Vol. 7, 469-480, April 1996 Cell Growth & Differentiation 469

A Novel Transforming Growth Factor �J Response ElementControls the Expression of the Connective TissueGrowth Factor Gene1

Gary R. Grotendorst,2 Hitoshi Okochi, andNobukazu HayashiDepartment of Cell Biology and Anatomy, University of Miami School of

Medicine, Miami, FL 33136

AbstractWe reported previously that transforming growth factor13 (TGF-13) selectively induced high levels of connectivetissue growth factor (CTGF) mRNA and protein inhuman skin fibroblasts. In this study, we investigatedthe molecular mechanism for TGF-fi regulation ofCTGF gene expression. Northern blot and run-ontranscription assays indicate that TGF-fi directlyactivates transcription of the CTGF gene. Fragments ofthe 5’ flanking region of the human CTGF gene werelinked to luciferase reporter constructs. TGF-fi induceda 25-30-fold increase in luciferase activity in NIH/3T3fibroblasts that had been transfected with thisconstruct compared with nontreated cells after 24 hincubation. Other growth factors, such as platelet-derived growth factor or fibroblast growth factor,caused only a 2-3-fold induction. This response toTGF-1J occurred only in human skin fibroblasts, fetalbovine aortic smooth muscle cells, and NIH/3T3fibroblasts but not in the epithelial cell lines tested.Analysis of deletion mutants indicated that animportant TGF-�J regulatory element is located between

positions -162 and -128 of the CTGF promotersequence. A fragment of the promoter containing thisregion conferred TGF-j3 induction to a SV4O

enhancerless promoter. Methylation interference andcompetition gel shift assays mapped a unique 13-

nucleotide sequence delineating a novel TGF-� cis-regulatory element. Point mutations in this regionresuft in a complete loss of the TGF-fi induction,identifying this sequence as a new TGF-fi responseelement.

Received 12/15/95; revised 1/25/96; accepted 2/1/96.The costs of publication of this article were defrayed in part by thepayment of page charges. This article must therefore be hereby markedadvertisement in accordance with 18 U.S.C. Section 1734 solely to mdi-cate this fact.I This work was supported by NIH Grant GM37223 and a grant fromFibroGen, Inc. (to G. A. G.).2 To whom requests for reprints should be addressed, at Department ofCell Biology & Anatomy, University of Miami School of Medicine, 1600NW. 10th Avenue, Miami, FL 33136. Phone: (305) 243-4471; Fax: (305)545-7166.

Introduction

The formation of new and regenerating tissue requires the

coordinate regulation of various genes that produce bothregulatory and structural molecules that participate in theprocess of cell growth and tissue organization. This processproceeds in a cascade, with ordered invasions by different

specialized cell types. Each cell type produces a set offactors that act to condition the site of regeneration and

repair, to recruit the next wave of cells, and to regulate theirbehavior during the next phase of the cascade. TGF-�3�appears to be a central regulatory component of this pro-cess. TGF-f3 is released by platelets (1) macrophages, andneutrophils (2, 3) that are present in the initial phases of the

repair process. TGF-�3 can act as a growth-stimulatory factorfor mesenchymal cells (4-6) and as a growth-inhibitory fac-tor for endothelial and epithelial cells (7-1 0). The growthstimulatory action of TGF-�3 appears to be mediated via anindirect mechanism of induced autocnne growth factors,such as PDGF-BB (5) or PDGF-AA (1 1 , 1 2), or CTGF (1 3, 14).

We initially identified CTGF, isolated from the conditionedmedia of human umbilical vein endothelial cells due to itsPDGF-Iike biological and immunological activities (1 5). CTGFis a member of a protein family that includes serum-inducedimmediate-early gene products such as Cyr6l (1 6) and Fisp

12 (17)/BIGM2 (18), a v-src-induced peptide (CEF-lO; Ref.

19) and a putative avian oncoprotein (nov; Ref. 20). Thesepeptides are characterized by an absolute conservation of 38cysteine residues that make up over 10% of the total aminoacid content. Fisp 12 (BIGM2) is likely to be the mousehomologue of CTGF because it has a 94% amino acid se-quence identity to CTGF. Twisted gastrulation (twg), a genethat functions to control cell fate during dorsal ventral patternformation in Drosophila embryogenesis, is more distantly

related to CTGF (21). Thus, CTGF-related gene productsappear to function in a wide variety of important biologicalprocesses, including normal embryonic development andtissue regeneration, as well as tumor formation and growth.

We wanted to better understand the mechanisms thatcontrol the expression of the CTGFgene. To accomplish this,we have examined the molecular mechanisms for the eleva-tion of CTGF transcripts by TGF-�3 in fibroblastic cells in vitro.The results of our experiments indicate that a major regula-

3 The abbreviations used are: TGF-p, transforming growth factor /3; PDGF,platelet-derived growth factor; CTGF, connective tissue growth factor�T/3RE, TGF-j3 response element; PAl, plasminogen activator inhibitor;DMEM-ITS, DMEM supplemented by insulin, transfemn, and selenium;NF, nuclear factor; TIE, TGF-/3 inhibitory element; EGF, epidermal growthfactor; FGF, fibroblast growth factor� PM, point mutation; PMSF, phenyl-methylsulfonyl fluoride.

24

244

24

0

Hours after ATGF�3 removal C 0 1 4 8 24 30 48

GIG F � � � �-- -

CHX CHx

T�m.(HRS) � � II BTGFP 8

T�m,(HRS) � � II

c�x CMXTaF� �

TIm.(MRS) � � � 3 � II -

ABCDEFGH

CTGF

CTGF�3 (lOng/mI)

Puromycin (lOOpg/mI)

Anisomycin (lOpg/mI)

CTGF

Fig. 1. A, prolonged induction of CTGF mRNA by short-term TGF-f3exposure. Confluent cultures of human skin fibroblasts were incubatedwith DMEM-ITS for 24 h prior to the addition of TGF-f3. After the treatmentwith i 0 ng/mI of TGF-13 for i h, cells were washed with PBS and incubatedwith DMEM-ITS for the indicated time periods. B, the effect of cyclohex-imide on induction of CTGF mRNA. Lanes A and H, nontreated controlcells at 4 and 24 h, respectively. Lane B, 4 h cycloheximide (1 0 j.tg/ml);Lane C, 4 h TGF-/3 (10 ng/ml); Lane D, 4 h cycloheximide (10 pg/mI) withTGF-j3 present for 1 h during hours 1 to 2 of cycloheximide exposure, RNAwas prepared at 4 h; Lane E, same as B, with RNA prepared 20 h afterremoval of cycloheximide; Lane F, 24 h TGF-/3 (10 �sg/mI); Lane G, sameas D, with RNA prepared 20 h after removal of cycloheximide (22 h afterremoval of TGF-f3). C, the effect of protein synthesis inhibitors on induc-tion of CTGF mRNA. Cells were treated with puromycin or anisomycin for4 h. TGF-/3 was added 1 h after protein synthesis treatment; then cellswere incubated for an additional 3 h, and the RNA was prepared at thattime.

470 TGF-(3 Response Element in the CTGF Gene

tory mechanism for the elevation of the CTGF mRNA by

TGF-p is increased transcription from the CTGF promoter

and have led to the identification of a new Tj3RE. The TI3RE

sequence is distinct from previously described TGF-f3 regu-

latory elements identified in the a2(I) collagen gene (22), the

plasminogen activator inhibitor (PA!- 1) gene (23), or the EGF

receptor gene (24).

ResultsProlonged Induction of CTGF mRNA by Short-TermTGF-j3 Exposure. Some immediate-early genes, such as

c-fos, that are induced by growth factors exhibit a short burst

of expression, although the growth factor remains present in

the media (25). Others, such as c-myc, exhibit a more pro-

longed period of expression (26), and some, such as JE/MCP-1 , remain induced as long as the growth factor is

present in the media (27). CTGF transcripts remain at high

levels for over 24 h after activation of the cells with TGF-p

(1 4). To determine whether the long term elevation of CTGF

transcripts was dependent on the continuous presence of

TGF-�, we performed the following series of experiments.

Confluent human skin fibroblasts were cultured in serum-free

DMEM-ITS for 24 h prior to adding TGF-f3. After a 1-h ex-

posure to TGF-p, cells were washed in PBS, and the mediawere replaced with DMEM-ITS, followed by different periods

of incubation. Northern blot analysis revealed that CTGF

mRNA was strongly induced from 4 to 30 h after TGF-�

removal (Fig. 1A). Next we investigated the ability of TGF-�

to induce long-term expression of the CTGF gene in the

presence of the protein synthesis inhibitor cycloheximide. As

shown in Fig. 18, a 1 -h stimulation by TGF-j3 in the presence

of cycloheximide was sufficient to induce CTGF mRNA 4 h

later as well as 24 h later. Cycloheximide alone induced a

transient increase in CTGF mRNA 4 h later, suggesting the

possibility of mRNA stabilization, as has been reported for

cycloheximide induction of other transcripts such as c-fos or

c-myc (28). However, a recent report by Edwards and Ma-

hadevan (29) indicated that the protein synthesis inhibitors

cycloheximide and anisomycin, but not puromycin, could act

to stimulate transcription of the c-fos and c-jun genes. We

investigated the ability of anisomycin and puromycin to ele-

vate CTGF transcripts or to inhibit TGF-p induction of CTGF

transcripts (Fig. 1C). Puromycin did not induce the

CTGF mRNA at any of the concentrations tested (up to 100

�g/ml), which were 1 0-fold higher than that needed to com-

pletely block protein synthesis in our cells. Even at this high

concentration, it had no effect on the ability of TGF-� to

induce the CTGF mRNA. Anisomycin did elevate CTGF tran-

scripts (Fig. 1C), as was seen with cycloheximide, although

TGF-p treatment raised the level of CTGF mRNA in the

presence of anisomycin. These data indicate that TGF-f3

directly regulates CTGF gene expression via a mechanism

that is independent of protein synthesis and that increased

transcription plays a role in this process.

To confirm that TGF-/3 induces increased transcription of

the CTGF gene, we performed nuclear run-on assays (Fig. 2).

Confluent human skin fibroblasts were cultured in DMEM-

ITS for 24 h, and their nuclei were isolated after a 4- or 24-h

exposure to TGF-p. We compared the level of CTGF tran-

- + - + - +

--++ --

- --- ++

�. a”--

scription with that of other “immediate-early genes” such as

c-fos, c-myc, JE, and Gro. TGF-f3 decreased c-myc gene

transcription 4 h after the addition to the cells, with the rate

returning to basal levels 20 h later. TGF-13 treatment elevated

CTGF gene transcription at 4 and 24 h after the addition the

cells, confirming that transcriptional activation by TGF-p is a

component of the mechanism for the large elevation in CTGF

mRNA detected in TGF-j3 activated fibroblastic cell types.

Isolation of the Human CTGF Gene. To study the pro-

moter of the CTGF gene, we needed to isolate genomic

clones containing the CTGF gene. This was accomplished by

first obtaining a fragment of the CTGF gene using PCR.

Using H02 and H03 primers, we recovered a 900-bp frag-

ment (H0900) that was 390-bp longer than predicted from

Cell Growth & Differentiation 471

4 Unpublished observations.

Cl)

I’�p

- I I

�Li�LL.

0I-l-

. 4� c-myc

#{149}JE.�. . KC/gro

S 0 CTGF

� �: BluescriptSK� GAPDH

Fig. 2. Nuclear run-on assays. Nuclei of human skin fibroblasts wereisolated from nonstimulated cells (C) or from TGF-f3-stimulated cells afterthe indicated periods of time. The labeled transcripts were hybridized to5 ,.L9 of plasmid containing an insert encoding the indicated cDNA, whichhad been previously spotted onto the nitrocellulose membrane. The in-serts (probes) for the various transcripts were as follows: fos, a i .3-kb PstIfragment of v-fos; myc, a 5.5-kb EcoRl fragment of v-myc; JE, a 0.8-kbPstI fragment; Gro, a 0.9-kb PstI fragment; CTGF, a 2.1-kb entire cDNA;and glyceraldehyde-3-dehydrogenase (GAPDH), a 550-bp HindIII-XbaIfragment from human GAPDH cDNA. All inserts were contained in theBluescript SK vector, which was used as a negative control.

the cDNA sequence. Nucleotide sequence analysis of this

fragment revealed the presence of a 387-bp intron in the

middle portion of the fragment. Using H0900 as a probe, we

isolated three phage clones from the human genomic library

that contained a 4.3-kb Xbal fragment that represented the

entire coding sequence of the CTGF gene and a large portionof the putative promoter region. As shown in Fig. 3, the CTGFgene has five exons and four introns and is similar in orga-

nization to the murine Fisp 12 gene (1 7). A TATA sequence ispresent 24 nucleotides upstream of the mRNA cap site, as

determined by oligonucleotide primer extension.4 The con-

sensus sequence of a CArG box, which is the inner core of

the serum response element characterized by CC(A/T)6GG,

is present between nucleotide position -380 to -390. Other

potential regulatory elements are also present, including a

CATbox (-58 to -53), three Spi sites (-42 to -37; - 1 5 to

- 1 0 and - 1 1 to -7), and two AP-l sites (-651 to -645 and

-291 to -285). The CTGF promoter has a NF-1 -like site

(TGGN6GCCAA), between positions - 194 to - 1 82, and a

TIE-like sequence (GNNTTGGtGa) between positions -128

to - 1 1 9. Both of these elements have single-base differ-

ences from the reported consensus sequences (30). DNA

sequence comparison showed that the human CTGF pro-

moter has an 80% sequence identity to the murine Fisp 12

promoter in the region 300 nucleotides 5’ flanking the tran-

scription start site (Fig. 3B). Further upstream regions exhibit

almost no similarity in DNA sequence.

TGF-13 Regulation of the CTGF Promoter. To determine

whether the 5’ flanking region of CTGF functions as a TGF-

u-inducible promoter, we constructed a fusion gene contain-ing the CTGF promoter (nucleotides -823 to +74) and the

coding region of the firefly luciferase gene in the vector

pGL2-basic. Luciferase activity was tested in a transient

transfection assay using NIH/3T3 cells. This construct con-

ferred a 25-30-fold induction of luciferase activity after a

24-h stimulation by TGF-f3 (1 0 ng/mI) compared with control

cultures (Table 1 ). We have reported previously that when

either CTGF protein or transcripts (1 4) were measured, other

growth factors, including PDGF, EGF, and FGF, stimulated

only a transient and low level induction. Using the luciferase

reporter assay, we found that growth factors that act via

tyrosine-based receptors, such as PDGF, EGF, and FGF,

stimulate only 2-3-fold induction of luciferase activity under

identical conditions (Table 1). When this promoter fragment

was cloned in the reverse orientation (+74 to -823), only

basal levels of luciferase activity were detected, and this level

was unaffected by TGF-13 or other growth factor treatment of

the cells. The same pattern of growth factor induction was

observed when human skin fibroblasts, or fetal bovine aortic

smooth muscle cells, were used instead of NIH/3T3 cells

(Table 1). TGF-!3 did not induce luciferase activity in several

epithelial cell lines (Table 1), demonstrating that TGF-� in-

duction of the CTGF gene appears to occur only in mesen-

chymal-derived connective tissue cells. The lack of any re-

sponse by the epithelial cells is not due to a lack of a TGF-f3

response because the growth of these cells is inhibited by

TGF-f3 (1 0 ng/mI).4 The induction of luciferase activity under

the control of the CTGF promoter only requires a brief ex-

posure of the cells to TGF-/3 because a 1 -h treatment of the

cells with TGF-�3 gives nearly the same fold induction at 4

and 24 h as cells continuously exposed to TGF-/3 (Table 2).

These results confirm the data from the Northern blots de-scribed previously and demonstrate that transcriptional reg-

ulation plays a primary role in the control of CTGF gene

expression by TGF-�.

Identification of a Region in the CTGF Promoter Re-quired for TGF-fi Induction. To delineate which region of

the promoter sequence is responsible for the induction by

TGF-f3, we constructed deletion mutants of the CTGF pro-

moter using PCR primers designed to delete the known

transcription factor consensus elements. Regions of the pro-

moter beginning at the 5-most region and moving toward

the transcription start site were systematically deleted

(Fig. 4A). No significant changes in the basal level of expres-

sion of the chimeric gene were detected for any of the

deletion constructs compared to the largest construct used

(-823 to +74). With respect to the TGF-f3 inducibility of theCTGF promoter, removal of the nucleotide sequence from

-823 to -363, which included an AP-1 site and the CaRG

box, had no significant effect on the induction of luciferase

activity by TGF-j3. An approximate 25% reduction was seen

B

A

-823 CTGF

TCTAGAGCTCGCGCG�.GCTCTAATACGACTCACTATAGGGCGTCGACTCGATCCCTTTT--TCTGGAAA.CATTGA-TGGCCACTCGTCCCTTGTCCTTGCCTATATAAAACTCCTACATA

TCT--TTCT-TCTCCCACTATATTCCCTGACAC--TTAGGCTTCTGAAGATAGCCATTTGGTCTGAACTCATAAACTTATTTTTCTA--GAAAACCATGCCCAGTAATACCCCTTGCCTG

-596 CTGF

-706 CTGT AP-1

TATTAAG--AGAAAACTAA GCA---AGAGTTTTGGA.AATCTGCCCCAGGAGACTGCATCCTGAGTCACACGAGTCTTTGTTCTCTTTCTTGTCCCAAAACCGTTACCTCAA.GTGA

CCAA-GATCATGCCTGTATTTCAGATACAA�.AGTTGCACATAGGAATTCTGGcaGGAGAGGAGGCATTTCAAATGGCTATAAGCAC-CCTTCTCCTCTCAGTAGAA.GAACACCAAGAG

-483 CTGF CkrGbox

CCTATAGCCTCT AAAACGCAAATCATTGCTAA000TTG000000AGAAACCTTTTCGAATTTTTTAGGAATTCCTGCTGTTTGCCTCTTCAGCTACCTACTTCCTAAAAAGGATG

ACTACAGCCCCGTAAAGAAAAAAAAAAAAAA.ATCCAAAACAAAGAAAAAGAAATATTTTTTTTAATTTCTAGGGGCCCATGGTATTTGCCTCTTGAGCTATTTG�.GTCTTGAGAAGTTTTAP-1

-368 CTGT AP-1

TATGTCAGTGGACAGAACAGGGCAAACTTATTCGAAAAAGAAATAA--GA.AATAATTGCCAGTGTGTTTATA�.ATGATATGAATCAGGAGTGGTGCGAAGAGGATA-GGAAAAAAAAATT

472 TGF-f3 Response Element in the CTGF Gene

22 74 85 70

Amino Acids

98

CTGF

FIsP

CTGF

CTCCAGTGA FISP

AGGCATCTTGAGG ATTTCAAATGGTTAAAAGCAACT-CACTCCTTTTCTACTCT-TTGGAGAGT’rTCAAGAG CTGF

FIsP

CTGF

FISP

CTGF

-251 CTGT NT-i-like TGF-� element

CTATTTGGTGCTGGAAATACTGCGCTTTTTTTTTCCTTTTTTTTTTTTT- -CTGT-GAGCTGGAGTGTGCCAGCTTTTTCAGACGGAGGAATGCTGAGTGTCAA0000TCAGGATCAATC CTGF

CTTCTTGGTGTTGTGCTGGAAACAC AGCGCCTTTTTTTTTTTTTTCCTGGCGAGCTh.AAGTGTGCCAGCTTTTTCAGACGGAGGAATGTGGAGTGTCAA0000TCAGGATCAATC FI5P

-134 CTGF

TIE-Uk. CATbox SP-1 TATAbox

CGGTGTGATTG�TG�GGCAGGAAGGTGGGG�.GGAATGCGAGGAATGTCCCTGTTTGTGTAGGACTCCATTCAGCTC�TTGGCGA.GCCGCGGCCGCCCGGAGCGTATAAAAGCCTCGGG CTGF********* *

CGGTGTGAGTTGATGAGGCAGGAAGGT0000AGGAATGTGAGGAATGTCCCTGTTTGTGTAGGACTTCATTCAGTTCTTTGGCGAGCCG GCTCCCGGGAGCGTATAAAAGCCA GCG TI SP

- 15 CTGF

sp-1 +2.CTGF

********* 9 ******** *9*9*

CCGCCCGCCT-AGTCTCACACAG---CTCTTC YISP

Fig. 3. A, structural organization of the CTGF gene. Exons are the indicated boxed regions and are numbered above. Solid areas in the boxes correspondto the open reading frame, and the numbers below indicate the number of amino acids encoded by that particular exon. B, comparison of the nucleotidesequence of the CTGF and Fisp 12 gene promoter regions. “, identical nucleotides. Known consensus elements are indicated by shaded regions and arelabeled. The site of transcription initiation is indicated at position + 1.

when the region from -363 to -276 was deleted, which

contained the second AP-1 site. Deletion of the DNA se-

quence (-276 to -128), which contains the NF-1-Iike Se-

quence, eliminated the TGF-f3 inducibility of the promoter,

suggesting that this region contained an important TGF-f3

response element. Taking advantage of two Bsml sites, wedeleted the nucleotides from -162 to -95, leaving the re-

mamning portions of the promoter intact. This construct

exhibited a complete loss of TGF-f3 inducibility, demonstrat-

ing that the nucleotides sequence between positions -162

and - 128 appears to be essential for the TGF-�3 induction ofluciferase activity. In this construct, the TIE-like site is

deleted, but the NF-1-like site is retained.

We next constructed a fusion gene by placing the nude-

otides from positions -275 to -106 of the CTGF promoter

upstream from an SV4O enhancerless promoter controlling a

luciferase structural element to determine whether this re-

gion of the promoter was sufficient to confer TGF-�3 induc-

ibility (Fig. 48). The 5V40 enhancerless promoter was not

regulated by TGF-�. However, the promoter containing the

CTGF sequences -275 to - 1 06 conferred nearly a 9-fold

induction after TGF-f3 treatment. Inversion of the fragment

resulted in little stimulation of luciferase activity after TGF-/3

treatment. These data confirm the results of our deletion

studies and indicate that sites located between nucleotides

-275 and - 1 06 of the CTGF promoter can act indepen-

dently as TGF-j3 regulatory elements.

Nuclear Protein Binding Sites in the TGF-p ResponseRegion of the CTGF Promoter. A series of competitive gelshift assays were performed to identify sites in this region of

the CTGF promoter that were binding nuclear proteins and to

determine if there were any differences between the gel shift

patterns of nuclear proteins isolated from control and TGF-j3

treated cells. Nuclei were prepared from control and TGF-�-

treated cells (10 ng/ml for 6 h) and the proteins extracted as

described under methods. A fragment of the promoter con-taming the nucleotides from -275 to - 1 09 was used as the

labeled probe. As seen in Fig. 5, an identical pattern of asingle shifted band was detected in the gel shift assay when

nuclear protein extracts were used from either control or

TGF-�3-treated cells. Studies comparing extracts prepared

from isolated nuclei versus whole cells or cytoplasm indicted

Table 1 Cell type and growth factor regulation of the CTGF promoter Table 2 Short-term TGF-f3 exposure stimulates long-term CTGFpromoter activity�

a Fold induction ofluciferase activity determined as described in the Table

1 legend and in “Materials and Methods.” NIH/3T3 cells were used forthese experiments. Results are the average of duplicate transfections,where the values varied by less than 10%. These studies were repeatedin three separate studies, all of which gave similar results.

1.8 ND 1.9

(1 .0) (1 .3) ___________________(1 .4) (1.8)

a No ADD, no additions; ND, not determined; HSF, human foreskin fibro-

blasts (primary); VSMC, fetal bovine aortic smooth muscle cells (primary);HBL100, human breast epithelial cell line (nontumorigenic); HEP G2, hu-man hepatic epithelial cell line (nontumorigenic).Numbers in parentheses, (23.4) fold induction over No Additions Controlluciferase activity, photons/mm x 10-6.

Cell Growth & Differentiation 473

A CTGF gene fragment extending from nucleotides position -823 to+74 was inserted in the pGL2-basic vector. Plasmids were transfectedwith lipofectin for6 h, and cells were incubated in DMEM-ITS for 16 h priorto the addition of growth factors. After a 24-h incubation, cell extractswere prepared, and luciferase activity was measured. Luciferase activitieswere normalized by measuring (3-galactosidase activity expressed from acotransfected IacZ expression vector, pSV-f3-galactosidase vector, andcompared between growth factor-treated cells and nontreated cells.These experiments were repeated five times with similar observations. Arepresentative experiment is shown that represents the average of dupli-cate transfections. Results varied by less than 1 0% of average value.

Cell No ADDS TGF-f3 PDGF FGF EGF

NIH/3T3 0.75 19.3 2.2 2.5 1.1(1 .0) (25.7) (2.9) (3.3) (1.4)

HSF 0.38 6.8 0.9 1 .2 0.8(1.0) (16.8) (2.4) (3.1) (2.2)

VSMC 0.37 7.1 1 .1 0.6 0.5(1 .0) (19.2) (2.9) (1 .6) (1.4)

HBL 100 0.99 1.4 ND 1.3 1.4(1 .0) (1 .4) (1 .3) (1.4)

HEPG2 1.36 2.4

that the protein was predominantly present in the nucleo-plasm (data not shown). The binding to these nuclear pro-teins appears to be specific because it can be competed witheither a fragment of the promoter, which is identical to theprobe (-275 to -106), or by a smaller fragment (-205 to-109) but not by nucleotides from -128 to +74 (Fig. 5).Because no significant difference was seen between thecontrol and TGF-/3 activated cells, we used nuclear extractsfrom nonactivated cells for all future studies, unless other-wise indicated. These data suggest that a TI3RE is locatedbetween nucleotide positions -204 and - 128. The region ofthe promoter from -204 to -1 28 contains the NF-1 -like andTIE- like sequences and a portion of the region deleted byBsml digestion in the P5 promoter construct that lacksTGF-f3 inducibility (Fig. 4A). We next compared the ability ofoligonucleotides that represented various portions of thesequence between positions -204 and - 109 to act as com-petitors in our gel shift assay. A diagram of the probe and thevarious competitor fragments used is illustrated in Fig. 6. Theresults of these studies demonstrate that any fragment thatcontained nucleotides from -1 69 to - 150 could act as aspecific competitor for the labeled promoter fragment con-taming bases from -204 to -1 06. Oligonucleotides thatcontained only the NF-1 like or TIE-like sites were not effec-tive as competitors in this assay.

To map the binding of nuclear proteins more precisely tothis and adjacent regions of the CTGF promoter and todemonstrate that specific sequences were responsible forthe binding in the gel shift assays, we performed sequenceanalysis using methylation interference assays. Initially, afragment of the promoter from positions -275 to -1 06 was

Duration of TGF-f3exposure

Time of assay ofluciferase activity

4h 24h

Continuous 3.8 21

lh 3.5 19

used to detect binding sites. The methylation pattern wasdetermined for the shifted band (Fig. 7A, Lane S), for the

non-shifted free probe (Fig. 7A, Lane F), and for the intactfree probe (Fig. 7A, Lane G). As we had seen in the compe-tition gel shift assays, the results of these studies indicatedthat neither the NF-1 like site or the TIE-like element wereinteracting with any nuclear proteins. We also did not detect

any evidence of protein binding to any sequence between

nucleotides -275 and - 1 06, except for the region located at

nucleotides from - 157 to - 147, where four G residues werenot methylated in the shifted band, indicating that this was anuclear protein binding site. Methylation sequence analysisof the complimentary strand indicated that two G residues,which were paired to C residues at positions - 153 and- 1 45, were also involved in nuclear protein binding (data not

shown). A smaller fragment of the region (nucleotides -169to -139) was then used to give better resolution of theinvolved G residues. The results of these studies (Fig. 7B)

confirmed our observations with the larger fragment andmap G residues on both strands that are located betweenpositions -157 and -145 upstream from the transcription

start. This sequence is in the same region of the promotermapped by the competition gel shift assays and is containedwithin the region (-204 to -128) delineated by the deletionmutant analysis in the luciferase reporter assays.

We then wanted to compare various oligonucleotides fromthis region of the promoter to determine which was mosteffective for binding to the nuclear proteins. Competitionbinding assays were conducted that compared the effective-ness of oligonucleotides containing the intact sequence de-termined from the methylation interference assays with oh-gonucleotides that contained only a portion of the sequence

determined by methylation interference and some of thecontiguous adjacent sequence. The concentrations of theohigos were adjusted so that equivalent amounts were usedas competitors. The results of these experiments are illus-trated in Fig. 8. The ohigonucleotide that represented theCTGF promoter sequence from positions -159 to -143 andcontained the intact sequence determined by methylationinterference sequence analysis was a highly effective com-petitor with 10 ng of ohigonucleotide completely blocking thebinding to the labeled probe (nucleotides - 169 to -139).Ohigonucleotides that lacked the 3-prime TC sequence of the

mapped element were 10-fold less effective competitors,and those lacking the 3-prime region from - 1 51 to -145

ANH3 Luciferase

p��23 � ���y74

P1 I � I HIjll�h �_.-�_

.363P211

.276

p3 1 F1�llHH �

-823

P51 I II I �

.162 to .95 Deleted

B __

CTGF sense ______________

CTGF antisense I�IF1 I SV4O-106 .275

�

U NH3

� TIE-LIKE

were 1 00-fold less effective competitors than the ohigonucle-otide containing the intact sequence. An ohigonucleotide that

contained only the region from -150 to - 145 and lacked the5-prime portion of the element was nearly 1000-fold lesseffective as a competitor. These data indicate that both ofthese regions of the sequence are important for recognitionby the nuclear proteins that are binding to this element.

To link this binding element to the TGF-f3 regulation of the

CTGF gene, we constructed intact promoters with PMs thatwere contained within the element from positions - 157 to- 145 (Table 3). The generation and fidelity of these PMconstructs were confirmed by DNA sequence analysis, andthe mutant promoters were tested in the luciferase reporterassay. We chose to alter G residues that had been identifiedby the methylation interference assay as sites that wouldmost likely have an impact on TGF-f3 responsiveness of thepromoter. Four different constructs were produced that con-tamed either single-base PM2 and PM4 changes or two-base PM1 and PM3 changes. None of the PMs significantlyaltered the basal level of expression of the CTGF promoter.However, all mutant constructs exhibited significant reduc-

T G F � Fig. 4. A, deletion analysis of- CTGF promoter-luciferase con-

___________________ structs. Known consensus so-I quences are indicated. NIH/3T3

- I fibroblasts were transfected asdescribed under experimentalprocedures. Ten ng/ml of TGF-p

28 were added for activation of thecells, and cell extracts were pro-

23 pared 24 h later. The amount ofluciferase activity was deter-mined for both basal levels of ex-

26 pression and after activation ofthe cells with TGF-p (10 ng/mI).These studies were repeated six

n 41 8 2� 20 times with similar results. Data‘,. . represent the average of dupli-

cate transfections with the mdi-cated construct performed in a

I .3 single set of experiments. B,

TGF-p response of an SV4O en-

I 5 hancerless promoter element-. luciferase reporter construct

containing nucleotides -275 to-106 ofthe CTGF promoter. Theindicated region of the CTGFpromoter was cloned in both ori-entations upstream from an

,� ,� SV4O enhancerless promoter,0.52 0.47 u.� w�ilch was driving the expression

of a luciferase reporter. Assayswere performed as described

n 5Q 5 1 � R Q above. Cellsweretreated with 10V. �1 . I � � ng/mI of TGF-f3 for 24 h prior to

extraction and analysis of lucif-erase activity. These experi-

0.49 1 .03 2.1 ments were repeated six timeswith similar results. Data repro-sent the average of duplicatetransfections of the indicatedconstruct from a single experi-mental set.

tions in the TGF-/3 inducibihity of the chimeric gene. The PM4construct, which contained a single change from the normalsequence with a T in place of a C at position - 153, exhibiteda 45% reduction in TGF-13 induction of luciferase activitycompared to the normal promoter. All three of theother constructs, PM1 (two G residues replaced withT at positions -1 55 and -1 57), PM2 (G replaced with a T atposition -148), and PM3 (two G residues replaced with

T at positions -148 and - 1 57), exhibited little if any detect-able induction of luciferase activity by TGF-/3 compared tothe normal promoter sequence. These results demonstratethat PMs in this region can inactivate the response of theCTGF promoter to TGF-/3 and identify this sequence as aT/3RE.

DiscussionThe regulation of CTGF gene expression is distinct from thatof other genes that have been demonstrated to be directlyregulated by growth factors in two ways: (a) the gene isstrongly induced by TGF-� but not by other growth factors,

474 TGF-p Aesponse Element in the CTGF Gene

LUCIFERASE ACTIVITY(Photons x 1O�)

0.45 12.6

0.48 11.0

HI1llMI� � 0.53 13.8

.128

P 4 flJ�fl�� .--_-� 0.47 0.61

Lu ciferase

ENHANCERLESS�__sv4o

��1O6

Ullil I SV4O

4-

0.41 0.62

I-.. N+ +0 0

00 asr’i r’�- -

1 2 3 4 5 6 7 8 9 10

Fig. 5. Specificity of binding and effect of TGF-/3 on the gel shift pattemof nuclear proteins that bind to the -275 to -106 region of the CTGFpromoter based on gel shift assays. A nucleotide fragment consisting ofthe region from -275 to - 1 06 of the CTGF promoter was end labeled with32p and purified for use in competitive gel shifts. Nuclei were isolated fromcontrol or TGF-f3-treated cells (1 0 ng/ml for 6 h) as described in “Materialsand Methods.” Gel shift assays were performed with proteins isolatedfrom control cell nuclei (Lanes 1-4 and 10) and TGF-f3-treated cell nuclei(Lanes 5-9). Lanes 1 and 8, the gel shifts with no competitor nucleotidepresent. Arrow, the specifically shifted band. Lanes 2-7, 9, and 10 contain

competitor oligonucleotides that were identical to the indicated regions ofthe CTGF promoter sequence. Specific competition was detected whenoligonucleotides containing sequences between -205 to -109 wereused. No difference was observed in either the gel shift pattem or com-petition binding activity of nuclear proteins isolated from either control orTGF-f3-treated cells.

Cell Growth & Differentiation 475

COMPETITOR OLIGONUCLEOTIDEas ‘a a� so \Cri © � r’i ©- �4 - - - -

� � � � � �z: ‘,� ‘,� ,,� ‘,� II� ‘1) ZC N N © t”” N ©z � r�lr�I r�i

TGF-B - - - -++

such as PDGF, FGF, or EGF. This is due to different signaltransduction pathways produced by activation of the TGF-�

receptor, which is a serine/threonine protein kinase (31 , 32)

compared to the tyrosine kinase receptors for PDGF, FGF or

EGF (33, 34) and to the T,3RE we have identified in the CTGF

promoter, which is not present in the promoter of any of thegrowth factor-regulated genes that have been reported to

date; and (b) the kinetics of CTGF expression are unique.

Many immediate-early genes, such as c-fos, c-jun, and c-

myc, and Cyr6l are transiently expressed during the G1

phase of the cell cycle or require continued presence of theinducing growth factor to maintain expression, such as JE. In

contrast to these observations, a brief exposure of the cells

to TGF-/3 (1 h) causes a prolonged activation of CTGF geneexpression that lasts for 48 h. The principal control mecha-nism for CTGF gene expression appears to be at the level oftranscription. Our studies with protein synthesis inhibitors

indicate that TGF-�3 can stimulate long-term activation (24 h)

of CTGF gene expression after a 1 -h exposure to TGF-� in

the absence of any protein synthesis at that time. The studies

with the CTGF promoter-luciferase reporter constructs mdi-

date that a brief exposure of the cells to TGF-j3 is sufficient tostimulate elevated transcription from this promoter for long

periods of time. Thus, CTGF may act as a mediator of long-term effects initiated by TGF-�.

The promoter of the human CTGF gene contains multiple

consensus elements that are characteristic of other growth

factor or serum-induced genes, including a CaRG box and

AP-1 and SP-1 sites. It has been reported that TGF-p can

activate expression of other immediate-early genes, such as

c-jun and jun-B (35), and of its own gene (36) through AP-1(IRE) sites. However, these elements do not appear to be

essential for TGF-J3 up-regulation of the CTGF gene because

a fragment of the promoter that does not contain any of these

elements (-275 to -109) can confer TGF-p induction on a

previously nonresponsive promoter (Fig. 4B). Two regions of

the promoter contained within this sequence are related but

not identical to the reported TGF-13 control elements, NF-1and TIE. NF-1 sequences have been reported to control, in

part, the up-regulation of the a2 I collagen gene by TGF-�

(22) and the PA!-1 gene (23). The TIE [consensus sequence

(GNNTTGGtGa); Ref. 30] are present in a number of genes,such as the stromelysin gene promoter (37), where this ele-

ment acts to repress transcription in response to TGF-p. A

closely related sequence is present in the c-myc promoter

O�CE), which inhibits the expression of this gene in responseto TGF-�3 (38). However, there are no sequences present in

the CTGF promoter that are identical to the consensus ele-

ments of either the NF-1 , TIE, or TCE sequences. Further-

more, the sequences present in the CTGF promoter that are

similar to these regulatory elements do not appear to playany active role in TGF-j3 response. The results of deletion

analysis using the luciferase reporter assay demonstrate that

fragments of the promoter that contain either the NF-1 -like

site (P5) or the TIE-like site (P4) are not inducible by TGF-f3

(Fig. 4A). Competitive gel shift and methylation interference

assays confirm that these regions of the promoter are notbinding any nuclear proteins in either control or TGF-f3-

treated cells. A sequence that lies between the NF-1 -like and

TIE-like sites is the region that is binding to nuclear proteins.

This region is deleted in all reporter constructs that lack

TGF-f3 responsiveness.The sequence we have identified as the Tf3RE element

(GTGTCAAGGGGTC) in the CTGF promoter has not been

described previously as a control element. The sequence is

absolutely conserved in the murine Fisp 12 gene. A search of

the GenBank promoter data base did not indicate any closely

related elements that have been identified at the current time.

We also did not find the sequence in the CyrSl promoter or

in the promoters of other immediate-early genes that have

been identified. Other TGF-j3-regulated genes, including

co!!agen, PA!- 1 , and fibronectin, do not contain this or any

closely related sequence in the promoter region. These data

indicate that the regulation of CTGF gene expression by

TGF-�3 may function by a distinct mechanism than otherTGF-f3-regulated genes. The proteins that bind to this se-

quence are constitutively present in the fibroblast nucleo-

plasm, as anticipated from the studies demonstrating that

the CTGF gene is directly regulated by TGF-f3 and can beinduced in the presence of protein synthesis inhibitors.

These data suggest that some type of posttranslational mod-

ification of these proteins is needed for the alteration in

transcription rates of the CTGF gene induced by TGF-f3. Thebiochemical basis for this is under investigation.

CTGF is a member of a growing family of related peptides.

All of the members of this gene family that have been iden-

205 �194

LANE

182 .169 .134 .128 �119 .109

1I��NF�1

I I IIlike NH3 TIE like

45 -

8 6

9

AGSF B

7

0C

0

Ce

competItor co.000�

a)C0C

O)OO)OO)i-O0 LOOL0000LO1#{149}- � T� 1� �- �

“-“ -“- � “- � “- �““

L() LOO)O)�LOLC)

0 0 CD CO C’sJ 0 0)c’J CsJ�-i-,--CsJw�-

4

1 2 3 4 5 6 7 89

3’ G S F

�___

Fig. 7. Methylation interference assay of the - 205 to - 1 09 region of theCTGF promoter. A, sequence analysis of the region from -205 to -109.The sequence from -200 to - 1 13 is shown. Assays were performed asdescribed in “Materials and Methods.” Lane G is the G sequence of theintact labeled probe; Lane S is the sequence of the shifted band; Lane Fis the sequence of the nonshifted free probe from the same sample. Theonly region containing missing G residues is from positions -157 to -147.B, sequence analysis ofthe region - 1 59 to - 1 42 using a smaller fragmentof the promoter (- 169 to - 1 39). Lanes are the same as in A. Armws,competed G residues in this sequence. S, G residues detected in analysisof complementary strand (data not shown). Symbols ‘ and # are fororientation with the sequence in A.

and brain of adult chickens but is expressed in the embryo at

high levels only in kidneys (20). We did not find that either

Cyr6l or nov is significantly induced by TGF-� in human skin

fibroblasts or NIH/3T3 cells. In adult mammals, CTGF isexpressed at high levels during wound repair (1 4) and at sitesof connective tissue formation in a variety of fibrotic disor-

ders (40).� The expression of CTGF transcripts in all of the in

Fig. 6. Competitive gel shift assays to delineate the binding site fornuclear proteins in the - 205 to - 109 region of the CTGF promoter. Anucleotide fragment consisting of the region from - 205 to - 1 09 of theCTGF promoter was end labeled with 32P and used in competitive gelshifts with the indicated oligonucleotides. Arrow, the specific gel shiftedband. A diagram of the sequences used indicates the position of theoligonucleotides used relative to the NF-1 - and TIE-like elements. Thenumbered fragments in the diagram indicate the lane number in thecompetitive gel shift assay with the specific nucleotide sequence mdi-cated above the lane (i.e., 3, - 205/ - 1 50). Unlabeled competitors wereused at a 250-fold molar excess over the labeled fragment. Only oligo-nucleotides containing the region from - 1 69 to - 1 50 acted as specificcompetitors. Neither the NF-1 - or TIE-like regions competed in this assay.

tified in vertebrates share high levels of amino acid sequence

homology and an absolute conservation of the number and

spacing of 38 cysteine residues. All of these peptides are

secretory proteins and have multifunctional domains sug-

gesting a mosaic structure and function of these molecules.

Sequence analysis of the various members indicates that all

have four distinct domains. These domains are (from the

amino to carboxyl terminal of the protein): (a) an insulin-like

growth factor binding protein-like domain; (b) a von Wille-brand type C repeat; (c) a thrombospondin type I repeat,which may be involved in glycosaminoglycan binding; and

(d) a COOH-terminal domain, which is homologous to otherextracellular mosaic proteins and shows potential structural

relationship to TGF-3 (see Ref. 39 for a detailed discussion ofthese relationships). Interestingly, the genomic organization

of the five exons of the CTGF gene corresponds with thecoding regions of these four domains with the first exon

encoding the signal peptide. The functional analysis of these

domains in the CTGF protein should provide more clues to its

biological activities.

With the exception of CTGF, which was originally identifiedby its mitogenic and chemotactic activity for connective tis-

sue cells (1 5), the biological activities of the other CTGF-

related gene products, such as Cyr6l/CEF1O or nov, is not

well defined. In the adult, the Fisp 12 and Cyr6l genes show

similar patterns of expression, with the highest levels de-

tected in the lung, although these are 10-20-fold below the

level of expression in induced fibroblasts in culture (1 6, 17).

476 TGF-f3 Response Element in the CTGF Gene

Similarly, nov is expressed at detectable levels in the lung 5 G. R. Grotendorst, K. Frazier, and M. Glassberg, unpublished observations.

.204 NF-l like .169 .134 -128 TIE like -106

II 7��i �x’/��/A

I,

-169

‘I

Illu Illu -134GAGGAATGCTGAGTGTCAAGGGGTCAGGATCAATCC

PROBE . . . .

.159/.143 I I

-162/.147 I

.163/.152 --150/.134 I I I

(9) P� (‘1 �‘.� � U) C’)c� - � � � -

U S S I I

0 � ..L � �c� � C’) 0 io�It) � . I0

- ,- �. �. � -

I I I S I U

0000e000000000000 00 000� 0� 0� � 00- - - �.

I 2 3 4 5 6 7 8 9101112131415

I � I

vivo studies we have made is highly correlated with the

expression or presence of TGF-f3. Injection of TGF-f3 into thesubdermal layer of newborn mouse skin induces a high level

of CTGF transcripts in fibroblasts present at the site within 24

h, based on in situ hybridization studies.6 !n vitro studies in

our laboratory with recombinant CTGF indicate that it is a

mitogen for connective tissue cells and that it may mediate

many of the effects of TGF-j3 on synthesis of extracellular

matrix proteins, including collagen and fibronectin. These

findings lead us to speculate that CTGF may be a mediator

of some of the action of TGF-f3 on connective tissue cells.

The coordinate expression of TGF-p and CTGF during

connective tissue regeneration in mammals supports a cas-

cade model where initiators of the process stimulate produc-

tion of secondary factors that sustain and regulate specific

cellular processes in the regenerating tissue. Our observa-

tions demonstrating the prolonged induction of CTGF ex-

pression after a transient exposure to TGF-� are consistent

with such a cascade model. Other experimental studies in

our laboratory indicate that a brief exposure of cells to TGF-p

are sufficient to induce biological effects that persist long

after the removal of the TGF-f3. TGF-j3 was initially described

based on its unique ability to stimulate the growth of normalrat kidney fibroblasts to grow in an anchorage-independent

culture system (agarose suspension; Ref. 41). We have in-

vestigated whether a transient exposure (1 h) of normal ratkidney fibroblasts to TGF-p is sufficient to stimulate DNA

synthesis in an anchorage-independent growth assays that

allows for recovery and washing of the cells to remove the

TGF-f3 (42). The results of these experiments indicate that a1-h exposure to TGF-p induces a 3.5-fold increase in DNA

synthesis 24 h later compared with control cells that were nottreated with TGF-j3. Continuous exposure to TGF-� under

these conditions resulted in a 6-fold increase in DNA syn-

thesis. These data indicate that TGF-�3 need not be contin-

uously present to induce biological effects on target cells,

and a 1 -h exposure of cells to TGF-f3 can result in growth

stimulation. The role of CTGF in this process in currently

under investigation.

Regulatory cascades involving TGF-f3 and CTGF-related

genes may play an important role in early development as

well as in wound repair. In the early Drosophila embryo, a

cascade of gene products controls pattern formation during

Cell Growth & Differentiation 477

6 G. A. Grotendorst, unpublished observations.

Fig. 8. Competitive gel shift titration assayofoligonucleotides in the T�RE. Overlappingand nonoverlapping oligonucleotides con-taming portions of the -157 to -145 regionof the CTGF promoter were tested in thecompetitive gel shift assay using a 32P-endlabeled human CTGF promoter fragment(-205 to -109). An oligonucleotide from-159 to -143, which contains the intactsequence, exhibits the highest affinity withcomplete competition at 1 0 ng. All otherfragments that contain only a portion of thissequence are less effective, with the - 150 to- 134 region being the least effective. Lanes14 and 15 are the NF-1- and TIE-like ele-ments, respectively, and show no competi-tion at 5000-fold molar excess of labeledprobe.

I I I I

I I I �

Ii I �

I � I � I

478 TGF-� Response Element in the CTGF Gene

Table 3 Effect of PMs in region -157 to -145 on TGF-(3 induction

Sequenceused

Luciferase expression (photons x 10�)

Basal exp. TGF-f3 FOLDinduced inductiona

NativeGTGTCI4AGGGGTC 0.17 ± 0.02 2.05 ± 0.08 14.7(100)

PM1tTtTCAAGGGGTCb 0.1 1 ± 0.03 0.13 ± 0.03 1 .2 (8)

PM2GTGTCAAGGtGTCb 0.21 ± 0.05 0.51 ± 0.05 2.3 (16)

PM3tTGTCAAGGtGTCb 0.1 1 ± 0.05 0.18 ± 0.03 1 .64 (11)

PM4GTGTtMGGGGTC�’ 0.16 ± 0.03 1 .04 ± 0.03 6.50 (44)

a Numbers on parentheses (16) are the percentage of fold induction rel-

ative to the native sequence.b Bases in bold, bases mapped by methylation interference.

PMs were generated by designing oligonucleotides with the desired PMlinked to a BSM-1 site identical in sequence to the site at positions -167to -161 . This was used with a second oligo to PCR amplify a mutantBSM-1 fragment region, which was subsequently isolated and cloned intoa BSM-1 -digested P0 (intact, native sequence) CTGF promoter fragmentcontained in the Bluescript plasmid. Recombinant clones were checkedby restriction digestion and the DNA sequence analysis of the entiresequence to ensure that only the PM had occurred. These inserts werethen cloned into the pGL2 vector. Transient transfections were performedwith NIH/3T3 cells, and the amount of luciferase activity was determinedin control (basal) and TGF-fractivated cells.

dorsal ventral axis organization (43, 44). Several of the genesthat have been identified in the dorsaVventral cascade of theDmsophila embryo, decapentaplegic (dpp; Ref. 45) andscrew (scw; Ref. 46), encode proteins that are members ofthe TGF-� superfamily and are closely related to mammalian

genes that have been identified as bone morphogenetic pro-teins (47). Other studies have indicated that the human bonemorphogenetic protein gene sequences can confer normaldorsal ventral patterning in the Drosophila embryo (48). An-other gene in the dorsaVventral cascade [twisted gastrulation(tsg)] is related to CTGF (21). Twisted gastrulation controlsthe cell fate of medial mesodermal elements in the Drosoph-

ila embryo, where it acts in coordinate fashion with the geneproducts of dpp and scw (21). Collectively, these observa-tions suggest that the genetic pathways controlling tissueformation are highly conserved and that the wound repaircascade may be derived from or possibly identical to thepathways that originated to specify mesodermal tissueformation and organization during embryogenesis in the

invertebrate.

Materials and Methodscell cultures. Human skin fibroblasts were grown from explants of skinbiopsy specimens. NIH/3T3 cells and Cos 7 cells were obtained from theAmerican Type Culture Collection (Aockville, MD). All cells were culturedin DMEM containing 10% FCS at 37#{176}Cin an atmosphere of 10% CO2 and90% air. Human skin fibroblasts were used prior to the sixth passage.

Growth Factors. TGF-f31 was a gift from Richard Assoian (Universityof Miami). Recombinant PDGF-BB was obtained from Chiron (Emeryville,CA). Purified murine EGF was purchased from Biomedical Technologies,Inc. (Stoughton, MA). Purified basic FGF was purchased from SigmaChemical Co. (St. Louis, MO).

RNA Isolation and Northern Blotting. Total ANA was isolated from

cultured cells by acid guanidium thiocyanate-phenol-chloroform extrac-tion, as reported previously (49). Total RNA was electrophoresed on an1.5% agarose/formaldehyde gel and transferred to nitrocellulose. TheCTGF probe was a 1 .1-kb fragment representing the CTGF open reading

frame obtained by PCA reaction using specific primers HOl (5’-cggaat-

tcgcagtgccaaccatgacc-3’) and H02 (5’-ccgaattcttaatgtctctcactctc-3’).Hybridizations were performed using 1 x lO6cpm/ml of these probeslabeled with [a-32P]dCTP by using a Random Primer DNA labeling kit

(Boehringer Mannheim Biochemicals, Indianapolis, IN). Autoradiography

was performed at -70#{176}Cfor6 to 72 h by using X-ray films and intensifying

screens.Isolation of Genomic clones and Sequence AnalysIs. Genomic

DNA was isolated from human skin fibroblasts as described previously(50). Using 4 ,L9 of genomic DNA as a template, a fragment of the CTGF

gene was amplified by PCR using primers H02 (5’-ccgaattcttaatgtctct-cactctc-3’) and H03 (5’-cggaattcctggaagacacgtttggc-3’). PCA products

were digested with EcoRl and subcloned into M13. Sequence analysis bythe dideoxy chain termination method (51) using the Sequenase kit (U.S.Biochemical Corp., Cleveland, OH) demonstrated a 900-bp fragment that

had a 387-bp intron in the middle portion. Using a Human Genomic

Ubrary in the Lambda FIX II vector(Stratagene, La Jolla, CA), we screened

approximately 1 x 106 recombinant phages with 32P-labeled 900-pgenomic DNA fragment as probe and isolated three phage clones thatcontained the CTGF gene.

Nuclear Run-On Assays. Nuclear run-on transcription assays were

performed as described previously (26) with a slight modification. Weprecipitated ANA by the addition of sodium acetate (pH 5.2) to a final

concentration of 0.3 M and ethanol to 70% instead of using trichloroaceticacid. The [�P]UTP-labeIed run-on transcripts were hybridized at 8 x 106

cpm/mI to filters on which 5 �g of corresponding cDNA was immobilized.

Luciferase Reporter Gene Assays. A fragment of the CTGF pro-

moter containing nucleotides -823 to +74 from one of the humangenomic clones was first cloned into the SacI-XhoI cloning site of pGL2-Basic vector(Promega). We used this construct (P0) as a template for PCR

and made deletion mutants with specific primers as follows: P1 contained

nucleotides from -638 to +74, P2 from -363 to +74, P3 from -276 to+74, and P4 from -128to +74. NIH/3T3 cells were transfected in a 6-well

plate with Lipofectin reagent (Ufe Technologies, Inc.) for 6 h. Each trans-fection included 2 �g of reporter plasmid and 0.5 �g of pSV-�3-gaIacto-sidase vector (Promega). Cells were incubated in serum-free DMEM-ITS

(Collaborative Biomedical Products) for 24 h after transfection, followedby incubation with growth factors for 4 or 24 h. Luciferase activity wasmeasured by using the Luciferase assay system (Promega) and a scintil-lation counter (Beckman LS6000SC) which had a single photon monitor.To normalize for differences in transfection efficiency, 13-galactosidaseactivity was measured using a chemiluminescent assay using Galacto-Light (TROPIX, Inc.).

Preparation of Nuclear Extracts. Nuclear extracts were prepared asdescribed by Abmayr and Workman (52). Confluent monolayers of cellswere scraped from tissue culture dishes using a cell scraper in Tris-buffered saline (0.125 M NaCI, 25 mp,i Tris, pH 7.4) and collected bycentrifugation at 1 500 x g for 5 mm. The cell pellets were resuspended inhypotonic buffer [10 mM HEPES (pH 7.9), 1 .5 m� MgCI2, 10 m� KCI, 0.2

mM PMSF, and 0.5 mM Dli] and homogenized with 10 strokes of a glassDounce homogenizer nuclei were isolated by centrifugation at 3300 x gfor 1 5 mm. Nuclear proteins were extracted by suspending the nuclei in an

equal volume of extraction buffer [20 mt�i HEPES (pH 7.9), 25% glycerol,1 .5 mM MgCI2, 0.8 M KCI, 0.2 mM EDTA, 0.2 mp�i PMSF, and 0.5 m� Dli].

The extract was dialyzed against 20 mt�i HEPES (pH 7.9), 20% glycerol,100 m� KCI, 0.2 mM EDTA, 0.2 mi,i PMSF, and 0.5 m� Dli before use. Theprotein concentration was determined using the BCA protein assay regent

(Pierce).Gel Mobllfty Shift Assays. Fragments of the CTGF promoter were

prepared by PCR or restriction endonuclease digestion of the promoter

fragment. Double-stranded oligonucleotides were prepared by annealingcomplementary single-stranded oligonucleotides. All oligonucleotidesand fragments were checked by electrophoresis in agarose or polyacryl-

amide gels prior to use. Aadiolabeled fragments of the CTGF promoterwere prepared by end-labeling with Klenow enzyme (Boehringer Man-heim) and polynucleotide kinase (Boehringer Mannheim). Labeled frag-

Cell Growth & Differentiation 479

ments were purified by electrophoresis in 2% agarose gels or 20% poly-

acrylamide gel before use in gel mobility shift assay.The binding reaction mixture contained 1 �g of nuclear extract protein

in 20 �iJ of 10 mM HEPES (pH 7.9), 5 mr�i Tris, 50 mr,i KCI, 0.1 mr�i EDTA,1 �L9 poly(dI-dC).poly(dl-dC) (Pharmacia), 10% glycerol, 300 �sg/mI BSA,

and 10,000 cpm 32P-labeled DNA probe. In competition assays, theindicated amount of unlabeled competitor DNA was added and incubated

at 4#{176}Cfor 2 h prior to adding the labeled probe. The labeled probe was

incubated for one h at 4#{176}Cin the reaction mixture. Electrophoresis was

performed using 5% polyacrylamide gel with 50 mr�i Tris, 0.38 M glycine,and 2 mr�i EDTA.

Methylatlon Interference Assay. End-labeled fragments of double-stranded oligonucleotides were prepared as described forthe gel mobility

shift assay. The oligonucleotides were methylated by dimethyl sulfate(Fisher Scientific)for 5 mm at room temperature. DNA-protein binding andgel mobility shift assays were performed as described above using largeamounts of labeled probe (lOOK cpm) and nuclear protein (20 pg). DNA

from shifted and nonshifted bands was purified and cleaved with piperi-dine (Fisher Scientific), and the samples were electrophoresed on a poly-acrylamide DNA sequencing gel. The sequences of the shifted and non-shifted fragments were compared with the intact probe sequence using

the same methods.

AcknowledgmentsWe thank Drs. Richard Voellmy, Nevis Fregien, and Richard Assoian for

critical and helpful discussions.

References1 . Assoian, R. K., Komoriya, A., Meyers, C. A., Miller, D. M., and Spom,M. B. Transforming growth factor-a in human platelets: identification of amajor storage site, purification, and characterization. J. Biol. Chem., 258:

7155-7160, 1983.

2. Assoian, R. K., Fleurdelys, B. E., Stevenson, H. C., Miller, P. J., Madtes,

D. K., Raines, E. W., Ross, R., and Spom, M. B. Expression and secretionof type 13transforming growth factor by activated human macrophages.

Proc. NatI. Acad. Sci. USA, 84: 6020-6024, 1987.

3. Grotendorst, G. R., Smale, G., and Pencev, D. Production of trans-forming growth factor f3 by human peripheral blood monocytes and neu-trophils. J. Cell. Physiol., 140: 396-402, 1989.

4. Childs, C. B., Proper, J. A., Tucker, R. F., and Moses, H. L Serumcontains a platelet derived transforming growth factor. Proc. NatI. Acad.

Sd. USA, 79: 5312-5316, 1982.

5. Leof, E. B., Proper, J. A., Gouston, A. S., Shipley, G. D., DiCorleto,P. E., and Moses, H. L Introduction of c-sis mRNA and activity similar to

platelet-derived growth factor by transforming growth factor-a: a pro-

posed model for indirect mitogenesis involving autocnne activity. Proc.NatI. Aced. Sd. USA, 83: 2453-2457, 1986.

6. Yaar, M., Peacocke, M., Cohen, M. S., and Gilchrest, B. A. Dissociation

of proto-oncogene induction from growth response in normal humanfibroblasts. J. Cell. Physiol., 145: 39-45, 1990.

7. Tucker, R. F., Shipley, G. D., Moses, H. L, and Holley, R. W. Growthinhibitorfrom BSC-1 cells is closely related to platelet type (3 transforminggrowth factor. Science (Washington DC), 226: 705-707, 1984.

8. Heimark, R. L, Twardzik, D. R., and Schwartz, S. M. Inhibition ofendothelial regeneration by type-a transforming growth factor from plate-lets. Science (Washington DC), 233: 1078-1080, 1986.

9. Takehara, K., LeRoy, E. C., and Grotendorst, G. R. TGF-(3 inhibition ofendothelial cell proliferation: alteration of EGF binding and EGF-inducedgrowth-regulatory (competence) gene expression. Cell, 49: 415-422,1987.

10. Coffey, R. J., Jr., Bascom, C. C., Sipes, N. J., Graves-Deal, R.,Weissman, B. E., and Moses, H. L Selective inhibition of growth-related

gene expression in murine keratinocytes by transforming growth factor (3.Mol. Cell. BioI., 8: 3088-3093, 1988.

11 . Battegay, E. J., Raines, E. W., Seifert, R. A., Bowen-Pope, D. F., and

Ross, R. TGF-j3 induces bimodal proliferation of connective tissue cells via

complex control of an autocrine PDGF loop. Cell, 63: 515-524, 1990.

12. Majack, R. A., Majesky, M. W., and Goodman, L V. Role of PDGF-Achan expression in the control of vascular smooth muscle cell growth byTGF-f3. J. Cell Biol., 1 1 1: 239-247, 1990.

13. Soma, V., and Grotendorst, G. R. TGFI3 stimulates primary human

skin fibroblasts DNA synthesis via an autocrine production of PDGF-related peptides. J. Cell. Physiol., 140: 246-253, 1989.

14. Igarashi, A., Okochi, H., Bradham, D. M., and Grotendorst, G. R.

Regulation of connective tissue growth factor gene expression in humanskin fibroblasts and during wound repair. Mol. Biol. Cell., 4: 637-645,

1993.

15. Bradham, D. M., Igarashi, A., Potter, R. L, and Grotendorst, G. R.Connective tissue growth factor a cysteine-rich mitogen secreted byhuman vascular endothelial cells is related to the SAC induced immediateearly gene product CEF-lO. J. Cell. Biol., 114: 1285-1294, 1991.

16. O’Brien, T. P., Yang, G. P., Sanders, L, and Lau, L F. Expression ofcyr6l, a growth factor-inducible immediate-early gene. Mol. Cell. Biol., 10:

3569-3577, 1990.

17. Ryseck, A. P., Macdonald-Bravo, H., Mattei, M. G., and Bravo, A.Structure, mapping and expression of fisp-12, a growth factor-induciblegene encoding a secreted cysteine-rich protein. Cell Growth & Differ., 2:

225-233, 1991.

18. Brunner, A., Chinn, J., Neubauer, M., and Purchio, A. F. Identificationofa genefamily regulated bytransforming growth factor-p. DNA Cell Biol.,10: 293-300, 1991.

19. Simmons, D. L, Levy, D. B., Yannoni, Y., and Erikson, R. L Identifi-

cation of a phorbol ester-repressible v-sm-inducible gene. Proc. NatI.Aced. Sci. USA, 86: 1178-1182, 1989.

20. Joliot, V., Martinerie, C., Dambrine, G., Plassiart, G., Brisac, M.,Crochet, J., and Perbal, B. Proviral rearrangements and overexpression ofa new cellular gene (nov) in myeloblastosis-associated virus type 1 -in-

duced nephroblastomas. Mol. Cell. Biol., 12: 10-21, 1992.

21 . Mason, E. D., Konrad, K. D., Webb, C. D., and Marsh, J. L Dorsalmidline fate in Drosophila embryos requires twisted gastrulation: a gene

encoding a secreted protein related to human connective tissue growth

factor. Genes & Dev., 8: 1489-1501 , 1994.

22. Oikarinen, J., Hatamochi, A., and de Crombrugghe, B. Separate bind-ing sites for nuclear factor 1 and a CCAAT DNA binding factor in themouse a2(1) collagen promoter. J. Biol. Chem., 262: 1 1064-1 1070, 1987.

23. Riccio, A., Pedone, P. V., Lund, L A., Olesen, T., Olsen, H. S., andAndreasen, P. A. Transforming growth factor (31 -responsive element:closely associated binding sites for USF and CCAAT-binding transcriptionfactor-nuclear factor 1 in the type 1 plasminogen activator inhibitor gene.Mol. Cell. Biol., 12: 1846-1855, 1992.

24. Hou, X., Johnson, A. C., and Rower, M. A. Induction of epidermalgrowth factor receptor gene transcription by TGF-(31 : association with

loss of protein binding to a negative regulatory element. Cell Growth &Differ., 5: 801-809, 1994.

25. Greenberg, M. E., and Ziff, E. B. Stimulation of 3T3 cells inducestranscription of the c-fos proto-oncogene. Nature (Lond.), 31 1: 433-438,

1984.

26. Cochran, B. H., Reffel, A. C., and Stiles, C. D. Molecular cloning ofgene sequences regulated by platelet-derived growth factor. Cell, 33:

939-947, 1983.

27. Rollins, B. J., Morrison, E. D., and Stiles, C. D. Cloning and expressionof JE, a gene inducible by platelet-derived growth factor and whoseproduct has cytokine-like properties. Proc. NatI. Aced. Sd. USA, 85:

3738-3742, 1988.

28. Greenberg, M. E., Hermanowski, A. L, and Ziff, E. B. Effect of proteinsynthesis inhibitors on growth factor activation of c-fos, c-myc and actingene transcription. Mol. Cell. Biol., 6: 1050-1057, 1986.

29. Edwards, D. A., and Mahadevan, L C. Protein synthesis inhibitorsdifferentially superinduce c-fos and c-jun by three distinct mechanisms:lack of evidence for labile repressors. EMBO J., 1 1: 2415-2424, 1992.

30. Edwards, D. A., and Heath, J. K. Regulation of transcription by trans-forming growth factor-(3. In: P. Cohen and J. G. Foulkes (eds.), The

Hormonal Control Regulation of Gene Transcription, pp. 333-347.

Amsterdam: Elsevier, 1991.

4&� TGF-f3 Response Element in the CTGF Gene

31 . Mathews, L S., and Vale, W. W. Expression cloning of an activinreceptor, a predicted transmembrane serine kinase. Cell, 65: 973-982,1991.

32. Ohtsuki, M., and Massagu#{233},J. Evidenceforthe involvement of proteinkinase activity in transforming growth factor-j3 signal transduction. Mol.Cell. Biol., 12: 261-265, 1992.

33. Williams, L T. Signal transduction by the platelet-derived growthfactor receptor. Science (Washington DC), 243: 1564-1570, 1989.

34. UlIrich, A., and Schlessinger, J. Signal transduction by receptors withtyrosine kinase activity. Cell, 61: 203-212, 1990.

35. de Groot, A. P., and Kruijer, W. Transcriptional activation by TGF-j31mediated by the DYAD symmetry element (DSE) and the WA responsiveelement (TRE). Biochem. Biophys. Res. Commun., 168: 1074-1081 , 1990.

36. Kim, S-J., Angel, P., Lafyatis, A., Hattori, K., Kim, K. Y., Spom, M. B.,Karin, M., and Roberts, A. B. Autoinduction of transforming growth factor

(31 is mediated by the AP-1 complex. Mol. Cell. Biol., 10: 1492-1497,1990.

37. Kerr, L D., Miller, D. B., and Matrisian, L M. TGF(31 inhibition of

transin/stromelysin gene expression is mediated through a Fos bindingsequence. Cell, 61: 267-278, 1990.

38. Pietenpol, J. A., Munger, K., Howley, P. M., Stein, A. W., and Moses,H. L Factor-binding element in the human c-myc promoter involved in

transcriptional regulation by transforming growth factor (31 and bythe retinoblastoma gene product. Proc. NatI. Acad. Sci. USA, 88:

10227-10231, 1991.

39. Bork, P. The modular architecture of a new family of growth regulatorsrelated to connective tissue growth factor. FEBS Lett., 327: 125-i 30,

1993.

40. Igarashi, A., Nashiro, K., Kikuchi, K., Sato, S., lhn, H., Grotendorst, G.R., and Takehara, K. Significant correlation between CTGF gene expres-sion and skin sclerosis in tissue sections from patients with systemicsclerosis. J. Invest. Dermatol., 105: 280-284, 1995.

41 . Delarco, J. E., and Todaro, G. J. Growth factors from murine sarcoma

virus transformed cells. Proc. NatI. Aced. Sd. USA, 75: 4001-4005, 1978.

42. Guadagno, T. M., and Assoian, A. K. G1/S control of anchorageindependent growth in the fibroblast cell cycle. J. Cell Biol., 115:

1572-1575, 1991.

43. Ferguson, E. L, and Anderson, K. V. Dorsal-ventral pattern formation

in the Drosophila embryo: the role of zygotically active genes. Curr. Top.

0ev. Biol., 25: 17-43, 1991.

44. St. Johnson, D., and Nusslein-Volhard, C. The origin of pattern andpolarity in the Drosophila embryo. Cell, 68: 201-219, 1992.

45. Ferguson, E. L, and Anderson, K. V. Decapentaplegic acts as a

morphogen to organize dorsal-ventral pattern in the Drosophila embryo.Cell, 71: 451-461, 1992.

46. Arora, K., Levine, M. S., and O’Connor, M. B. The screw gene en-codes a ubiquitously expressed member of the TGF-(3 family required for

specification of dorsal cell fates in the Drosophila embryo. Genes & Dev.,

8: 2588-2601, 1994.

47. Celeste, A. J., lannazzi, J. A., Taylor, A. C., Hewick, A. M., Rosen, V.,Wang, E. A., and Wozney, J. M. Identification of transforming growthfactor (3 family members present in bone-inductive protein purified from

bovine bone. Proc. NatI. Aced. Sd. USA, 87: 9843-9847, 1990.

48. Padgett, A. W., Wozney, J. M., and Gelbart, W. M. Human BMP

sequences can confer normal dorsal ventral patterning in the Drosophilaembryo. Proc. NatI. Acad. Sci. USA, 90: 2905-2909, 1993.

49. Chomczynski, P., and Sacchi, N. Single-step method of ANA isolationby acid guanidmnium thiocyanate-phenol-chloroform extraction. Anal. Bio-chem., 162: 156-159, 1987.

50. Sambrook, J., Fritsch, E. F., and Maniatis, T. Molecular Cloning: ALaboratory Manual, Ed. 2, pp. 14-19. Cold Spring Harbor, NY: Cold

Spring Harbor Laboratory, 1989.

51 . Sanger, F., Nicklen, S., and Coulsen, A. A. DNA sequencing withchain-terminating inhibitors. Proc. NatI. Acad. Scm. USA, 74: 5463-5467,

1977.

52. Abmayr, S. M., and Workman, J. L. Preparation of nuclear andcytoplasmic extracts from mammalian cells. In: F. M. Ausubel, A. Brent,A. E. Kingston, D. D. Moore, J. G. Seidman, J. A. Smith and K. Struhl

(eds.), Current Protocols in Molecular Biology, Vol. 2, Suppl. 1 1 , pp.

12.1 .1-12.1 .9. New York: Greene Publishing and Wiley-lnterscience,1990.