Regulation of Epidermal Differentiation by a Distal-less
Homeodomain Gene
Maria I. Morasso,* Nedialka G. Markova,* and Thomas D. Sargent*
Laboratory of Molecular Genetics, National Institute of Child Health and Human Development and *Laboratory of Skin
Biology, National Institute of Arthritis and Musculoskeletal and Skin Diseases, National Institutes of Health, Bethesda,
Maryland 20892
Abstract. The Distal-less-related homeodomain gene
Dlx3 is expressed in terminally differentiated murine
epidermal cells. Ectopic expression of this gene in the
basal cell layer of transgenic skin results in a severely
abnormal epidermal phenotype and leads to perinatal
lethality. The basal cells of affected mice ceased to proliferate, and expressed the profilaggrin and loricrin
genes which are normally transcribed only in the latest
stages of epidermal differentiation. All suprabasal cell
types were diminished and the stratum corneum was reduced to a single layer. These data indicate that Dlx3
misexpression results in transformation of basal cells
into more differentiated keratinocytes, suggesting that
this homeoprotein is an important regulator of epidermal differentiation.
PIDERMALdifferentiation is a multistage process during which basal keratinocytes cease to proliferate
and progressively differentiate as they migrate
through the spinous, granular, and cornified layers of the
epidermis (Watt, 1989; Fuchs and Byrne, 1994). During this
process the cells express specific structural and enzymatic
markers that are characteristic and obligatory features of
each successive differentiation stage. These markers include the keratin proteins of the intermediate filaments
(Steinert and Freedberg, 1991), the intermediate filament-associated protein filaggrin (Dale et al., 1993), and the cell
envelope proteins involucrin (Rice and Green, 1979) and
loricrin (Hohl and Roop, 1993). Ultimately, activation of
keratinocyte-specific transglutaminases leads to a covalent
cross-linking of the cornified cell envelope proteins and the
formation of the protective stratum corneum (Polakowska
and Goldsmith, 1991). The cornified cells, or squames, are
the primary barrier protecting the vertebrate body from the
environment, and are continually shed, to be replaced by
new squames from lower strata. Thus, the epidermis arises
from a population of cells that undergo a regular program
of differentiation throughout the life of the organism.
The regulation of epidermal specific genes appears to be
achieved primarily at the level of RNA synthesis. The
transcription of the basal keratins K5 and K14 is markedly
reduced in the spinous layer, coinciding with induction of
the suprabasal keratins K1 and K10. Subsequently, there is
decreased expression of keratins K1 and K10 coupled to
Address all correspondance to Maria I. Morasso, National Institutes of
Health, National Institute of Child Health and Human Development,
Laboratory of Molecular Genetics, Building 6B, Room 412, Bethesda,
MD 20892. Tel.: (301)496-0369. Fax: (301) 496-0243.
the transcriptional activation of the late differentiation
markers profilaggrin and loricrin (Fuchs and Byrne, 1994).
Several transcription factors have been implicated in the
control of specific structural genes expressed in keratinocytes. For some of these, such as AP1, AP2, and KRF-1,
binding sites in the regulatory regions of target genes have
been identified (Blumenberg, 1993). In addition, other
transcription factors have been shown to be present in the
epidermis, including the POU familygenes Sknla/i, EPOC-1
and Oct6, and the putative zinc finger factor basonuclin
(Andersen et al., 1993;Yukawa et al., 1993;Faus et al., 1994;
Tseng and Green, 1992). Some of these have been found to
affect gene expression in cultured epithelial cells (Andersen et al., 1993; Faus et al., 1994).
Homeodomain genes, particularly those located in the
homeotic complex (HOM-C) of Drosophila melanogaster,
and their vertebrate orthologues, the Hox genes (McGinnis and Krumlauf, 1992), have been shown to be important
in regulating embryonic pattern formation. These genes
may play a role in skin development, particularly in its spatially differentiated aspects, such as dorsal/ventral differences or hair patterns. In fact, some of the vertebrate Hox
genes are transcribed at significant levels in developing
skin (Detmer et al., 1993; Mathews et al., 1993), although
the cell type specificity has not been determined for most
of these. Homeodomain genes that are not part of the
HOM/Hox complexes might also be expected to play a role
in skin patterning and cytodifferentiationof epidermis. An
example of such a homeodomain gene that is expressed in
epidermis is Distal-less 3 (Dlx3).I
1. Abbreviations used in this paper: Dlx3, Distal-less 3; PCNA, proliferating cell-specific nuclear antigen.
(c) The Rockefeller University Press, 0021-9525/96/12/1879/9 $2.00
The Journal of Cell Biology, Volume 135, Number 6, Part 2, December 1996 1879-1887 1879
Distal-lesswas originally identified in Drosophila, where
it is important in patterning the legs and sensory appendages (Cohen et al., 1989). The murine Distal-less family
comprises six individual members, Dlxl-6 ( Robinson et al.,
1991; Simeone et al., 1994; Boncinelli, 1994) and interestingly, Dlx3 differs from the other members of this family
because it is not detected in the central nervous system
(Boncinelli, 1994). In mouse skin, Dlx3 is expressed in the
suprabasal cells of the interfollicular epidermis and the
matrix cells of the hair follicles (Robinson and Mahon,
1994; Morasso et al., 1994). To gain insight into the possible role of the Dlx3 gene in skin development, we have ectopically expressed the Xenopus Dlx3 orthologue (XDlx3)
in mouse basal keratinocytes under the control of a human
keratin K5 promoter (Ohtsuki et al., 1992; Byrne and
Fuchs, 1993). Ectopic expression of Dlx3 resulted in profound morphological and biochemical changes in the epidermis. The superficial epidermal phenotype of affected
animals was shiny, sticky, and wet. At the histological
level, basal cell proliferation was inhibited, and suprabasal
cell layers were dramatically reduced. Late differentiation
markers loricrin and profilaggrin were severely altered:
expression was both reduced in granular cells and ectopically activated in basal cells. Mobility shift and footprinting experiments revealed the presence of a binding site for
Dtx3 in the proximal promoter region of the human profilaggrin gene. These results suggest an important role for
Dlx3 in the process of skin differentiation, which could be
exerted in part by a direct interaction between Dlx3 and
regulatory elements of structural genes such as profilaggrin.
Materials and Methods
Transgenic Constructs and Mice
The K5/XDlx3 construct was generated by subcloning a 905-bp K5 promoter sequence (Ohtsuki et al., 1992; Byrne and Fuchs, 1993) upstream of
the coding sequence of the XDlx3 gene (Dirksen et al., 1994). Downstream of the termination codon, the construct had a segment of the human Bglobin gene which included the second intron with splice donor and
acceptor sites and the polyadenylation sequence. Transgenic mice were
generated according to Hogan et al. (1986), and were isolated by cesarean
removal at E18.5. Transgenic fetuses and littermates were fixed in 4%
paraformaldehyde and skin samples were dissected from the left abdominal area. These samples were embedded in paraffin and sectioned (10 I~M).
Immunocytochemistry
Reactions were done with sections of abdominal skin with polyclonal
rabbit anti-mouse antibodies against keratins 5, 14, 1, and 10, loricrin, fll-
aggrin; kindly provided by Stuart Yuspa (National Cancer Institute, Bethesda, MD). The secondary antibody was an alkaline phosphatase-eonju-
gated anti-rabbit IgG from Boehringer Mannheim (Indianapolis, IN), and
was developed with the BM-purple alkaline phosphatase substrate also
from Boehringer Mannheim. Staining of sections with antibody for the
proliferating cell nuclear antigen were carried out according to Hansen
and Tennant (1994).
In Situ Hybridization
Sense and antisense RNAs were generated for XDtx3 and mouse loricrin
and filaggrin. The construct for RNA synthesis of XDlx3 included the
complete coding sequence for the gene (Dirksen et al., 1994). The construct for loricrin was made by subcloning a 0.62-kb PCR fragment spanning from amino acid 349 to the polyadenylation signal (Mehrel et al.,
1990) into pCRII (Invitrogen, San Diego, CA). The filaggrin construct, a
0.3-kb fragment from the coding region, was provided by Dr. S. Yuspa.
Digoxigenin-labeledcRNA probes were synthesized in vitro from the corresponding templates for XDlx3, profilaggrin and loricrin, and used for
hybridization in situ to sections of nontransgenic littermate and transgenic
skin according to Schaeren-Wiemers and Gerfin-Moser (1993).
For Dlx3, a 1.1-kb cDNA insert (Robinson and Mahon, 1994) was transcribed to generate a [35S]UTP-labeled probe to be used on skin from 2-d-old
mice, the slides were coated with emulsion and developed after 5 d exposure (Molecular Histology, Inc., Gaithersburg, MD).
Cornified Envelope Preparation
Cell envelopes were prepared according to Hohl et al. (1991). Briefly, the
epidermis was obtained by heat separation from skin samples of normal
and affected littermates. The epidermis was stirred gently in extraction
buffer containing 100 mM Tris, pH 8.5, 2% SDS, 20 mM D'IW and 5 mM
EDTA, and boiled for 10 rain at 96C. The cornified envelopes were collected by precipitation at 14,000 rpm for 10 min and reextracted once the
same way. After resuspension in the extraction buffer the cornified envelopes were visualized under a phase contrast microscope.
Electron Microscopy
Routine transmission electron microscopy analysis was performed by Advanced Biotechnologies Inc. (Columbia, MD). Briefly, skin samples were
fixed in glutaraldehyde, postfixed in osmium tetroxide, stained with uranyl acetate, dehydrated, and embedded in Spurr's plastic resin. The blocks
were ultrathin sectioned at a thickness of 60--80 rim, were poststained with
lead citrate, and examined under the transmission electron microscope.
Figure 1. In situ hybridization of Dlx3 probe to neonatal (2-d) mouse epidermis. Transcripts were detected in the granular cell layer and
ia the hair follicle matrix cells. (A) Bright field with eosin-hematoxyfin staining. (B) Dark field view of the same section, c, cornified; g,
granular; s, spinous; b, basal layers. Bar, 20 p,m.
The loumal of Cell Biology, Volume 135, 1996 1880
Mobility Shift Assays
A recombinant XDIx3 protein was generated by expressing the complete
coding sequence for XDlx3 (834 bp), cloned in the EcoRI site of the expression vector pET-28a (Novagen, Madison, WI). Recombinant XDIx3
was purified according to the manufacturer's instructions. Oligonucleotide
Fla sequence: CCCTAGGCITCATTATCTCTI'CGAATCCC (profilaggrin promoter sequence underlined); mutated oligonucleotide Fla: CCC-
TAGGCTI'CACTCACAGTTCGAATCCC; TFIID consensus oligonu-
deofide: GCAGC~ATATAAGGTGAGGTAGGA. POU domain consensus
oligonucleotide: CTAGAGGATCCATGCAAATGGATCCCCGGGT-
ACCGAGCTC. Binding assays were performed with 1-2 I.Lgof XDIx3 recombinant protein and 4 x 104 cpm (~1 ng) of gel-purified 5' end-labeled
double-stranded DNA fragments. The reactions were carried out in 20 ~l
containing 10 mM Tris (pH 7.5), 65 mM NaC1, 5 mM DTT, 5 mM
MgCI2, 0.05% NP-40, 10% glycerol, 1 mg/ml BSA, and 25 ~g/ml poly (dI-dC) as a carrier for 30 min at 4C. In competition experiments a 100-fold
molar excess of the cold competitor was preincubated with the extracts for
30 rain at 4C before adding the labeled DNA fragment. The DNA-protein complexes were resolved on 6% polyacrylamide gels in 0.5x TBE
buffer. Gels were dried and exposed overnight at room temperature.
DNase 1 Footprint
Footprints were done on the proximal promoter region of filaggrin with
DNaseI/SureTrack Footprinting kit from Pharmacia (Uppsala, Sweden).
The DNA probes were amplified by PCR using a 5' end-labeled oligonucleotide and wild-type or mutated profilaggrin promoter constructs as
templates. DNA was gel purified and 1.5 x 104 cpm were used per reaction and treated according to the manufacturer's instructions. The same
oligonucleotides were used to amplify a profilaggrin sequence with mutations in the Dlx3 binding site.
Results
Localization of Dlx3 Expression to Granular Cells
It has been shown previously that Dlx3 is actively expressed in epidermis (Robinson and Mahon, 1994). Furthermore, this expression is clearly confined to the suprabasal compartment of this tissue, as shown by Northern
blot analysis of fractionated epidermis (Morasso et al.,
1994). To further sublocalize the expression of Dlx3, in
situ hybridization of 35S-labeled antisense probe to a section of abdominal 2-d mouse epidermis was carried out
(Fig. 1). The highest expression of Dlx3 was detected in
the upper spinous and granular layers, although limited
expression in lower spinous cell layers cannot be ruled out.
No specific hybridization was detected in parallel sections
using sense probe (data not shown).
Targeted Misexpression of Dlx3 to Basal Cells
The restriction of Dlx3 expression to the suprabasal layers,
predominantly to the granular cells, raised the possibility
that this homeodomain protein might be involved in the
regulation of the later stages of epidermal differentiation.
To test this hypothesis, we targeted Dlx3 transcription to
the basal keratinocytes of mouse skin by expressing the
Xenopus orthologue of Dlx3 under the control of a 905-bp
5' flanking element from the human keratin K5 gene (Fig.
2 A). In transfection experiments, Ohtsuki et al. (1992) have
demonstrated that this 905-bp element was preferentially
active in epithelial ceils. In transgenic studies, Byrne and
Fuchs (1993) showed that a 6.5-kb human K5 region encompassing the 905-bp sequence was faithfully expressed
in mouse epidermis. As shown in Fig. 2 C, the 905-bp K5
fragment was uniformly active only in the basal cell layer,
Figure2. Transgene construct, expression, and gross phenotype.
(A) K5-XDlx3 transgenic construct. Hybridization of sense (B)
and antisense (C) XDIx3 probe to abdominal skin sections(10 IxM)
of affected transgenic E18.5 embryos. Expression was confined
to the basal layer. (D) Appearance of severely affected K5/XDlx3
transgenicembryo (left)and normal littermate (right).Bar, 40 i~m.
indicating that this region contains the necessary elements
to restrict the epidermal expression of the Xenopus Dlx3
gene to the basal cells in transgenic animals. Hybridization
was also detected in hair follicles (data not shown). The
Morasso et al. Regulation ofEpidermal Differentiation by Dlx3 1881
Figure3. Histology and ultrastructure of normal and transgenic E18.5 epidermis. (A and B) Hematoxylin staining of normal (A) and affected (B) skin. The thickness of the differentiated layers in the transgenic skin was substantially diminished when compared to normal skin. In addition, basal layer cells appeared flattened, and many nucleated cells were seen in the upper strata of affected skin (arrows). (C and D) Electron microscopy of normal (C) and affected transgenic (D) epidermis, showing a dramatic reduction of the stratum
corneum in the affected skin. The affected stratum corneum appeared to be partially separated from underlying cells, which were highly
vacuolated. (E and F) Preparations of cornified cell envelopes isolated from control (E) and transgenic (F) skin. Arrow in E indicates a
typical cornified envelope. In F, the arrow indicates insoluble material isolated from affected skin. V, vacuoles; S, separation of stratum
corneum, c, cornified; g, granular; s, spinous; b, basal layers. In D, granular and spinous cell morphology is abnormal (dashedline).Bars:
(A) 20 I~m;(O) 1.5 I~M.
Xenopus orthologue of Dlx3 was originally named Xdll2
(Dirksen et al., 1994; Papalopulu and Kintner, 1994), but
to avoid confusion we will refer to this gene as XDIx3.
XDlx3 and murine Dlx3 have identical homeodomains,
except for two conservative substitutions in helix 1. All of
the nonhomeodomain homology elements are highly conserved between the two genes. Furthermore, the expression patterns of XDIx3 and Dlx3 are very similar in frog
and mouse (Robinson and Mahon, 1994; Dirksen et al.,
1994), and as we have reported recently, the regulatory elements of the Xenopus gene function in transgenic mice to
produce an expression pattern essentially identical to that
of the endogenous mouse Dlx3 (Morasso et al., 1995).
Phenotype of Transgenic Mice
Transgenic mice expressing the K5/XDlx3 construct tended
to be smaller, but did not display obvious body or limb deformities, and were of similar developmental stage compared to the unaffected siblings, as judged from anatomical
features such as digit morphology (data not shown). Affected mice could be immediately recognized by highly abnormal skin, which was partially transparent, shiny, sticky,
and wet suggesting a disturbed barrier function (Fig. 2 D).
The phenotype was variable in severity, and moderately
affected embryos showed a more pronounced phenotype
in the ventral epidermis than on the dorsal side. Severely
TheJournalofCellBiology,Volume135,1996 1882
Figure 4. Immunocytochemistry of
epidermal markers in E18.5 abdominal skin from normal and transgenic
mice. Sections from normal littermate (A-F) and transgenic (G-L)
mice were stained with anti-keratin 5
antibody (K5; A and G), anti keratin
14 antibody (K14; B and H), anti-keratin 1 antibody (K1; C and /), anti-keratin 10 antibody (K10; D and J),
anti-filaggrin antibody (FIL; E and
K) and anti-loricrin antibody (LOR;
F and L). The staining of keratins
was relatively unchanged with respect to localization and intensity
(A-D, G-J), whereas staining for ill-
aggrin (E and K) and loricrin(Fand L)
was markedly reduced in the granular
layer of the affected compared to the
nontransgenic epidermis. Bar, 40 txm.
Morassoet al.RegulationofEpidermalDifferentiationbyDlx3 1883
Figure5. Detection of profilaggrin and loricrin transcripts. Sections of normal (A and B) and affected transgenic (C and D) epidermis were hybridized in situ with digoxigenin-labeled antisense
probes for profilaggrin (A and C) and loricrin (B and D). Ectopic
expression of both genes was visible in the basal cells of affected
epidermis (C and D). Bar, 40 ~m.
affected neonates were immediately cannibalized by the
mothers. To circumvent this, pups were delivered by cesarean section at about E18.5. Attempts to rescue affected
animals with foster mothers were not successful. From a
total of 96 pups examined, eight exhibiting a clear epidermal phenotype were studied. The data presented in this
paper were derived from severely affected animals.
Histology
Sections of E18.5 abdominal skin, stained with hematoxylin (Fig. 3), clearly revealed a pronounced reduction in
skin thickness. Instead of the typical columnar appearance
in normal epidermis (Fig. 3 A), the basal epidermal layer
of affected skin was disorganized and most of the cells appeared flattened (Fig. 3 B). All suprabasal strata were substantially diminished, and there was a noticeable increase
of nucleated (parakeratotic) cells in the upper layers. The
number of hair follicles was reduced from 5.1 _ 0.4 (75 total) to 1.2 ___0.3 (26 total) follicles/mm in sections of normal vs affected epidermis. There was no evidence for lymphocyte infiltration or other inflammatory responses in
the affected epidermis or underlying dermis. In most sections, the dermal layer also appeared to be diminished
compared to controls, resulting in an overall thinning of
the skin.
Fig. 3, C and D show electron micrographs of ultrathin
abdominal skin sections from normal and affected littermates, respectively. The most striking abnormality induced by XDlx3 misexpression was in the stratum corneum, which was reduced from the multilayered structure
of normal epidermis to a single layer. This abnormal cornifled layer appeared to be partially detached from the underlying cells, which were highly vacuolated. Cornifled envelopes can be isolated from skin by extraction with SDS
at high temperature (Hohl et al., 1991). Such an extraction
Figure 6. Proliferating cell nuclear antigen (PCNA) in basal
keratinocytes. Sections of abdominal skin from (A) control littermate and (B) affected transgenic E18.5 mice stained with a
monoclonal antibody recognizing PCNA (see Materials and
Methods). Staining was clearly visible in the basal layer of control
epidermis, but was weaker and present in fewer basal cells (open
triangles)in the affected skin. PCNA-positive cells were also detected in the suprabasal strata of the transgenic epidermis (closed
triangle).Bar, 20 I~m.
of affected skin resulted only in the isolation of a disorganized insoluble material containing few if any cornified envelopes (Fig. 3, E and F).
Expression of Epidermal Genes
Immunocytochemistry and in situ hybridization were used
to evaluate the expression of epidermal markers in the
skin of affected mice and normal siblings. Basal cell-specific keratins K5 and K14 were present at similar levels in
transgenic and normal skin, indicating that the basal cells
continued to express the major structural proteins characteristic of this layer (Fig. 4, A and B, G and H). The suprabasal keratins K1 and K10, were also correctly expressed
at high levels. For all keratins tested, expression was confined to a narrower zone in the transgenic skin, probably
reflecting the general decrease in the number of suprabasal layers. In contrast, the expression of the two major
markers for granular cells, filaggrin and loricrin, was significantly disturbed in transgenic epidermis. A patchy and
relatively weak staining throughout the epidermis was observed with both filaggrin and loricrin-speeific antibodies
(Fig. 4, E, F, K, and L). 'This indicated that misexpression
of XDlx3 resulted in both downregulation of filaggrin and
loricrin in granular layer cells and ectopic synthesis of these
TheJournalofCellBiology,Volume135,1996 1884
proteins in lower strata. To determine whether these effects were due to deregulated transcription, in situ hybridization was performed. The hybridization of antisense
profilaggrin and loricrin RNA probes is shown in Fig. 5.
Comparison of these results revealed that while profilaggrin (Fig. 5 A) and loricrin (Fig. 5 B) transcripts were uniformly and abundantly expressed in the granular cells of
unaffected siblings, only a few isolated cells expressing
these genes were detected in the granular layers of the
transgenic skin (Fig. 5, C and D). However, profilaggrin
and loricrin transcripts were expressed in the basal cells
and in the lowest spinous layer.
In normal epidermis, cell division is primarily confined
to the basal layer (Watt, 1989; Fuchs and Byrne, 1994). Although the transgenic basal cells expressed the K5 and
K14 markers, the perturbance of the typical basal layer
configuration and the flattened morphology suggested
that these cells might have acquired a more differentiated
character, one aspect of which is the cessation of proliferation. To evaluate this, sections were stained with an antibody recognizing the proliferating cell-specific nuclear antigen (PCNA; Hansen and Tennant, 1994). This antigen is
most abundant at S phase, although it is also found in
other phases of the cell cycle (Kurki et al., 1986). As
shown in Fig. 6, this marker was detected at much higher
levels in nontransgenic basal cells compared to the epidermis of affected mice, indicating that XDlx3 misexpression
correlated with the suppression of basal cell proliferation.
Interestingly, PCNA staining in the dermis of affected skin
was also drastically reduced compared to controls.
Interaction of XDlx3 with the Profilaggrin
Promoter Region
In normal epidermis, the profilaggrin and loricrin genes
become activated in the granular layer, in the same cells in
which Dlx3 is maximally expressed. In view of this correlation, one possible explanation for the ectopic activation of
profilaggrin and loricrin in basal cells forced to produce
the XDlx3 protein could be that one or both genes were
directly regulated by this homeoprotein. To investigate
this possibility, bacterially expressed XDlx3 protein was
used in mobility shift and footprinting studies with the
proximal promoter of the profilaggrin gene. This region
has been shown to be sufficient to confer high levels of keratinocyte-specific transcription to a reporter gene in cultured cells (Jang et al., 1996). A series of overlapping oligonucleotides spanning the 120-bp proximal promoter of
the profilaggrin gene (Fig. 7 A) was tested for binding to
recombinant XDlx3 homeoprotein.
From these probes (indicated in Fig. 7 A), the oligonucleotide F1, which contains an AT-rich potential homeodomain-binding motif (Gehring et al., 1994), was found to
produce two complexes (data not shown). A smaller oligonucleotide (Fla) containing only this putative binding site,
yielded an identical mobility shift pattern (Fig. 7 B). Both of
the shifted bands (lane 2, arrowheads) were completely
eliminated by competition with unlabeled probe (lane 3).
Competition was not observed with oligonucleotides in
which the AT-rich core nucleotides had been mutated
(lane 4) or with a consensus binding site for POU domain
proteins (lane 5)or an unrelated AT-rich sequence, the conFigure 7. Detection of a XDlx3 binding site in the profilaggrin
proximal promoter. (A) Sequence of the proximal promoter region of the profilaggrin gene. The specific oligonucleotides used
in mobility shift assays (F1-F5, Fla, E) are indicated by the arrowed lines below the sequence. Functional transcription factor
binding sites are indicated, including the region protected from
DNase I by binding to XDIx3. (B) Mobility shift analysis. 1-2 txg
of recombinant XDlx3 were combined with end-labeled double-stranded oligonucleotide under the conditions described in Materials and Methods. Lane 1, oligonucleotide probe Fla alone.
Lane 2, probe Fla with recombinant XDIx3 protein. Lanes 3--6,
competition with a 100-foldmolar excess of unlabeled oligonucleotide Fla, mutated oligonucleotide Fla, Oct consensus sequence
oligonucleotide, and TFIID consensus oligonucleotide, respectively. P, probe. Shifted complexes are indicated by triangles. (C)
DNasel footprinting assay. An end-labeled DNA fragment encompassing the profilaggrin gene region between -150 and +47
was incubated with recombinant XDlx3 and treated with DNasel.
Lane 1, G+A ladder. Lane 2, no XDlx3 protein added. Lane 3,
XDlx3 footprint. Lane 4, competition with a 100-fold molar excess of oligonucleotide Fla. Lane 5, footprinting reaction carried
out with a profilaggrin probe in which the AT-rich XDIx3 binding
site was mutated (see Materials and Methods). The protected
bases are indicated by the vertical bar, and a hypersensitive site
by an asterisk. Position of profilaggrin gene base pairs +1 and
-75, corresponding to A, are indicated to the left.
Morassoetal.RegulationofEpidermalDifferentiationbyDlx3 1885
sensus TFIID motif (lane 6). These results were confirmed
by DNaseI footprint analysis, and the results are shown in
Fig. 7 C. A single protected region was detected that coincided with the sequence of the probe used in the bandshift
analysis (CATTATCTC; lane 3). The protection was abolished when the binding assay was done in the presence of a
100-fold excess of unlabeled oligonucleotide Fla (lane 4),
or when mutations identical to those used in Fig. 7 B were
introduced in the site (lane 5).
Discussion
In this study we show that the misexpression of the suprabasal cell-specific homeobox gene XDlx3 in the basal cells
of transgenic mice has two important effects. First, the targeted basal keratinocytes are transformed into a more differentiated phenotype. Second, the overall structure of the
epidermis, particularly of the more differentiated strata, is
severely disrupted. As a result of ectopically expressing
XDlx3 in basal cells, the granular cell phenotype is prematurely initiated, as judged by the cessation of cell division
and the activation of the genes encoding the major structural proteins characteristic of this cell type, filaggrin and
loricrin. We show that the proximal promoter region of
the profilaggrin gene includes a binding site for XDlx3,
suggesting a direct interaction between XDIx3 and this
gene. These findings strongly support the conclusion that
Dlx3 functions in vivo to regulate the differentiation of
epidermal ceils.
It is instructive to compare the XDIx3 misexpression
phenotype to that of epidermis in which other regulators
of cell growth and differentiation have been altered. The
basal-cell targeted expression of lymphoid enhancer factor 1
affects hair follicle and tooth formation, but does not otherwise disrupt the epidermal differentiation or proliferation programs (Zhou et al., 1995). Another example is the
overexpression of cyclin D1 using a bovine keratin K5 promoter, which results in hyperproliferation, but does not interfere with the differentiation of stratified epithelia (Robles et al., 1996).
Especially interesting is the comparison of the XDlx3
misexpression phenotype to that in which retinoid signaling has been disrupted by dominant negative retinoic acid
receptors driven by K14 or K1 promoters. Targeting such
molecules to basal cells resulted in thinning of the epidermis, delay in hair follicle formation, and virtual elimination of filaggrin gene expression (Saitou et al., 1995).
Expression of a similar dominant negative retinoic acid receptor in suprabasal cells resulted in abnormal lipid deposition and concomitant disruption of the epidermal barrier
function, but did not lead to alterations in the cell type
specificity of gene expression (Imakado et al., 1995). Thus,
retinoid signaling appears to function in the coordination of
epidermal development, whereas DIx3 seems to more directly
control the spectrum of gene expression in suprabasal cells.
How does misexpression of XDlx3 in basal keratinocytes result in the pathological phenotype of the epidermis? Affected animals had reduced levels of filaggrin and
loricrin in the granular cells. These are the two major proteins of the differentiated keratinocyte, and are thought to
play important roles in the alignment of keratin filaments
and formation of the moisture barrier provided by the
cornified cell envelope (Dale et al., 1978; Mehrel et al.,
1990). Judging from the phenotype observed when these
genes are down-regulated in skin disorders, it is unlikely
that the reduced levels of these proteins alone can account
for the morphology of affected skin. For example, congenital genodermatosis icthyosis vulgaris is accompanied by a
drastic reduction in the number of keratohyalin granules,
and in many patients no profilaggrin mRNA or protein
can be detected (Nirunsuksiri et al., 1995). Histologically
the disease is characterized by hyperkeratosis rather than
reduction of the cornified layers and the symptoms are
very mild compared to the disruption of epidermis resulting from XDlx3 misexpression. Therefore the severe skin
phenotype in the XDlx3 transgenic mice probably has a
more complex explanation.
One possibility is that affected basal cells are physically
disrupted by accumulation of ectopic profilaggrin and/or
loricrin, leading to secondary defects. Support for this
interpretation comes from observations indicating that
inappropriate expression of epidermal structural proteins
interferes with the normal function of the keratinocyte.
For example, Carroll et al. (1995) have reported that prolonging integrin gene expression beyond the basal-suprabasal transition, by forcing ectopic expression via an involucrin gene promoter, results in an abnormal epidermal
phenotype resembling psoriasis. Also, inappropriate expression of suprabasal keratins K1 and K10 interferes with
the proliferation of keratinocyte-derived tumor cell lines,
suggesting that the incorporation of these proteins into the
intermediate filament cytoskeleton is incompatible with mitosis, a key feature of cells in the basal compartment (Kartasova et al., 1992). Thus, it is possible that premature expression of the profilaggrin gene might itself be the basis
for much of the observed epidermal dysplasia. Forced expression of filaggrin in cultured cells can lead to collapse
of the intermediate filament network, disruption of the nuclear membrane, vacuolization, and eventually cell death
(Dale et al., 1997; L.-G. Kim, L.-S. Park, and P.M. Steinert,
personal communication). The primary effect of XDlx3
could be limited to activation of profilaggrin expression in
basal cells, while the other features of affected skin are
secondary consequences of the presence of this protein.
A second alternative to account for the skin abnormalities is that misexpression of XDlx3 could result in a general and premature activation of keratinocyte differentiation, including suppression of cell division which reduces
the supply of cells entering the suprabasal compartment.
Premature initiation of the granular cell differentiation
program at the basal cell level could make it impossible to
properly organize the structural elements required for cornifled cell formation. Differentiation could also be fundamentally incompatible with basal cell proliferation, reducing
the supply of cells capable of migration into the suprabasal
strata. One argument in favor of this interpretation comes
from the fact (Fig. 6) that little PCNA staining was observed in the basal layer of affected skin.
The differentiation process by which the epidermal keratinocyte is transformed from a columnar mitotically active basal cell into the flattened anucleated cornified cell
has only begun to be understood. The data presented in
this paper show that Dlx3 plays an important role in mediating this program.
The Journal of Cell Biology, Volume 135, t996 1886
We thank Drs. J. Compton, A. Dlugosz, J. Franklin, C. Nocente, P. Steinert, and S. Yuspa for helpful discussions and invaluable assistance and advice in this work; Dr. S. Yuspa for supplying the epidermal antibodies and
filaggrin construct; and Dr. M. Blumenberg for providing us with the K5
promoter region. We are grateful to Drs. B. Dale, R. Presland, R. Lewis,
P. Fleckman, L.-G. Kim, L.-S. Park, and P. Steinert for sharing their unpublished results.
Received for publication 15 September 1996 and in revised form 29 October 1996.
References
Andersen, B., M.D. Schonemann, S.E. Flynn, R.V. Pearse, H. Singh, and M.G.
Rosenfeld. 1993. Skn-la and Skn-li: two functionally distinct Oct-2-related
factors expressed in epidermis. Science (Wash. DC). 260:78-82.
Blumenberg, M. 1993.Molecular Biology of Human Keratin Genes. In Molecular Biology of the Skin. M. Darmon and M. Blumenberg, editors. Academic
Press, CA. 1-32 pp.
Boncinelli, E. 1994.Early CNS development: Distal-less related genes and forebrain development. Curr. Opin. NeurobioL 4:29-36.
Byrne, C., and E. Fuchs. 1993. Probing keratinocyte and differentiation specificity of the human K5 promoter in vitro and in transgenic mice. Mot. Cell.
BioL 13:3176-3190.
Carroll, J.M., M.R. Romero, and F.M. Watt. I995. SuprabasaI integrin expression in the epidermis of transgenic mice results in developmental defects and
a phenotype resembling psoriasis. Cell 83:957-968.
Cohen, S.M., G. Br6nner, F. Ktittner, G. Jtirgens, and H. J~ickle. 1989.Distal-less encodes a homeodomain protein required for limb development in
Drosophila. Nature (Lond.). 338:432-434.
Dale, B.A.. K.A. Holbrook, and P.M. Steinert, 1978. Assembly of stratum corneum basic protein and keratin filaments in macrofibrils. Nature (Lond.).
276:729-731.
Dale, B.A.. R.B. Presland, P. Fleckman, E. Kam, and K. Resing. 1993. Phenotypic expression and processing of filaggrin in epidermal differentiation. In
Molecular Biology of the Skin. M. Darmon and M. Blumemberg, editors.
Academic Press. 79-106 pp.
Dale, B.A., R.B. Pressland, S.P. Lewis,R.A. Underwood, and P. Fleckman. 1997.
Transient expression of epidermal filaggrin in cultured cells causes collapse
of intermediate filament networks with alteration of cell shape and nuclear
integrity. J. Invest. Dermatol. In press.
Detmer, K., H.J. Lawrence, and C. Largman. 1993. Expression of class I homeobox genes in fetal and adult murine skin.Z Invest. Dermatol. 101:517-522.
Dirksen, M.-L., M.I. Morasso, T.D. Sargent, and M. Jamrich. 1994. Differential
expression of a Distal-less homeoboxgene Xdll-2 in ectodermal cell lineages.
Mech. Dev. 46:63-70.
Faus, O., H. Hsu, and E. Fuchs. 1994. Oct-6: a regulator of keratinocyte gene
expression in stratified squamousepithelia. Mot. Cell. Biol. 14:3263-3275.
Fuchs, E., and C. Byrne. 1994.The epidermis: rising to the surface. Curt Opin.
Genet. Dev. 4:725-736.
Gehring, W.J., M. Affolter, and T. Btirglin. 1994. Homeodomain proteins.
Annu. Rev. Biochem. 63:487-526.
Hansen, L.A., and R. Tennant. 1994. Focal transgene expression associated
with papilloma development in v-Ha-ras transgenic TG.AC mice. Mot. Carcinog. 9:143-154.
Hogan, B., F. Costantini, and E. Lacy. 1986. Manipulating the Mouse Embryo:
a Laboratory Manual. Cold Spring Harbor Laboratory Press, Plainview,NY.
Hohl, D., T. Mehrel, U. Lichti, M.L.Turner, D.R. Roop, and P.S. Steinert. 1991.
Characterization of human loricrin. Structure and function of a new class of
epidermal cell envelope proteins..L Biol. Chem. 266:6626--6636.
Hohl, D., D. Roop, and D. Loricrin. 1993.In Molecular Biology of the Skin. M.
Darmon and M. Blumemberg, editors. Academic Press. 151-179 pp.
Imakado, S., J.R. Bickenbach, D.S. Bundman, J.A. Rothnagel, P.S. Attar, X.-J.
Wang, V.R. Walczak, S. Wisniewski,J. Pote, J.S. Gordon, et al. 1995.Targeting expression of a dominant-negative retinoic acid receptor mutant in the
epidermis of transgenic mice results in lossof barrier function. Genes & Dev.
9:317-329.
Jang, S.-I., P. Steinert, and N. Markova. 1996. AP1 activity regulates the specificity of expression from the proximal promoter of the human profilaggrin
gene in cultured keratinocytes. J. BioL Chem. 271:24105-24114.
Kartasova, T., D.R. Roop, and S.H. Yuspa. 1992. Relationship between the expression of differentiation-specific keratins 1 and 10 and cell proliferation in
epidermal tumors. MoL Carcinogenesis. 6:18--25.
Kurki, P., M. Vanderlaan, F. Dolbeare, G.Gray, and E.M. Tan. 1986. Expression of proliferating cell nuclear antigen (PCNA)/cyclinduring the cell cycle.
Exp. Cell Res. 166:209-219.
Mathews, C.H.E, K. Definer, HJ. Lawrence, and C. Largman. 1993.Expression
of the Hox 2.2 homeobox gene in murine embryonic epidermis. Differentiation. 52:177-184.
Mehrel T., D. Hohl, M. Rothnagel, D. Longley,C. Bundman, C. Cheng, U. Lichti,
M. Bisher, A. Steven, P. Steinert, et al. 1990. Identification of a major keratinocyte cell envelope protein, loricrin. Cell. 61:1103-1112.
McGinnis, W., and R. Krumlauf. 1992. Homeobox genes and axial patterning.
Cell. 68:283-302.
Morasso, M.I., M. Jamrich, and T.D. Sargent. 1994. The homeodomain gene
XenopusDistal-less-like-2 (Xdll-2) is regulated by a conserved mechanism in
amphibian and mammalian epidermis. Dev. BioL 162:267-276.
Morasso, M.I., K.A. Mahon, and T.D. Sargent. 1995. A Xenopus Distal-less
gene in transgenic mice: conserved regulation in distal limb epidermis and
other sites of epithelial-mesenchymal interaction. Proc. Natl. Acad. Sci.
USA. 92:3968-3972.
Nirunsuksiri, W., R.B. Presland, S.G. Brumbaugh, B.A. Dale, and P. Fleckman.
1995. Decreased profilaggrin expression in icthyosis vulgaris is a result of selectively impaired posttranscriptional control. J. BioL Chem. 270:871--876.
Ohtsuki. M.. M. Tomic-Canic, I.M.Freedberg, and M. Blumenberg. 1992. Nuclear proteins involved in transcription of the human K5 keratin gene. J, Invest. DermatoL 99:206-215.
Papalopulu, N., and C. Kintner. 1994. Xenopus Distal-less related homeobox
genes are expressed in the developing forebrain and are induced in planar
signals. Development. 117:961-975.
Polakowska, R.P., and L.A. Goldsmith. 1991. The cell envelope and transglutaminases. In Physiology, Biochemistry and Molecular Biology of the
Skin. L.A. Goldsmith, editor. Oxford University Press, New York, Oxford.
168-204 pp.
Rice, R., and H. Green. 1979. Presence in human epidermal cells of a soluble
protein precursor of the cross-linked envelope: activation of the cross linking by calcium ions. CeIL 18:681~i94.
Robinson, G.W., and K.A. Mahon. 1994. Differential and overlapping expression domains of Dlx-2 and Dlx-3 suggest distinct roles for Distal-less homeobox genes in craniofacial development. Mech. Dev. 48:199-215.
Robinson, G.W., S. Wray, and K.A. Mahon. 1991. Spatially restricted expression of a member of a new family of murine Distal.less homeobox genes in
the developing forebrain. New BioL 3:1183-1194.
Robles, A.I., F. Larcher, R.B. Whalin, R. Murillas, E. Richie, I.B. Gimenez-
Conti, J.L. Jorcano, and CJ. Conti. 1996. Expression of cyclin D1 in epithelial tissues of transgenic mice result in epidermal hyperproliferation and severe thymic hyperplasia. Proc. Natl. Acad, ScL USA. 93:7634-7638.
Saitou, M., S. Sugai, T. Tanaka, K. Shimouchi, E. Fuchs, S. Narumiya, and A.
Kakizuka. 1995. Inhibition of skin development by targeted expression of a
dominant-negative retinoic acid receptor. Nature (Lond.). 374:15%162.
Schaeren-Wiemers, N., and A. Gerfin-Moser. 1993. A single protocol to detect
transcripts of various types and expression levels in neural tissue and cultured cells: in situ hybridization using digoxigenin-labelled cRNA probes.
Histochemistry. 100:431-440.
Simeone, A., D. Acampora, M. Pannese, M. D'Esposito, A. Stornaivolo, M.
Gulisano, A. Mallamaci, K. Kastury, T. Druck, K. Heubner, et al. 1994.
Cloning and characterization of two members of the vertebrate DIx gene
family. Proc. Natl. Acad. ScL USA. 91:2250-2254.
Steinert, P., and I. Freedberg. 1991. Molecular and cellular biology of keratins.
In Physiology, Biochemistry and Molecular Biology of the Skin. L.A. Goldsmith, editor. Oxford University Press, New York, Oxford. 113-147 pp.
Tseng, H., and H. Green. 1992. Basonuclin: a keratinocyte protein with multiple paired zinc fingers. Proc. Natl. Acad. Sci. USA. 89:10311-10315.
Watt, F.M. 1989. Terminal differentiation of epidermal keratinocytes. Curr.
Opin. Cell Biol. 1:110%1115.
Yukawa, K., T. Yasui, A. Yamamoto, H. Shiku, T. Kishimoto, and H. Kikutani
1993. Epoc-l: a POU-domain gene expressed in murine epidermal basal cells
and thymic stromal cells. Gene (Amst.). 133:163-169.
Zhou, P., C. Byrne, J. Jacobs, and E. Fuchs. 1995. Lymphoid enhancer factor 1
directs hair folliclepatterning and epithelial cellfate. Genes & Dev. 9:570583.
Morasso et al. Regulation of Epidermal Differentiation bY Dlx3 1887

