EXPERIMENTAL
CELL
RESEARCH
193,
93-100
(1991)
induction of Extracellular Matrix Gene Expression in Normal Human Keratinocytes by Transforming Growth Factor ,O Is Altered by Cellular Differentiation THOMAS M. VOLLBERG, SR., MARGARET D. GEORGE,AND ANTON M. JETTEN' Cell Biology Group, Laboratory of Pulmonary Pathobiology. National Institute of Environmental P.0. Bon 12233, Research Triangle Park, North Carolina 27709
Changes in epithelial substrata have been related to the cellular capacity for proliferation and to changes in cellular behavior. The effect of TGF/31 on the expression of the basement membrane genes, fibronectin, laminin Bl, and collagen (~1 (IV), was examined. Northern analysis revealed that treatment of normal human epidermal keratinocytes with 100 pM TGF/31 increased the expression of each extracellular matrix (ECM) gene within 4 h of treatment. Maximal induction was reached within 24 h after treatment. The induction of ECM mRNA expression was dose dependent and was observed at doses as low as l-3 pM TGF/31. Incremental doses of TGF@l also increased cellular levels of fibronectin protein in undifferentiated keratinocytes and resulted in increased secretion of fibronectin. Squamous-differentiated cultures of keratinocytes expressed lower levels of the extracellular matrix RNAs than did undifferentiated cells. Treatment of these differentiated cells with TGFPl induced the expression of fibronectin mRNA to levels seen in TGF&treated, undifferentiated keratinocytes but only marginally increased the expression of collagen (Al and laminin Bl mRNA. The increased fibronectin mRNA expression in the differentiated keratinocytes was also reflected by increased accumulation of cellular and secreted fibronectin protein. The inclusion of cycloheximide in the protocol indicated that TGFP induction of collagen (Al mRNA was signaled by proteins already present in the cells but that TGFP required the synthesis of a protein(s) to fully induce expression of fibronectin and laminin Bl mRNA. The differential regulation of these genes in differentiated cells may be important to TGFfi action in regulating reepithelialization. (c 1991 Academic Press, Inc.
INTRODUCTION The proliferation of the epidermal epithelium is tightly controlled to ensure that cell renewal and cell ’ To whom correspondence dressed.
and
reprint
requests
should
be ad-
Health Sciences,
loss occur at the same rate. During wound healing this process of cell growth and maturation is altered so that potential stem cells may migrate to the denuded area [l]. Once in place the proliferation of these cells is accelerated to ensure the replacement of the lost stem cell population and the regeneration of the protective layers of differentiating cells. At the molecular level these events would appear to result from the interplay of a variety of growth factors which influence, both positively and negatively, the proliferation and maturation of the cells [2]. In uitro, epidermal keratinocytes require epidermal growth factor and insulin for proliferation [2, 31. Confluence, phorbol esters, and increased calcium concentration can induce these cells to undergo terminal cell division and to express a number of differentiation markers [3-61. Transforming growth factors /3 (TGFPs), a family of hormonally active polypeptides, have been implicated as controlling factors in both cellular proliferation and differentiation (see review [7, 81). TGFPs are synthesized as prepropeptides and secreted as inactive homodimers which are activated by proteolysis. TGFPs elicit in responsive cells changes in proliferation, differentiation, and genetic expression [7, 91. The response is cell specific; some cells are stimulated to proliferate, whereas others are growth arrested by TGF/3 [g-11]. Three types of high-affinity receptors for TGFP have been identified on the surface of responsive cells [la]. The intracellular signaling which is triggered by the binding of TGFfl to these receptors has not been elucidated. In epithelial cells TGF/3 Type 1 (TGFPl) inhibits cell proliferation and induces a change in the undifferentiated cells from an angular, raised morphology to a rounded, flattened shape [ 131. In human epidermal keratinocytes TGFP inhibition of cell proliferation is reversible and does not lead to differentiation [9,14]. This is in contrast to other epithelial cell systems where TGFP has been described as an inducer of terminal differentiation [15, 161. In keratinocyte cultures TGF(31 stimulates cell production of fibronectin protein and in-
93 All
Copyright 0 1991 rights of reproduction
0014-4827/91 $3.00 by Academic Press, Inc. in any form reserved.
94
VOLLBERG.
GEORGE.
creases cell motility [18, 191. Because of TGFPs effects on epithelial proliferation and cellular behavior, a physiologic role for TGFP has been proposed in the control of postinjury regenerative growth of skin epithelia [6, 13, 18, 201. The binding and interaction of epithelial cells with extracellular matrix proteins influences their growth and differentiation. For example, the binding of fibronectin to cellular receptors of the integrin family inhibits the differentiation of keratinocytes [21]. Further, the binding of fibronectin to epithelial cancer cells has been shown to promote cell spreading and motility in competitive antagonism with the binding of laminin to the cells [22]. Other studies have demonstrated the requirement of cell motility for sustained proliferation of keratinocytes grown in colonies [23]. In fibroblasts TGFP has been shown to induce the mRNA expression of the extracellular matrix genes, collagen I and fibronectin [20, 24-281. Thus, we hypothesized that TGF/3 could be altering keratinocyte extracellular matrix expression as an intermediate step in affecting cell behavior. In this study the expression of the extracellular matrix genes, fibronectin, laminin Bl, and collagen cul(IV), is demonstrated to be inducible by TGFP in undifferentiated keratinocytes. Since differentiated cells would be present at the wound border, we also examined TGFfi effects in differentiated cells and report that these cells responded to TGFP by increasing production of fibronectin but that the process of differentiation caused the cells to acquire resistance to TGFP induction of laminin Bl and collagen al(IV) expression. The induction of ECM proteins and the alteration of this induction by differentiation may represent a mechanism by which TGFP alters epithelial cell behavior to effect reepithelialization during the healing process. MATERIALS
AND METHODS
Normal human keratinocytes were obtained as cryopreserved primary culture cells from Clonetics Corp. (San Diego, CA) and cultured according to the suppliers protocols in KGM serum-free medium (Clonetics) in tissue culture plastic flasks and dishes (Costar, Cambridge, MA). Subculture of the cells was continued for a maximum of three passages. On the final subculture before use the cells were seeded at a density of 2-3 X lo3 cells/cm’ into either 60.mm dishes or 150-mm dishes for protein or RNA analysis, respectively. IJndifferentiated cells represented conditions under which culture was continued for 5-6 days and the cell population had expanded to 8-10 X lo3 cells/cm2 before treatment or collection. The number of cells in these cultures was approximately 20% of the confluent cell density. Differentiated cells were obtained by culturing cells to confluent density over a lo- to 12.day period and were allowed morphologic differentiation over 223 additional days of culture before treatment or collection. Expression of ECM proteins by human keratinocytes was measured by Western blot analysis. Following treatment, the protease inhibitors, leupeptin and aprotinin, were added to the conditioned media at 10 Gg each per milliliter and the media were stored at -20°C. The cultures were rinsed twice with ice-cold PBS. The cells were
AND
JETTEN
scraped into PBS and pelleted by centrifuging 6OOgfor 10 min at 4°C. The pellet was resuspended in 0.5 ml of 50 mM TrissCl, pH 8.0, 125 mM NaC1, 0.5% (v/v) NP-40, 0.1 mM PMSF, and 10 pg/ml each of aprotinin and leupeptin (Sigma Chemical Co., St. Louis, MO) andcell lysis was allowed to occur for 30 min on ice. The cell lysate was then sonicated at 20 PW for three 10-s bursts and microcentrifuged (10,OOOg) for 10 min at 4°C. Protein in the cell lysate supernatant was quantitated with the Bio-Rad protein assay (Bio-Rad Laboratories, Richmond, CA) using bovine serum albumin as a standard. Protein samples were brought to 1X with SDSPAGE sample buffer, heatdenatured, and electrophoresed in a SDS-lo% polyacrylamide slab gel as described by Laemmli [29]. The proteins were electroblotted to a 0.45.brn Nitroplus 2000 membrane (Micron Separations, Inc., Westborough, MA) as described [30]. Fibronectin was immunodetected by a 1:X0 dilution of human fibronectin antiserum (rabbit, Biomedical Technologies, Inc., Stoughton, MA) using Auroprobe BL plus kit for immunogold/silver staining (Janssen Biotech, Olen, Belgium) in a protocol supplied by ,Janssen Biotech. Expression of mRNA for ECM genes was examined in a Northern blot protocol using cDNA probes. Total cellular RNA was isolated from 5-10 150.mm dishes by scraping the cells into guanidium isothiocyanate lysis buffer and pelleting of the RNA through a CsCl cushion [31]. The RNA was solubilized in 100 mM sodium chloride/l mM EDTA/lO mM TrisHCl, pH 8.0, extracted with chloroform:isoamyl alcohol (24:1), and precipitated in 250 mM sodium acetate/70% ethanol at ~20°C. The purified RNA was dissolved in water and quantity and purity were assessed by absorbance at 260 nm (1 A,,,,, unit = 40 pg single-stranded RNA/ml) and at 280 nm (A,,,,,/A,,,, greater than 1.8 for pure RNA). Total RNA (lo-20 p(p) was separated by electrophoresis in a 0.66 M formaldehyde/l% agarose gel [32] and transferred to Nytran nylon membrane (Schleicher & Schuell, Keene, NH) in a Vacublot apparatus using the manufacturer’s protocol (American Bionetics, Inc., Hayward, CA). Nucleic acid was permanently fixed to the membrane by uv cross-linking [33]. Ribosomal RNA and a 0.24. to 9.5kb RNA ladder (BRL, Life Technologies, Inc., Gaithersburg, MD) were used as size markers, RNA blots were hybridized with 224 rig/ml heat-denatured DNA probe in hybridization buffer (50% formamide/5X SSPE/2x Denhardt’s solution/l% SDS/ 100 pg/ml denatured salmon testes DNA) at 42°C for 16-20 h after prehybridization overnight at 42°C in the hybridization buffer. Following hybridization ECM probes were washed to a final stringency of 60°C in 0.1X SSC/O.l% SDS. RNAs hybridizing to cDNA were detected by autoradiography with enhancing screens (Lightning Plus, DuPont Co., Wilmington, DE) and Kodak XAR-5 film (Eastman Kodak, Rochester, NY) at --70°C. The expressions of the various RNAs were quantitated from the developed autoradiograms by densitometry with the Bio-Rad Model 620 video densitometer and 1-D analyst software (Bio-Rad Laboratories). For reuse of blots with alternative probes, previously detected probes were erased from the blot by washing at 70-75°C in 5 mM TrissHCI, pH 8.0/0.2 mM EDTA/O.lX Den hardt’s solution/0.05% (w/v) sodium pyrophosphate prior to reprobing. Radioactively labeled DNA prohes were synthesized from purified cDNA inserts using a random prime labeling kit (Boehringer-Mannheim Biochemicals, Indianapolis, IN) according to the manufacturer’s protocol with [tu-32P]dCTP (-3000 Ci/mmol, Amersham Corp., Arlington Heights, IL). The specific activities of the DNA probes were in the range of 4-8 X 10’ cpm per microgram DNA. Human fibronectin cDNA [34] was a 1.7.kb PatI-PuuII restriction fragment of plasmid pFH1.54 (from Dr. A. Kornblihtt). Human laminin Bl cDNA [35] was a 1.2.kb EcoRI restriction fragment of plasmid ~1236 (from Dr. Y. Yamada). The transglutaminase Type I probe was the EcoRI cDNA insert of clone pTG7 [36]. Purified cDNA inserts of the squamous-cell-specific clones SQlO and SQ37 were prepared as described previously [37]. Human collagen cxl(IV) cDNA [38] was a 2.6.kb PsB restriction fragment of plasmid pHT-21 (from Dr. A. Ganguly and Dr. D. Prockop). Chicken glyceraldehyde S-phosphate dehy-
TGFli
INDUCTION
OF ECM
HOURS IOOpH TGF-&
0 ”
4 +
8 +
12 +
18 +
24 +
45 +
9.5kb -
24 -
Fibronectin
7.5 -
285 4.4 z 9.5 7.5 285
Lwdnin
81
4.4 zz 9.5 75-
Coliagen
al(lY)
4.4 -
GENES
95
IN KERATINOCYTES
that which produced a similar image with the fibronectin probe. The dose-dependent. effect of the TGFfl treatment was measured at 24 h by analysis of fibronectin, laminin Bl, and collagen al(IV) mRNA expression (Fig. 2). AS was seen during the time course ECM gene expression was increased in TGF@-treated keratinocytes (Fig. ZA). Laminin Bl and collagen al(IV) mRNA once again required longer exposure of the au~oradiograms for detection and therefore were relatively less abundant than fibronectin mRNA in the untreated and TGFpl-treated keratinocyt.es (exposures of 5 and 3 days, respectively, versus 14 h). Expression of GPDH mRNA in these cells was unaffected by the TGFfl treat,ment. When the expression of the mRNAs was quantitated by densitometry of the autoradiograms, the induction of ECM gene
GPDH
FIG. 1. Time course of TGF8 induction of extracellular matrix genes in human keratinocytes. IJndifferentiated cultures of normal human epidermal keratinocyt.es were treated with 100 pA4 porcine platelet TCF@l (R&D Systems, Minneapolis, MN) and RNA was isolated after the indicated incubation time. Control cultures received no treatment and were harvested at the indicated times. Total RNA (10 pg) from each sample was examined by Northern analysis for expression of ECM genes as described under Materials and Methods. Each probe was hybridized individually with the blot and stripped before use of the next probe. The aut(~radiographs were exposed as follows: ~bronec~in, 8 h; laminin Bl, 5 days; collagen cul(fV), 7 days; and GPDH, 18 h.
drogenase (GPDH) was a 1.12-kb &I mid pGAD 28 [ 391.
restriction
fragment
TGF-p,
A
phff 0
1
3
IO
30
100 3m
Fibroneciin
Laminin
Bi
of plas-
RESULTS
In order to examine if TGFP was inducing the expression of ECM genes in human epidermal keratinocytes, the mRNA expression of the extracellular matrix proteins, laminin, collagen cyl(IV), and fibronectin, was measured by Northern analysis. Total cellular RNA was collected during a 45-h time course following treatment of undifferentiated keratinocyte cultures with 100 pM TGFPl. In untreated control cells (0 and 24 h no addition, Fig. 1) each of the ECM genes were expressed and each was induced to higher mRNA expression by TGFP treatment. Densitometric scanning of the autoradiograms revealed that fibronectin mRNA levels increased 50% within 4 h and were increased maximally (24-fold) at 45 h after treatment. The expression of laminin Bl and collagen alfIV) mRNA was also induced by TGF@ during the time course, although the levels of these mRNAs were maximal at 18 h (5-fold and l7-fold increases, respectively). Both laminin Bl and collagen (ul(IV) mRNA appear to be less abundant than fibronectin mRNA since the laminin Bl and collagen cul(IV) autoradiograms required exposures of 15 and 20 times
GPDH
FIG. 2. Relation of TGF$ dose to ECM gene expression. Following 24 h of treatment with the indicated doses of TGFBl, total RNA was isolated from undifferentiated normal human keratinocytes and examined for expression of ECM genes. (A) Northern blot. Total RNA (10 rgi from each sample was fractionated and transferred to a Nytran membrttne. Expression of each mRNA was probed as described in t.he legend t.o Fig. 1. The aut.oradiographs were exposed as fotlows: fibronectin, 14 h; laminin Bl, 3 days; collagen al(W), 5 days; and GPDH, 24 h. (B) Quantitation of the induction of mRNA expression. The autoradiograms were densitometrically scanned as described under Materials and Methods. The expression of each mRNA is plotted relative to its own expression in untreated keratinocytes. The RNA loaded to each lane of the gel was equalized by comparing the expression of GPDH mRNA in each sample.
96
VOLLBERG, U Hours 1OOpM TGF-fl,
0 -
GEORGE,
D 40 +
0 -
48 +
48 _ Fibronectin
Laminin El
Collagen al(W)
SQlO
Transglutaminase Type 1
GPDH
FIG. 3. Effect of differentiation on the regulation of extracellular matrix genes by TGF@. 1Jndifferentiated (U) and differentiated (D) cultures of normal human epidermal keratinocytes were obtained as described under Materials and Methods. Cells were treated with 100 pM TGFfil as indicated and total RNA (10 pg each) was analyzed by Northern blot analysis as indicated in the legend to Fig. 1.
expression was concentration-dependent and evident at concentrations as low as 1 pM TGF/31 (Fig. 2B). Maximal expression of ECM genes was induced by TGFP at a concentration of lo-30 pM (11.9-, 4.6-, and 10.8.fold increases for fibronectin, laminin Bl, and collagen oll(IV) mRNAs, respectively). This induction was halfmaximal for fibronectin and collagen ~ul(1V) expression at approximately 3 pM TGFPl and for laminin Bl mRNA at 1 pM TGFPl. In the preceding experiments the effects of TGFfi on the induction of ECM gene expression were measured in undifferentiated, proliferating keratinocytes which are susceptible to TGFP-induced cell growth arrest. To determine if the increase in ECM gene expression was correlated with the arrest of cell growth (as opposed to other actions of TGFP), keratinocytes were cultured to confluent cell density. Under these conditions the cells lose all proliferative potential and terminally differentiate. Total cellular RNA was then analyzed for the expression of ECM genes in these nonproliferative cells. The differentiated character of the keratinocytes was confirmed by their expression of the transglutaminase Type I and the squamous-cell-specific mRNAs, SQlO and SQ37 (Fig. 3, lanes 3-5). Total RNA from undifferentiated cells expressed very low levels of these mRNAs (Fig. 3, lanes 1-2). The expression of the fibronectin mRNA in the differentiated keratinocytes was sixfold
AND
.JETTEN
less than the level of expression seen in untreated, proliferating, undifferentiated cells. The expression of laminin Bl and collagen oll(IV) mRNA was not detected in the untreated, differentiated keratinocytes. Thus, growth arrest as caused by cell confluence was insufficient for the induction of ECM gene expression in the keratinocytes and, in fact, reduced the expression of fibronectin, collagen al(W), and laminin Bl mRNA. Next, differentiated keratinocytes were treated with TGFfll for 48 h to determine if TGF/31 could induce ECM genes in these already growth-arrested cells. Fibronectin, laminin Bl, and collagen cul(IV) mRNA were each increased by the treatment with 100 pM TGFPl. After TGF/3 treatment the fibronectin message in the differentiated cells reached a level of expression similar to that seen in the TGFfll-treated, undifferentiated cells (sevenfold and fourfold, respectively, above the expression seen in untreated, undifferentiated cells). In contrast, the TGFB induction of laminin Bl and collagen (ul(IV) mRNA in the differentiated keratinocytes resulted in levels of expression which were, respectively, 0.6 and 1.2 relative to their expression in untreated, undifferentiated keratinocytes. Untreated, differentiated keratinocytes collected at the same time as the TGF@ltreated differentiated cells (Fig. 3, lane 5) expressed the same levels of ECM mRNA as differentiated cells collected prior to the 48-h TGFP treatment (Fig. 3, lane 3). The induction of fibronectin, collagen al(IV), and laminin Bl mRNA in response to TGF/X was tested in the presence of the protein synthesis inhibitor, cycloheximide, to determine if TGFfl was inducing these changes directly or indirectly through the synthesis of cellular protein(s). Detectable mRNA levels were seen for fibronectin, collagen al(IV), laminin Bl, and GPDH in the controls as well as in the TGFpl-induced cells (Fig. 4). Densitometry of GPDH mRNA levels in
2.5pglml
1OOpM TGF-P, Cycloheximlde
_ -
+ -
_ +
+ + Fibronectin
Laminin Bl
Collagen al(W)
GPDH
FIG. 4. Etfect of cycloheximide treatment on the TGF@ induction of fibronectin and laminin Bl mRNA expression. Undifferentiated normal human keratinocytes were untreated or treated with 100 pM TGFDI in the absence or presence of 2.5 pg/ml cycloheximide for 7 h. This concentration of cycloheximide inhibited protein synthesis by greater than 95%. Northern analysis of 10 pg total RNA from each treatment was as described in the legend to Fig. 1.
TGFd
INDUCTION
OF ECM
each condition varied by less than 20%. When undifferentiated keratinocytes were treated for 8 h with 100 pM TGF/31, fibronectin mRNA levels increased 2.3fold over the level in untreated cells. Laminin Bl expression was increased 1.9-fold by the TGF/31 treatment. Inclusion of cycloheximide (2.5 pg/ml) in the medium with TGF/31 partially blocked the increase in fibronectin mRNA and ablated the increase in laminin Bl mRNA expression (1.3 for fibronectin and 1.0 for laminin Bl relative to control cells). Cycloheximide alone decreased laminin Bl expression to 0.9 of the level in untreated cells and caused fibronectin expression to fall to 0.7 of the untreated level. Constitutive expression of fibronectin may require a labile factor whose absence in the cycloheximide-treated cells may account for the reduced response to TGFP treatment. TGFP induced collagen crl(IV) mRNA expression 14fold over the untreated controls in the absence of cycloheximide. In the presence of cycloheximide, the collagen cul(IV) expression was equally induced by treatment with TGF/3 (l&fold over the untreated control). Cycloheximide alone had no effect on collagen ul(IV) mRNA expression (1.4 of the mRNA level in untreated cells). In addition to inducing the levels of ECM gene expression in cellular mRNA, TGFP treatment of the keratinocytes increased the level of cell-associated ECM protein. Increased immunoreactivity to anti-fibronectin antibodies was seen after 24 h of treatment with TGFPl (Fig. 5). As shown for the mRNA expression of fibronectin, TGFPl induction of fibronectin protein was dose dependent. Maximal levels of the protein were detectable at TGFPl concentrations greater than 10 PM. No increase in fibronectin protein was discernible after treatment with 1 pM TGFfll. Next the effect of differentiation on the expression of fibronectin protein was examined (Fig. 6). At 48 h after treatment with 100 pM TGFP the induction of cell-associated fibronectin was still evident in the undifferentiated cells (Fig. 6A). Lysates of differentiated NHEK cells expressed fibronectin protein at levels similar to those seen in the proliferating undifferentiated cells. Treatment of these cultures with 100 pM TGFP increased the amount of fibronectin in these differentiated cells. Untreated cells collected after the same 48-h period showed no change in fibronectin from the levels seen at the time of culture when TGFP treatment was initiated. Upon collection of these cells for preparation of cell lysates, the conditioned media was reserved. The amount of fibronectin protein was measured in conditioned media from each culture (Fig. 6B). Fibronectin in the conditioned media was detected in all of the samples from both proliferative cultures and confluent, growth-arrested, differentiated cultures. More fibronectin protein accumulated in the medium from the proliferating cultures than from the squamous, differentiated cultures (Fig. 6B). This is
GENES
97
IN KERATINOCYTES TGFp (PM) 0
30
3
I
IO
300
100
0 MW(X10-3)
A
-
200
-65
B
-
200
-
65
-45
FIG. 5. Increased accumulation of cell-associated fibronectin by increasing doses of TGFB. Undifferentiated normal human keratinocytes in a 60-mm culture dish were treated with varying doses of TGF/fl as indicated for 24 h. The cells were lysed by scraping directly into electrophoresis sample buffer and analyzed for intracellular fibronectin in an immunoblot (A) as described under Materials and Methods. (R) Coomassie blue staining of the SDS-PAGE gel.
in agreement with the expression of fibronectin mRNA under these culture conditions (Fig. 3). TGFP treatment increased the secretion of immunoreactive fibronectin from both proliferating and differentiated keratinocytes (Fig. 6B).
DISCUSSION TGF/?l treatment induces reversible growth arrest [9, 141 and a morphologic alteration [13] of keratinocytes in proliferative, undifferentiated cultures. These changes have been shown to occur independently from the process of keratinocyte differentiation [ 131. Studies from Watt and co-workers have indicated that cellular adhesion to extracellular matrix proteins are important to the control of keratinocyte behavior [21,40]. Nickoloff et al. [ll] noted increased motility and increased fibronectin synthesis of keratinocytes in response to TGFP. Others have noted the effects of TGFP on the expression of extracellular matrix genes in fibroblasts [ 12,20,25-281. These findings led us to hypothesize that TGF/3 could be affecting keratinocyte behavior by regulating extracellular matrix gene expression. Thus, we examined the expression of three components of the epithelial basement membrane, fibronectin, laminin Bl, and collagen al(IV), in keratinocytes. Each of these
98
VOLLBERG,
A 1OOpM TGF-0,
GEORGE,
B -- U - +
0
IJ
D + _
1 OOpM TGF-P,
--+o+-
id
- 200K
- 200K
- 92.5
- 92.5
-69 -46 “.
D
- 69 :i
- 46
FIG. 6. TGFB induced production and secretion of fibronectin protein in undifferentiated and differentiated keratinocytes. Undifferentiated (IJ) and differentiated (D) cultures of normal human keratinocytes were cultured in GO-mm dishes as described under Materials and Methods. After exchange to fresh KGM medium the cells were untreated (-) or treated (+) with 100 pM TGFPl for 48 h. An additional control (0) represents the cells and conditioned media of differentiated cells prior to the 48 h of treatment. (A) Cell-associated fibronectin. Cell lysate protein (5 fig) from each treatment was prepared and analyzed for fibronectin in an immunoblot protocol as described under Materials and Methods. (B) Fibronectin in conditioned medium. Protein in conditioned medium was precipitated in 10% (xv/ v) TCA with 10 pg bovine serum albumin as a carrier and washed at 4°C sequentially with 100% ethanol and acetone to remove residual acid. The protein pellets were dried briefly under vacuum and solubilized in electrophoresis sample buffer. The cell density of each culture was estimated from the concentration of protein in the lysate and the amount of conditioned medium which was applied to the electrophoresis gel was adjust accordingly. The volumes of conditioned medium represented by each sample were, from left to right, respectively, 0.250, 0.223, 0.084, 0.077 and 0.057 ml. Immunoblot analysis for the detection of fibronectin was as described above. Tailing of the immunoreactive fibronectin seen in each of the conditioned medium samples is likely due to the degradation in the medium during culture. A normal rabbit serum control for the primary antibody was completely negative in this immunoblot protocol. The migration ofprotein molecular weight markers is indicated in the right hand margin.
genes was expressed at a relatively low level in undifferentiated cells and TGFP treatment of the keratinocytes increased their level of expression. The induction of keratinocyte ECM expression by TGFP was time and dose dependent, suggesting that changes in ECM expression are part of the cellular process in which TGFP alters keratinocyte behavior. These alterations in ECM expression appeared to coincide with the time course for changes in cellular morphology (results not shown). Induction of ECM gene expression was measurable within 4 h and remained above basal levels for at least 48 h. Inhibition of cell growth [ 141 and alteration of keratinocyte morphology occur at picomolar concentrations of TGFP. Additionally, at 24 h after treatment, increasing concentrations of TGFP resulted in an increasing number of morphologically altered cells
AND
CJETTEN
([13], results not shown). The expression of ECM mRNA was enhanced at 1 pM TGFP and was maximal at 30-100 pM TGFP. Accumulation of intracellular fibronectin was first observed at 3 pM TGFP and increased with increasing TGFP dose. This was accompanied by a stimulation in the secretion of fibronectin to the extracellular environment. Thus, TGF@ treatment led to changes in ECM production and changes in cell behavior with similar onset of effect for time and dose. Cellular interactions with the ECM proteins in the absence of TGFP have been reported to cause changes in keratinocyte behavior similar to those observed in TGF&treated cells. Particularly, the interaction of fibronectin with cellular integrin was reported to suppress keratinocyte proliferation and to inhibit the expression of differentiation [al]. These studies utilized plasma fibronectin, one of several fibronectins produced by the differential splicing of a single gene. The concentrations of plasma fibronectin which produced maximal effects (75 pg/ml) were greater than can be expected to be generated by the TGFB-treated keratinocytes in the conditioned media collected in this study (33 pg total protein/ml in the most protein-rich sample). However, Shoji et al. have reported that cellular fibronectins have far greater chemotactic activity than plasma fibronectin [41]. Their data would indicate that keratinocyte-derived fibronectin could be effective in altering cell behavior at the levels generated in response to TGFP. The changes in ECM gene expression were not simply related to the arrest of cell growth which results from the TGF@ treatment. In contrast to TGFP-treatment, confluence arrest of the keratinocyte cultures resulted in the reduction of ECM mRNA expression. This arrested state led to the differentiation of the keratinocytes in the confluent cultures. Previous studies have shown that keratinocytes in confluent culture lose all colony-forming efficiency and, thus, undergo terminal cell division as part of their differentiation [42]. In these differentiated keratinocytes fibronectin mRNA expression was greatly induced in response to TGFP treatment. In contrast, TGFB caused only very modest increases of the expression of laminin Bl and collagen ~ul(1V) in the differentiated cultures. The increase of fibronectin expression must be due to the TGF/3 responsiveness of differentiated keratinocytes in these cultures since the magnitude of the increase cannot be accounted for by undifferentiated cells which, if present, would represent only a small part (~2%) of the culture. The relatively minor changes in laminin Bl and collagen oll(IV) expression were such that it could represent the effect of TGFP on a small percentage of cells within the culture. Thus, the differentiation of the keratinocytes results in a differential induction of ECM genes by
TGF/j
INDUCTION
OF ECM
TGF@. In ho, differentiated keratinocyt,es would be present at the wound border. This region of the wound would then be expected to have proportionally more of its TGFP-induced ECM protein as fibronectin than the center of the wound where expression of laminin and collagen IV would be induced in the undi~erent,iated keratinocytes which migrated to the area. Fibronectin and laminin have been shown to be antagonistic to one another in effecting cell motility and cell adhesion [ 221. Therefore, this differential induction of ECM genes in the differentiated cells may be important to TGFP action in influencing the migration of stem cells from the area of the wound border to the center of the wound. The TGF/3 dose response of the ECM induction indicated that these changes in gene expression are signaled through high-affinity TGFfl receptors. In fibroblasts, the induction of collagen I and fibronectin mRNA by TGFP requires protein synthesis in order to express the TGFP effect, [27]. In the keratinocytes suppression of protein synthesis by cycloheximide largely ablated the TGF/3 effect on fibronectin and laminin Bl mRNA expression. Rossi et al. [24] have shown that induction of collagen 0~2 [I] in fibroblasts by TGF@ is dependent on the presence of a nuclear factor 1 (NF-1) binding site in the sequence of this gene’s promoter. Dean and coworkers [43-451 have recently reported the presence of a NF-1 binding site in the promoter of t,he fibronectin gene which together with two CAMP response elements and a binding site for the transcription factor SP-1 results in the TGF/3 induct,ion of this gene. Thus, the TGF/3 induction of fibronectin in the keratinocytes is likely to require for its full effect the synthesis of one or more of the transcriptional factors which int,eract with its promoter. The decreased expression of fibronectin in cycloheximide-treated control cells also indicates the lability of a factor which is required for the maintenance of the basal expression of this gene. Laminin Bl would appear to be maintained by stable factors which are unaffected by cycloheximide treatment. However, TGFP induction of laminin Bl, like the induction of collagen I and fibronectin, required synthesis of a protein factor(s). It. is t.empting to speculate that the induction of a single transcriptional factor could signal t.he TGF,i3 induction of expression of the family of ECM genes. However, the induction of collagen rul(IV) expression was independent. of protein synthesis and, therefore, TGF@ is inducing the expression of this gene through another mechanism. The apparent direct effect of TGFB on this gene makes further study of collagen al(IV) regulation important to understanding the intracellular signaling of this hormone. Other changes in TGFfl gene regulation of ECM expression as influenced by cellular differentiation will be important for understanding the physiologic role of TGF/3 in epidermal keratinocytes.
GENES
IN KERATINOCYTES
99
Our appreciation is Given to Drs. S. Bernacki, K. Marvin, C. Nervi, and N. Saunders for helpful discussions. We gratefully acknowledge the clerical help of Ms. Helena Bonner. Also, Drs. V. Kalter and S. Randell are thanked for their comments on the manuscript.
REFERENCES 1. 2. 3.
Woodley, D. T., O’Keefe, E.
CELL
RESEARCH
193,
93-100
(1991)
induction of Extracellular Matrix Gene Expression in Normal Human Keratinocytes by Transforming Growth Factor ,O Is Altered by Cellular Differentiation THOMAS M. VOLLBERG, SR., MARGARET D. GEORGE,AND ANTON M. JETTEN' Cell Biology Group, Laboratory of Pulmonary Pathobiology. National Institute of Environmental P.0. Bon 12233, Research Triangle Park, North Carolina 27709
Changes in epithelial substrata have been related to the cellular capacity for proliferation and to changes in cellular behavior. The effect of TGF/31 on the expression of the basement membrane genes, fibronectin, laminin Bl, and collagen (~1 (IV), was examined. Northern analysis revealed that treatment of normal human epidermal keratinocytes with 100 pM TGF/31 increased the expression of each extracellular matrix (ECM) gene within 4 h of treatment. Maximal induction was reached within 24 h after treatment. The induction of ECM mRNA expression was dose dependent and was observed at doses as low as l-3 pM TGF/31. Incremental doses of TGF@l also increased cellular levels of fibronectin protein in undifferentiated keratinocytes and resulted in increased secretion of fibronectin. Squamous-differentiated cultures of keratinocytes expressed lower levels of the extracellular matrix RNAs than did undifferentiated cells. Treatment of these differentiated cells with TGFPl induced the expression of fibronectin mRNA to levels seen in TGF&treated, undifferentiated keratinocytes but only marginally increased the expression of collagen (Al and laminin Bl mRNA. The increased fibronectin mRNA expression in the differentiated keratinocytes was also reflected by increased accumulation of cellular and secreted fibronectin protein. The inclusion of cycloheximide in the protocol indicated that TGFP induction of collagen (Al mRNA was signaled by proteins already present in the cells but that TGFP required the synthesis of a protein(s) to fully induce expression of fibronectin and laminin Bl mRNA. The differential regulation of these genes in differentiated cells may be important to TGFfi action in regulating reepithelialization. (c 1991 Academic Press, Inc.
INTRODUCTION The proliferation of the epidermal epithelium is tightly controlled to ensure that cell renewal and cell ’ To whom correspondence dressed.
and
reprint
requests
should
be ad-
Health Sciences,
loss occur at the same rate. During wound healing this process of cell growth and maturation is altered so that potential stem cells may migrate to the denuded area [l]. Once in place the proliferation of these cells is accelerated to ensure the replacement of the lost stem cell population and the regeneration of the protective layers of differentiating cells. At the molecular level these events would appear to result from the interplay of a variety of growth factors which influence, both positively and negatively, the proliferation and maturation of the cells [2]. In uitro, epidermal keratinocytes require epidermal growth factor and insulin for proliferation [2, 31. Confluence, phorbol esters, and increased calcium concentration can induce these cells to undergo terminal cell division and to express a number of differentiation markers [3-61. Transforming growth factors /3 (TGFPs), a family of hormonally active polypeptides, have been implicated as controlling factors in both cellular proliferation and differentiation (see review [7, 81). TGFPs are synthesized as prepropeptides and secreted as inactive homodimers which are activated by proteolysis. TGFPs elicit in responsive cells changes in proliferation, differentiation, and genetic expression [7, 91. The response is cell specific; some cells are stimulated to proliferate, whereas others are growth arrested by TGF/3 [g-11]. Three types of high-affinity receptors for TGFP have been identified on the surface of responsive cells [la]. The intracellular signaling which is triggered by the binding of TGFfl to these receptors has not been elucidated. In epithelial cells TGF/3 Type 1 (TGFPl) inhibits cell proliferation and induces a change in the undifferentiated cells from an angular, raised morphology to a rounded, flattened shape [ 131. In human epidermal keratinocytes TGFP inhibition of cell proliferation is reversible and does not lead to differentiation [9,14]. This is in contrast to other epithelial cell systems where TGFP has been described as an inducer of terminal differentiation [15, 161. In keratinocyte cultures TGF(31 stimulates cell production of fibronectin protein and in-
93 All
Copyright 0 1991 rights of reproduction
0014-4827/91 $3.00 by Academic Press, Inc. in any form reserved.
94
VOLLBERG.
GEORGE.
creases cell motility [18, 191. Because of TGFPs effects on epithelial proliferation and cellular behavior, a physiologic role for TGFP has been proposed in the control of postinjury regenerative growth of skin epithelia [6, 13, 18, 201. The binding and interaction of epithelial cells with extracellular matrix proteins influences their growth and differentiation. For example, the binding of fibronectin to cellular receptors of the integrin family inhibits the differentiation of keratinocytes [21]. Further, the binding of fibronectin to epithelial cancer cells has been shown to promote cell spreading and motility in competitive antagonism with the binding of laminin to the cells [22]. Other studies have demonstrated the requirement of cell motility for sustained proliferation of keratinocytes grown in colonies [23]. In fibroblasts TGFP has been shown to induce the mRNA expression of the extracellular matrix genes, collagen I and fibronectin [20, 24-281. Thus, we hypothesized that TGF/3 could be altering keratinocyte extracellular matrix expression as an intermediate step in affecting cell behavior. In this study the expression of the extracellular matrix genes, fibronectin, laminin Bl, and collagen cul(IV), is demonstrated to be inducible by TGFP in undifferentiated keratinocytes. Since differentiated cells would be present at the wound border, we also examined TGFfi effects in differentiated cells and report that these cells responded to TGFP by increasing production of fibronectin but that the process of differentiation caused the cells to acquire resistance to TGFP induction of laminin Bl and collagen al(IV) expression. The induction of ECM proteins and the alteration of this induction by differentiation may represent a mechanism by which TGFP alters epithelial cell behavior to effect reepithelialization during the healing process. MATERIALS
AND METHODS
Normal human keratinocytes were obtained as cryopreserved primary culture cells from Clonetics Corp. (San Diego, CA) and cultured according to the suppliers protocols in KGM serum-free medium (Clonetics) in tissue culture plastic flasks and dishes (Costar, Cambridge, MA). Subculture of the cells was continued for a maximum of three passages. On the final subculture before use the cells were seeded at a density of 2-3 X lo3 cells/cm’ into either 60.mm dishes or 150-mm dishes for protein or RNA analysis, respectively. IJndifferentiated cells represented conditions under which culture was continued for 5-6 days and the cell population had expanded to 8-10 X lo3 cells/cm2 before treatment or collection. The number of cells in these cultures was approximately 20% of the confluent cell density. Differentiated cells were obtained by culturing cells to confluent density over a lo- to 12.day period and were allowed morphologic differentiation over 223 additional days of culture before treatment or collection. Expression of ECM proteins by human keratinocytes was measured by Western blot analysis. Following treatment, the protease inhibitors, leupeptin and aprotinin, were added to the conditioned media at 10 Gg each per milliliter and the media were stored at -20°C. The cultures were rinsed twice with ice-cold PBS. The cells were
AND
JETTEN
scraped into PBS and pelleted by centrifuging 6OOgfor 10 min at 4°C. The pellet was resuspended in 0.5 ml of 50 mM TrissCl, pH 8.0, 125 mM NaC1, 0.5% (v/v) NP-40, 0.1 mM PMSF, and 10 pg/ml each of aprotinin and leupeptin (Sigma Chemical Co., St. Louis, MO) andcell lysis was allowed to occur for 30 min on ice. The cell lysate was then sonicated at 20 PW for three 10-s bursts and microcentrifuged (10,OOOg) for 10 min at 4°C. Protein in the cell lysate supernatant was quantitated with the Bio-Rad protein assay (Bio-Rad Laboratories, Richmond, CA) using bovine serum albumin as a standard. Protein samples were brought to 1X with SDSPAGE sample buffer, heatdenatured, and electrophoresed in a SDS-lo% polyacrylamide slab gel as described by Laemmli [29]. The proteins were electroblotted to a 0.45.brn Nitroplus 2000 membrane (Micron Separations, Inc., Westborough, MA) as described [30]. Fibronectin was immunodetected by a 1:X0 dilution of human fibronectin antiserum (rabbit, Biomedical Technologies, Inc., Stoughton, MA) using Auroprobe BL plus kit for immunogold/silver staining (Janssen Biotech, Olen, Belgium) in a protocol supplied by ,Janssen Biotech. Expression of mRNA for ECM genes was examined in a Northern blot protocol using cDNA probes. Total cellular RNA was isolated from 5-10 150.mm dishes by scraping the cells into guanidium isothiocyanate lysis buffer and pelleting of the RNA through a CsCl cushion [31]. The RNA was solubilized in 100 mM sodium chloride/l mM EDTA/lO mM TrisHCl, pH 8.0, extracted with chloroform:isoamyl alcohol (24:1), and precipitated in 250 mM sodium acetate/70% ethanol at ~20°C. The purified RNA was dissolved in water and quantity and purity were assessed by absorbance at 260 nm (1 A,,,,, unit = 40 pg single-stranded RNA/ml) and at 280 nm (A,,,,,/A,,,, greater than 1.8 for pure RNA). Total RNA (lo-20 p(p) was separated by electrophoresis in a 0.66 M formaldehyde/l% agarose gel [32] and transferred to Nytran nylon membrane (Schleicher & Schuell, Keene, NH) in a Vacublot apparatus using the manufacturer’s protocol (American Bionetics, Inc., Hayward, CA). Nucleic acid was permanently fixed to the membrane by uv cross-linking [33]. Ribosomal RNA and a 0.24. to 9.5kb RNA ladder (BRL, Life Technologies, Inc., Gaithersburg, MD) were used as size markers, RNA blots were hybridized with 224 rig/ml heat-denatured DNA probe in hybridization buffer (50% formamide/5X SSPE/2x Denhardt’s solution/l% SDS/ 100 pg/ml denatured salmon testes DNA) at 42°C for 16-20 h after prehybridization overnight at 42°C in the hybridization buffer. Following hybridization ECM probes were washed to a final stringency of 60°C in 0.1X SSC/O.l% SDS. RNAs hybridizing to cDNA were detected by autoradiography with enhancing screens (Lightning Plus, DuPont Co., Wilmington, DE) and Kodak XAR-5 film (Eastman Kodak, Rochester, NY) at --70°C. The expressions of the various RNAs were quantitated from the developed autoradiograms by densitometry with the Bio-Rad Model 620 video densitometer and 1-D analyst software (Bio-Rad Laboratories). For reuse of blots with alternative probes, previously detected probes were erased from the blot by washing at 70-75°C in 5 mM TrissHCI, pH 8.0/0.2 mM EDTA/O.lX Den hardt’s solution/0.05% (w/v) sodium pyrophosphate prior to reprobing. Radioactively labeled DNA prohes were synthesized from purified cDNA inserts using a random prime labeling kit (Boehringer-Mannheim Biochemicals, Indianapolis, IN) according to the manufacturer’s protocol with [tu-32P]dCTP (-3000 Ci/mmol, Amersham Corp., Arlington Heights, IL). The specific activities of the DNA probes were in the range of 4-8 X 10’ cpm per microgram DNA. Human fibronectin cDNA [34] was a 1.7.kb PatI-PuuII restriction fragment of plasmid pFH1.54 (from Dr. A. Kornblihtt). Human laminin Bl cDNA [35] was a 1.2.kb EcoRI restriction fragment of plasmid ~1236 (from Dr. Y. Yamada). The transglutaminase Type I probe was the EcoRI cDNA insert of clone pTG7 [36]. Purified cDNA inserts of the squamous-cell-specific clones SQlO and SQ37 were prepared as described previously [37]. Human collagen cxl(IV) cDNA [38] was a 2.6.kb PsB restriction fragment of plasmid pHT-21 (from Dr. A. Ganguly and Dr. D. Prockop). Chicken glyceraldehyde S-phosphate dehy-
TGFli
INDUCTION
OF ECM
HOURS IOOpH TGF-&
0 ”
4 +
8 +
12 +
18 +
24 +
45 +
9.5kb -
24 -
Fibronectin
7.5 -
285 4.4 z 9.5 7.5 285
Lwdnin
81
4.4 zz 9.5 75-
Coliagen
al(lY)
4.4 -
GENES
95
IN KERATINOCYTES
that which produced a similar image with the fibronectin probe. The dose-dependent. effect of the TGFfl treatment was measured at 24 h by analysis of fibronectin, laminin Bl, and collagen al(IV) mRNA expression (Fig. 2). AS was seen during the time course ECM gene expression was increased in TGF@-treated keratinocytes (Fig. ZA). Laminin Bl and collagen al(IV) mRNA once again required longer exposure of the au~oradiograms for detection and therefore were relatively less abundant than fibronectin mRNA in the untreated and TGFpl-treated keratinocyt.es (exposures of 5 and 3 days, respectively, versus 14 h). Expression of GPDH mRNA in these cells was unaffected by the TGFfl treat,ment. When the expression of the mRNAs was quantitated by densitometry of the autoradiograms, the induction of ECM gene
GPDH
FIG. 1. Time course of TGF8 induction of extracellular matrix genes in human keratinocytes. IJndifferentiated cultures of normal human epidermal keratinocyt.es were treated with 100 pA4 porcine platelet TCF@l (R&D Systems, Minneapolis, MN) and RNA was isolated after the indicated incubation time. Control cultures received no treatment and were harvested at the indicated times. Total RNA (10 pg) from each sample was examined by Northern analysis for expression of ECM genes as described under Materials and Methods. Each probe was hybridized individually with the blot and stripped before use of the next probe. The aut(~radiographs were exposed as follows: ~bronec~in, 8 h; laminin Bl, 5 days; collagen cul(fV), 7 days; and GPDH, 18 h.
drogenase (GPDH) was a 1.12-kb &I mid pGAD 28 [ 391.
restriction
fragment
TGF-p,
A
phff 0
1
3
IO
30
100 3m
Fibroneciin
Laminin
Bi
of plas-
RESULTS
In order to examine if TGFP was inducing the expression of ECM genes in human epidermal keratinocytes, the mRNA expression of the extracellular matrix proteins, laminin, collagen cyl(IV), and fibronectin, was measured by Northern analysis. Total cellular RNA was collected during a 45-h time course following treatment of undifferentiated keratinocyte cultures with 100 pM TGFPl. In untreated control cells (0 and 24 h no addition, Fig. 1) each of the ECM genes were expressed and each was induced to higher mRNA expression by TGFP treatment. Densitometric scanning of the autoradiograms revealed that fibronectin mRNA levels increased 50% within 4 h and were increased maximally (24-fold) at 45 h after treatment. The expression of laminin Bl and collagen alfIV) mRNA was also induced by TGF@ during the time course, although the levels of these mRNAs were maximal at 18 h (5-fold and l7-fold increases, respectively). Both laminin Bl and collagen (ul(IV) mRNA appear to be less abundant than fibronectin mRNA since the laminin Bl and collagen cul(IV) autoradiograms required exposures of 15 and 20 times
GPDH
FIG. 2. Relation of TGF$ dose to ECM gene expression. Following 24 h of treatment with the indicated doses of TGFBl, total RNA was isolated from undifferentiated normal human keratinocytes and examined for expression of ECM genes. (A) Northern blot. Total RNA (10 rgi from each sample was fractionated and transferred to a Nytran membrttne. Expression of each mRNA was probed as described in t.he legend t.o Fig. 1. The aut.oradiographs were exposed as fotlows: fibronectin, 14 h; laminin Bl, 3 days; collagen al(W), 5 days; and GPDH, 24 h. (B) Quantitation of the induction of mRNA expression. The autoradiograms were densitometrically scanned as described under Materials and Methods. The expression of each mRNA is plotted relative to its own expression in untreated keratinocytes. The RNA loaded to each lane of the gel was equalized by comparing the expression of GPDH mRNA in each sample.
96
VOLLBERG, U Hours 1OOpM TGF-fl,
0 -
GEORGE,
D 40 +
0 -
48 +
48 _ Fibronectin
Laminin El
Collagen al(W)
SQlO
Transglutaminase Type 1
GPDH
FIG. 3. Effect of differentiation on the regulation of extracellular matrix genes by TGF@. 1Jndifferentiated (U) and differentiated (D) cultures of normal human epidermal keratinocytes were obtained as described under Materials and Methods. Cells were treated with 100 pM TGFfil as indicated and total RNA (10 pg each) was analyzed by Northern blot analysis as indicated in the legend to Fig. 1.
expression was concentration-dependent and evident at concentrations as low as 1 pM TGF/31 (Fig. 2B). Maximal expression of ECM genes was induced by TGFP at a concentration of lo-30 pM (11.9-, 4.6-, and 10.8.fold increases for fibronectin, laminin Bl, and collagen oll(IV) mRNAs, respectively). This induction was halfmaximal for fibronectin and collagen ~ul(1V) expression at approximately 3 pM TGFPl and for laminin Bl mRNA at 1 pM TGFPl. In the preceding experiments the effects of TGFfi on the induction of ECM gene expression were measured in undifferentiated, proliferating keratinocytes which are susceptible to TGFP-induced cell growth arrest. To determine if the increase in ECM gene expression was correlated with the arrest of cell growth (as opposed to other actions of TGFP), keratinocytes were cultured to confluent cell density. Under these conditions the cells lose all proliferative potential and terminally differentiate. Total cellular RNA was then analyzed for the expression of ECM genes in these nonproliferative cells. The differentiated character of the keratinocytes was confirmed by their expression of the transglutaminase Type I and the squamous-cell-specific mRNAs, SQlO and SQ37 (Fig. 3, lanes 3-5). Total RNA from undifferentiated cells expressed very low levels of these mRNAs (Fig. 3, lanes 1-2). The expression of the fibronectin mRNA in the differentiated keratinocytes was sixfold
AND
.JETTEN
less than the level of expression seen in untreated, proliferating, undifferentiated cells. The expression of laminin Bl and collagen oll(IV) mRNA was not detected in the untreated, differentiated keratinocytes. Thus, growth arrest as caused by cell confluence was insufficient for the induction of ECM gene expression in the keratinocytes and, in fact, reduced the expression of fibronectin, collagen al(W), and laminin Bl mRNA. Next, differentiated keratinocytes were treated with TGFfll for 48 h to determine if TGF/31 could induce ECM genes in these already growth-arrested cells. Fibronectin, laminin Bl, and collagen cul(IV) mRNA were each increased by the treatment with 100 pM TGFPl. After TGF/3 treatment the fibronectin message in the differentiated cells reached a level of expression similar to that seen in the TGFfll-treated, undifferentiated cells (sevenfold and fourfold, respectively, above the expression seen in untreated, undifferentiated cells). In contrast, the TGFB induction of laminin Bl and collagen (ul(IV) mRNA in the differentiated keratinocytes resulted in levels of expression which were, respectively, 0.6 and 1.2 relative to their expression in untreated, undifferentiated keratinocytes. Untreated, differentiated keratinocytes collected at the same time as the TGF@ltreated differentiated cells (Fig. 3, lane 5) expressed the same levels of ECM mRNA as differentiated cells collected prior to the 48-h TGFP treatment (Fig. 3, lane 3). The induction of fibronectin, collagen al(IV), and laminin Bl mRNA in response to TGF/X was tested in the presence of the protein synthesis inhibitor, cycloheximide, to determine if TGFfl was inducing these changes directly or indirectly through the synthesis of cellular protein(s). Detectable mRNA levels were seen for fibronectin, collagen al(IV), laminin Bl, and GPDH in the controls as well as in the TGFpl-induced cells (Fig. 4). Densitometry of GPDH mRNA levels in
2.5pglml
1OOpM TGF-P, Cycloheximlde
_ -
+ -
_ +
+ + Fibronectin
Laminin Bl
Collagen al(W)
GPDH
FIG. 4. Etfect of cycloheximide treatment on the TGF@ induction of fibronectin and laminin Bl mRNA expression. Undifferentiated normal human keratinocytes were untreated or treated with 100 pM TGFDI in the absence or presence of 2.5 pg/ml cycloheximide for 7 h. This concentration of cycloheximide inhibited protein synthesis by greater than 95%. Northern analysis of 10 pg total RNA from each treatment was as described in the legend to Fig. 1.
TGFd
INDUCTION
OF ECM
each condition varied by less than 20%. When undifferentiated keratinocytes were treated for 8 h with 100 pM TGF/31, fibronectin mRNA levels increased 2.3fold over the level in untreated cells. Laminin Bl expression was increased 1.9-fold by the TGF/31 treatment. Inclusion of cycloheximide (2.5 pg/ml) in the medium with TGF/31 partially blocked the increase in fibronectin mRNA and ablated the increase in laminin Bl mRNA expression (1.3 for fibronectin and 1.0 for laminin Bl relative to control cells). Cycloheximide alone decreased laminin Bl expression to 0.9 of the level in untreated cells and caused fibronectin expression to fall to 0.7 of the untreated level. Constitutive expression of fibronectin may require a labile factor whose absence in the cycloheximide-treated cells may account for the reduced response to TGFP treatment. TGFP induced collagen crl(IV) mRNA expression 14fold over the untreated controls in the absence of cycloheximide. In the presence of cycloheximide, the collagen cul(IV) expression was equally induced by treatment with TGF/3 (l&fold over the untreated control). Cycloheximide alone had no effect on collagen ul(IV) mRNA expression (1.4 of the mRNA level in untreated cells). In addition to inducing the levels of ECM gene expression in cellular mRNA, TGFP treatment of the keratinocytes increased the level of cell-associated ECM protein. Increased immunoreactivity to anti-fibronectin antibodies was seen after 24 h of treatment with TGFPl (Fig. 5). As shown for the mRNA expression of fibronectin, TGFPl induction of fibronectin protein was dose dependent. Maximal levels of the protein were detectable at TGFPl concentrations greater than 10 PM. No increase in fibronectin protein was discernible after treatment with 1 pM TGFfll. Next the effect of differentiation on the expression of fibronectin protein was examined (Fig. 6). At 48 h after treatment with 100 pM TGFP the induction of cell-associated fibronectin was still evident in the undifferentiated cells (Fig. 6A). Lysates of differentiated NHEK cells expressed fibronectin protein at levels similar to those seen in the proliferating undifferentiated cells. Treatment of these cultures with 100 pM TGFP increased the amount of fibronectin in these differentiated cells. Untreated cells collected after the same 48-h period showed no change in fibronectin from the levels seen at the time of culture when TGFP treatment was initiated. Upon collection of these cells for preparation of cell lysates, the conditioned media was reserved. The amount of fibronectin protein was measured in conditioned media from each culture (Fig. 6B). Fibronectin in the conditioned media was detected in all of the samples from both proliferative cultures and confluent, growth-arrested, differentiated cultures. More fibronectin protein accumulated in the medium from the proliferating cultures than from the squamous, differentiated cultures (Fig. 6B). This is
GENES
97
IN KERATINOCYTES TGFp (PM) 0
30
3
I
IO
300
100
0 MW(X10-3)
A
-
200
-65
B
-
200
-
65
-45
FIG. 5. Increased accumulation of cell-associated fibronectin by increasing doses of TGFB. Undifferentiated normal human keratinocytes in a 60-mm culture dish were treated with varying doses of TGF/fl as indicated for 24 h. The cells were lysed by scraping directly into electrophoresis sample buffer and analyzed for intracellular fibronectin in an immunoblot (A) as described under Materials and Methods. (R) Coomassie blue staining of the SDS-PAGE gel.
in agreement with the expression of fibronectin mRNA under these culture conditions (Fig. 3). TGFP treatment increased the secretion of immunoreactive fibronectin from both proliferating and differentiated keratinocytes (Fig. 6B).
DISCUSSION TGF/?l treatment induces reversible growth arrest [9, 141 and a morphologic alteration [13] of keratinocytes in proliferative, undifferentiated cultures. These changes have been shown to occur independently from the process of keratinocyte differentiation [ 131. Studies from Watt and co-workers have indicated that cellular adhesion to extracellular matrix proteins are important to the control of keratinocyte behavior [21,40]. Nickoloff et al. [ll] noted increased motility and increased fibronectin synthesis of keratinocytes in response to TGFP. Others have noted the effects of TGFP on the expression of extracellular matrix genes in fibroblasts [ 12,20,25-281. These findings led us to hypothesize that TGF/3 could be affecting keratinocyte behavior by regulating extracellular matrix gene expression. Thus, we examined the expression of three components of the epithelial basement membrane, fibronectin, laminin Bl, and collagen al(IV), in keratinocytes. Each of these
98
VOLLBERG,
A 1OOpM TGF-0,
GEORGE,
B -- U - +
0
IJ
D + _
1 OOpM TGF-P,
--+o+-
id
- 200K
- 200K
- 92.5
- 92.5
-69 -46 “.
D
- 69 :i
- 46
FIG. 6. TGFB induced production and secretion of fibronectin protein in undifferentiated and differentiated keratinocytes. Undifferentiated (IJ) and differentiated (D) cultures of normal human keratinocytes were cultured in GO-mm dishes as described under Materials and Methods. After exchange to fresh KGM medium the cells were untreated (-) or treated (+) with 100 pM TGFPl for 48 h. An additional control (0) represents the cells and conditioned media of differentiated cells prior to the 48 h of treatment. (A) Cell-associated fibronectin. Cell lysate protein (5 fig) from each treatment was prepared and analyzed for fibronectin in an immunoblot protocol as described under Materials and Methods. (B) Fibronectin in conditioned medium. Protein in conditioned medium was precipitated in 10% (xv/ v) TCA with 10 pg bovine serum albumin as a carrier and washed at 4°C sequentially with 100% ethanol and acetone to remove residual acid. The protein pellets were dried briefly under vacuum and solubilized in electrophoresis sample buffer. The cell density of each culture was estimated from the concentration of protein in the lysate and the amount of conditioned medium which was applied to the electrophoresis gel was adjust accordingly. The volumes of conditioned medium represented by each sample were, from left to right, respectively, 0.250, 0.223, 0.084, 0.077 and 0.057 ml. Immunoblot analysis for the detection of fibronectin was as described above. Tailing of the immunoreactive fibronectin seen in each of the conditioned medium samples is likely due to the degradation in the medium during culture. A normal rabbit serum control for the primary antibody was completely negative in this immunoblot protocol. The migration ofprotein molecular weight markers is indicated in the right hand margin.
genes was expressed at a relatively low level in undifferentiated cells and TGFP treatment of the keratinocytes increased their level of expression. The induction of keratinocyte ECM expression by TGFP was time and dose dependent, suggesting that changes in ECM expression are part of the cellular process in which TGFP alters keratinocyte behavior. These alterations in ECM expression appeared to coincide with the time course for changes in cellular morphology (results not shown). Induction of ECM gene expression was measurable within 4 h and remained above basal levels for at least 48 h. Inhibition of cell growth [ 141 and alteration of keratinocyte morphology occur at picomolar concentrations of TGFP. Additionally, at 24 h after treatment, increasing concentrations of TGFP resulted in an increasing number of morphologically altered cells
AND
CJETTEN
([13], results not shown). The expression of ECM mRNA was enhanced at 1 pM TGFP and was maximal at 30-100 pM TGFP. Accumulation of intracellular fibronectin was first observed at 3 pM TGFP and increased with increasing TGFP dose. This was accompanied by a stimulation in the secretion of fibronectin to the extracellular environment. Thus, TGF@ treatment led to changes in ECM production and changes in cell behavior with similar onset of effect for time and dose. Cellular interactions with the ECM proteins in the absence of TGFP have been reported to cause changes in keratinocyte behavior similar to those observed in TGF&treated cells. Particularly, the interaction of fibronectin with cellular integrin was reported to suppress keratinocyte proliferation and to inhibit the expression of differentiation [al]. These studies utilized plasma fibronectin, one of several fibronectins produced by the differential splicing of a single gene. The concentrations of plasma fibronectin which produced maximal effects (75 pg/ml) were greater than can be expected to be generated by the TGFB-treated keratinocytes in the conditioned media collected in this study (33 pg total protein/ml in the most protein-rich sample). However, Shoji et al. have reported that cellular fibronectins have far greater chemotactic activity than plasma fibronectin [41]. Their data would indicate that keratinocyte-derived fibronectin could be effective in altering cell behavior at the levels generated in response to TGFP. The changes in ECM gene expression were not simply related to the arrest of cell growth which results from the TGF@ treatment. In contrast to TGFP-treatment, confluence arrest of the keratinocyte cultures resulted in the reduction of ECM mRNA expression. This arrested state led to the differentiation of the keratinocytes in the confluent cultures. Previous studies have shown that keratinocytes in confluent culture lose all colony-forming efficiency and, thus, undergo terminal cell division as part of their differentiation [42]. In these differentiated keratinocytes fibronectin mRNA expression was greatly induced in response to TGFP treatment. In contrast, TGFB caused only very modest increases of the expression of laminin Bl and collagen ~ul(1V) in the differentiated cultures. The increase of fibronectin expression must be due to the TGF/3 responsiveness of differentiated keratinocytes in these cultures since the magnitude of the increase cannot be accounted for by undifferentiated cells which, if present, would represent only a small part (~2%) of the culture. The relatively minor changes in laminin Bl and collagen oll(IV) expression were such that it could represent the effect of TGFP on a small percentage of cells within the culture. Thus, the differentiation of the keratinocytes results in a differential induction of ECM genes by
TGF/j
INDUCTION
OF ECM
TGF@. In ho, differentiated keratinocyt,es would be present at the wound border. This region of the wound would then be expected to have proportionally more of its TGFP-induced ECM protein as fibronectin than the center of the wound where expression of laminin and collagen IV would be induced in the undi~erent,iated keratinocytes which migrated to the area. Fibronectin and laminin have been shown to be antagonistic to one another in effecting cell motility and cell adhesion [ 221. Therefore, this differential induction of ECM genes in the differentiated cells may be important to TGFP action in influencing the migration of stem cells from the area of the wound border to the center of the wound. The TGF/3 dose response of the ECM induction indicated that these changes in gene expression are signaled through high-affinity TGFfl receptors. In fibroblasts, the induction of collagen I and fibronectin mRNA by TGFP requires protein synthesis in order to express the TGFP effect, [27]. In the keratinocytes suppression of protein synthesis by cycloheximide largely ablated the TGF/3 effect on fibronectin and laminin Bl mRNA expression. Rossi et al. [24] have shown that induction of collagen 0~2 [I] in fibroblasts by TGF@ is dependent on the presence of a nuclear factor 1 (NF-1) binding site in the sequence of this gene’s promoter. Dean and coworkers [43-451 have recently reported the presence of a NF-1 binding site in the promoter of t,he fibronectin gene which together with two CAMP response elements and a binding site for the transcription factor SP-1 results in the TGF/3 induct,ion of this gene. Thus, the TGF/3 induction of fibronectin in the keratinocytes is likely to require for its full effect the synthesis of one or more of the transcriptional factors which int,eract with its promoter. The decreased expression of fibronectin in cycloheximide-treated control cells also indicates the lability of a factor which is required for the maintenance of the basal expression of this gene. Laminin Bl would appear to be maintained by stable factors which are unaffected by cycloheximide treatment. However, TGFP induction of laminin Bl, like the induction of collagen I and fibronectin, required synthesis of a protein factor(s). It. is t.empting to speculate that the induction of a single transcriptional factor could signal t.he TGF,i3 induction of expression of the family of ECM genes. However, the induction of collagen rul(IV) expression was independent. of protein synthesis and, therefore, TGF@ is inducing the expression of this gene through another mechanism. The apparent direct effect of TGFB on this gene makes further study of collagen al(IV) regulation important to understanding the intracellular signaling of this hormone. Other changes in TGFfl gene regulation of ECM expression as influenced by cellular differentiation will be important for understanding the physiologic role of TGF/3 in epidermal keratinocytes.
GENES
IN KERATINOCYTES
99
Our appreciation is Given to Drs. S. Bernacki, K. Marvin, C. Nervi, and N. Saunders for helpful discussions. We gratefully acknowledge the clerical help of Ms. Helena Bonner. Also, Drs. V. Kalter and S. Randell are thanked for their comments on the manuscript.
REFERENCES 1. 2. 3.
Woodley, D. T., O’Keefe, E.
