј-upstream region from the human K5 gene can direct the expression of a LacZ reporter in a tissue-specific fashion in transgenic mice. Similar results were obtained with the human type I K14 and K10 genes (29,30), but not with the type II K1 gene (31), whose faithful regulation seems to necessitate sequences located outside of the proximal 5Ј-upstream sequence (32). Here, we show that the proximal 5.2 kb of 5Ј-upstream sequence from the dominant K6 isoform gene in human skin, K6a (4), does not support consistent expression of a heterologous reporter sequence at a detectable level in any tissue of adult transgenic mice (with the potential exception of vibrissae; see below). In separate studies, we found that the presence of the 3Ј-untranslated region of the human K6a gene in the context of the KT1 and KT2 transgene constructs did not alter the expression pattern of a distinct coding sequence (a mutant K6a cDNA) in transgenic mice (33). We therefore conclude that when assessed in transgenic mice, the constitutive aspect of human K6a expression necessitates sequences that are located: (i) upstream from the proximal 5.2 kb of 5Ј-upstream sequence; (ii) distal to the 3Ј-noncoding region; and/or (iii) in introns, as is the case for the simple epithelial K18 gene (34). The organization of regulatory sequence elements in the human K6a gene thus appears distinct from that documented for the evolutionary related K5 gene as well as other major keratin genes that are constitutively expressed in skin epithelia.
More consistent expression of the KT1 transgene occurs in vibrissae follicles, although it still represents a small fraction of the K6-positive tissue. This may imply that the regulatory sequences involved in directing constitutive expression of K6 are somewhat distinct between hair follicles and vibrissae. Alternatively, however, it could be that the KT1 transgene activity is "constantly induced" at a low level by the mild frictional trauma incurred due to the frequent rubbing of this area associated with grooming. Of all the transgene constructs tested in our studies, indeed, the KT1 was the most responsive to trauma.
The results reported here contrast with those reported for the bovine K6␤ gene, which encodes a keratin protein most related to human K6 in its predicted amino acid sequence and expression pattern (the BK6␤ gene was originally designated BKIV*; see Ref. 8). Two groups observed a near-tissue-specific expression of heterologous coding sequences in transgenic mice when using either 5.2 or 8.8 kb of 5Ј-upstream sequence from the BK6␤ gene (8,35). The occurrence of such significant differences is surprising, given the extensive homology in the proximal 5Ј-upstream sequences of the human K6a, K6b, and bovine K6␤ genes (data not shown; Ref. 38). Multiple K6 isoform genes have been identified in both the human and bovine genomes (4,36,37), and we do not know whether the bovine K6␤ gene is the actual ortholog of human K6a. This notion could explain the differences observed in the activity of the 5Ј-upstream sequence of these genes in transgenic mice. An in-depth comparison of the promoter sequences of these genes should provide significant insights into the unique aspects of the regulation of the human K6a gene.
The keratin 6 gene(s) are co-expressed with the K16 and/or K17 genes as type I keratin partners in stratified epithelia under basal or challenged conditions (see Introduction). We previously reported that a full-length genomic clone (11 kb) containing the entire human K16 gene yielded cell type-specific expression in the trunk skin of transgenic mice under both basal and injury conditions (17), but not in specialized skin epithelia such as foot pad epidermis and nail matrix. As these studies did not address the contribution of the various segments of the human K16 gene to the pattern of expression observed in transgenic mice, the organization of regulatory sequences in the human K6a and K16 genes can not be compared at the present time.
Role of the Proximal 5Ј-Upstream Sequence in K6a Gene Expression after Various Acute Stimuli-We demonstrated here that the proximal 960 bp of 5Ј-upstream sequence in the human K6a gene successfully mediates the rapid induction of a heterologous reporter gene in adult transgenic mouse skin after acute injury or treatment with appropriate chemical inducers, while the proximal 550-bp segment can not. At least when studied in transgenic mice, therefore, we conclude that cis-acting sequences located between Ϫ550 and Ϫ960 bp in the human K6a gene are necessary for its induction when subjecting stratified epithelia to a variety of acute stimuli (injury, 12-O-tetradecanoylphorbol-13-acetate, RA, DNFB). Moreover, these regulatory sequences are at least partly distinct from those underlying its constitutive expression in the relevant epithelia. Whether these inducible elements activate transcription by acting directly on core promoter elements or alternatively by negating a repressor element located within the proximal 550-bp segment remains to be defined. Moreover, given our observation that the product of the KT3 transgene is spatially restricted to the suprabasal layers after its induction ( Fig. 2), as is the case for the K6 isoforms after injury to human skin (11), we also conclude that the critical elements controlling the cell type specificity of human K6a expression are likely to be present within the proximal 960 bp of its 5Ј-upstream sequence. These data extend the findings of Ramirez et al. (8), who observed an induction of a LacZ reporter transgene featuring 8.8 kb of 5Ј-upstream sequence from the bovine K6␤ gene after treatment with 12-O-tetradecanoylphorbol-13-acetate and RA and after injury to the skin. On the other hand, our conclusions differ from those reached by Jiang et al. (39,40), who found that the proximal 390 bp of 5Ј-upstream sequence from the human K6b gene conferred positive and cell typespecific expression of a CAT reporter in human keratinocytes in culture, a context that allegedly mimicks hyperproliferation (14). The "promoter region" of many keratin genes has been found to behave differently when transfected in cultured cell lines compared with when stably integrated within the mouse genome (e.g. Refs. 18, 34, and 41), a notion that may be at play here. Other explanations for this discrepancy include the existence of distinct regulatory mechanisms for human K6a and K6b (see Ref. 4) or alternatively, the potential presence of strong silencer element(s) located between Ϫ390 bp and Ϫ550 bp in both these genes.
We observed a much stronger induction of transgene expression following acute chemical induction or injury to adult transgenic mice bearing a construct featuring 5.2 kb of human K6a 5Ј-upstream sequence compared with those having shorter 5Ј sequences (Table II; Fig. 3). For each acute challenge tested (PMA, RA, injury), indeed, the extent of LacZ induction in mouse skin showed a similar dependence upon the amount of 5Ј-upstream sequence in the transgene. This notion suggests that the relevant regulatory elements in the proximal core promoter are subject to positive regulation by powerful enhancer element(s) located between Ϫ2500 and Ϫ5200 bp in the human K6a gene. Our data also suggest that the molecular mechanisms that trigger K6 induction after acute stimuli in skin may differ to some extent from those underlying its sustained expression in chronic lesions such as those typical of psoriasis and benign and malignant neoplasia. Further characterization of the human K6a gene in transgenic mice should enable us to define the identity and mode of action of the various functional elements involved in controlling the complex regulation of this gene.
It should been emphasized that the results reported here apply to post-natal mouse skin and that our interpretation of the expression pattern is based on the comparison of the distribution of the transgene product with that of mouse K6 protein(s). Transient expression of K6 has been detected in epidermis at a late stage of human fetal development (week 36; see Ref. 7). Studies are in progress to examine whether the [5Ј hK6a]-LacZ transgene is expressed pre-natally in developing hair follicles or epidermis in our lines of transgenic mice. At another level, a close examination of the pattern of [5Ј hK6a]-LacZ transgene expression in adult transgenic mouse skin suggests that after induction not all suprabasal keratinocytes show a ␤-galactosidase-positive nucleus, whereas the majority of them stain positive for mouse K6 protein(s) under the same conditions (e.g. Fig. 2). As apparent from previous transgenic mouse studies (18), the ␤-galactosidase protein may be relatively short-lived in skin keratinocytes, even when targeted to the nucleus (this study). Given that keratin proteins are very stable in epithelial cells (7), a survey of the mouse K6 mRNA(s) distribution would provide a more suitable reference against which to compare the distribution of the transgene product. In a parallel set of transgenic mouse studies involving the same promoter sequences, we found that after induction by PMA application (33) or injury (data not shown), a Myc epitopetagged transgenic keratin protein shows more consistent expression in suprabasal epidermis. A characterization of the mouse K6 isoform family has yet to be completed and should enable the design of suitable probes for the specific detection of K6 mRNA(s). These issues are of significant importance for our understanding of the control of de novo keratin gene transcription at a spatial and temporal levels in stratified epithelia subjected to various types of challenges. This information is also needed to better exploit the 5Ј-upstream region of the human K6a gene for inducible expression or inducible gene rearrangements in stratified epithelia of transgenic mice.
Is K6 Induction a Faithful Marker of Hyperproliferation in Stratified Epithelia?-Induction or enhancement of K6 and K16 expression often accompanies enhanced mitotic activity in stratified epithelia (9), such that the former has often been taken as direct evidence for the latter. Various lines of evidence suggest, however, that induction of K6/K16 expression and enhanced cell proliferation in stratified epithelia may be triggered by distinct signaling pathways. Thus, the expression of K6 and K16 is restricted to post-mitotic, suprabasal keratinocytes under conditions featuring enhanced proliferation in skin (e.g. psoriasis, carcinoma, and after injury; see Refs. 10, 11, and 42). In addition, it has been shown that the population of suprabasal keratinocytes expressing K6 at the wound edge following skin injury clearly extends away from a narrower zone of tissue containing basal keratinocytes with enhanced mitotic activity (see Ref. 2). The evidence introduced here shows that consistent with the rapid appearance of K6 of the K16 proteins in wound edge epidermis after injury to human (11) and mouse skin (data not shown), the induction of the [hK6a 5Ј]-LacZ transgenes occurs within 2.5 h after injury to transgenic mouse epidermis. Yet, an enhancement of mitotic activity in the basal epidermal layer at the wound edge is first detectable at ϳ20 -30 h after injury to mouse and human skin (see Refs. 2 and 11 and references therein). Taken together, these observations point to the existence of significant differences at a spatial and temporal levels with regards to K6 induction and enhanced keratinocyte proliferation at the wound edge. This evidence derived from in vivo studies corroborate previous findings dissociating K6 expression from mitotic activity in ex vivo cultures of epidermal keratinocytes and corneal epithelial cells (14,42,43). At another level, we also found that expression of the KT2 transgene was sporadic at best in chemically induced skin papillomas as well as in the chronic skin lesions of K16-overexpressing transgenic mice. Yet in both circumstances the skin lesions feature abundant K6 protein levels and a lymphocytic infiltration in the context of a markedly thickened, hyperproliferative epidermis. Based on the frequent presence of ␤-galactosidase activity in the uppermost layers of lesional epidermis (e.g. Fig. 2M), it appears likely that these lesions featured transgene expression at an earlier stage of their development. However, the construct tested (KT2) apparently lacks the regulatory sequences required for a sustained expression in a K6-like fashion in chronic hyperproliferative lesions. Collectively, these observations provide strong evidence that enhanced K6 expression can be dissociated from enhanced keratinocyte proliferation in stratified epithelia in vivo, suggesting that the regulatory pathways involved are at least partially distinct.