Keratinocyte-Releasable Stratifin Functions as a Potent Collagenase-Stimulating Factor in Fibroblasts

Termination of wound healing requires a fine balance between collagen deposition and its hydrolysis. To dissect the underlying control mechanisms for this process, we established a keratinocyte/fibroblast co-culture system and subsequently demonstrated more than a 10-fold increase in collagenase expression in fibroblasts co-cultured with keratinocytes relative to that of control cells. This finding was further confirmed in fibroblasts grown in a keratinocyte/fibroblast collagen-GAG gel. The efficacy of keratinocyte-derived collagenase stimulatory factors on collagenase activity was evaluated, and the results showed that only conditioned medium derived from fibroblasts co-cultured with keratinocytes was able to break down markedly type I collagen to its one-quarter and three-quarter fragments of both alpha (alpha1 and alpha2) and beta (beta1.1 and beta1.2) chains. The results of a dose-response experiment showed that keratinocyte-conditioned medium (KCM) stimulates the expression of collagenase mRNA by dermal fibroblasts in a concentration-dependent fashion. In a similar experiment, the results of a time-response experiment revealed that KCM treatment increases the expression of collagenase mRNA in dermal fibroblasts as early as 6 h and reaches its maximum level within 24-48 h. Considering that this keratinocyte-releasable factor has a potent collagenase stimulatory effect on fibroblasts, which favors the resolution of accumulated type I and type III collagen found in fibrotic

Termination of wound healing requires a fine balance between collagen deposition and its hydrolysis. To dissect
the underlying control mechanisms for this process, we established a keratinocyte/fibroblast co-culture system
and subsequently demonstrated more than a 10-fold increase in collagenase expression in fibroblasts co-cultured
with keratinocytes relative to that of control cells. This finding was further confirmed in fibroblasts grown in a
keratinocyte/fibroblast collagen-GAG gel. The efficacy of keratinocyte-derived collagenase stimulatory factors on
collagenase activity was evaluated, and the results showed that only conditioned medium derived from fibroblasts
co-cultured with keratinocytes was able to break down markedly type I collagen to its one-quarter and three-quarter
fragments of both a (a1 and a2) and b (b1.1 and b1.2) chains. The results of a dose?response experiment showed
that keratinocyte-conditioned medium (KCM) stimulates the expression of collagenase mRNA by dermal fibroblasts
in a concentration-dependent fashion. In a similar experiment, the results of a time?response experiment revealed
that KCM treatment increases the expression of collagenase mRNA in dermal fibroblasts as early as 6 h and
reaches its maximum level within 24?48 h. Considering that this keratinocyte-releasable factor has a potent
collagenase stimulatory effect on fibroblasts, which favors the resolution of accumulated type I and type III
collagen found in fibrotic tissue, we referred to this protein as a keratinocyte-derived anti-fibrogenic factor (KDAF).
In a series of chromatography experiments and a direct trypsin digestion of the proteins and subsequent peptide
mapping, a keratinocyte-derived collagenase-stimulating factor turned out to be a releasable form of stratifin, also
known as 14-3-3 r protein. To validate this finding, stratifin cDNA was cloned into a pGEX-6P-1 expressing vector
and more than 50 mg of recombinant stratifin was generated and used to treat fibroblasts with various
concentrations for 24 h. The results of northern analysis showed a remarkable dose?response increase in the
expression of collagenase mRNA in stratifin-treated fibroblasts relative to that of the control. This finding was
consistent with that obtained from collagenase activity assay. In conclusion, we identified a keratinocytereleasable
form of stratifin in KCM that mimics the collagenase stimulatory effect of KCM for dermal fibroblasts.
This finding suggests that stratifin is likely to be, at least, one of the KDAFs found in KCM.
Key words: cell?cell interaction/collagenase/fibroblasts/keratinocyte/protein 14-3-3 s/skin/stratifin
J Invest Dermatol 122:1188 ?1197, 2004
Epidermal?mesenchymal communication is critical in exchanging
information between keratinocytes and fibroblasts
in skin morphogenesis during its development and probably
also in maintaining the integumentary structure of skin in
adults (Johnson-Wint, 1988). Disruption of this interaction,
as is the case with delays in epithelialization during the
process of wound healing due to either infection or severity
of injury, increases the frequency of developing fibrotic
conditions. This is based on the fact that when keratinocytes
form an epithelial layer on the wound within 2?3 wk,
only one-third of the anatomically site-matched wounds
become fibrotic; however, this increases to 78% when a
wound is epithelialized later than 21 d. This signifies that in
the absence of epithelialization, extracellular matrix (ECM)
continues to accumulate until dermal fibroblasts receive
signal(s) from epidermal cells to slow down the dynamic
process of healing that leads to maturation and remodelling
of the healing wound (Machesney et al, 1998). Thus, a fine
balance between synthesis of ECM and degradation by a
large family of enzymes known as matrix metalloproteinases
(MMP) is a key factor in maintaining the structural integrity
of normal skin.
MMPs represent a group of diverse proteolytic enzymes
involved in ECM turnover and connective tissue remodelling
during physiological conditions such as embryonic growth
and development, uterine involution, bone growth and
resorption, and wound healing (Nagase and Woessner,
1999). The MMP family consists of 25 zinc- and calciumdependent
proteinases in the mammalian system. Accord-
Abbreviations: ECM, extracellular matrix; MMP, matrix metalloproteinase;
KCM, keratinocyte-conditioned medium; KDAF, keratinocyte-
derived anti-fibrogenic factor; KSFM, keratinocyte serum-free
medium; MALDI, matrix-assisted laseradsorption
Copyright r 2004 by The Society for Investigative Dermatology, Inc.
1188
ing to their substrate specificity, primary structure, and
cellular localization, these enzymes can be divided into at
least five different subfamilies of closely related members
known as collagenases, gelatinases, stromelysins, matrilysins,
and membrane-type MMPs (reviewed in Murphy et al,
2002). The level of MMP expression in normal cells is low
and this allows healthy connective tissue remodelling. But
an imbalance in expression of MMPs has been implicated in
a number of pathological conditions such as dermal fibrosis
(Ghahary et al, 1996), rheumatoid arthritis, atherosclerosis,
pulmonary emphysema, and tumor invasion and metastasis
(Birkedal-Hansen et al, 1993; Nagase and Woessner, 1999).
Epidermal?mesenchymal interactions play a critical role
in controlling the expression of MMPs during development
and healing of skin. An increase in MMP expression has
been implicated with tissue degradation and remodelling
during tumor invasion and wound healing. In both conditions,
cell interaction between either fibroblasts and tumor
cells or fibroblasts and keratinocytes results in an increase
in MMP production. DeCastro et al (1996) purified a tumor
cell surface protein known as extracellular matrix metalloproteinase
inducer (EMMPRIN) and showed its efficacy in
promoting the expression of MMP-1, gelatinase A, and
stromelysin-1 by fibroblasts. These investigators also
demonstrated the presence of EMMPRIN in the normal
epidermis and raised the possibility of its involvement in the
regulation of matrix remodelling at the epidermal?dermal
interface. Kertinocytes themselves also produce MMPs in
response to various stimulatory factors such as cytokines
and growth factors (Lyons et al, 1993). It was shown that
interleukin-1b (IL-1b) and a, as well as epidermal growth
factor, stimulate the expression of MMP-9 and MMP-1 in rat
mucosal keratinocytes (Lyons et al, 1993). Similarly for
fibroblasts, cytokines and growth factors such as IL-1b,
platelet-derived growth factor, epidermal growth factor,
TNF-a and purified tumor promoters have remarkable
effects in controlling the signal mechanism involved in the
regulation of MMP gene expression (Johson-Wint, 1988). It
has also been shown that keratinocytes cultured on type IV
or type I collagen produced more collagenase than did
those cultured on laminin or in the absence of matrix
(Petersen et al, 1990). Interestingly, MMP-1 expression was
induced in primary keratinocytes by contact with native, but
not denatured, type I collagen or basement membrane
proteins (Pilcher et al, 1999). Sudbeck et al (1994) confirmed
this finding and further showed that only migrating
keratinocytes expressed MMP-1 and this expression was
downregulated when cells contacted laminin-1. This finding
further suggests that contact with only collagen is not
sufficient to induce MMP-1 expression in keratinocytes.
Riikonen et al (1995) used three different osteogenic cell
lines with distinct patterns of putative collagen receptors
and showed that the presence of integrin a2 b1 was a
positive regulator of collagenase. This finding, therefore,
suggests that the level of MMP-1 expression is regulated by
the collagen receptor a2 b1 integrin. It is generally accepted
that MMPs produced by keratinocytes facilitate epithelial
migration, whereas MMPs expressed by dermal fibroblasts
promote tissue remodelling (Salo et al, 1991).
Although several investigators have suggested that the
epidermal?dermal interface might be a site of continuous
cellular communication regulating collagen breakdown, the
identities of the involved factors have not yet been
determined. Thus, to understand the importance of the
mesenchymal?epithelial communication in the physiological
and pathological process of dermal healing, we conducted
a series of experiments to test our working hypothesis that a
keratinocyte-derived anti-fibrogenic factor(s) (KDAF) might
function as a wound healing stop signal(s) by modulating
the expression of key extracellular proteins such as
collagenase (MMP-1) and/possibly collagen type I and type
III in dermal fibroblasts. As shown in this report, sequential
chromatography of keratinocyte-conditioned medium
(KCM) identified a protein with an apparent MW between
30 and 50 kDa with strong collagenase stimulatory activity.
After further purification, peptide mapping, and amino acid
sequencing of the purified KDAF, the protein was identified
as stratifin, which is also known as 14-3-3 s protein. This
was then confirmed by showing that recombinant stratifin
has potent collagenase stimulatory effects on dermal
fibroblasts.
Results
Identification of a KDAF for dermal fibroblasts To
identify a KDAF for dermal fibroblasts, we established a
keratinocyte/fibroblast co-culture system in which keratinocytes
and fibroblasts are grown in the upper and lower
chambers of this system, respectively. As an index for the
anti-fibrogenic effects of keratinocyte-derived factor(s) on
dermal fibroblasts, total RNA was extracted from fibroblasts
grown in the lower chamber, and the expression of
collagenase (MMP-1) mRNA was evaluated by northern
analysis. Fibroblasts grown either alone or in a fibroblast/
fibroblast co-culture system were also used as controls.
As shown in Fig 1A, collagenase mRNA expression is
increased more than 10-fold in fibroblasts co-cultured with
keratinocytes (lane F/K) relative to that of fibroblasts grown
in a fibroblast/fibroblast (lane F) co-culture system. No
detectable level of collagenase mRNA was found in
keratinocytes grown in the co-culture system (lane K).
To further confirm this finding, a condition modelling a 3D
in vivo dermal environment was prepared from a keratinocyte/
fibroblast collagen-GAG gel. The skin substitute was
kept submerged in 49% DMEM, 49% KSFM and was
supplemented with 2% FBS for one week and then raised to
the air?liquid interface for another week. Multi-layers of
keratinocytes and a collagen-populated fibroblast lattice
formed epidermal and dermal components of this skin
substitute, respectively. The results of northern analysis (Fig
1B) clearly show that fibroblasts isolated from this skin
substitute (lane F/K) expressed more than 10 times as much
collagenase transcript as that obtained from the fibroblast/
fibroblast skin substitute (lane F/F). The expression of
collagenase mRNA isolated from the epidermal layer of this
substitute (lane K) was undetectable. To correlate the
expression of collagenase mRNA with collagenase activity,
we then evaluated the efficacy of KDAF on collagenase
activity in conditioned medium obtained from either
keratinocytes alone (K), fibroblasts alone (F), or the
keratinocyte/fibroblast (K/F) co-culture system using type I
122 : 5 MAY 2004 PROTEIN 14-3-3 s AS COLLAGENASE-STIMULATING FACTOR 1189
collagen as a substrate. Type I collagen with no treatment
(collagen) was also included as a control. The rationale for
performing this experiment is based on the fact that the
members of the collagenase subfamily such as interstitial
collagenase (collagenase 1 or MMP-1), neutrophil collagenase
(collagenase 2 or MMP-8), and collagenase 3 are the
only known enzymes to cleave efficiently native triple-helical
fibrillar collagen type I, II, and III at a single site and thereby
generate one-quarter and three-quarter fragments (Murphy
et al, 2002). Thus, the presence of these fragments in the gel
is considered to be an index for collagenase activity. The
results of this experiment, shown in Fig 1C, reveal that only
conditioned medium derived from fibroblasts co-cultured
with keratinocytes was able to break down markedly type I
collagen to its one-quarter and three-quarter fragments of a
(a1 and a2) and b (b1.1 and b1.2) chains of type I collagen.
A series of experiments were then designed to further
characterize the efficacy of KDAF on collagenase expression
and to demonstrate whether the presence of fibroblasts
is needed for the release and efficacy of this factor.
For the KDAF dose?response experiment, conditioned
medium derived from mono-cultured keratinocytes was
collected and the expression of collagenase mRNA was
determined in dermal fibroblasts receiving various volumes
of KCM. This result revealed that KCM stimulated the
expression of collagenase mRNA by dermal fibroblasts in a
concentration-dependent fashion (Fig 2A). This was not due
variation in total RNA loading as re-hybridization of the
same blot with cDNA specific for 18S ribosomal RNA shows
similar loading. In a similar experiment, the results of a time?
response experiment revealed that KDAF, which is present
in KCM, increases the expression of collagenase mRNA in
dermal fibroblasts as early as 6 h and reaches its maximum
level within 24?48 h (Fig 2B). The findings of these
experiments indicate that expression of KDAF by keratinocytes
is independent of a co-culture system because
conditioned medium derived from keratinocyte monoculture
can also stimulate collagenase mRNA expression
in dermal fibroblasts.
To correlate keratinocyte differentiation with KDAF
release, KCM was collected every 48 h for up to 24 d and
the stimulation of collagenase mRNA expression was
evaluated for each time point collection. In general, the
results of Fig. 2C revealed a greater expression of
collagenase mRNA in fibroblasts treated with each collection
relative to that of the control. To correct for RNA
loading, the signals for collagenase mRNA expression and
18S ribosomal RNA were determined by densitometry and
the ratio of collagenase/185 was calculated. This finding
indicates a greater expression of collagenase mRNA in
response to KCM collected on days 6?10 as well as 18?24,
a time at which keratinocytes become differentiated (Fig 2C,
right panel). We have previously demonstrated that a
differentiation marker, involucrin, is expressed between
days 5 and 14 of culturing keratinocytes in test medium.
The expression of involucrin mRNA remained high up to day
24 examined (Ghahary et al, 2001).
KDAF isolation and its identification as stratifin The
factor was purified by a three-stage chromatography
protocol as discussed in Experimental Procedures. At the
last stage of chromatography, a total of 22 fractions were
collected and evaluated for their collagenase mRNA
activities using dot blot analysis. Cells treated with KCM
and non-conditioned medium (NCM) were also included
and served as positive and negative controls. As shown in
Fig 3A, a peak of collagenase-stimulating activity appeared
from fraction 10, peaked on 13 and 16, and disappeared
after fraction 19. To identify this activity, fractions 13?16 with
the highest collagenase activities were subjected to protein
fractionation by using PAGE. Candidate protein bands (Fig
3B, i) as well as a 50 kDa protein (arrow head) related to the
active fractions were then eluted, pooled, and subjected to
Figure 1
Efficacy of keratinocyte-derived conditioned medium on the
expression of collagenase in dermal fibroblasts. Panel A shows
the expression of collagenase mRNA in keratinocyte co-cultured with
fibroblasts (lane K), fibroblasts co-cultured with either keratinocytes (F/
K) or fibroblasts (lane F). To further confirm this finding, a keratinocyte/
fibroblast collagen-GAG gel was prepared. Fibroblasts (F/K) and
keratinocytes (K) grown in this system as well as fibroblasts cocultured
with fibroblasts (F/F) were separately isolated and evaluated
for collagenase mRNA expression by northern analysis (Panel B). Panel
C, collagenase activity in conditioned medium obtained either from
keratinocytes alone (K), fibroblasts alone (F), or keratinocyte/fibroblast
(K/F) co-culture was evaluated in triplicate using type I collagen as a
substrate. Type I collagen with no treatment (collagen) was included as
a control. Note that only conditioned medium derived from keratinocytes
was able to break down markedly type I collagen to its onequarter
and three-quarter fragments of the a (a1 and a2) and b (b1.1
and b1.2) chains of type I collagen.
1190 GHAHARY ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY
further evaluation. Direct trypsin digestion of the proteins
and subsequent peptide mapping of the recovered peptides
by a combination of mass spectrometry and Edman
degradation identified the 30 kDa protein as 14-3-3 s
protein also known as stratifin. As shown in Table I, the
sequence of identified peptides matched that of stratifin
(but not other members of the 14-3-3 protein family). When
the 51 kDa protein shown in Fig 3B was excised and
sequenced, however, it turned out to be a 51 kDa fragment
of the cytovillin protein. The apparent size of this protein
was estimated to be about 70?75 kDa. As reported by
Turunen et al (J Biol Chem 264: 16727?16732, 1989), this
protein is a microvillar cytoplasmic peripheral membrane
protein, which is prominently found in syncytiotrophoblasts
of the placenta. For this reason, in this study, the
collagenase stimulatory effect of only the 30, but not 51
kDa protein was further evaluated. Thus, to confirm this
finding, total keratinocyte RNA was extracted and RT-PCR
was performed using PCR primers specific for stratifin. We
then cloned stratifin cDNA into a pGEX-6P-1 expressing
vector and as shown in Fig 4 (lane 4), the isolated stratifin
was more than 95% pure. To validate the efficacy of the
collagenase stimulatory effects of stratifin in fibroblasts,
cells were treated with various concentrations of recombinant
stratifin protein for 24 h (Fig 4B). The results of
Northern analysis showed that fibroblasts express collagenase
mRNA in a dose-responsive manner when incubated
with increasing doses of recombinant stratifin protein. At the
highest dose used, there was more than a 10-fold increase
in collagenase expression relative to untreated control cells
(lane 6 vs lane C). To further examine whether this increase
is consistent with collagenase activity, fibroblasts were
treated with either nothing (Fig 4 C, lane U) or purified
recombinant stratifin (Fig 4 C, lane S), and the result
showed that conditioned medium derived from stratifintreated
fibroblasts was able to break down almost
completely type I collagen to its one-fourth and threefourths
fragments of a (a1 and a2) and b (b1.1 and b1.2)
chains of type I collagen. These findings suggest that the
extracellular form of stratifin functions as a potent collagenase-
stimulating factor for dermal fibroblasts.
Discussion
To identify and characterize any keratinocyte-releasable
factors, which may function as a stop signal for wound
healing, we explored the importance of the fibroblast/
Figure 2
Dose- and time-dependent collagenase stimulatory
effect of KCM on dermal fibroblasts. Panel A,
dermal fibroblasts were treated with various volumes
of KCM (expressed as a percentage of total volume of
KCM added) for 24 h. Total RNA was then extracted
and subjected to dot blot analysis. The blots were
initially hybridized with collagenase cDNA and subsequently
with a cDNA specific for 18S ribosomal RNA
used as a control for RNA loading. Panel B, dermal
fibroblasts were treated with KCM for 0, 3, 6, 12, 24,
and 48 h. Total RNA was then extracted and subjected
to Northern analysis using collagenase cDNA and 18S
ribosomal RNA cDNA as the probes. Panel C shows
the expression of collagenase mRNA in fibroblasts
grown in KCM collected at indicated time points. Prior
to collecting KCM for KDAF purification, KSFM
supplemented with EGF and pituitary extract was
exchanged with our test medium consisting of 49%
DMEM, 49% KSFM, and 2% FBS with no additives.
The conditioned medium was then collected every
48 h thereafter up to 24 d and tested for a collagenase
stimulatory effect on fibroblasts by dot blot analysis
using 18S ribosomal RNA as loading control. To
correct for RNA loading, the expression of collagenase
and 18S was determined by densitometry and
the ratio of collagenase mRNA/18S ribosomal RNA
was calculated and depicted in Panel C, right panel.
122 : 5 MAY 2004 PROTEIN 14-3-3 s AS COLLAGENASE-STIMULATING FACTOR 1191
keratinocyte interaction in ECM modulation. In this study,
we found a keratinocyte-releasable collagenase-stimulating
factor for dermal fibroblasts grown in either a two-chamber
co-culture or three-dimensional organotypic co-culture
system. The collagenase-stimulating activity of this factor
was characterized under several experimental conditions
and subsequently identified as a keratinocyte-releasable
form of stratifin or 14-3-3 s. We also demonstrated that
purified recombinant stratifin markedly increases the expression
of MMP-1 mRNA in fibroblasts and this was
consistent with collagenase activity found in conditioned
medium derived from stratifin-treated fibroblasts. This is a
novel role for this protein. The 14-3-3 proteins are a class of
highly conserved molecular chaperones. They are a
ubiquitous family of acidic eucaryotic proteins. There are
seven known mammalian isoforms, a/b g, e, p, s, t, y::, and z:
(Martens et al, 1992). Since the discovery of the first 14-3-3
protein in 1967 (Moore et al, 1967), the members of the 14-
3-3 protein family have been repeatedly re-discovered
based on their new biological activities, primarily in signal
transduction pathways. They have been identified as
activators of tryptophan and tyrosine hydroxylase (Ichimura
et al, 1987; Ichimura et al, 1988) and PKC inhibitors (Toker et
al, 1990). Subsequent studies identified the 14-3-3 proteins
as molecules that interact with PKCs, Raf family members,
and now more than 100 other intracellular proteins with
critical biological functions (Craparo et al, 1997; Yaffe,
2002), including cellular response to DNA damage and cell
cycle regulation (Hermeking et al, 1997; Chan et al, 1999;
Laronga et al, 2000).
The collagenase-stimulating activity of keratinocytereleasable
14-3-3 s was surprising. This is because, of all
the members of the protein 14-3-3 family, the s form is
reported to be vital in preventing mitotic catastrophe after
DNA damage (Chan et al, 1999). The 14-3-3 s protein is also
reported to be a p53-regulator that inhibits G2/M progression
(Hermeking et al, 1997). All of these biological
activities, however, are regarded as being intracellular
interactions and functions of the 14-3-3 proteins. A
releasable form of 14-3-3 protein was shown to be present
in cerebrospinal fluid (CSF) and shown to be associated
with prion diseases such as Creutzfeldt?Jakob disease and
other neurological disorders (Boston et al, 1982; Kurohara
et al, 1999). Stratifin was also included in a catalogue of
proteins found to be secreted by epidermal keratinocytes
(Katz and Taichman, 1999), but no physiological function
was assigned to these proteins. Our finding therefore,
regarding the collagenase stimulatory effect of a keratinocyte-
releasable stratifin for fibroblasts, is the first indication
of a relevant extracellular biological function for at least one
member (s) of this important family of proteins.
As mentioned above, 14-3-3 proteins are considered to
be intracellular proteins with many critical biological
activities. It is therefore assumed that these proteins lack
known amino-terminal ER signal peptides and, as such, the
mechanism by which these proteins become releasable has
yet to be explored. As is the case for annexin II, 14-3-3
proteins cannot pass through the classic endoplasmic
reticulum secretory pathway in order to be released into
CSF and KCM (Boston et al, 1982; Siever and Eridson,
1997) and therefore, they must be released via another
Figure 3
Identification of KDAF by a sequential chromatography. Proteins in
the media collected from KCM every 48 h over a 24-d period were
precipitated by 65% ammonium sulfate and subjected to a three-stage
chromatographic purification. Panel A shows the effects of 22
chromatography eluted samples (sample index) on the expression of
collagenase mRNA (upper panel) and 18S ribosomal RNA (middle
panel) in dermal fibroblasts. Cells treated with keratinocyte conditioned
(KCM) and non-conditioned medium (NCM) were also included and
served as positive and negative controls. Panel B shows the protein
patterns of active fractions 13?16 determined by their collagenase
activities (Panel A) analyzed by SDS-PAGE. In this panel, protein
marker and column fractions containing candidate KDAF protein bands
with 30 and 50 kDa are indicated (arrows on the right).
Table I. Determination of amino acid sequences of candidate
KDAF protein
Amino acid sequence of KDAF
Start-End Peptide
12?27 LAEQAERYEDMAAFMK
69?82 SNEEGSEEKGPEVR
170?195 LGLALNFSVFHYEIANSPEEAISLAK
225?248 DNLTLWTADNAGEEGGEAPQEPQS
Candidate KDAF protein bands were excised from the gel, and
proteomic analysis of amino acid sequences was performed. The
sequences of four peptides listed in the table were matched to those of
stratifin (14-3-3 s protein)
1192 GHAHARY ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY
mechanism that remains to be elucidated. There is now
supporting evidence to indicate that this is not unique to the
14-3-3 protein as there are several well-known releasable
proteins, such as IL-1 (Andree et al, 1992; Corradi et al,
1995), fibroblast growth factor-2 (Albuquerque et al, 1998),
and endothelial cell growth factor (Jaye et al, 1986), which
lack signal peptides and yet are still considered to be
releasable proteins. In an earlier report, we found a secreted
form of annexin II in KCM (Karimi-Busheri et al, 2002). At
that time we assayed for the presence of lactate dehydrogenase
(LDH) in the medium, as a marker of cell lysis
(Korzeniewski and Callewaert, 1983; Decker and Lohmann-
Matthes, 1988), and established that the release of this
protein into the media was not the result of keratinocyte
lysis. Since the same batches of KCM were used in both
studies, we can rule out cell lysis as the cause for release of
stratifin into the KCM. Furthermore, accidental release of
stratifin is also unlikely due to the fact that recombinant
stratifin mimics the collagenase stimulatory function of
KDAF found in keratinocyte conditioned medium.
As overexpression of MMPs has been implicated with
non-healing wounds, one may speculate that the release of
KDAF and subsequent stimulation of MMP-1 in dermal
fibroblasts causes wound healing retardation. The result of
our longitudinal experiments revealed that this is unlikely to
be the case as differentiated keratinocytes express a
markedly high level of KDAF mRNA relative to proliferating
keratinocytes grown in KSFM, a well-established culture
condition in which keratinocytes have the capacity to
proliferate. A densitometry analysis on expression of KDAF
in keratinocytes grown in test medium revealed that the
status of cell differentiation is correlated with KDAF
expression. Although differentiation markers were not
determined during this longitudinal study, our previous
study demonstrated that keratinocytes grown in test
medium express involucrin as early as day 5. This is
consistent with our data shown in Fig 2 revealing that
keratinocytes express greater levels of KDAF between days
6?10 (early stage of differentiation) and 18?24 (late stage of
differentiation) examined. It is not clear why keratinocytes
express KDAF at these time points. Thus, further study is
needed to explore the mechanism by which KDAF is greatly
expressed during these particular time periods. We have
previously demonstrated an inverse regulation between
keratinocyte-releasable wound healing promoting factors
such as TGF-b1 and differentiation of keratinocytes (Ghahary
et al, 2001). This finding indicated that a mono-, but not
multi-layer, form of keratinocyte releases both mitogenic
and collagen-stimulating factors for dermal fibroblasts.
Interestingly, when keratinocyte differentiation was induced
by growing these cells in a test medium consisting of 49%
KSFM, 49% DMEM, and 2% fetal bovine serum (Ghahary
et al, 2001), the level of TGF-b-1 was strongly suppressed in
differentiated keratinocytes at the later time points. Instead,
keratinocytes expressed a very high level of involucrin
mRNA, which has been used as a differentiation marker.
Involucrin is an intermediate differentiation marker that is
expressed mainly by suprabasal keratinocytes (Schoop
et al, 1999).
The importance of mesenchymal?epithelial interactions
in physiological (embryonic development) and pathological
(tumorigenesis) conditions has recently been reviewed
by Angel and Szabowski (2002). Using an in vitro skin
equivalent similar to the one we used in this study, they
elucidated a critical role of the AP-1 complex of C -Jun and
Jun B in the mesenchymal?epithelial interaction in skin by
regulating the expression of IL-1-induced KGF and GMCSF
in fibroblasts. As suggested by these investigators,
appropriate interaction of mesenchymal?epithelial cells is
critical for a proper histoarchitecture of the epidermis. In
our study, we demonstrated that interaction between
keratinocytes and fibroblasts may, in fact, control the extra-
Figure 4
Cloning, expression, and purification of stratifin. The cDNA for
keratinocyte-derived stratifin was prepared from total RNA extracted
from human keratinocytes, expressed in E. coli and affinity purified as
described in Experimental Procedures. Panel A shows the pre-stained
protein marker (lane M), pattern of total protein expressed by GSTstratifin
transformed BL-21-DE3 bacteria (lane 1), extraction of GSTstratifin
expressed in BL-21-DE3 cells (lane 2), affinity purified
GST-stratifin fusion protein (lane 3), and recombinant purified stratifin
(lane 4). Panel B shows the efficacy of either no treatment (C), 49%KCM
(lane KCM) or various concentrations (0.1, 0.25, 0.5, 1.0, 2.0, and 3 mg
per mL, lanes 1?6, respectively) of purified stratifin protein on the
expression of collagenase mRNA in dermal fibroblasts evaluated by
northern analysis. Panel C, collagenase activity in conditioned medium
obtained from stratifin treated (lane S) or untreated (lane U) fibroblasts
was evaluated according to the procedure described in the Experimental
Procedures using type I collagen as a substrate. Type I collagen
with no treatment (C) was also included as a negative control. Note that
conditioned medium derived from stratifin-treated fibroblasts completely
breaks down type I collagen to its three-quarter fragments of the a
(a1 and a2) and b (b1.1 and b1.2) chains of type I collagen.
122 : 5 MAY 2004 PROTEIN 14-3-3 s AS COLLAGENASE-STIMULATING FACTOR 1193
cellular matrix modulation through keratinocyte-releasable,
collagenase-stimulating, 14-3-3 s for dermal fibroblasts.
Considering the fact that 80% of the total protein in skin is
type I collagen, and under physiological conditions degradation
of interstitial collagen (type I, II, and III) is specifically
initiated by collagenase, a tight balance between synthesis
and degradation of collagens is essential for the integrity of
dermal architecture (Johnson-Wint, 1980). In fact, it is well
established that when keratinocytes migrate beneath
wound edges and the wound surface is covered by reepithelialization
in o21 d, only one-third of the anatomically
matched wounds become fibrotic with a characteristic
feature of collagen accumulation, whereas this increases
to 78% when a wound is re-epithelialized in 421 d. This
signifies that re-epithelialization is required for remodelling
of the granulation tissue into a more normal neo-dermis,
during which time the synthesis and degradation of key
ECM proteins such as collagenase, type I, and type III
collagen are modulated (Garner, 1998; Machesney et al,
1998). Thus, our findings suggest that upon epithelialization,
keratinocytes may in fact release stratifin in order to
modulate the expression of ECM by dermal fibroblasts
during the phase of tissue remodelling. Although the
expression of other 14-3-3 proteins in keratinocytes has
been reported (Olsen et al, 1995), it remains to be seen
whether the releasable forms of these proteins have a
similar collagenase-stimulating effect on dermal fibroblasts.
In conclusion, the finding of this study for the first time
suggests and confirms that recombinant 14-3-3 s mimics
the collagenase stimulatory effect of KCM in fibroblasts and
this emphasizes the critical role of mesenchymal?epithelial
cell interaction in ECM modulation in these cells. Finally, it
remains to be seen whether keratinocyte-releasable 14-3-3
s is solely responsible or if it represents a subset of the total
KDAF activity found in KCM.
Materials and Methods
Clinical specimens and cell culture Following informed consent,
skin punch biopsies were obtained from patients undergoing
elective reconstructive surgery, under local anesthesia, according
to a protocol approved by the University of Alberta Hospitals
Human Ethics Committee. Biopsies were collected individually and
washed three times in sterile Dulbecco''s modified Eagle''s medium
(DMEM) (Gibco, Grand Island, New York) supplemented with
antibiotic?antimycotic preparation (100 mg per mL penicillin, 100 mg
per mL streptomycin, 0.25 mg per mL amphotericin B) (Gibco).
Specimens were dissected free of fat and minced into small pieces
less than 0.5 mm in diameter, washed six times with DMEM, and
distributed into 60
15 mm Petri dishes. Cultures of fibroblasts
were established as previously described (Karimi-Busheri et al,
2002). Upon reaching confluence, the cells were released by
trypsinization, split for subculture at a ratio of 1:6, and reseeded
into 75 cm2 flasks. Fibroblasts at passages 3?7 were used in all
experiments conducted in this study.
To establish cultured keratinocytes, the procedure of Rheinwald
and Green (1975) was used for cultivation of human foreskin
keratinocytes using serum-free keratinocyte medium (Gibco)
supplemented with bovine pituitary extract (50 mg per mL) and
EGF (5 ng per mL). These additives were used only to establish
keratinocytes in cultures. Thus, to eliminate any effects of EGF
and/or pituitary extract on our findings, obtained from co-culturing
keratinocytes with fibroblasts, keratinocyte serum-free medium
(KSFM) supplemented with EGF and pituitary extract was
exchanged with our test medium consisting of 49% DMEM, 49%
KSFM, and 2% FBS with no additives, a medium in which
keratinocytes undergo differentiation with time. In this system, both
keratinocytes and fibroblasts remain viable and any factor released
from keratinocytes can diffuse through the 0.4 mm porous
membrane separating the two chambers. Primary cultured keratinocytes
at passages 3?5 were used.
Keratinocyte fibroblast co-culture system In order to identify a
keratinocyte-derived anti-fibrogenic factor for dermal fibroblasts,
we had previously established (Karimi-Busheri et al, 2002) a
keratinocyte/fibroblast co-culture system in which keratinocytes
and fibroblasts were grown in the upper and lower chambers of the
system, respectively. Since there was only a permeable membrane
separating these cells, fibroblasts in the lower chamber could be
exposed to any soluble factor, which could be released from
keratinocytes. As an index for the anti-fibrogenic effects of
keratinocyte-derived factor(s) on dermal fibroblasts, total RNA
was extracted from fibroblasts grown in the lower chamber and the
expression of collagenase mRNA was evaluated by northern
analysis. Fibroblasts grown alone and fibroblasts grown in a
fibroblast/fibroblast co-culture system were also used as controls.
Extraction of cellular RNA and northern analysis To identify the
presence of any factor, which might function as an anti-fibrogenic
factor for dermal fibroblasts, keratinocytes in the top chamber and
fibroblasts from the bottom chamber were harvested separately
and pelleted by centrifugation at 300 g for 10 min. Pellets were
then lysed with 500 mL of 4 M guanidium isothiocyanate (GITC)
solution and total RNA from each group was isolated by the
guanidium isothiocyanate/CsCl procedure of Chomczynski and
Sacchi [1987] using phenol: chloroform (1:1), followed by chloroform:
isoamyl alcohol (49:1). Total RNA from each individual
fibroblast culture was then separated by electrophoresis (10 mg
per lane) on a 1% agarose gel containing 2.2 M formaldehyde and
blotted onto nitrocellulose filters. Filters were then baked under
vacuum for 2 h at 801C and pre-hybridized in a solution containing
50% formamide, 0.3 M sodium chloride, 20 mM Tris HCl (pH 8.0), 1
mM EDTA, 1
Denhardt''s solution (1
¼0.02% bovine serum
albumin, Ficoll and polyvinylpyrrolidone), 0.005% salmon sperm
DNA, and 0.005% poly (A) for 2?4 h at 451C. Hybridization was
performed in the same solution at 451C for 16?20 h using cDNA
probes for either human collagenase, 18S ribosomal RNA or 14-3-
3 s. The probes were labelled with 32P-a-dCTP (DuPont Canada,
Streetsville, Mississauga, Ontario, Canada) by nick-translation.
Filters were initially washed at room temperature with 2
SSC
(1
¼0.15 M sodium chloride, 0.015 M sodium citrate) and 0.1%
SDS for 30 min, then for 20 min at 651C in 0.1
SSC and 0.1%
SDS solution. Autoradiography was performed by exposing Kodak
X-Omat film (Eastman Kodak, Rochester, NY) to the nitrocellulose
filters at 701C in the presence of an enhancing screen. The cDNA
probes for collagenase and 18S ribosomal RNA were obtained
from the American Type Culture Collection (Rockville, Maryland).
The cDNA for 14-3-3 s was obtained by extracting keratinocyte
total RNA and was amplified by RT-PCR in a procedure described
in the following sections.
Collagenase activity To determine the effects of KDAF on
collagenase activity, conditioned medium from either keratinocytes
alone, fibroblasts alone, or fibroblasts co-cultured with keratinocytes
was replaced with serum-free medium supplemented with 25
mg per mL conconavalin A (to increase enzyme production)
and 0.5% (wt/vol) lactalbumin hydrolysate (Sigma Aldrich, Inc.,
St. Louis, MO). After 2 d, the conditioned medium from each
experimental condition was collected, centrifuged at 1000
g for
10 min., and stored at 41C. In another experiment, the total
proteins in the conditioned media derived from purified recombinant
stratifin-treated (S) and untreated (U) fibroblasts were also
collected and evaluated for collagenase activity. The collagenase
assay was carried out as described elsewhere (Khorramizadeh
1194 GHAHARY ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY
et al, 1999) with slight modifications. Briefly, the total proteins
presented in conditioned media were precipitated by ammonium
sulfate and the precipitates collected by centrifugation, dissolved
in assay buffer (0.05 M Tris, 0.2 M NaCl, 0.005 M CaCl2, 0.02%
sodium azide, pH 7.4), and then dialyzed overnight against 4 liters
of the same buffer. The final volume of each sample was adjusted
according to the cell number. Procollagenase was activated
proteolytically with trypsin (10 mg per mL), and soybean trypsin
inhibitor (100 mg per mL) was used to inactivate the trypsin. Acetic
acid soluble collagen (50 mg in 25 mL) from bovine skin was
incubated with the activated enzyme solution in the presence of
1 M glucose for 15?24 h. The products from digested collagen
were then separated by electrophoresis using 5% acrylamide gel
containing sodium dodecyl sulfate (SDS). The gels were stained
with Coomassie blue and the b1.1 (3/4) and b1.2 (3/4) fragments
were identified as KDAF-induced collagenase digestive products.
Isolation of KDAF The media collected from keratinocyte-conditioned
medium every 48 h over a 24-d period were subjected to a
65% ammonium sulfate precipitation followed by centrifugation at
10,000
g for 15 min. The pellet was re-suspended in a minimum
volume of buffer A (10 mM sodium phosphate, pH 7.3, 150 mM
NaCl, and 4 mM 2-mercaptoethanol, protease inhibitor cocktail
(Sigma, St Louis, Missouri)) and dialyzed overnight at 41C in the
same buffer. A total of 75 mL of extract, 285 mg, was applied on a
5 mL Amersham, Biosciences, Piscataway, NJ HiTrap SP column,
washed with 25 mL of buffer A, and eluted with a 90 mL linear
gradient of 150?800 mM NaCl in thirty 3 mL fractions. Fractions
containing KDAF activity eluted at approximately 0.35 M salt
concentration. They were pooled and concentrated by 65%
ammonium sulfate. The precipitate was dissolved in 500 mL of
buffer A and 50 mL was applied onto a Superdex-75 PC 3.2/30 gel
filtration column attached to a SMART micropurification system
(Amersham Pharmacia Biotech). Proteins on the column were
eluted with buffer A and collected in 80-mL aliquots. KDAF activity
eluted in fractions 7 and 8. This process was repeated several
times. Pooled active fractions were dialyzed against buffer B
(50 mM Tris-HCl, pH 7.5, 50 mM NaCl, 1 mM DTT and protease
inhibitors) and loaded onto a Mono Q PC 1.6/5 SMART column
(Amersham Pharmacia Biotech). Protein was eluted at a flow rate
of 100 mL per min with a 3 mL linear gradient of 50?800 mM NaCl in
thirty 100 mL fractions. The highest collagenase mRNA-stimulating
activity was eluted in fractions 13?16 and these fractions were
subsequently examined by electrophoresis on a 10% SDSpolyacrylamide
gel and stained with Coomassie Blue.
Mass spectrometry Proteins in candidate bands with apparent
sizes of 30 and 50 kDa were excised from SDS/polyacrylamide
gels were subjected to trypsin digestion according to a published
procedure (Shevchenko et al, 1996) and MS analysis (Dai et al,
1999). Samples from candidate protein bands were prepared
according to ACB Proteomics Resource Laboratory''s in-house
protocols. In brief, these protocols include washing, reduction,
alkylation, tryptic digestion, and extraction of tryptic peptides from
the gel spots. Matrix-assisted laser adsorption mass spectrometry
(MALDI MS) and matrix-assisted laser adsorption/ionization
post-source decay mass spectrometry (MALDI PSD MS) were
performed on a Voyager Elite MALDI MS instrument (Voyager
Elite, PerSeptive Biosystem, Framingham, Massachusetts) equipped
with a delayed extraction (DE) device. A two-layer method was
used for MALDI MS analysis in which 1?2 mL of first layer solution
(10 mg per mL of 4-hydroxy-a-cyanocinnamic acid (HCCA) per mL
of 20% methanol/acetone (vol/vol)) was deposited onto a probe
tip, and evaporated to form a thin matrix layer, and then 0.5?1 mL of
gel extract from 50% acetonitrile or 40% methanol saturated by
HCCA was deposited onto the first layer, allowed to air dry, and
washed three times with water. The PSD spectra were recorded
in the PSD mode of the Voyager Elite instrument. Nanoelectrospray
(NanoES) ion trap MS was performed on an Esquire-LC
ion trap spectrometer (Hewlett-Packard, Reno, Nevada) with
NanoES interface. Spectra were acquired over the mass range
200?2200 kDa.
Peptide extracts were analyzed on a Bruker REFLEX III time of
flight mass spectrometer (Bremen/Leipzig, Germany, Serial# FM
2413) using MALDI in positive ion mode. The obtained peptide
maps were used for database searching to identify proteins.
Furthermore, for each sample, 1?3 selected peptides were
fragmented using MALDI MS/MS analysis carried out on a PE
Sciex API-QSTAR pulsar (MDS-Sciex, Toronto, Ontario, Canada,
Serial# K0940105). The obtained partial sequence information for
each peptide was used to either confirm or correct the previously
obtained results from the peptide map search.
Cloning, expression, and purification of recombinant KDAF To
clone stratifin (14-3-3 protein s isoform) cDNA, total RNA was
prepared by the acid?guanidium?phenol?chloroform method from
human keratinocytes. cDNA was then synthesized with oligo(dT)
primer and MMLV reverse transcriptase (Gibco-BRL Grand Island,
NY). Samples were then incubated at 421C for 60 min, and the
reaction was terminated by heating at 701C for 15 min and followed
by rapid chilling on ice. PCR reaction was carried out using KDAFs
primer (sense: 50-GAATTCCCCAGAGCCATGGAGAGAGCC-30;
antisense: 50-CTCGAGTGGCGGGCAACACTCAGCTC-30). PCR reaction
was carried out for 30 cycles and the PCR product was
separated by electrophoresis on 1% agarose gel. The separated
DNA product was stained with ethidium bromide and visualized
under UV light.
DNA in the agarose gel was purified with a QIAEX II gel
extraction kit according to the manufacturer''s instructions (Qiagen
Inc., Mississauga, Ont., Canada). Purified DNA was then digested
with EcoRI/XhoI for 2 h at 371C. The digested products were
separated by electrophoresis on a 1% agarose gel, and the
specific DNA band related to KDAF-s was purified by the QIAEX II
gel extraction kit. Finally, the purified DNA was ligated into a pGEX-
6P-1 expressing vector using GST fusion protein (Amersham,
Biosciences).
For bacterial transformation, the ligated products were transformed
to competent XL0blue-1 cells with the regular heat-shock
transformation method. Positive clones were identified by the size
of restriction enzyme-digested products. DNA sequence was
confirmed by fluorescence dNTP sequence analysis. The plasmid
DNA containing KDAF-s was transformed into protein expressing
bacteria BL-21 (DE3) (Novagene, Madison, WI).
For protein expression, a single positive clone was grown in 100
mL of LB medium containing 50 mg per mL of ampicillin for 4?6 h at
291C until an OD600 nm of 0.4?0.6 was reached. Bacteria were then
diluted to 1:10 with fresh LB medium grown in the presence of 0.1
mM of IPTG for 24 h. For protein purification, bacteria were
collected by centrifugation and lysed with 50 mM Tris-HCl (pH 7.4)
containing 10 mM EDTA, 5 mM EGTA, Protease cocktail (Sigma),
1% Triton X-100, and 0.5% IGEPAL CA630. Cell lysate was passed
through a Glutathione Sepharo

Aziz Ghahary*, Feridoun Karimi-Busheri?, Yvonne Marcoux*, Yunyaun Li*, Edward E Tredget*, Ruhangiz Taghi Kilani, Liang Li?, Jing Zheng?, Ali Karami*, Bernd O Keller? and Michael Weinfeld?