Basal Keratinocytes Contribute to All Strata of the Adult Zebrafish Epidermis

The epidermis of terrestrial vertebrates is a stratified epithelium and forms an essential protective barrier. It is continually renewed, with dead corneocytes shed from the surface and replaced from a basal keratinocyte stem cell population. Whilst mouse is the prime model system used for epidermal studies, there is increasing employment of the zebrafish to analyse epidermis development and homeostasis, however the architecture and ontogeny of the epidermis in this system are incompletely described. In particular, it is unclear if adult zebrafish epidermis is derived entirely from the basal epidermal stem cell layer, as in the mouse, or if the most superficial keratinocyte layer is a remnant of the embryonic periderm. Furthermore, a relative paucity of cellular markers and genetic reagents to label and manipulate the basal epidermal stem cell compartment has hampered research. Here we show that the type I keratin, krtt1c19e, is a suitable marker of the basal epidermal layer and identify a krtt1c19e promoter fragment able to drive strong and specific expression in this cell type. Use of this promoter to express an inducible Cre recombinase allowed permanent labelling of basal cells during embryogenesis, and demonstrated that these cells do indeed generate keratinocytes of all strata in the adult epidermis. Further deployment of the Cre-Lox system highlighted the transient nature of the embryonic periderm. We thus show that the epidermis of adult zebrafish, as in the mouse, derives from basal stem cells, further expanding the similarities of epidermal ontogeny across vertebrates. Future use of this promoter will assist genetic analysis of basal keratinocyte biology in zebrafish.


Introduction
All organisms exist in environments that are hostile, changeable and with physical and chemical properties at odds with the conditions found within constituent cells. To ensure normal cellular and organ function, organisms must generate and maintain a precise internal environment buffered from external conditions, with pH, osmolarity and ion concentration tightly regulated. The primary organ system that retains water and electrolytes and protects from the environment is the epidermis, which acts as an impermeable barrier elaborated with channels regulating passage of ions and small molecules including water [1]. Furthermore the epidermis in all organisms must also protect from outside-in assaults such as infectious agents, chemical insults, UV irradiation and physical damage [2].
An effective epidermal barrier is an ancient invention, and thus there are a number of features of the epidermis conserved across the vertebrate phylum [3]. The epidermis of adult vertebrates is stratified, composed of a basal (inner-most) stem cell layer residing on a basement membrane and overlaid by increasingly more mature superficial epidermal cell layers. Keratinocytes maintain adhesion through adherens junctions and desmosomes, mediated through the cadherin protein family, whilst impermeability of the tissue to small molecule diffusion is afforded by apically localised tight junctions in more superficial strata [2]. In addition to these features, terrestrial vertebrates require protection from dehydra-tion. This is achieved through an additional lipid-rich superficial cornified layer, in which epidermal cells have undergone a process of terminal differentiation to become flattened, dead corneocytes, lacking organelles but containing heavily cross-linked insoluble protein structures and are enmeshed in an envelope of lipid lamellae. Desquamation leads to a continual loss of corneocytes superficially, with epidermal cell number maintained by replenishment from the basal stem cell layer [4]. In aquatic vertebrates such as fish there is no cornified layer and all cells of the epidermis remain metabolically active [1].
During mouse development, the basal compartment is established from the ectoderm as a monolayer during embryogenesis at E8.5, and then generates an overlying periderm layer [5]. This periderm constitutes the first front of the differentiating keratinocyte program. As development proceeds, this bilayered epidermis becomes multi-layered and individual strata are established. Eventually the periderm becomes the first cornified layer, is subsequently sloughed and replaced from intermediate cells of the underlying layer [6,7]. In contrast to mouse, fish development occurs entirely externally and embryos are thus exposed to hypotonic conditions at early stages. To buffer against this, a cell lineage segregates prior to gastrulation, and generates a specialised epithelial layer, the Enveloping Layer (EVL) which constitutes a periderm monolayer encasing the entire embryo following gastrulation [8], and which contains membrane localised compo-nents of the tight junction complex [9]. A basal epidermal layer is established later during gastrulation. Similar to mouse and in contrast to the EVL, this basal layer is derived from the ventral non-neural ectoderm [10], and comes to reside under the EVL following gastrulation. Thus by somitogenesis stages, there is a bilayered epidermis covering the entire embryo, with each layer derived from different cell precursors.
The zebrafish epidermis remains bilayered until around 15 dpf (days post fertilisation) when intermediate layers become apparent [11]. The origin of these intermediate strata has not been demonstrated but is assumed to derive from the basal compartment. Whilst it has long been assumed that the EVL layer is lost during embryogenesis, mirroring the periderm of the mouse embryo, this has not been demonstrated [8]. Indeed mosaic cell labelling in zebrafish demonstrated that EVL cells persist until at least 9 dpf [12], however their fate beyond larval stages is unknown.
Analysis of murine epidermal development has been aided by a number of markers of different layers and stages of epidermal development. Much use has been made of the dynamic expression of keratins, which temporally and spatially reflect the differentiation status of epidermal cells. Keratin proteins form intermediate filaments within epithelial cells (cyto-or soft keratins), and also generate nail and hair structures decorating the epidermis (hard keratins) [13]. The human genome contains at least 54 keratin genes which encode proteins classified as either type I (acidic pI) or type II (basic pI) [14]. Together these form heterodimers which aggregate to generate the intermediate filaments of the cytoskeleton or of the hard cornifications of epidermal modifications [13]. While over half of keratin genes in the human genome are dedicated to the latter function, the numerous cytokeratins are expressed in specific domains of different epithelia. Simple epithelia of different organs express Keratin8 and 18, with certain simple epithelial tissues additionally labelled by other keratin pairs. Stratified epithelia express a distinct array of keratins, which again are specific to both tissue type and constituent strata. For example, Keratin5 and Keratin14 form a heterodimer specific to the basal cell layer of the epidermis, whilst the suprabasal layers are labelled by Keratin 1 and Keratin10 [13,14]. Thus through in situ or antibody labelling, specific keratins have been used to characterise different epidermal layers, whilst cloning of the promoter regions of specific mammalian keratins has allowed misexpression of proteins in specific epidermal compartments.
Zebrafish contain much fewer keratin genes than mammals, reflecting the absence of hard epidermal modifications such as hair and nails. 16 type I and 7 type II keratins have been identified in the zebrafish genome [15], however a detailed survey of their expression in epithelia has not been conducted. It is likely that, as in mouse, they will show cell type specificities. Indeed the zebrafish keratin4 gene has been shown to be expressed predominantly in the EVL [16] and the keratin4 promoter has been used extensively to label the EVL from early stages [17]. To date a compact and robust promoter labelling basal epidermal cells is not available. A promoter from the gene encoding the transcription factor DNp63 has been used to label basal keratinocytes at 5 dpf but displays low and variable expression levels [18]. As the basal layer contains the epidermal stem cells, a strong and robust promoter to label these cells would be invaluable to researchers wishing to manipulate and study the epidermal stem cell niche in the zebrafish model system.
Here we demonstrate that the type I keratin encoding gene krtt1c19e, is broadly expressed in the basal epidermal cell layer, and identify an upstream promoter region which recapitulates this expression pattern in germline transgenics. Using this and the krt4 promoter to drive an inducible form of Cre recombinase in the two cell layers of the embryonic epidermis, we trace the fate of the periderm in zebrafish and the origin of the external epidermal cell layer in adult zebrafish. We show for the first time that the EVL is transient and is shed gradually during metamorphic stages, being replaced from basally derived cells. Thus all strata of the adult zebrafish epidermis can be reconstituted from the basal compartment, highlighting that a system of epidermal renewal from a basal stem cell compartment likely existed prior to adaption to land.

Fish Husbandry and Cre Reporter Line
Zebrafish were housed in the IMCB Zebrafish facility and embryos used for analysis obtained through natural crosses and staged according to Kimmel et al. (1995) [19]. For lineage analysis, we used the ubi:switch reporter line Tg(23.5ubb:loxP-EGFP-loxP-mCherry) cz1701 [20]. All animal procedures in this study complied with the National Advisory Committee For Laboratory Animal Research (NACLAR) Guidelines set out by the Agri-Food and Veterinary Authority (AVA) of Singapore and was overseen by the Institutional Animal Care and Use Committee (IACUC) of the Biological Resource Centre BRC (IACUC Protocol Number: 090435).

RNA Isolation and RT-PCR
RNA was isolated using Trizol (Invitrogen) from WT samples or water control, and cDNA generated using Superscript III Reverse Transcriptase (Invitrogen). Detection of krtt1c19e by PCR used the following forward and reverse oligos: F: CTCTTGAGAAAGC-CAATGCTG; R: ACCTGTCCACTCATTTGATCG. a-actin was amplified as a positive control using these primers: F: TGGCATTGCTGACCGTATGC; R: GTCATGGACGCC-CATTGTGA.

Transgene Construction, Embryo Injection and Transgenesis
To generate transgene constructs, the promoters of both krtt1c19e and krt4 [17] were cloned into a 59 entry clone of the Tol2Kit [21], and all transgene constructs generated by Gateway recombination using LR Clonase II Plus enzyme (Invitrogen). Both the krtt1c19e:Cre ERt2 and krt4:Cre ERt2 constructs carried the myl7:egfp transgene in cis as a means to identify carriers. Plasmid DNA for injection was isolated using HiSpeed Plasmid Midi Kit (Qiagen). To generate transgenic zebrafish lines, 30 ng/ml plasmid DNA and 30 ng/ml tol2 RNA were co-injected into 1-cell stage embryos using a PLI-100 microinjector (Harvard Apparatus).

Genetic Labelling through Tamoxifen Treatment and Transplantation
Permanent labelling of keratinocytes in krt4:cre Ert2 ; ubi:switch and krtt1c19e:cre Ert2 ; ubi:switch transgenics was achieved by pulsed activation of Cre ERt2 with 4-hydroxytamoxifen as per Lee et al, (2013) [25]. For each timecourse experiment, we imaged 8 fish at 4 different timepoints and repeated the experiement 3 times per genotype.
To generate clones of genetically labelled basal keratinocytes, deep cells from dome stage krtt1c19e:lyn-tdtomato; krt4:lyn-egfp double transgenic embryos were transplanted to age-matched wild-type hosts layer as described [26]. 5 individuals were imaged at 7 dpf and then assessed every 4 days for evidence of eGFP expression in EVL cells.

Imaging
Confocal images were taken on a Zeiss LSM700 or an Olympus BX61 Fluoview microscope, whilst brightfield or Nomarski images were taken on a Zeiss AxioImager M2. Low-magnification fluorescent images of adult transgenic fish were taken on a Leica MZ16FA. In all experiments, multiple individuals or sections were examined and representative micrographs imaged.
To follow the expansion or loss of mCherry expressing cells upon permanent Cre-lox mediated labelling, floxed individuals were imaged every 4 days from 12 days onwards. Fish were anaesthetized with 0.02% Tricaine (buffered to pH 7.0) and mounted in 3% methyl cellulose. A post-anal region was iteratively re-imaged using a DM6000B Leica microscope. After imaging, the embryos were washed in facility water and returned to the fish facility.

Identification of krtt1c19e as a Predominantly Basal Epidermal Marker
To identify genes predominantly expressed in the basal epidermis, we exploited the lateral displacement of basal epidermal cells by the posterior lateral line primordium during its migration. This displacement generates a gap in the staining pattern of basal epidermal markers such as DNp63 ( Figure S1A-A'), whilst overlying EVL cells are unaffected (see timelapse movies in [27]). We scanned an online database of wholemount in situ hybridisation patterns for epidermally expressed genes which show a gap in staining corresponding to the primordium. The type I keratin designated krtt1c19e [15] showed an appropriate pattern at 24 hpf [28], and the expression of this gene was further assessed. RT-PCR indicated expression at all stages of development up to adulthood, including the 2-cell stage ( Figure S2A). Such early expression, indicative of maternal mRNA contribution, was confirmed by in situ hybridisation (data not shown). Using a fulllength in situ probe, specific expression in the epidermis was first noted weakly by 15 hpf (Figure S2B, C), and was detected from 24 hpf up to at least 5 dpf ( Figure 1A, 1F and 1I), with strong signal noted from 48 hpf onwards. We demonstrated expression was predominantly in basal epidermal cells and not EVL cells from 24 hpf onwards in a number of ways. Firstly, through Nomarski optics, we could identify unlabelled EVL cells overlying the labelled cells ( Figure 1B; Figure S2E, S2G). Secondly, we could see labelled cells outline the primordium ( Figure 1C; Figure S2F). Thirdly, we noted that the staining appeared to label only a single layer of the bilayered epidermis (for example of the cornea [29]; Figure 1G). Finally, immunostaining krtt1c19e fluorescent in situs with the basal keratinocyte nuclear marker, DNp63, showed consistent co-labelling of cells in a lateral view (24 hpf - Figure 1D). Cryosections of 24 and 48 hpf co-stained embryos indicated labelled cells occurred in a monolayer ( Figure 1E-E'; 1H-H'), found below the (DNp63 and krtt1c19e negative) EVL layer, as visualised by DAPI counterstaining and Nomarski optics. Curiously we observed that the larval fin fold epithelium does not express krtt1c19e at any stage ( Figure 1F; Figure S2H), despite the presence of basal keratinocytes here. We also noted a small number of EVL cells were faintly stained using the more sensitive fluorescent in situ hybridisation ( Figure S2D), suggesting that some EVL cells do express very low levels of krtt1c19e or a transcript with some sequence homology. In situ probes against either the first or second half of the krtt1c19e cDNA show an identical pattern, albeit weaker than the full length ( Figure S3A-F).

Isolation of a krtt1c19e Promoter and Generation of Germline Transgenics
Given that krtt1c19e is highly and predominantly expressed in basal keratinocytes, we sought to isolate its regulatory regions as a means of driving exogenous genes in this cell layer. This gene sits within a type I keratin gene cluster on chromosome 19 (Figure 2A), with the start of transcription only 3.9 kb downstream from the end of the preceding gene (cki - Figure 2B). This suggested krtt1c19e might possess a relatively small promoter. We cloned 3.9 kb upstream from the start of translation of the krtt1c19e gene into a 59 gateway entry vector of the Tol2kit [21], and recombined it upstream of eGFP ( Figure 2C). Transient analysis upon injection of this construct into embryos showed limited epidermal expression at 24 hpf, which became extremely widespread from 48 hpf onwards ( Figure 2D). To ascertain if the cells labelled were indeed basal cells as expected, we counterstained injected embryos with basal and EVL specific markers. From 24 hpf to 72 hpf, embryos contained eGFP cells which were DNp63 positive ( Figure 2E-G), and were outlined with strong E-cadherin staining at 72 hpf ( Figure 2H), both indicative of basal keratinocyte expression. At all stages however, we also noted GFP labelling of overlying ZO-1 positive EVL cells ( Figure 2I-K), confirmed by eGFP cell borders bisecting underlying DNp63 nuclei ( Figure 2L). Thus the krtt1c19e promoter can transiently drive gene expression in both cell layers of the embryonic and larval epidermis.
Due to the variable transgene copy number partitioned to cells upon transient injection, promoters often show activity in ectopic locations, not representative of their true activity domains. To test this, we generated germline transgenics from the injected embryos and identified a number of founder lines, the majority of which showed identical expression patterns. We could first detect eGFP weakly at 36 hpf in the epidermis, but by 48 hpf expression is robustly detected in keratinocytes over the head and body of the fish, but excluded from the fin fold epithelium (Figure 3A), mirroring the endogenous krtt1c19e mRNA. We observed similar expression pattern at all larval stages to 7 dpf ( Figure 3B). The promoter drove eGFP expression strongly and predominantly in the basal DNp63 positive keratinocytes at both 48 hpf to 7 dpf with the EVL unlabelled, as demonstrated by immunofluorescent staining of cryosections ( Figure 3C-D'). To better visualise basal keratinocyte morphology, we used this promoter to label the cell membrane with lyn-tdTomato. The lyn tag acts as a myristoyla- tion and palmitoylation substrate, and once modified will direct proteins to the cell membrane [30]. We crossed the resulting transgenic line with another transgenic line expressing membrane tethered eGFP in the EVL using the krt4 promoter. At both early ( Figure 3E  This further highlighted the specific basal expression of the krtt1c19e promoter. We raised the krtt1c19e:egfp transgenic larvae to adulthood, and noted GFP expression is sustained, albeit faintly, in the trunk epidermis ( Figure 4A), whilst there was now strong GFP signal associated with the neuromasts of both the body and head lateral line systems ( Figure 4B-C). Co-staining of krtt1c19e:lyn-tdtomato adults with antibodies against p63 and tdTtomato indicated that this labelling was restricted to p63 positive epidermal cells surrounding the neuromast ( Figure S5A-B''). In contrast to the larval expression pattern, there was strong epidermal expression in all fins ( Figure 4A). Immunostaining of cryosectioned transgenics was used to determine which epidermal cell layers were GFP positive. Counterstaining with DAPI, DNp63, Phalloidin and Pancadherin demonstrated that the promoter labels the basal and suprabasal layers of the epidermis, but does not label the most superficial layer of the epidermis at trunk or fin locations ( Figure 4D-G'''). This indicates that the krtt1c19e promoter can drive expression in basal (or suprabasal) epidermal layers at all stages, but is inactive in the superficial EVL.

The EVL is a Transient Structure and is Replaced during Metamorphosis
Having a means to label the basal keratinocytes prompted us to test if these cells eventually generate the entire epidermis as in mammals, or if the EVL is retained throughout the life of the fish. We used two methods to test if the superficial EVL is derived from the basal cells. Firstly we employed a transplantation approach to generate clones of basal cells carrying transgenes labelling the basal keratinocytes and EVL cells. To do this, we used the krtt1c19e:lyn-tdtomato; krt4:lyn-egfp double transgenic fish described above and the fact that the EVL lineage restriction to superficial periderm cells occurs prior to 4 hpf. At this time point deep cells of the blastula are unable to give rise to periderm cells of the larva [8]. We thus transplanted deep cells from double transgenic embryos into unlabelled hosts at dome stage (4.3 hpf) (see Figure 5A for experimental outline). We confirmed at embryonic and larval stages that whilst the hosts contained a mosaically labelled basal epidermal layer ( Figure 5B, D), there had been no transplantation of EVL cells, as all cells in this layer were unlabelled (therefore host derived - Figure 5C, D). Daughter cells of these basal cells will share the same genotype including the krt4:lyn-egfp transgene, and thus EVL cells will be marked by lyn-eGFP if, and only if, they are derived from basal cells. By 26 dpf we could observe expression of lyn-eGFP in the superficial layer of the epidermis ( Figure 5F). In all cases, eGFP positive EVL clones were found above tdTomato labelled basal cells ( Figure 5E, G). Together this strongly suggests that at least a proportion of the EVL is replaced by basal keratinocytes and that the embryonic EVL is transient.
To confirm this conclusion, we used the Cre-lox system to permanently label both basal keratinocytes and EVL. We generated krtt1c19e:cre Ert2 and krt4:cre Ert2 transgenic lines and crossed these to the ubi:switch Lox reporter line [20]. Upon activation of Cre recombinase, this reporter switches from expression of eGFP to mCherry. We treated krt4:cre Ert2 ; ubi:switch larvae with 4-Hydroxytamoxifen (4-OHT) from 30 to 78 hpf to activate Cre recombinase mediated switching. This labelled the EVL specifically with mCherry at larval stages ( Figure 6A-B'). Similarly treatment of krtt1c19e:cre Ert2 ; ubi:switch larvae with 4-OHT for 90 minutes at 72 hpf led to robust labelling of the basal epidermis when visualised at 7 dpf ( Figure 6C-D''). We followed larvae over the following 2 months and noted that EVL cells labelled at early larval stages do persist until approximately 1 month, but found that they are gradually lost from the epidermis at metamorphosis stages. Clones thinned out from 15 dpf onwards such that at 1 month only a handful of cells remain, and by 42 dpf, we could not observe labelled cells in the majority of individuals ( Figure 6E-E'''). This indicated that the vast majority (and likely, all) EVL cells are not permanent. To confirm that the superficial cells are being replaced from basal keratinocytes, we followed clones of mCherry positive cells from the floxed krtt1c19e:cre Ert2 ; ubi:switch transgenics. In contrast to EVL labelled clones, we noted both expansion and stratification of basal clones ( Figure 6F-F'''). To demonstrate that these cells expand to contribute to the most superficial stratum, we sectioned through mCherry labelled clones of krtt1c19e:cre Ert2 ; ubi:switch double transgenics at 42 dpf and immunostained for mCherry, eGFP and DNp63. We observed that within both the trunk ( Figure 6G-G'') and fin ( Figure 6H-H'') epidermis, basal cells clones contribute to all layers of the epidermis, including the most superficial cells (as marked by DAPI), suggesting that not only is the EVL lost, but it can be replaced from basal cells. Note that each clone also contains EVL cells which are not floxed, expressing eGFP rather than mCherry (example shown in Figure 6G). Due to the mosaic nature of the Cre mediated recombination, it is not possible to determine if such cells represent perduring embryonic EVL cells or are derived from neighbouring unfloxed basal keratinocytes. Nonetheless our krt4:cre Ert2 ; ubi:switch data conclusively shows that the vast majority of embryonic EVL cells are lost within the first month after fertilization.

Discussion
There is an increasing use of the zebrafish system in the epidermal field, underscored by a growing awareness of the similarities between mammalian and piscine epidermis at the genetic, cellular and organ levels. Being aquatic, fish have no need for a lipid rich water-impermeable stratum corneum, which evolved after the transition to land [3], and this lack of cornification represents the major difference between mouse and fish epidermis. However we propose that the fish presents an excellent system to study the biology of basal/suprabasal keratinocytes and the dermis, in isolation from secondary cornification defects. Indeed ultrastructural and marker analysis [11,31] identified many anatomical and cytological similarities between fish and mammalian skin, including stratification, adhesion structures and dermal stroma content. Furthermore, genetic analyses, driven in part by genetic screens [32,33], have led to the growing awareness that the genetic basis of zebrafish epidermal development shares much with mouse and human genetics, and can model human disease. Attenuated function of kindlin-1 [34], integrina3, laminina5 [35], irf6 [36], epcam [37], lgl2 [31], ikk1 [12], col17a1 [38] and spint1a [39] all generate defects in the zebrafish epidermis or periderm which are comparable to loss in mammalian systems, underscoring conservation of gene function. Indeed in some cases the mutants model the human disease accurately. For example, mutations in the zebrafish orthologues of human FRAS1, FREM1 and FREM2 recapitulate the epidermal blistering of distal appendages seen in Fraser Syndrome [35]. Importantly, the basal epidermal transcription factor DNp63, which has a mitogenic role in mammalian basal keratinocytes, has an identical function in zebrafish. Loss of zebrafish DNp63 reduced basal keratinocyte proliferation and lead to blistering and epidermal lysis [10,40], and shows strong phenotypic parallels with the human ectodermal dysplasia, ectrodactyly, and cleft palate syndrome [41]. Thus a basal epidermal stem cell population, and the genetic programs deployed therein, were an ancient invention in the vertebrate lineage.
In addition to the basal layer, the epidermis of zebrafish embryos contains a periderm layer, which is derived from the first cell lineage to segregate during development -the enveloping layer. Previously assumed to be lost during development, it had been shown that the EVL is retained until at least late larval stages [12], thus making its final fate unclear. It was possible that the lineage was retained throughout life and EVL cells were replaced by cell division within the stratum as seen in the epidermis of  Drosophila [42]. We have shown for the first time using a Cre/ Lox approach that, as in the mouse, the EVL/periderm is gradually lost in zebrafish albeit at a relatively late developmental stage. Both transplantation and employment of the Cre/Lox system demonstrated that the cells of the most superficial stratum are reconstituted from basal cells. These replacement superficial cells, derived from an alternate lineage, display all the hallmarks of the original EVL cells, namely labelling by the krt4:lyn-egfp transgene and presence of apical microridges. Thus we have shown that keratinocytes of all strata of the adult epidermis derive from the basal keratinocytes, and a basal keratinocyte layer able to contribute daughter cells to all epidermal strata likely evolved before the transition of vertebrates to land.
We noted a number of secretory cells in the epidermis are unlabelled in the floxed krtt1c19e:cre Ert2 ; ubi:switch transgenics, however this was expected as they segregate very early from epidermal progenitors within the non-neural ectoderm, prior to our Cre labelling [43,44,45]. Whilst the loss of the original EVL cells at metamorphosis was gradual we have not yet quantified the rate of turnover of the adult epidermis. Experiments are currently underway using later pulses of 4-OHT to measure this. It would be of further interest to employ the reagents we have presented here to assess the behaviour of different lineages during wound healing and assess if there is an increased turnover of cells in different strata. Indeed preliminary characterisation of the wound healing process in adult zebrafish shows extremely rapid re-epithelialisation [46]. Our reagents now permit dissection of the strata driving barrier re-establishment.
Finally, we have identified that krtt1c19e is strongly and predominantly expressed in basal epidermal cells. We observed some unexpected regionalisation of the epidermis through our in situ analysis, with krtt1c19e expression excluded from the larval fins, despite the presence of p63 positive basal epidermal cells in this location. Such a sub-regionalisation had not been described before and we are unsure of its significance. It was confirmed by use of the krtt1c19e promoter, which curiously showed a converse expression pattern in the adult with basal epidermal cells of the fins and surrounding neuromasts strongly labelled, whilst trunk epidermal cells weakly labelled. We surmise the promoter might be responding to an inhibitory signal from the larval fin apex, and is de-repressed in the adult fin and neuromast region. However it occurs, it demonstrates that the zebrafish epidermis shows regionalisation, mirroring the regionalisation of mammalian epidermis, where areas such as the palms are labelled by specific keratins [13].
It is impossible to assign homology of the krtt1c19e gene to a mammalian orthologue, as most zebrafish keratin proteins are more closely related to each other than any are to mammalian homologues. This precludes definitive orthology assignment based on phylogenetics, and may be due to an origin from tandem duplications or through intra-specific sequence convergence [15]. Nonetheless, the krtt1c19e promoter provides a useful means to drive strong expression in basal keratinocytes, allowing and precise mis-expression of genes in this stem cell type. Whilst this promoter presents a useful reagent, we suspect it may not be complete as there is a delay in observable eGFP levels compared to krtt1c19e in situ signals, and fortunately we do not observe maternal contribution of egfp which would be expected for a complete promoter. Nonetheless, it will assist imaging of basal keratinocyte morphology in real-time in a vertebrate model. The tools we present will assist the growing number of researchers using the zebrafish system to study epidermal development.