Stimulation of Fibroblast Cell Growth, Matrix Production, and Granulation Tissue Fortnation by Connective Tissue Growth Factor

Ken Frazier, Shawn Williams, Devashish Kothapalli, Helene Klapper, and Gary R. Grotendorst
Department of Cell Biology and Anatomy, University of Miami School of Medicine, Miami, Florida, U.S.A.

Connective tissue growth factor (CTGF) is a 36- to 38-kDa peptide that is selectively induced by transfornUng growth factor-fJ (TGF-fJ)in fibroblastic cell types. We compared the biologic activities of CTGF with TGF-J3 on fibroblasts in culture and in animal models of fibroplasia. CTGF was active as a mitogen in monolayer cultures of normal rat kidney fibroblasts. CTGF did not stimulate anchorage-independent growth ofNRI{ fibroblasts, however, or inhibit the growth of mink lung epithelial cells, distinguishing CTGF's growth-regulatory activities from those ofTGF-J3. In NRK fibroblasts, both TGF-J3 and CTGF significantly increased the transcripts encoding 0:1 type I collagen, as integrin, and fibronectin. Stimulation of type I collagen and fibronectin protein synthesis by TGF-J3 and CTGF was confirmed by

pulse labeling of cells with [35S]methionine. Subcutaneous injection of TGF-J3and CTGF into neonatal NIH Swiss resulted in a large stimulation of granulation tissue and fibrosis at the site of injection. Itt siuc hybridization studies revealed that TGF-J3 injection induced high levels of CTGF mRNA in the dermal fibroblasts at the injection site, demonstrating that TGF-{3 can induce the expression of CTGFin connective tissue cells ill vivo. No CTGF transcripts were detected in the epidermal cells in either control or TGF-{3-injected skin or in fibroblasts in control (saline-injected) skin. These results demonstrate that, like TGF-I3, CTGF can induce connective tissue cell proliferation arrd extracellular rrratzix synthesis. Key words: TGF-J3lcollagenlwomld repairlfihrosis. J Invest Dermatol. 107:404-411, 1996

issue regeneration and repair proceed in a cascade fashion beginning with a coagulation and inflammatOI), phase, folJowed by granulation tissue formation and finally extracellulnr matrix deposition and termination of the response. Peptide growth factors play a central role in this process. It is likely that factors released by platelets and inflammatory cells serve as initiators of the regeneration/repait response. Similarly, wound repair disorders, as well as organ-specific fibrosis, may be caused by dysfunctional cascades. One of the principal regulatory factors that appears to function as an initiator in these processes is rransforming growth factor-{3 (TGF-f3) (Arnento and Beck, 1991; Rnghow, 1991; Wahl, 1992; Roberts and Sporn, 1993). TGF-,B has been shown to act as a potent stimulatory signal for connective tissue formation during wound repair and in fibrotic conditions. Elevated TGF-f3 mRNA or protein levels have been documented in tissue during normal wound repair (Igarashi et al, 1993; Levine et al, 1993), and fibrotic disorders of the skin (Kulozik et ill, 1990; Peltonen et ai, 1990; Smith and Lek.oy, 1990) and internal organs and tissues (Nagy et ill, 1991; Kagami et al, 1993; Bahadori et ill, 1995). The increased fibrotic tissue has been attributed to several actions of TGF-,B, including: (i) increased Manuscript received February 23, 1996; revised April 29, 1996; accepted for publication May 21, 1996. . Reprint requests to: Dr. Gary R.. Grotendorst, Department of Cell Biology and Anatomy, University of Miami School of Medicine (R-124), 1600 NW 10th Avenue, Miami, FL 33136. Abbreviations: CTGF, connective tissue growth factor; rCTGF, recombinant connective tissue growth factor.


fibroblast proliferation (Del.arco and Todaro, 1978; e( ill, 1984; Leof el ill, 1986; Soma and Grorcndorst, 1989; Ishikawa dill, 1990; Battegay ct ill, 1990); (ii) elevated synthesis of extracellular matrix components including fibronectin, type I collagen, inregrins, laminin, and glycosarninoglycans production (Ignotz and Massague, :1986; Roberts ct al, 1986; Raghow dill, 1987; Varga et. III, 1987; Penttinen ct ai, 1988), and (iii) decreased degradation of extracellular matrix due to direct inhibition of protease activity and stimulation of the synthesis of protease inhibitors (Laiho et al, 1986; Lnnd et al, 1987; Kerr ct ill, 1990). Previous studies have demonstnated that '.1 large portion of the TG F-{3 induction of matrix protein synthesis is not shared by other growth factors such as fibroblast growth factor (FGF) or platelet derived growth factors (PDGF) (Ignotz and Massague, 1986; Roberts e/ ill, 1986; Penttinen et ai, 1988). Connective tissue growth factor (CTGF) is a cysteine-rich mitogenic peptide that was originally identified as a growth factor secreted by vascular endothelial cells in culture (Bradham et ill, 1991). CTGF is selectively induced in fibroblasts after activation with TGF-{3 (Soma and Grotendorst, 1989; Igarashi e( ill, 1993). CTGF is a member of a family of peptides that include seruminduced gene products ceflO (Simmons et ai, 1989), cyro l (O'Brien et al, 1990), flsp12/ {3IG Ml (Brunner et ai, 1991; Ryscck et Ill, 1991) and 11 chicken transforming gene, nov (joliet et al, 1992). CTGF also shares significant sequence homology with a Drosopliiln gene product, twisted gastrulation (twg) (Mason et ill, 1994), that determines cell fates during dorsal/ventral pattern formation in the embryo. Previous studies have demonstrated coordinate expression of TGF-/31 and CTGF in granulation tissue beds during wound


• Copyright

© 1996 by The Society for Investigative





107, NO.3




repair (Igarashi et til, 1993) and found that dermal fibroblasts in sc1er:oderma lesions overexpress CTGF (Igarashi et ai, 1995). The CTGF mRNA is selectively induced in fibroblasts by TGF:"'{3, but not by other growth factors, such as epidermal growth factor (EGF), PDGF, or FGFs (acidic FGF, basic FGF) (Igarashi et al, 1993). Protein synthesis is not required for TGF-J! stimulation of CTGF gene expression (Igarashi et al, 1993; Grotcridorst et al , 1996), indicating that the CTGF gene is directly regulated by TGF-/3. We have now identified a novel TGF-{:l response element (T/31lE) in the CTGI' promoter (Grotendorst et al, 1996), which begins to define the molecular basis for the selective regulation of tins gene by TGF-{3.The Tf3RE sequence present in the human and mouse CTGF genes is not present in the promoters of other TGF-/3-regulafed gene that has been described to date. Because of the pleiotropic actions of TGF-f3 on cells and the selective iudu.ction ofCTGF by TGF-f3 in fibroblasts, we wanted to compare the biologic activities of CTGF with TGF-f3. Until now, the limited availability of natural CTGF made it difficult to perform studies on the biologic actions of the molecule. We have produced recombinant CTGF to address this question. Om studies indicate that recomb-inant CTGF is mitogenic for Nl~K fibroblasts in monolayer culture. TGF-f3 in the presence of EGF and serum is capable of stimulating anchorage-independentgl'owt.h of NIU( fibroblasts (DeLarco and Todaro, 1978; Assoian ct al, 1994), but CTGF does not substitute for TGF-{3 in this assay. Also, the growth-inhibitory actions ofTGF-f3 on epithelial cells (Tucker et al ; 1984; Ogawa and Seyedin, 1991) are not shared by CTGF. CTGF induces elevated expression and synthesis of both collagen and fibronectill in fibroblasts in culture and induces fibroplasia in the skin of newborn mice. These results demonstrate that CTGF stimulates fibroblast proliferation and extracellular matrix synthesis in a manner similar to TGF-{3. CTGF does not share TGF-f3's ability to stimulate anchorage-independent growth of normal fibroblasts or act as a growth inhibitor for epithelial cells. MATERIALS AND METHODS

49F fibroblasts as target cells as described previously (Soma and Grotendorst, 1989). Anchorage-independent gr-owth assays were performed essen-

tially as described' by Guadagno and Assoian (1991). Growth Inhibition Assays The mink lung epithelial cell line, MvlLu (ATTC No. CCL 64), was used as targer for growrh-inhibitory assays with TGF-/3 and CTGF. Assays were performed as described by Ogawa and Scycdin (1991). Extracellular








'were grown

to conAucncc


with 50/0 feral bovine


and chenserum starved in DM.EM with 1% bovine serum albumin for 24 h. Growth factors were added to the cell cultures; total cellular RNA was extracted after 24 .h, and northern blot analysis was performed as described previously (Igarashi et al, 1993). To ensure that equivalent amounts of total RNA were added to each lane on a gel, RNA was quantitated by .A260/2S0 ratios. and equivalent transfer was assured by comparing ribosomal 28S and 18S RNA bands in each lane after staining with ethidiurn bromide. As an additional control, blots were reprobed with an actin cDNA probe. Double-stranded cDNA fragments used for probes were labeled with 2 p]dCTP using a random prime labeling kit (Boehringer Mannhcim, Indianapolis, IN). The CTG!' probe was derived from a ·.1.1-kb human eDNA fragment that encompassed the open reading frame of the CTGF transcript. The TGF-{31 probe was a Lfl-kb IVm:! fragment derived 6'0111 <J 2.0-kb human TGF-J31 cDNA present (G.L Bell, H.H. Medical Insritutc, University of Chicago). The cd.-type 1 human collagen probe was derived from a J.5-kh open reading frame fragment at the 3' end (AT'CC No. 61323). The t1'5 intcgrin probe was produced from a cDNA insert containing a portion of the human eDNA containing the open reading frame (R. Assoian, University of Miami). The human fibronectin probe was a O.9-kb EwIU/Hi"dlII fragment derived 6'0111 a 2.2-kb eDNA clone containing the 3' region of the open reading frame provided by F. Woessner (University of Miami). The human actin (control RNA) probe was purchased from Oncor Co. (Gaithersburg, MD).


Cell Cultures Sf9 cells were grown in spinner and monolayer culture and maintained in TNM-FH media JGracc's Insect Culture Media, (GIBCO, Gaithersburg, .MD) with 3.33 g lactalbumin hydrolysate per 1111, 3.33 g yeastolatc per rnl, S'v" fetal bovine serum, 5 'v" Nuscrum (Collaborative Biomedical, Bedford, MA), 10 JLg gentamicin per rnl, and O. '1.% Pluronic F68 (spinner cultures only)]. High Five cells (Invirrogcu, San Diego, CAl were grown in shaker and monolayer culture and maintained in ExCel1405 serum-free insect cell media (JRH Bioscience, Lenexa, KS). NR:I;C49F and Mv1Lu cells were obtained from American Type Culture Collection (Rockville, MD) and maintained in Dulbeccos modified Eagle's medium (DMEM) with 5'){,fetal bovine serum and 10 JLg gentamicin per 1111. Sources of Growrh Factors TGF-J3 isolated from platelets was a gift from Richard Assoian (University of Miami). Recombinant platelet-derived growth factor EE was obtained fi:0111 Chiron (Emeryville, CAl. Purified murine EGF was purchased from Biomedical Technologies, Inc. (Stoughton, Ma). Samples were assayed for endotoxin using an endotoxin enzymeIinkcd immunosorbent assay kit (Sigma, St. Louis, MO). Recombinant CTGF was produced in the laboratory using a bncculoviral expression system employing the plslucbac II transfer plasmid (Invitrogen) containing the CTGF open rending frame, cotrausfcctcd with linearized wild-type AcMNPV into Sf') cells according to prescribed techniques (Ausubel "I nl, 1990). For large scale production of recombinant CTGF (rCTGF), High Five cells were grown in shaker cultures to a density 0(3.5 X :J 0" ce:llsper ml and infected with rCTGF baculovirus at an MOl = 7. The recombinant CTGF was purified using heparin Sepharose affinity chromatography. Peak fracrions containing rCTGF were determined by immunoblotting and COO.l11assie staining of sodium dodecyl sulfate (SDS)-polyacrylamidc gel electrophoresis gels. Gel Electrophoresis and Imrrruuoblorrtng Electrophoresis was performed using 12'X, polyacrylamide gels containing SDS as, described by Laernmli ~1970). Immunobloccing was performed by electroblotting the proteins in the acrylarnide gel to a nitrocellulose membrane as described previously (Soma and Grotendorst, 19-89). Mitogenic and Anchorage-Independent Growth Assays Mitogenic assays were performed in monolayer cultures using 48-well plates and NRK

Stfmularlon of Matrix Protein Synthesis Cell cultures were grown to . confluence, then serum-starved in DMEM and insulin, transferrin, and selenium (ITS, CollaborutivcP ... csearch, Bedford, MA) for 18 h. Cultures were then placed in DMEM containing 100 JLgbovine serum albumin pcr 1111 and either 10 ug TGF-{3 per ml, 20 ng CTGF per rnl, or nothing (negative control) and incubated for 22 h. After this time, cultures were 35 pulse labeled 'with 50 {LCi [ S]med1ionille per 1111 in methionine-free DMEM for 2 h in the presence of the indicated growth factor. Cell proteins were extracted from the cell pellet with 0.1 % Triton Xl 00, 0.5 111M phenylmethylsulfonyl fluoride, and leupeptin for 10 min. Fibronecrin was immunoprecipitared using Protein A-SephaJ"ose using prescribed methods (Roberts et al, 1988). Type I collagen was analyzed by overnight pepsin digestion (5 JLgper ml in 'I M acetic acid) in the presence oflO JLgof carrier bovine type I co llagen. Samples used for the. .i1l1.11TUnoprec::ipitation and pepsin digestion were adjusted so that equivalent amounts of total incorporated C5S]meth,ionine were used in each sample. SDS-polyacrylamide gel electrophoresis was performed, and samples were autoradiographed. Subcutaneous Administration N1J-l Swiss mice were injected in JLIof total volume of either saline (II = 15); 200 or 800 ng ofPDGF 10, 20, 50, 100,200,400, or 800 on a previous protocol (Roberts
after the last injection, and

of Growth Factors to Mice Neonatal the nape of the neck daily for 3 d with 20 control (n = 12); 400 or 800 ng ofTGF-/3 BB (n = 12),800 ng ofEGF (n = 8) or ng of recombiuanr CTGF (n = 38), based ct nt, 1.986). Animals were sacrificed 24 h
biopsies of the area containing and



the injection site were removed. Sections were prepared for histopathologic examination after routine formalin.fixation, processing, and staining with Mayer's hcmatoxvlin and eosin.
III Siru

Hybridization NIH Swiss neonatal mice were injected as described aboveexcept that.samples of dennis and subcutis were taken 6:01l1 the site of injection 8 h after the second injection. These tissue samples were immediately placed in 4.0% paraforrnaldehyde for 1.5 h and then flashfrozen and embedded. Sections were cur at 5 JLI11 and placed on TESPAcoated slides (Oucor, Gaithersburg. MD.) [II situ hybridization for CTGF mRNA was performed using previously described methods (Pava et al, 1990). The slides were dipped in photographic emulsion (Ilford K-5. Polysciencc) and incubated for 8 d at 4°C. Slides were then developed and sections counter stained in Mayer's hematoxylin and eosin. RESULTS Production of Recomb ill ant CTGF Recombinant CTGF was produced using a baculoviral vector system that contained the open reading frame of the human CTGI' transcript under the regulation



ET ilL








LL en o

0 l-



o W U. o u, > o .... rJ) ::::> C
U .... I I

oo l-

C) I-




CTGF, based on amino acid cornposmon and peptide sequence analysis. The doublet pattern observed in the recombinant material is similar to that of the natural product isolated from human skin fibroblast or endothelial cell-conditioned medium (Fig 1). The CTGF is greater than 95'Yc, pure based on quantitative scans of the Coornrnassic-stained gels. The basis for the multiple bands appears to be due to differences in glycosylation as detected by differences in reactivity with concanavalin A (Grotendorst GR, WilHams 5, Segarurn P, unpublished observations). Mitogenic Activity of CTGF We initially tested the biologic activity of the CTGF with NRK fibroblasts using a mitogenic assay performed on monolayer cultures. CTGF stimulated a concentration-dependent increase in DNA synthesis in the monolayer cultures, with a maximal stimulation of 6-fold over non-treated cultures at concentrations of 20-50 ng rCTGF per 1111(Fig 2A). This activity was enhanced by the presence of small amounts of EGF (Fig 2A), although less than that observed when EGF was added with TGF-f3 (Fig 2C). Because CTGF exhibited a high affinity for heparin, and because previous studies have demonstrated that heparin-binding growth factors such as acidic FGF can exhibit much higher mitogenic activity in tbe presence of heparin, we examined the effects of heparin 011 CTGF activity. Addition of 10 /.Lg heparin per ml to the medium resulted in a significant increase in the amount of mitogenic activity exhibited by CTGF (3-fold over non=beparin-treated) with no significant change in the concentrations of CTGF required to give half-maximal or maximal stimulation of DNA synthesis (Fig 2B). Lastly, we compared the activity of TGF-/3 alone or in combination with either EGF or CTGF in the NIUZ monolayer mitogenic assay (Fig 2C). TGF-/3 alone exhibited little activity in onr DNA synthesis assay. A large synergistic stimulation occurred with low concentrations ofEGF, as has been, previously reported (Delarco and Todaro, 1978; Assoian et at, 1984). The presence ofCTGF stimulated only a slight increase in the ability of TGF-f3 to induce DNA synthesis in these cultures (the arroul in the figure indicates baseline of CTGF activity with no TGF-/3 present). These data suggest that at least some part of the synergistic action ofTGF-f3 and EGF is shared by CTGF and EGF, but that there is no synergistic activity between TGF-f3 and CTGF in this 'Issay. Inability of CTGF to Stimulate Anchorage-Independent Growth or Act as a Growth Inhibitor for Epithelial Cells TGF-f3 was originally identified due to its ability to stimulate the anchorage-independent growth of normal fibroblasts (Delarco and Todaro, 1978). While other growth factors are required for cell division under these conditions (Assoian e/ al, 1984), 110 other growth factor has been found to possess this unique activity. TGF-/3 has also been shown to act as a specific inhibitor of epithelial cells


302114Coornassie R-250 Imrnunoblot anti-PDGF

Figure 1. Gel electrophoretic analysis of CTGF. rCTGF produced in High Five cells by a bacculovirus expression system was purified by affinity chromatography on heparin Sepharose. The purity of the rCTGF is demonstrated in lane 2 of the fiji pond where 5 /Lgof rCTGF are compared to Bio-Rad protein standards (5 /Lg/ea). Comparison ofrCTGF with CTGF secreted by human skin fibroblasts (HSF) after activation with TGF-/3 and human umbilicalvein endothelial cells (HUVE) and rPDGF BB (right panel], All samples represent approximately 10 ng of CTGF or PDGF. Peptides were detected after transfer to nitrocellulose by an anti-human PDGF IgG and an alkaline phosphatase-conjugated second antibody as described in
Materials alld Methods.

of the baculoviral polyhedron gene promoter. CTGF was purified by a one-step isolation procedure using hcparin-Sepharose affinity chromatography as described in Materials and Methods. The purity of the rCTGF was determined by SDS-PAGE and staining with Coornassie Blue K-250. As shown in Fig 1, two peptides are detected that co-migrate with bands detected in a parallel immunoblot performed on the same material. Both of these peptides are


100 '7 0

-+ Heparin( 10 ug/ml)






-a- EGF 0.2 ng/ml


Q. >, .<: l-





>, .<: IM




No Additions
75 100 125


>, .<: I-

>< E 0.. o


"4 10
01'" 0



CTGF 20 ng/ml



No Additions










[CTGF] (ng/mll

[CTGFJ (ng/mll

[TGF-BJ (ng/ml)

Figure 2. Mitogenic activity of CTGF. (A) Mitogenic activity of CTGF alone and in the presence of EGF for NRK fibroblasts. Cells were assayed as described in Materials and Methods. The indicated amounts of CTGF were added to the media alone (0) or in the presence of EGF (0.2 ng per ml, .). (B) Effectof heparin on the mitogenic activity of CTGF on NRK cells. Cells Were assayed as in A. The indicated amounts of CTGF were added alone (0) or in the presence of heparin (10 peg per 1111, .) Synergism ofTGF-B1 and EGF but not CTGF to stimulate DNA synthesisof NRK cells in monolayer cultures, The indicated amounts of pure TGF-/3were added alone (0), ill the presence ofCTGF (20 ng per ml, .) or EGF (0.2 ng per ml, 0). A'TOIII points to baseline of CTGF stimulation.. Error bars, mean:': SD.














"' «

0,4 0.3 0.2 0.1 0,0 2.5 5.0


'" -S


.I: C.

Figure 3. Effect of TGF-f3 and CTGF on anchorage-independent growth of NRK. fibroblasts and inhibition of growth of MvlLu lung epithelial cells. (A) Anchorage-independent growth assays were performed essentially as described by Guadagna anti Assoian (1991). TGF-{3 induced anchorage-independent growth in a conccntration-

'" .s:: a.



[Growth Factor] (ng/rnl)



[Growth Factor] (ng/l11l)

dependent manner (0) whereas CTGF did not (e). (B) Growth inhibition assays were performed as described by Ogawa and Seyedin (J 991). TGF-{3 caused a concentration-dependent inhibition of the growth of the Mv l Lu cells (0), whereas CTGF had 110 effect on the growth of the lung epithelial cell line (el. En'or bars, mean ± SD.

(Tucker et al, 1984), and vascular endothelial cells (Heimark et al, 1986; Muller et al, 1987; Takehara et al, 1987). We wanted to com.pare the activity of CTGF with that of TGF-/3, both in an ancl~orage-il1dependent growth assay with NRK fibroblasts and a growth inhibition assay with Mv l Lu epithelial cells. As expected, TGF-/3 stimulated the growth of NRK fibroblasts in a concentration-dependent manner (Fig 3A). CTGF had 110 activity in this assay. Furthermore, addition of CTGF with TGF-/3 did not augrnent the activity found with TGF-{3 alone. A similar outcome occurred in the growth-inhibitory assay using MvlLu cells as targets (Fig 3B). TGF-{3 was a potent and effective inhibitor of proliferation of these lung epithelial cells in monolayer culture, whereas CTGF had no effect either as a growth inhibitor or growth stitnulator. These data. demonstrate that CTGF cannot substitute for TGF-/3 in either of these assays and indicate that TGF-/3 exhibits biologic activities that are distinctfrom those of CTGF. Elevation of Matrix Protein Transcripts and Matrix Protein Synthesis by rCTGF TGF-/3 stimulates the expression of collagen, fibronectin, and integrin genes in fibroblasts. We wanted to com.pare the ability ofCTGF with TGF-/3 to elevate mRNA coding for cd type I collagenvfibronectin, or as integrin in the monolayer

cultures of NRK. cells. Treatment of cells with TGF-{3 and CTGF induced a marked upregulation of cd -type I collagen, fibronectin, and as integrin transcripts as detected by northern blot analysis (Fig 4). TGF-{3 also stimulated a large increase in the level of CTGF transcripts consistent with our previous observations. No detectable changes in CTGF message were observed in the CTGFtreated cultures. Control experiments examining ribosomal RNA by ethidium bromide staining indicated that equivalent amounts of total RNA were present in each of the samples. Quantitative scans of the blots demonstrate a variation of ± 15% in the actin signal, which served as a second control for RNA loading. The matrix protein-encoding genes (0'1 type I collagen and fibronectin) exhibit 4- to 12-fold levels of induction (400 to 1200%), and the CTGF transcript was induced 20-fold (2000%) after TGF-{3 treatment compared to control cultures. The transcript for as integrin was induced 3-fold by TGF-/3 and nearly IS-fold by CTGF. The rate of synthesis of collagen and fibronectin was then examined in both TGF-{3- and CTGF-treated monolayers ofNRK fibroblasts. Cells were starved in serum-free DMEM containing ITS for 18 h. Type I collagen and fibronectin were followed by either immunoprecipitation (fibronectin) or pepsin digestion (collagen), followed by SDS-polyacrylamide gel electrophoresis and autoradiography (Fig 5). The results of these studies demonstrate that both type I collagen and fibronectin synthesis were stimulated by TGF-/3 and CTGF compared to control cultures. Quantitative densitometric analysis of the films indicate that, based on the


Experiment # Fibronectin


TYPE I collagen

4. Norchenn blot analysis for extracellular matrix protein in cultured fibroblasts. Confluent cultures of N1U{ fibroblasts were placed in serum-free media for 24 h prior to the addition of either buffer (control lane 1), lOng TGF-{3 per rnl , or 10 ng rCTGF per ml for 24 h. Northern blot analysis was then performed on totall(_NA a;de'scribed .in Mntcrials and 1I11'i.//Ods.

Figure 5. Stimulation of fibroncctin and type I collagen synthesis by TGF-f3 and CTGF. Confluent monolayer cultures of N1U{ fibroblasts were stimulated for 24 h with either TGF-{3 (10 ng per ml) or CTGF (20 ng per 111.1) pulse labeled for 2 It with [,15S]l11crhiolline (50 !LCi per ml) in and methionine-free media. Fibronectin and collagen synthesis was determined as described in Materiats and Methods. Two separate experiments were performed and arc indicated as experiments 1 and 2. Arrows indicate fibroucctin (lap 1'([1Ic1) or collagen a-chains (botton: pancl}.







Figure 6. Skin and subcutis from NIH Swiss neonatal mice after injection of saline control or growth factors. Mice were injected one time daily for 3 d and the tissue was harvested and processed as described in Materials alld Methods. (A) saline; (B) 800 ng rPDGF BB; (e) 800 ng EGF; (D) 800 ng TGF-~; (E) 400 ng rCTGF daily for 3 d. Epidermis (E), dermis (D), hair follicles (F), and granulation tissue (G) are indicated. Scale bar, 200 /-LU!.


107, NO.3







Figure 7. III sit" hybridization for'CTGF tnRNA in TGF~fl-jnjected neonncal muninc skin and subcutis. TGF~Bl (800 ng) was subcutaneously injected into the nape of the neck of neonatal NJH Swissmice daily (sec Materials IIlId Methods). III situ hybridizationusingCTGF riboprobes in both antisense (A) and (B) and sense (C) orientationswas performedunder high stringencyconditionsfollowedby'an 8-d exposure prior to development of the emulsion (see Materials. alld Methods), (A) TGF-f3 il\jectedmouse skin; O,.roll~Spoint to somestr.ongly positivefibroblasts granulatiou tissue. (B) An~senseCTGF probe, 111 saline-injected mouse skin (no.signal IS detected). (C) TGF-f3-ulJectedmouse 5k111, sense-orientedCTGF negative control (no signal IS detected). F, hair follicles; G, granulati.ontissue. Scale bar, 50 Mm.

average of the two experiments, TGI~-,Bstimulated a 3-fold increase in the synthesis of fibronecrin and a 2.S-fold increase in collagen synthesis. CTGF stimulated a S-fold increase in both fibronectin and. collagen. Thus, with respect to the induction of m~ltrix production, CTGF has actions that are similar to TGF-,B and distinct from other growth factors such as PDGl~ or FGF. Fibroplasia Induced by Subcuraneous Injection of Growth Factors The injection of TGF-{3 into the dermal/subcuticular area of skin in neonatal NIH Swiss mice induces a rapid and dramatic increase in the formation of granulation tissue primarily composed of connective tissue cells and large amounts of extracellular matrix (Roberts et al, 1986). We compared the actions of TGF-{3, PDGF BB, EGF, and CTGF in this illl,il!o model of fibrosis and wound rep au' (Fig 6). After two injections, gross nodules were apparent in both the TGf-{3- andCTGF-injected mice and continued to enlarge throughout the remainder of the experimental protocol. Both TGF-{3 and CTGF induced similar alterations of the normal tissue, including large increases in the number of connective tissue cells and extracellular matrix material. Dosage studies indicated that the optimal concentration of CTGF for induction of fibroplasia was 400 ng/site compared with 800 ng/site for TGF-l3, which is consistent with the previously reported dose response for TGF-,B injection Ul mice. The amount of granulation tissue formed at sites injected with 800 ng of CTGF was consistently less than in those with 400, ng/site. Lower concentrations of CTGF induced significant granulation tissue formation with some effect detectable at amounts as low as 50 Ilg per injection. Prolonging the interval from the last injection to sacrifice and histologic examination to 7 d in both theTGF-f3- and C'I'Gl--injecred mice resulted in resorption of granulation tissue. Injection of PDGF BB at 800 ng/site pro-

.duced only a mild granulation tissue response, similar to that produced by CTGF levels of 50 rig/site. Injection of EGF did not result in detectable fibroplasia, and the connective tissue in these tissue samples had an appearance similar to tbat of saline-injected controls. The EGF-injected specimens, however, exhibited hyperkeratosis of the epidermis and degenerative changes in the hair follicular epithelium, demonstrating that the EGF was biologically active and produced effects on the epidermal cells at the sites of injection. The area of granulation tissue was measured in slides from the mice injected with TGF-/3, CTGF, or PDGF. The average area of.fibroplasia in the specimens from the TGF-,B-injected mice was 28.8 n11112, the C'TGf=injected mice it was 22.4 mm ", and in .in the PDGF-injected specimens it, was 3.4 1111112. TGF-f3 Induction of CTGF mRNA in Skin Fibroblasts III. Vivo. To determine whether the CTGF transcript was induced in the fibroblasts participating in the formation of the granulation tissue induced by TGF-{3 injection, we performed ill. situ hybridizati.on for CTGF message on sections of the subcutis obtained fr0111 saline- and TGF-{3-illjected mice (Fig 7). Animals were injected with 800 ng ofTGF-Bl followed by a second injection 24 h later. The tissue fr0111 the injection site was harvested and prepared 8 h after the second injection. Control experiments using saline only were performed i.n parallel. A strong signal was detected in the TGF-,B-injected dermis when hybridization was performed using an antisense CTGF riboprobe (Fig 7A). Hybridization controls of TGF-,B-injected tissue sections with sense strand CTGF riboprobes were negative, with no significant background hybridization (Fig 7C). No CTGF transcripts were present in either noninjected skill (not shown) or saline control-injected skin sites (Fig 7B). The only cells expressing the CTGF transcripts were connective tissue cells








(fibroblasts). No cells in the epidermis were positive for CTGF transcripts in either control or TGF-{3-injected skin. Cells in the deeper regions of the skin such as adipocytes were also negative for CTGF transcripts in both control and TGF-/3-injected skin sites. Surprisingly, although the CTGF gene is constitutively expressed by vascular (human umbilical vein) endothelial cells in culture, we did not detect CTGF transcripts in clearly identifiable endothelial cells in either large vessels or capillaries. These results indicate that the principal source of CTGF ill 1Ji-IJO are connective tissue cells activated with TGF-{3. DISCUSSION The data from our studies indicate that CTGF acts to stimulate cell division and extracellular matrix production by NlU{ fibroblasts in culture. The mitogenic activity of CTGF is augmented by EGP and heparin but not by TGF-/3. With respect to matrix protein synthesis, we find that CTGF stimulates an increase in the expression of o t-type 1 collagen and fibronectin mRNAs, which results in increased synthesis of these proteins. Injection of CTGF into the skin of neonatal mice stimulates the formation of a matrix-rich granulanion tissue. Compared to other known growth factors, CTGF's biologic actions are most similar to those of TGF-{3. Nonetheless, CTGF does not substitute for TGF-{3 in anchorageindependent growth assays, nor does it act to inhibit the growth of epithelial cells whose growth is inhibited by TGF-{3. Thus, CTGF appears to mimic some of the actions of TGF-/3 with regard to the stimulation of connective tissue cell growth and the synthesis of extracellular matrix but is not a substitute for TGF-_B in other assays. Previous studies have indicated that TGF-{3 can stimulate the growth of various fibroblastic cell types by the induction ofPDGF genes (Lcof et ai, 1986; Battegay et al, 1990; Ishikawa et al, 1990). This may be a part of the mechanism for TGF-{3 stimulation of connective tissue cell growth, however, it does not appear that PDGF plays a direct role in the stimulation of extracellular matrix protein production induced by TGF-{3 (Penttinen et al; 1988). Thus, whereas ill IJiIJO studies indicate that PDGF alone can accelerate granulation tissue formation in normal animals (Grotendorst et al, 1985), the type of tissue induced by PDGF injection has much less extracellular matrix than the tissue induced by TGF-/3 or CTGF injection. We feel there is a significant physiologic difference between the injection model [initially described by Roberts et a! (1986) and employed in the studies described here] and wound models. In the injection model, only a minimal amount of trauma occurs at the site, and growth factors must act alone or in concert with factors that are constitutively present in interstitial fluids. In contrast, growth factors added to wound models in animals with normal healing act in an environment containing a milieu of factors whose primary functions are to stimulate tissue regeneration and repair. Thus, agents that may play important roles during wound healing would not necessarily stimulate a large amount of connective tissue formation in the injection model. This may explain the PDGF and EGF results reported here. These constraints make it all the more remarkable that CTGF induces the formation of connective tissue in a manner that closely resembles that induced by TGF-{3. Such observations support the possibility that CTGF may playa role in TGF-{3-mediated formation of granulation tissue. The coordinate expression of TGF-/3 and CTGF during the wound repair cascade (Igarashi et nl, 1993) helps to functionally link the inflammatory phase (where TGF-{3 is expressed) and the granulation tissue formation phase (where CTGF is expressed) of this complex process. During the course of examination of the expression of growth factor genes in fibrotic disorders we have found CTGF expression to be closely linked to the expression of TGF-{3. Por example, CTGP is overexpressed in fibroblasts in the dermis of patients with scleroderma (Igarashi et ai, 1995), a systemic fibrotic disorder characterized by the overproduction of collagen (LeRoy et ai, 1988). Overexpression of TGF-{3 also has been reported in scleroderma lesions (Kulozik et al, 1990; Peltonen et ai, 1990; Smith and LeRoy, 1990), suggesting a causative association

between TGF-{3 and the increased deposition of collagen. One significant problem with linkil,g the presence of TGF-/3 transcripts in a tissue withT'GF-{3 activity is that TGF-{3 is made in a latent form that must be processed to obtain the active protein (Lawrence et al, 1985; Pircher et al, 1986; Odekon et al, 1994). Most of the TGF-{3 in tissue and produced by cells in culture is in this latent form (Flaumenhaft et al ; 1993). Consequently, the presence of transcripts does not always correlate well with the amount of active TGF-/3 present (Flanders et al, 1989). Because the methods employed to effectively extract growth factorsfrom tissue also convert latent TGF-{3 to the active form, it is difficult to determine how much active TGF-/3 is present in a tissue. Our previous studies on fibroblastic cells in culture and the data reported in this manuscript demonstrate that the expression of CTGF is dependent 011 the presence of active TGF-{3. We have not found factors other than TGF-{3 that induce a significaut level of expression of the CTGF gene. Furthermore, other genes, such as PAI-1, whose transcription is stimulated by TGF-{3 are equally induced by other growth factors such as PDGF (Grotendorst, unpublished observations). These studies suggest that the presence ofCTGF transcripts may be useful as a marker for the presence of active TGF-{3 in a tissue. In summary, these data indicate that CTGF acts to stimulate fibroblastic cell proliferation and extracellular matrix synthesis. Because CTGP is selectively induced in connective tissue cells by TGF-{3, it is attractive to speculate that CTGF may function as an autocrine mediator for some of the biologic actions of TGF-{3 on fibroblasts. The restricted activity of CT'GF on fibroblasts is consistent with tills model. It is also consistent with the observations that the expression of the TGF-_B and CTGF genes are coordinately linked during normal wound repair. Furthermore, in a wide variety of fibrotic conditions in which TGF-{3 is found to be expressed, such as scleroderma, fibrotic liver disease, and human atherosclerotic plaques, we find high levels of expression ~!f the CTGF gene (Grotendorst, unpublished observations). These observations support a cascade model for connective tissue formation, shared by normal and pathologic processes. In tills model, initiators acting during the acute response to injury induce factors whose function is to maintain the repair process and allow for the maturation of the forming connective tissue. In this regard, the pleotropic actions ofTGF-{3 make it well suited to act as ,U1 initiator of the repair process. The restricted actions of CTGF suggest this peptide functions as an autocrine and paracrine signaling molecule to maintain and perhaps amplify and synchronize the response of fibroblastic cells in the tissue.

This research 'JltlS sllpported by Notiollal Institutes of Healtl, Cralll CM] 7223, a devtlopmenta! tllllardFolil the Svlvcste: Callcer Cell/,er, alld a gralltfrolll FibroCell, Inc. (to C.R. C.). KF. is slipported Ill' Nntionnl lnstitntes <1Hea/l" Traillillg Grallt IUW7057-0,/ A 1 alit! (III illstitlltional fellowship FOIil tlie Ullil/ersill' Mialll;.




Arnenro EP, Beck LS: TGF-{3 and wound healing. CilM 1'0'"1115)'1111' 57:115-129, '1991 1 Assoian R.K, Grorendorst GR, Miller DV, Sporn M.B: Cellular transformation by coordinated action of three peptide growth factors from hurnnn platelets Nnfllre 309:804 -806,1984 Ausubel PM, Brent R., Kingston I'-E, 'Moore DD, Seidman JG, Smith JA, So-uh! K (eds.) Purification and transfcctirm of bnculoviral DNA tor generating rcccrubi11:U1t viruses. In: I-I Piwuicn-Worms. Current Protocols ill Moicrniar i3io/"SY. Vol. 2. Wiley Inrcrscicncc, New York, pp )6.10.1.-8. 1990 Balmdori L, Milder J, Gold L, Berney M: Active macrophage associated TGF-~ co-localizes with type 1 procollagcu gene expression in atherosclerotic human pulmonary arteries. 11,"./ p"tll,,1 146: 1140-1149, i995 Batteguy Ej, Raines lEW, Seifert itA. Bowen-Pope DP, RQssR: Transforming grQwf.h factor beta induces bimodal proliferation of connective tis.. cells via complex sue control of an autocrinc platelet-derived growth factor loop. Cei! 63:5"15-524, 1990 Bradham OM., lgurshi A, Potter RL. Grotcndorst GR: Connective tissue growth factor: a cysteine-rich mitogen secreted by human vascular endothelial is related to the SRC-induccd immediate early t;Cl1C product CEP-'llJ,'/ Cell Biol l'I4:12851294, 1991












A, Chiuu J. Neubauer M., Purchio AF: ldenuficatiou of a gene- family regulated by transforming growth factor beta. DNA C"II 8io/l O:293~300. 1991 DeLan;oJE. Todaro GJ: Growth factors fro III III 11rill c sarcoma virus-rrnusformcd cells. l'/'()( Not! liwd Sri USA 75:4001-4005. 1978 Fava RA. Casey 1'1'. Wilcox] , Pelton It W. Moses I-IL. Nanney LB: Synthesis of tr~1I1s_formillg: growth factor-beta 1 by mcgukuryocytcs and its localizarcn to megakaryocyte and platelet alpha-granules. lJ/,'od 76: 194()-1955, 1990 FTandersKC. Tho-upson NL, Cisscl OS. Van Obbcrghcu Schilling E, Baker CC. Kass ME. Ellingsworth LR, robert» AD. Sporn MB: Transforming growth factor bern-L: histochemical Iocnlizariou with antibodies to different cpitopcs.J Ceil Bioi 108:653-660, 1989 Flaw'uenhaft R. Kojima S., Abe M, Rifkin DB: Activnrion of latent n-ansforrning gro\tvth factor beta. lid" Phnnnncol 24:51-76. 1993 Grotcndorst GR. Okochi H. Hayashi N: A novel TGF-J3 response element controls the expl-c!'sion of'rhc connective tissue growth factor (CTGI::) gcne. Cell Crowth Offer 7:469-480.1996 Grotendorst GR, Martin GR, Pencev D, Sadek j, Harvey AK: Stimulation of grauulation tissue formation by pluccler-dcrivcd growth factor in normal and diabetic ra".) Clill 11""'.1'1 76:2323-2329. '1985 Guadagno TM. Assoian 1<1(: Gl/S control of anchorage independent growth in the f,brohlast cell cycle.) Cell DivI1'l5:1572-1575. '1991 Hehnark RL. Twardzik Dlx., Schwartz SM: lnhibirion of endothelial regeneration by type-beta transforming growth factor from platelets. Scicllce 233:1078-1080. 1986 Igar~lshi A, Nnsluro K. Kikuchi K, Saro S" 11111 H. Grorcudorst GR, Takeharu K: Signif:ic:lIlt correlation between CTG.F gene expression and skin sclerosis ill tissue sections 6_'0111 patients with systemic sclerosis. J Invest Dcnnntol 105:280-284. '1995 Igarashi A. Okoohi H. Bradham OM. Grotcndorst Gil..: Regulation of connective tissue growth factor gene expression in human skin fibroblasts and during wound repair. Mol Bioi Cd/4:637-645, 1993 Ignotz ItA. Massaguc J: Transforming growth factor-beta stimulates the expression of fibroncctin and collagen aud their incorporation into the extrnccllulnr matrix. ) Bioi Clunn 261 :4337-4345.198(, Ishikawa 0, LeRoy Ee, Trojanowska M: Mitogenic effect of transforming growth factor bctn-t 011 human fibroblasts involves the induction of platelet-derived growth factor alpha-rcccprorsv I Cdl l'h)'siol 145: 181-186. 199(1 Joliot V. Marinerie C, Dnmbrinc G. Plassiart G. Brisac M. Crochet}. Pcrbnl B: Proviral re~lIT~lI1gel11entsand overexprcsston of a new cellular gene (nov) in myeloblastosis-associatcd virus rypc-I induced ncphroblastomns. AI}(11Cell Bioi '12:"10-21, J992 Kag<ll11i S, Border W A. Ruoslnhti E, Noble NA: Coordinated Expression .of beta 'I in.tegrins and trausfonuing growth factor-bora induced matrix proteins in glomerulonephritis. LIII, I""csi ()9:68-76. 1993 Kerr LD. M:iller DB, Mnrrisinn LM: TGF,Bl inhibition of trausin/srromelysin gene expression is mediated through a Pes binding: sequence. Cell 61 :267-278, 1990 Kulozik M. Hogg A, Lankat-Buttgercir B, Krieg T: Co-localizauon ofTGF-J3 and type I procollngcn mRNA in tissue sections of puricnrs with systemic sclerosis.) Ctin lnvest 85:917-922, 1990 Laenll11Ji UK: Cleavage of structural proteins during the assembly of the head of bacteriophage T.1• Nonnv 227:680-685, ·1970 Laiho M, Snkseln 0, Andreasen PA, Kcski-Oja ]: Enhanced pcoductin and extracellular deposition of the endothelial-type plasminogen activator inhibitor in cultured human lung fibroblasts by transforming growth factor {3. J Cutt BioI '103:24032410.1986 Lawrence DA. Pircher I" Jullien P: Conversion of a high molecular weight TGI:;-bcta from chick embryo fibroblasts into a low molecular weight active rgf-beca under acidic conditions. B;I)(;II(;'III Bil)pl!y.~ Res Comlllllli 133:t026-'-' 034. 1985 LeofEB.Proper ,1A. Goustou AS, Shipley GD, Dicorlero PE, Moses I-IL: Induction of c-xis 111_H.NA and activity similar to PDGf by TGF/3: ;1 proposed model fOT" indirect mitogenesis involving autocrine activity. Pvoc Nat! Acnd Sci USA 83: 2453-2457, 1986 LeroyEC. Black C, Plciscluunjer R.Jablollska S. Krieg T. Medsger TAJr.. Rowell N. Wol.lhcim F: Scleroderma (systemic sclerosis): classification, subsets and pathogenesis.) Rhcnnunol 15:202-205, 1988 Levine JI:1, Moses 1-11., Gold Ll, Nanney L13: Spatial and tcmproal patterns of immunoreactive nuustorruing growth factor beta 1.! beta 2, beta 3 during: excixional wound repair. A",) PIIII",1 143:-368-380. 1993 Brunner

Lund LH... Riccio A, Andreasen PA. Nielsen LS, Kristensen P, Laiho M. Sakscln 0, , Blasi F. 0:1110 K: Transforming growth facror-B is a strong and fast acting positive regulator of the level of rypc-t plosminogcu nctivneor-Inlubitor mRNA in "\\/1-38 human lung fibroblasts. EMBO) 6:1281-J 286, 1%7 Mason ED. Konrad KD. Webb CD. Marsh jL: Dorsal midline fate in Drosophila embryos requires twisted ,---~(JstrHlflti(JlI, a gene encoding a secreted protein rclnred to human connective tissue growth Elcror. Grl/{:s D(~l!8:1489-1501, t 994 Mu.ller G! Behrcus ] , Nussbaumer U. Bohlen P, Birchmeier \X/: Inhibitory action of TGF-{3 ()J) endothelial cells. Pro c Nml Aend S ..i USA 84:5600-5604. 1987 Nagy P, Schaff Z, Lapis K: Immunohistochemical detection ofTGF-131 in fibrotic liver disease. Hq)(lIo/"SI' 14:269-273, '1991 O'Brien 1~P. Yang GP. Sanders L. Lay LF: Expression of cyrfil. a growth factorinducible iuuucdiate early gene. Mol Cdl Well 0:3569-3577. 1990 Odekon LE. Blasi F. Rifkin DB: Requirement for receptor-bound urokinase. in plnsrniu-dcpcndent cellular conversion of lnrenc trausfonuing growth factor beta to transforming growth factor beta.) Cell l'ft)'sin/l 58:398-407. 1994 Ogawa Y. Scycdin SM: Purification of transforming growth factors {3l and {32 from bovine bone and cell culture assays. Ln: 0 Barnes, JP Mather, GI-f Snro (cds.) }Vfcflwds (!f Ellz)"lIoloSY, Vol 198. Academic Press. New York, '199', pp 317-327 1'01t0l1el1,1. Knhari Li jaakkola S. Kahar-i V-M. Varga). Uitco Ljirucncz SA: Evaluation ofTGF-J3 and type 1 procollngen gene expression in fibrotic skin diseases by ill siu. hybridization.) 1",'cSI DC,.,IIIIIIII 94:365-371. 1990 Pcutriucn It.P, Kobayashi S, Bornstein P: Trnnsfonning growth factor {3 .iucrenses mRNA for matrix proteins both in the presence and ill the absence of changes in mRNA srabiliry. Proc Nail licod Sd USA 85:'1 105-1108, 1988 Pirchcr R, jullieu P. Lawrence DA: /3-lTansfonnillg growth factor is stored in human platelets us a latent high molecular weight complex. Biochcm Bioph)'s N.. s emili/II/II c 136:30-37,\986 . Raghow R: Role of c-ansforming growth factor /3 in repair and fibrosis. CItes! 99s:61.-65, 199'1 [tagh ow It, Postlethwaite A.E, Keski-Oj" j, Moses l-IL. Kaug AH: Transforming growth factor beta increases steady stare levels of rypc 1 procollogen and fibrouecrin ml"tNAs posttranslationally in cultured fibroblasts. J elill Invest 79:1285-1288, 1987 Roberts CJ. Birkcnmeir TM,McQailhllt JJ. Akiyama SS, Yamada SS. Chell WT. Yamada K.M. McDonald JA: Transforming growth factor beta srimularcs the expression of fibronectiu and of both subunits of the human fibrouccti» receptor by cultured human lung fibroblastsv] Bioi CIt,,", 263:4586-4592.1988 Roberts AB. Sporn M.ll: Physiological actions and clinical applications cf transforruing growth factor bet" (TGF-J3). Croll'llt Fllel"rs 8:1-9, '1993 Roberts AB, Sporn M.B,AssoianRK. SmithJM. Roche NS. W,rkdield LM. Heine UI. Liotta LA. Fnlsnga V, Kelul JH. Fauci AS: Transforming growth facror rypc f3: rapid induction of fibrosis and angiogenesis i/I Pillj) and sriniulatiou of collagen formation in-vitro. Prot: Ntlfl Jlend Sci USA 83:4'167-4"171, 1.986 Ryscck R-P. Macdonald-Bravo I-I. Mattei M-G, Bravo It: Structure, Mapping and expression of Fisp-1.2. a growth factor inducible gene encoding a cysteine-rich protein. Cell Cr""'I1t DiP'er 2:225-233. 199'1 Simmons DL. Levy DB. Yannoni Y, Erikson lt.L: ldenrification of a phorbol ester-repressible v-src-inducible gene. Pro .. Nail Acari Sci USA 86:1178-1182, 1989 Smith EA, LeRoy EC: A possible role for rransfonuiug growth f.1CtOT beta ill systemic sc\erosis.j I""esl DCI'IIwl,,1 95:125s-127s. 1990 Soma Y. Grorcndorsr GR: Transforming growth factor beta stimulates primary hurunn skin fibroblast DNA synthesis via an autocrinc production of peptides.) Cdll'III'sioI140:24(>-253, '1989 Takehnra K, Lcft.oy EC. Grorendorst GR: TGF-/3 inhibition of endothelial cell proliferation: nlrerarion of EGF hinding and EGF-il1duced growth-regulntcry (competence) gene expression. Cell 49:415-422. 1987 Tucker R'P, Shipley GO, Moses HL. Holley R W: Growth inhibitor from BSC-1 cells closely related to platelet type beta transforming growth factor. Sdel1c{1226:705707,1984 Varga ]. Rosenbloom J. Jimenez SA: Transforming growth factor hera C~lUSCS a persistent increase in stcadv-srare amounts for type I and type III collagen and fibroucctiu IlIRNAs_in normal dermal human fibroblasts. Bioc/ICII'j 297:597-604. '1987 Wahl SM: Tmll~formillg growth factor beta in inflammation. a cause and a cure. ) 1"""11",,/12:6'1-74, 1992

This document is a scanned copy of a printed document. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material.

Sign up to vote on this title
UsefulNot useful