Changing Patterns of Gap J unctional Intercellular
Communication and Connexin Distribution
in Mouse E pidermis and Hair F ollicles
During E mbryonic Development
R . CHOUDHRY,1J .D. PIT T S,2and M.B. HODGINS1*
1Department of Dermatology, University of Glasgow, Glasgow, Scotland
2Beatson Institutefor Cancer Research, GarscubeEstate, Bearsden, Glasgow, Scotland
ABST R ACT
embryonic days 12 (E 12) and 16, regular arrays of
epidermal placodes on the mystacial pad develop
into whisker follicles. T his system was chosen for
analysis of gap junctional intercellular communi-
cation during differentiation. T he patterns of
communication were studied by microinjection
of the tracers L ucifer yellow-CH (LY-CH) and
neurobiotin (NB), while immunofluorescent stain-
ing was used to study distribution of connexins
26 and 43. E xtensive communication was seen
between keratinocytes in developing hair pegs
or, in later-stage hair follicles, in the germinative
matrix. Coupling between adjacent hair pegs via
interfollicular epidermis was not observed. Cou-
pling also became restricted as follicular cells
differentiated to form outer root sheath, inner
root sheath, and hair shaft. E xtensive gap junc-
tional coupling is characteristic of keratinocytes
that are rapidly proliferating (as in hair pegs and
germinative matrix). F ollicular keratinocytes
commence differentiation shortly before restric-
tion of gap junctional coupling becomes evident.
Dermal mesenchymal cells undergoing different
modes of differentiation also exhibit differences
in gap junctional coupling, as evidenced by poor
transfer of LY-CH between cells in dermal conden-
sations of hair follicles compared with extensive
transfer elsewhere in the dermis. LY-CH and NB
were not transferred between epidermal or fol-
licular epithelium and mesenchyme, arguing
against a direct role for gap junctions permeable
to known second messenger molecules or nucleo-
tides in epithelial-mesenchymal interactions in
this system. T he distribution of connexins 26 and
43 in epidermis and hair follicles changed during
differentiation but there was no correlation with
changing patterns of dye transfer, indicating an
unexpected degree of complexity in the relation-
ship between gap junctional intercellular commu-
nication and connexin protein distribution dur-
ing development. Dev. Dyn. 1997;210:417–430.
?1997 Wiley-L iss, Inc.
In the mouse embryo betweenK ey words: gap-junctions; connexins; skin; hair
follicle; epidermis; embryo; develop-
INT R ODUCT ION
Development of vertebrate skin appendages involves
complex cellular interactions within the epithelium
and between the epithelium and mesenchyme (Sengel,
1976; Kratochwil et al., 1996). These interactions in-
volve cell adhesion molecules (Chuong and Edelman,
1985; Hirai et al., 1989; Zhang et al., 1996), extracellu-
lar matrix components (Goetinck and Carlone, 1988;
J iang and Chuong, 1992; Zhang et al., 1996), and
secreted growth factors (Hogan et al., 1994; Finch et al.,
1995; Guo et al,. 1996; TingBerreth and Chuong
1996a,b). Gap junctions provide opportunities for fur-
ther forms of control in tissue differentiation (Pitts et
al., 1988a,b; Katz, 1995; Paul et al., 1995; Bruzzone et
al., 1996; Kumar and Gilula, 1996) and generation of
pattern (Wolpert, 1989).
Gap junctional communication, which is widespread
in animal tissues, exposes cells to a particular form of
environmental influence arising from the fluctuating
concentrations of small ions and molecules within the
cytoplasms of surrounding cells (Lawrence et al., 1978;
Lowenstein, 1979). The rapid movement of the second
messengers and other small ions and molecules down
concentration gradients between cells, generates a form
of homeostasis within groups of cells joined by gap
junctions. In cells in culture, this leads to coordination
and uniformity of phenotype and tends to inhibit the
expression of cellular differences (Pitts et al., 1988a,b).
It has therefore been suggested that the formation of
communication compartments (groups of cells joined to
each other by gap junctions but not to cells in different
compartments) is likely to be associated with the
expression of new phenotypes during morphogenesis
and patterning (Pitts et al., 1988a,b; Kam and Hodgins,
Grant sponsor: bbsrc; Grant number: 17/S03206.
*Correspondence to: M.B. Hodgins, Department of Dermatology,
University of Glasgow, Glasgow G12 8QQ, Scotland.
Received 24 March 1997;Accepted 30August 1997
DE VE L OPME NTAL DYNAMICS 210:417–430 (1997)
?1997 WILEY-LISS, INC.
1992; Serras et al., 1993). Although studies in several
systems have provided general support for these ideas,
thereis very littledetailedinformation relating changes
in gap junctional intercellular communication to spe-
cific stages in tissue morphogenesis. Much recent work
has concentrated on documenting the expression pat-
terns of connexins, a family of gap junction proteins
containing at least 13 distinct types that are expressed
during embryogenesis in complex tissue-specific pat-
terns (Risek et al., 1992, 1994; Goliger and Paul, 1994;
Dahl et al., 1996). Much less is known about the
relationship between theobserved patterns of connexin
staining and the compartmentation of gap junctional
communication. It is often assumed that connexin
staining is sufficient evidence of functional coupling,
but as gapjunctions may adopt open or closedconforma-
tions, this assumption may not bejustified.
The mouse embryo whisker follicle is a particularly
suitable mammalian system for analysis of cellular
interactions involved in pattern formation and differen-
tiation. Interactions between mesenchyme and epithe-
lium and interactions within the epithelium are all
important during organogenesis (Sengel, 1976). Indi-
vidual whisker follicles develop from epidermal plac-
odes established in a simple pattern of rows in the
mystacial pad (Davidson and Hardy, 1952). Epidermal
cells giverisetotheprogenitor cells of thetwomain cell
lineages of the follicle, the outer root sheath and the
germinative matrix. The outer root sheath is appar-
ently continuouswith theepidermis, whereasthegermi-
native matrix in the hair bulb is a population of rapidly
proliferating keratinocytes that gives rise to the six
morphologically distinct cell types of the hair shaft and
inner root sheath during anagen (Hardy, 1992). A
condensation of mesenchymal cells, which interacts
with the epidermal placode to establish the follicle,
eventually forms the dermal sheath of the follicle and
dermal papilla. The dermal papilla, which is enveloped
by the hair bulb, plays an instructive role in the growth
of the whisker (Oliver, 1970; Reynolds and J ahoda,
We have compared patterns of gap junctional cou-
pling, using the fluorescent tracer Lucifer yellow-CH
(LY) and neurobiotin (NB), with the patterns of con-
nexin staining during development of whisker follicles
in the mouse embryo. We show that morphogenesis is
associated with separation of differentiating cells into
new communication compartments and with changes
in expression of connexins 43 and 26. However, func-
tional compartmentation is not explained by the differ-
ential expression of thesetwoconnexins. Our studies do
not support a role for gap junctions in epithelial-
R E SULT S
Development of Hair F ollicles in the Mouse
At embryonic days 12 (E12) to 13, the whisker pad
epidermis is stratified into three cell layers (including
the outer periderm layer) with a pattern of hair pegs at
different stages of differentiation according to position
within the rows (Hardy, 1992). The pegs range from
simple epidermal thickenings associated with underly-
ing condensations of dermal cells (Hardy’s stage 1) to
protrusions of circumferentially arranged epithelial
cell surrounded by densely packed dermal cells (stages
2/3a). By E14, the outer columnar cells of the upper
follicle have the appearance of outer root sheath, while
a well-formed bulb with germinative matrix surrounds
the dermal papilla in some follicles (stages 3b/4). By
E16, the squamous layers of the interfollicular epider-
mis are cornified and several rows of fully formed
whisker follicles are present, some with hair shafts
reaching theskin surface(stages 5/6). However, even at
this stage, some late-developing follicles at the early
stages of differentiation are alsoobserved. Perhaps the
most striking contrast between embryonic and adult
follicles is the higher density of mesenchymal cells in
theembryonic dermal sheath and papilla.
Analysis of J unctional Communication
by Dye Injection
LY-CH (or NB, see below) was injected into cells at
different sites as described in Experimental Proce-
dures. After fixation (and fluorescein-avidin staining
for NB), all dye spreads were analysed either after
embedding and cutting complete sets of serial sections
or by complete sets of optical sections imaged by
confocal microscopy.By following thestructuresthrough
serial sets of sections, it was possible to establish an
accurate morphological description of each section. To
facilitate interpretation, morphological features have
been marked on the selected sections presented in the
J unctional Communication in the
Interfollicular E pidermis
When injected into a basal epidermal cell at E12,
LY-CH spread only to neighbouring basal and su-
prabasal cells (Fig. 1A,B). Dye did not spread into the
periderm (outermost) layer. At E14, after cornification
of the outermost epidermal cells, LY-CH injected intoa
basal or suprabasal cell spread vertically through all
living layers, leaving sharp boundaries at the dermo-
epidermal junction and cornified layer (Fig. 1C,D).
Lateral spread of dye was not restricted by such
morphologically identifiable boundaries but was no
greater, and often less in extent, than the vertical
spread. This is consistent with the previous report of
Kam et al. (1986), which suggested that the epidermis
is divided into many small communication compart-
ments. However, the spreads seen at this stage of skin
development do not suggest specific boundaries, just
limitedlateral coupling.Therestriction in lateral spread
of LY-CH was more pronounced within the basal layer
(Fig. 1C,D). When LY-CH was injected into an epider-
mal cell close to the edge of a developing hair peg
(Hardy stage 2), there was limited transfer with some
CHOUDHRY ET AL.
injection into non-follicular epidermis (e) next to a hair peg (hp) in E12 WP
skin. Note the absence of dye transfer between epidermal keratinocytes
and the periderm (p) and between the epidermis (e) and underlying
dermis (d). (C,D) E14 whisker pad (WP) skin. Vertically unrestricted
LY-CH transfer is shown between all epidermal layers (e) except for the
outermost layer. Dye spread is more restricted laterally in the lower
(A,B) Fluorescence and phase contrast images of an LY dyeepidermal layers. Note the absence of dye transfer into dermal mesen-
chyme (d). A well-developed hair follicle (hf) is also visible. (E,F) Hardy’s
(stage 2) hair peg showing limited transfer of LY-CH between a hair peg
and adjacent epidermis (e) at E12. Note the absence of dye transfer
between epithelial cells and the dermis (d). Dotted lines mark the
boundaries between epidermis and dermis or hair follicle epithelium and
dermis. Scale bar ? 30 µm.
GAP J UNCTIONS AND CONNEXINS
sharp differences in labelling between adjacent cells
Communication Between F ollicular
and Interfollicular E pidermis
and Within Developing Hair Pegs
When LY-CH was injected intoa keratinocyte within
the epidermis overlying a stage 2 hair peg, there was
more extensive spread through the epidermal cells
overlying the peg than into surrounding interfollicular
epidermis (Figs. 2A–F, 8A). At this stage, dye also
spread well down into the cells of the peg itself (Fig.
2A–F). However, dye did not spread laterally into
adjacent hair pegs or into their overlying epidermis.
Further evidence confirming these conclusions was
obtained by analysis of LY-CH and NB spreads using
confocal microscopy (Figs. 5, 8A).
At E12, LY-CH injected into a hair peg showed the
keratinocytes within the peg were very extensively
coupled (Fig. 3A–D). Dye was seen to spread from the
injected cell into as many as 300 or more neighbours.
This extensive region of coupling included uniform,
cuboidal cells in the core and advancing tip of the peg
and more superficial and peripheral cells that were
already showing signs of morphological differentiation
toa morecolumnar shape(Figs. 3C,D, 8).
By E14, in moredevelopedfollicles, transfer of LY-CH
becamerestricted both within theperipheral, columnar
layer and between this and the inner core (Fig. 3E,F).
Dye injected intoa columnar cell was seen tospread to
only 20–25 neighbours, all within theperipheral colum-
nar layer (appearanceof a compartment).
At E14, in the most advanced follicles, keratinocytes
of thegerminativematrix in thenewly formedhair bulb
were extensively coupled (Fig. 3G,H). LY-CH often
spread from the injected cell right around the lower
bulb, particularly through cells immediately adjacent
to the mesenchymal dermal papilla. All lower bulb
keratinocytes werecoupled, including thosecontinuous
with outer root sheath. Coupling only becamerestricted
at the upper edge of the bulb above the proliferation
zone, where cells start to form the shaft and inner root
sheath (Fig. 3G,H). Dye spread was not observed
between germinative matrix and the dermal papilla or
surrounding dermal connectivetissuesheath cells (Fig.
Communication Within the Mesenchyme
Follicular mesenchymal cells, including those of the
dermal papilla, showedmarkedly restricteddyespread,
particularly in the cells layers adjacent tothe follicular
epithelial tissue(Fig. 4A–J ). Moreextensivedyespread
was seen between non-follicular mesenchymal cells at
E16 (Fig. 4I,J ). At no stage did LY-CH spread from
mesenchymal cells tofollicular or non-follicular epider-
Table 1 summarises the results from complete sets of
serial sections of 37 LY-CH injections. Similar results
were obtained from a further 38 injections of LY-CH
intostage 2/3 hair pegs and 16 injections intostage 5/6
follicular bulb keratinocytes (germinative matrix) that
were analysed in whole mounts by confocal microscopy.
Comparison of Dye Spreads Using NB and LY-CH
There is evidence (Penn et al., 1994) that some gap
junctions have channels that are more permeable to
positively charged NB (molecular weight, 287) than
negatively charged LY-CH (molecular weight, 443). We
therefore examined the patterns of communication by
NB injections in situations where LY-CH revealed
potentially important boundaries to dye spread. Injec-
tions (total of five) of developing follicular pegs (stage2)
at E13 showed extensive dye transfer between the
overlying epidermis and the peg. Figure 5B shows a
series of offset optical sections imaged by confocal
microscopy moving down from the epidermis on the left
into the peg on the right (for comparison, an LY-CH
injection has been analysed by thesameprocedure, Fig.
5A). There is only limited transfer from the overlying
epidermal cells to the surrounding epidermis and no
transfer to adjacent pegs (which can be seen in indi-
vidual images but not after overlaying). Again there is
no transfer from the epithelial cells of the peg to the
surrounding dermal structures. Similarly, NB (like
LY-CH, Fig. 3G) injected into a germinative matrix
keratinocyte of a stage 5/6 follicle spread extensively
through thegerminativematrix but not intothedermal
papilla or dermal sheath (data not shown). In all
injections, thepattern of NB spread was very similar to
that of LY-CH.
E xpression of Connexins 26 and 43
in Interfollicular E pidermis and E arly Hair Pegs
At E12, antibodies to connexin 43 stained basal and
suprabasal epidermal layers (Fig. 6A). By E13.5, stain-
ing was more intense in the suprabasal layer than in
thebasal layer, wherestaining was restricted mainly to
apical surfaces of the keratinocytes. Staining of the
lateral surface was variable and appeared to fluctuate
along the length of the sections with a periodicity of
approximately six cells. A similar pattern of staining
was maintained throughout development (Fig. 6B,C),
although with increasing age, connexin 43 expression
in thebasal layer appeared todecline.
At E12, the distribution of connexin 26 in the epider-
mis (Fig. 7A) was similar tothat of connexin 43, but as
epidermal stratification and squamous differentiation
progressed, connexin 26 became restricted tothe outer-
most living layer of cells (Fig. 7B). Double staining
established that by E17 there was little if any overlap
between the zones of connexin 43 and connexin 26
expression (data not shown). The transition from con-
nexin 43 toconnexin 26 staining in theupper epidermal
layers didnot correspondtoa barrier for vertical spread
of LY-CH (compareFigs. 1C, 6C, and 7B).
In early hair pegs (stage2), connexin 26 was not seen
(data not shown), whereas connexin 43 was much less
than in overlying epidermis (Fig. 6B).
CHOUDHRY ET AL.
2). Sections were cut tangentially to the surface of the epidermis (Fig. 8).
(A,B) LY has spread downward from the injection site (E,F) into the basal
epidermis (e) at the top of a hair peg (hp), but does not spread laterally into
the epidermis at the top of an adjacent hair peg. d, dermis. (C,D) LY-CH
(A–F) LY-CH injection into epidermis above a hair peg (stagespread restricted to lower epidermal layers (e) overlying hair peg (hp). d,
dermis. (E,F) more extensive lateral spread of LY-CH in upper epidermal
(e) layers. Note the absence of epithelial-mesenchymal LY-CH transfer.
Dotted lines mark the boundaries between epidermis and dermis or hair
follicle epithelium and dermis. Scale bar ? 30 µm.
GAP J UNCTIONS AND CONNEXINS
horizontally and vertically between epithelial cells of an E12 (stage 2) hair
peg into most superficial cells (A,B) and among cells at the advancing tip
of the peg (hp) (C,D). Note the absence of dye spread into the
mesenchymal condensation. (E,F) Longitudinal section of an E14 (stage
3b) follicle. The dye injection site (arrowhead) in the phase contrast image
is in the outer columnar cells (or) in the upper part of the follicle. Note the
absence of dye transfer into cells in the inner core of this follicle and into
the dermal sheath (ds). (G,H) Tangential section through the bulb of a
(A–D) Sequential images showing extensive LY-CH spreadwell-developed E14 (stage 5) whisker follicle. LY-CH injection in the
epithelial germinative matrix (gm) cells of this follicle shows extensive dye
transfer between these cells from the base of germinative matrix toward
the upper part where dye spread appears to be restricted. Note the
absence of dye spread into the dermal papilla (p) and dermal sheath (ds).
d, non-follicular dermis. In further serial sections (not shown), the
unlabelled, dark cell (arrow) appeared to be continuous with inner root
sheath. Dotted lines mark the boundaries between epidermis and dermis
or hair follicle epithelium and dermis. Scale bars ? 30 µm.
E xpression of Connexins 26 and 43
in Hair F ollicles
Connexin 43 appeared in the keratinocytes forming
the central core of more developed hair pegs by E14
(Fig. 6C). Connexin 43 staining was extensive in the
keratinocytes of the matrix in the hair bulb and of the
outer root sheath by E16 (Fig. 6D,E). This staining
decreased and eventually was lost from cells as they
formed the differentiated structures of the inner root
sheath, especially Henley’s layer, where staining was
greatly reduced by the level of mid-bulb (Fig. 6F). In
contrast, connexin 26 staining in thehair bulbwas first
seen in cells forming Henley’s layer of the inner root
sheath (Fig. 7C). Connexin 26 staining was also ob-
served in the hair shaft (Fig. 7D). The switch from
connexin 43 toconnexin 26 corresponds tothe region in
the upper bulb where intercellular transfer of LY-CH
becomes restricted (compareFigs. 3G and 7C and 8B).
Table2 summarises theresults of connexin 43 and 26
antibody in embryonic skin and hair follicles.
Fig. 4 (A,B) E16 (stage 6) whisker follicle. LY-CH injection in the
fibroblasts of dermal papilla (dp). Restricted intercellular dye spread is
seen in these cells. Some extracellular leakage of the dye outlining
unlabelled cells is also apparent. Note the absence of dye transfer into
surrounding germinative matrix epithelial cells (gm). d, non-follicular
dermis. (C,D) Longitudinal section of a late developed E14 (stage 3b)
follicle. Injection site in the mesenchymal connective tissue sheath cells
(ds) lying adjacent to the epithelial cells is indicated by an arrowhead. A
markedly restricted dye spread within connective tissue sheath cells and
the absence of dye transfer into closely associated epithelial cells is
apparent. (E,F) Longitudinal section of a late developed E16 (stage 3c)
follicle. LY-CH injection in dermal connective tissue sheath (ds) is in cells
lying adjacent to the epithelial cells (arrowhead). Note a restricted dye
spread in these cells and the absence of dye transfer into neighbouring
epithelial cells. (G,H) Tangential section of an E16 follicle (stage 5). Dye
injection (arrowhead) is in outer layers of the dermal sheath (ds) lying
adjacent to non-follicular mesenchyme. (I,J) Section through non-
follicular dermis in E16 WP skin. Note the extensive dye transfer in
mesenchymal cells. Dotted lines mark the boundaries between epidermis
and dermis or hair follicle epithelium and dermis. Scale bar ? 30 µm.
GAP J UNCTIONS AND CONNEXINS
In recent years immunohistochemistry has been
widely used toexamine the distribution of connexins in
embryonic and adult tissues (Risek et al., 1992, 1994;
Butterweck et al., 1994; Goliger andPaul, 1995; Dahl et
al., 1996). Thereis evidencethat in somecircumstances
connexin staining indicates theextent of gap junctional
communication (Dahl et al., 1996). This study, however,
shows that the relationship between connexin staining
and functional intercellular coupling is more complex.
In the fully formed hair follicle, extensive spread of
LY-CH and NB through the germinative matrix is
associated with connexin 43, whereas in the early hair
peg, extensive spread of both dyes occurs between cells
that express little of this connexin. On the other hand,
connexin 26 is prominent in theinner layer of theinner
root sheath where there is little dye coupling. In
general, there is a better correlation between the stage
of differentiation and connexin expression, with con-
nexin 43 being associated with earlier stages wherecell
proliferation occurs and connexin 26 being associated
with later stages at theonset of terminal differentiation
(Goliger and Paul, 1994; Risek et al., 1994). Dye
coupling in the absence of connexin 26 and 43 could be
dependent on thepresenceof connexins 31, 31.1, 37, 40,
or 45, which have been reported tobe present in mouse
or rat skin at a late embryonic stage (mouse E19) and
postnatally (Butterweck et al., 1994; Goliger and Paul,
1994, 1995). However, it is more difficult to explain
connexin 26 in junctional plaques between follicular
inner root sheath cells in theabsenceof dyecoupling, or
the switch from connexin 43 to 26 in upper epidermal
layers, which is not associated with any restriction in
transfer of LY-CH between basal and uppermost com-
partments. It may be that other components play some
essential role in formation of functional gap junction
channels, that junctions in some tissues are closed, or
that connexins havefunctions additional totheir rolein
gapjunctions. Whatever theexplanation, it is clear that
any description ofthepatterns ofgapjunctional commu-
nication must bebased on functional assays.
Relatively few attempts have been made to measure
gap junctional coupling in intact tissues (Warner and
Lawrence, 1982; Kam et al., 1986; Serras and Van den
Biggelaar, 1987; Kalimi and Low, 1988; Salomon et al.,
1988; Kam and Pitts, 1989; Gevers and Timmermans,
1991; Kam and Hodgins, 1992; Laird et al., 1993;
Serras et al., 1993; Goliger and Paul, 1995). The
present study on mouse embryo epidermis and hair
follicleprovides themost detailed account todateof the
changes in patterns of junctional communication as a
small population of precursor cells develops into an
anatomically complex structure.
As the initial pattern of whisker placodes becomes
established, cells within the hair peg and overlying
epidermis are well coupled by gap junctions. There is
limited coupling of follicular and interfollicular epider-
mis at this stage, and dyes do not spread through the
epidermis intoadjacent follicles. This segregation could
be due toexpansion of a single epidermal compartment
or represent specificity of junction formation between
cells destinedtoforma hair peg andthoseofinterfollicu-
TABL E 1. Summary of Dye Injection R esults
the dye spread
Outer root sheath cells/
sheath, hair shaft
tive tissue sheath
Suprabasal and basal
Basal layer and
Hair pegs (stage 2–3a)E12–E13.56 Epithelial follicular ker-
Outer columnar cells
Outer root sheath (stage
cells (stage 4–6)
E14–E156Lower and midbulb ker-
Dermal papilla (stage
Dermal sheath (inner-
most layer next toepi-
Dermal sheath (outer-
most layer continuous
dermal cells) (stage
E14–E163Dermal papilla Restricted
E142 Dermal sheath cellsRestrictedDermal sheath cells/
E14–E163 Dermal sheath cellsRestrictedDermal sheath/dermal
CHOUDHRY ET AL.
lar epidermis. Laird et al. (1993) observed dye transfer
from apical ectodermal ridge into surrounding epider-
mis of mouse limb buds, but Serras et al. (1993) found
restricted transfer of LY-CH between feather placodes
and adjacent epidermis in thechick embryo. Theexten-
sive coupling within defined borders coincides with a
phase of rapid growth of the hair peg (Wessels and
Roessner, 1965; Hardy, 1992; Choudhry, unpublished
data, 1997) and could be important in coordinating cell
division, restricting proliferation tothe peg. Consistent
with this suggestion, the rapidly proliferating cells of
the germinative matrix in the follicle bulb (formed
later) are also extensively coupled. Follicular keratino-
cytes become less well coupled as they differentiate.
First, cells forming theouter root sheath abovethebulb
become uncoupled from the innermost core of cells.
Then, as cells generated in the germinative matrix
differentiate into hair cortex and inner root sheath,
coupling is progressively lost. There is little or no dye
coupling between cells in the different lineages (Kam
and Hodgins, 1992), but this specificity is not apparent
until the morphologies of the cells are clearly distinct.
Wepreviously hypothesised that subcompartmentation
of gap junctional communication might be prerequisite
for expression of different phenotypes, but the data
presented here show that segregation of coupling is a
Our observations also point to a restriction of gap
junctional coupling associated with mesenchymal and
epithelial cell differentiation.Cellsofthedermal conden-
sations packedarounddeveloping hair pegs werepoorly
coupled to more distant mesenchyme, which in agree-
ment with previous observations is extensively coupled
(Kam and Pitts, 1989). This restriction of coupling in
the dermal condensation coincides with a decrease in
mitotic activity (Wessels and Roessner, 1965).
Throughout this study there has been no instance of
LY-CH spreadbetween dermal andepidermal or follicu-
lar epithelial cells. Injections of LY-CH have been made
closetoand on both sides of interfaces between interfol-
licular epidermis anddermis, early hair peg andmesen-
chymal condensation, follicular germinativematrix and
dermal papilla, and outer root sheath and dermal
sheath.LY-CH spreadbetween epithelial andmesenchy-
mal cells in these tissues is rare or nonexistent. Kam
and Hodgins (1992), Laird et al. (1993), and Serras et
(left) into a hair peg (right) showing injections of LY-CH (A) and NB (B) into
E13 epidermis above a stage 2 hair peg. Both tracers spread extensively
within the epidermis over the hair peg (left) and down into the peg itself
Serial, overlapping confocal images moving from epidermis
(right).There is no evidence that either tracer is transferred into dermal
cells. Arrows indicate the boundaries to the LY-CH and NB spread
coinciding with the basement membrane zone between hair peg and
dermis. Scale bars ? 20 µm.
GAP J UNCTIONS AND CONNEXINS
yellow: connexin 43 fluorescein. False red: tissue background. (A) E12
epidermis showing a uniform distribution of connexin 43 in all epidermal
layers. (B) In E13.5 epidermis, connexin 43 staining is most extensive in
basal layers. Connexin 43 is weakly expressed in a developing (stage 2)
hair peg. (C) In E14 epidermis, connexin 43 expression is strong in middle
epidermal layers. Note the absence of staining in outermost differentiated
layers. Also note connexin 43 expression only in the inner core of a more
Two colour fluorescence images of connexin 43 staining. False developed (stage 4) hair peg. (D) In tangential section of E16 well-
developed (stage 6) follicle, connexin 43 staining is extensive in the hair
bulb and outer root sheath keratinocytes. (E,F) In a transverse (E) and
tangential (F) section through an E16 (stage 6) follicle, connexin 43 is
weakly expressed in keratinocytes differentiating to form the inner root
sheath structure. Note the loss of connexin 43 staining from inner root
sheath at the level of mid-bulb. Scale bar ? 20 µm.
CHOUDHRY ET AL.
yellow: connexin 26 fluorescein. False red: tissue background. (A) In E12
epidermis, connexin 26 is uniformly distributed in all epidermal layers. (B)
In E16 epidermis, connexin 26 is only expressed in the outermost
differentiated epidermal layers. (C,D) In a longitudinal section (C) and
Two colour fluorescence images of connexin 26 staining. False transverse section (D) of an E16 well-developed (stage 6) whisker follicle,
connexin 26 is only present in the cells of Henley’s layer of the inner root
sheath. In D, also note connexin 26 expression in the keratinized hair
GAP J UNCTIONS AND CONNEXINS
al. (1993) alsofailedtoobserveepithelial andmesenchy-
mal transfer of LY-CH. Kam et al. (1986) reported
occasional transfer of LY-CH from epidermis to dermis
of newborn mouse, but this was more frequent in the
Er/Er mutant (Kam and Pitts, 1989). The similar
results obtained with injections of LY-CH and NB in the
present study indicate that lack of dye spread is un-
likely to be due to gap junctions that, although perme-
able to intracellular metabolites, have poor permeabil-
ity to LY-CH (Penn et al., 1994). Direct epithelial-
mesenchymal cell contacts across the basement
membrane zone at sites of inductive interactions have
been observed in a number of embryonicorgans, includ-
ing mouse hair follicle (see Hardy et al., 1983 and
references therein), but it seems unlikely from the
evidence available that such contacts would involve
epithelial-mesenchymal coupling by gap junctions per-
meable tointracellular metabolites in the size range of
NB or LY-CH (i.e., inositol phosphates, cyclic AMP, or
other nucleotides). However, electrical coupling has
been reported between cells that failed to transfer
LY-CH (Warner and Lawrence, 1982; Kalimi and Lo,
1988). It is also possible that inductive signals gener-
ated at sites of focal epithelial-mesenchymal cell con-
tact couldbepropagatedwithin thehair peg or germina-
tivematrix via gap junctions.
Finally, in both epidermis and developing hair pegs,
occasional isolated cells were seen that appeared not to
becoupled toadjacent cells (Fig. 3A). Similar cells were
alsoobserved by Serras et al. (1993) in embryonic chick
epidermis. The identity of the uncoupled cells is un-
known; they could be melanocytes or possibly keratino-
cytes at a particular stageof thecell cycle.
E XPE R IME NTAL PR OCE DUR E S
E mbryonic Whisker Pad Skin E xplants
Female ICR mice were time mated. Presence of a
uterine plug was defined as day 0 of gestation. Animals
were killed between E12 and E17, and embryos were
dissected out. In each experiment, one fetus was fixed
in 4% paraformaldehyde in phosphate-buffered saline
(PBS) for a further determination of gestational age by
anatomical characteristics. Other fetuses were killed
and upper lip skin was dissected out. Skin explants
were placed on Nucleopore filters (Sterilin, Hounslow,
UK) and floated in Ham’s F-12/Dulbecco’s modified
Eagle’s medium (50/50 v/v mixture; GibcoBRL, Paisley,
UK ) supplemented with 1% glutamine, penicillin/
streptomycin, and fungizone. Explants were main-
tained at 37°0C in a humidified incubator in an atmo-
sphereof 5% CO2in air for 2–3 hr beforedyeinjection.
Intracellular Injection of LY-CH and NB
This was carried out as described previously (Kam et
al., 1986; Kam and Hodgins, 1992) with modifications.
Tissues attached to Nucleopore filters were submerged
in Dulbecco’s modified Eagle’s medium buffered with 25
mM Hepes at 37°C and viewed under a stereo micro-
scope. The tissue was impaled by the microelectrode
containing LY-CH (dilithium salt, 4%), and intracellu-
lar location of the electrode tip was detected by a
change in resistance between injecting and reference
electrodes. LY-CH was then iontophoresed in 0.5-sec
pulses of 10 nA at 1 Hz for 2.0 min intotheinjected cell.
The injected specimen was examined immediately un-
der a fluorescent microscope to determine approxi-
mately the nature and extent of dye spread. The
specimen was then fixed in 4% paraformaldehyde in
PBS. Intracellular injections showed distinct localisa-
tion of dye within cell nuclei, while extracellular injec-
tions clearly showed dye spread through intercellular
spaces and no nuclear staining (Kam and Hodgins,
1992). In some injections, a small amount of extracellu-
lar dye leaked from the electrode during impalement or
withdraw but was distinguished from intracellular
fluorescence. NB was injected similarly but with the
opposite polarity. After fixation as above, the injected
tissue was permeabilised with 1.0% Triton ?100 and
stained with fluorescein labelled avidin.
Analysis of LY-CH and NB Transfer
Injected tissue whole mounts were examined further
with BIORad MRC 600 confocal laser scanning micro-
scope (BioRad, Hemel Hempstead, UK) to confirm the
site of intracellular injection and the extent of dye
2A–F, and 3A–D, 5) and stage 5/6 hair follicle bulbs (B: Fig. 3G,H).
Diagonal hatching: zones of dye spread. Broken horizontal lines: planes
of sections shown in Figs. 2A, 2C, 2E, 3A, and 3C in A. Sites of dye
injection: square in Fig. 1E, circle in Fig. 2, and triangle in Fig. 3A–D.
Summary of dye spreads in stage 2 hair pegs (A: Figs. 1E–F,
CHOUDHRY ET AL.
transfer in threedimensions (krypton/argon laser, exci-
tation 488 nm, emission filter 522 nm). In some cases
this analysis provided sufficient resolution, but to ob-
tain more precise information about the relationship
between zones of dye transfer and tissue architecture,
sections of LY-CH embedded tissue were prepared as
previously described (Kam and Hodgins, 1992). Speci-
mens were dehydrated, embedded in LR-White resin,
and serially sectioned at 3 µm. Sections were photo-
graphed using phase contrast and epifluorescence on
Kodak Ektachrome 400 film (Rochester, NY) with an
exposure time of 2.0 min. Colour photographic slides
were then scanned into Photoshop, composite images
were assembled with Micrografx Windows Draw, and
printed on a Kodak Coloreaseprinter.
Mouse embryo upper lip skin was embedded in OCT
compound and frozen in liquid nitrogen. Sections 5- to
7-?mthick werecut in a cryostat andfixedin acetoneat
room temperature for 10 min and blocked (20 min) in
20% normal horse serum or normal goat serum diluted
in PBS. Sections were incubated with one of the follow-
ing antibodies for 2 hr at room temperature: a 1/1,000
dilution of rabbit polyclonal antipeptide-antiserum
against amino acids 101–119 of rat connexin 26 (S.
J amieson, personal communication, 1995), a 1/200 dilu-
tion of an affinity-purified rabbit polyclonal antipeptide
antiserum against amino acids 108–122 of mouse con-
nexin 26 (Goliger and Paul, 1994), a 1/1,000 dilution of
a monoclonal mouse antipeptide-antiserum against
amino acids 252–271 of rat connexin 43 (Beyer et al.,
1989; ZymedLaboratories, San Francisco, CA), or 1/200
dilution of an affinity-purified rabbit antipeptide-
antiserum against aminoacids 252–271 of rat connexin
43 (Beyer et al., 1987). In negative controls primary
antibody was omitted. After washing in PBS, the sec-
tions were incubated with fluorescein-conjugated anti-
rabbit and anti-mouse immunoglobulins (Vector Labo-
ratories, Peterborough, UK) for 30–45 min, washed in
PBS, mounted in Vectashield (Vector), and examined by
epifluorescence and with a BioRad MRC 600 confocal
microscope. The confocal microscope was used to scan
immunofluorescent slides in both green (excitation 488
nm, emission filter 522 nm for fluorescein) and red
(excitation 568 nm, emission filter 585 nm for back-
ground). Scanning of tissue autofluorescence at 585 nm
in frozen sections provided the equivalent of a counter-
stain to visualise tissue architecture and illustrate
localisation of fluorescein-labelled immunoglobulins.
Dual false-colour images were transferred into Micro-
grafx PhotoMagic, and colour intensities were in-
creased to give a sufficiently strong red tissue back-
ground without obscuring the fluorescein (this results
in fluorescein appearing yellow in the colour plates).
Images were than assembled in Micrografx Windows
Draw and printed on a Kodak Coloreaseprinter.
ACK NOWL E DGME NT S
We thank Rita MacKay for technical assistance, Dr.
S. J amieson and Professor D. Paul for the gift of
connexin antibodies, and Drs. M. Finbow and C.A.B.
TABL E 2. Summary of Connexin Antibody Staining R esults
Time in development
Connexin 43 staining
E12–E13 (stage 2–3a)
Epidermis Hair pegsHair follicles
E14–E15 (stage 3c–6)Suprabasal?/?
E16–E17 (stage 3c–6) Suprabasal?
Connexin 26 staining
E14–E15 (stage 3c–6)—Henley’s, IRS??
E16–E17 (stage 3c–6)Suprabasal??
ORS, outer root sheath; IRS, inner root sheath; GM, germinativematrix.
GAP J UNCTIONS AND CONNEXINS
J ahoda for helpful discussions. We also thank Dr. D.
Neil for theloan of an oscilloscope.
R E F E R E NCE S
Beyer EC, Paul DL, Goodenough DA. Connexin 43: A protein from rat
heart homologous to a gap junction protein from liver. J . Cell Biol.
Beyer EC, Kistler J , Paul DL, Goodenough DA. Antisera directed
against connexin 43 peptides react with a 43-KD protein localised to
gap junctions in myocardium and other tissues. J . Cell Biol.
Bruzzone R, White TW, Paul DL. Connections with connexins: The
molecular basis of direct intercellular signalling. Eur. J . Biochem.
Butterweck A, Elfgang C, Willeck K, Traub O. Differential expression
of the gap junction proteins connexin -45, -43, -40, -31 and -26 in
mouseskin. Eur. J . Cell Biol. 1994;65:152–163.
Chuong CM, Edelman GM. Expression of cell adhesion molecules in
embryonic induction. II. Morphogenesis of adult feathers. J . Cell
Dahl E, Winterhager E, Reu B, Traub O, Butterweck A, Willecke K.
Expression of gap junction proteins connexin 31 and connexin 43
correlates with communication compartments in extraembryonic
tissues and in the gastrulating mouse embryo respectively. J . Cell
Davidson P, Hardy MH. The development of mouse vibrissae in vivo
and in vitro. J .Anat. 1952;86:342–356.
Finch PW, Cunha GR, Rubin J S, Wong J , Ron D. Pattern of keratino-
cyte growth factor and keratinocyte growth factor receptor expres-
sion during mouse fetal development suggests a role in mediating
Gevers P, Timmermans LPM. Dyecoupling and theformation and fate
of the hypoblast in the teleost fish embryo Barbus chonchonius.
Goetinck PFM, Carlone DL. Altered proteoglycan synthesis disrupts
feather pattern formation in chick embryonic skin. Dev. Biol.
Goliger J A, Paul DL. Expression of gap junction proteins Cx26,
Cx31.1, Cx37 and Cx43 in developing and mature rat epidermis.
Dev. Dyn. 1994;200:1–13.
Goliger J A, Paul DL. Wounding alters epidermal connexin expression
and gap junction mediated intercellular communication. Mol. Biol.
GuoL, Degenstein L, Fuchs E. Keratinocyte growth factor is required
for hair development but not for wound healing. Genes Dev.
Hardy MH, Van Exan RJ , Sontegard KS, Sweeny PR. Basal lamina
changes during tissue interactions in hair follicles—An in vitro
study of normal dermal papillae and vitamin-A induced glandular
morphogenesis. J . Invest. Dermatol. 1983;80:27–34.
Hardy MH. (1992). The secret life of hair follicle. Trends Genet.
Hirai Y, Nose A, Kobayashi S, Takeichi M. Expression and role of
E-and P-cadherin adhesion molecules in embryonic histogenesis. II.
Skin morphogenesis. Development 1989;105:271–277.
Hogan BLM, Blessing M, Winnier GE, Suzuki N, J ones CM. Growth
factors in development: The role of TGF? related polypeptide
signalling molecules in embryogenesis. Development Suppl. 1994;
J iang TX, Choung CM. Mechanism of skin morphogenesis, I. Analysis
with antibodies toadhesion molecules tenascin, N-camandintegrin.
Dev. Biol. 1992;150:82–98.
Kalimi GH, Low CW. Communication compartments in the gastrulat-
ing mouseembryo. J . Cell Biol. 1988;107:241–255.
KamE, Hodgins MB. Communication compartments in hair follicles &
their implication in differentiative control. Development 1992;114:
Kam E, Pitts J D. Tissue specific regulation of junctional communica-
tion in the skin of mouse foetuses homozygous for the repeated
epilation (ER) mutation. Development 1989;107:923–929.
Kam E, Melville L, Pitts J D. Patterns of junctional communication in
skin. J . Invest. Dermatol. 1986;87:748–753.
Katz LC. Co-ordination of vertebrate cellular assemblies by gap
junction. Semin. Dev. Biol. 1995;6:117–125.
Kratochwil K, Dull M, Farinas I, Galceran J , Grossched R. Lef-1
expression is activated by BMP-4 and regulates inductive tissue
interactions in tooth and hair development. Genes Dev. 1996;10:
Kumar NM, Gilula NB. Thegap junction communication channel. Cell
Laird DW, Yancey SB, Bugg L, Revel J P. Connexin expression and gap
junction communication compartments in the developing mouse
limb. Dev. Dyn. 1993;195:153–161.
Lawrence TS, Beers WH, Gilula NB. Transmission of hormonal
stimulation by cell-cell communication. Nature1978;325:60–62.
Lowenstein WR. J unctional intercellular communication and the
control of growth. Biochim. Biophys.Acta. 1979;560:1–65.
Oliver RF. The induction of hair follicle formation in the adult hooded
rat by vibrissa dermal papillae. J . Embryol. Exp. Morphol. 1970;23:
Paul DL, Bruzzone RL, Gimlich RL, Goodenough DA. Expression of a
dominant negativeinhibitor of intercellular communication in early
Xenopus embryo causes delamination and extrusion of cells. Devel-
Penn A, Wong ROL, Shatz CJ . Neuronal coupling in the developing
mammalian retina. J . Neurosci. 1994;14:3805–3815
Pitts J D, Finbow ME, Kam E. J unctional communication and cellular
differentiation. Br. J . Cancer 1988a;58:52–57.
Pitts J D, Kam E, Morgan D. The role of junctional communication in
growth control and tumorigenesis. In: Hertzberg E, J ohnson RG,
eds. Modern Cell Biology. Volume 7: Gap J unctions. New York: Alan
R. Liss, 1988b:397–409.
Reynolds AJ , J ahoda CAB. Hair matrix germinative epidermal cells
confer follicle inducing capabilities on dermal sheath and high
passagedermal papilla cells. Development 1996;122:3085–3094.
Risek B, Klier FG, Gilula NB. Multiple gap junction genes are utilized
during rat skin and hair development. Development 1992;116:639–
Risek B, Klier FG, Gilula NB. Developmental regulation and struc-
tural organisation of connexins in epidermal gap junctions. Dev.
Salomon D, Saurat J H, Meda P. Cell to cell communication within
intact human skin. J . Clin. Invest. 1988;82:248–254.
Sengel P. Morphogenesis of skin. Cambridge: Cambridge University
Serras F, Van den Biggelaar J AM. Is a mosaic embryoalsoa mosaic of
communication compartments. Dev. Biol. 1987;120:132–138.
Serras F, Fraser S, Choung CM.Asymmetricpatterns of gapjunctional
communication in developing chicken skin. Development 1993;119:
TingBerreth SA, Chuong CM. Sonic hedgehog in feather morphogen-
esis: Induction of mesenchymal condensation and association with
cell death. Dev. Dyn. 1996a;207:157–170.
TingBerreth SA, Chuong CM. Local delivery of TGF?2 can substitute
for placode epithelium toinduce mesenchymal condensation during
skin appendagemorphogenesis. Dev. Biol. 1996b;179:347–359.
Warner AE, Lawrence PA. Permeability of gap junctions at the
segmental border in insect epidermis. Cell 1982;28:243–252.
Wessels E, Roessner D. Non proliferation in dermal condensation of
mousevibrissaeand pelagehairs. Dev. Biol. 1965;12:419–433.
Wolpert L. Positional information revisited. Development Suppl. 1989;
Zhang HY, Tipml R, Sasaki T, Chu ML, Ekbolm P. Fibulin-1 and
Fibulin-2 expression during organogenesis in the developing mouse
embryo. Dev. Dyn. 1996;205:348–364.
CHOUDHRY ET AL.