Decorin Fragments Inhibiting Cell Regulatory Factors

This invention relates to cell biology and more specifically to the control of cell proliferation by inhibiting cell regulatory factors.

Proteoglycans are proteins that carry one or more glycosaminoglycan chains. The known proteoglycans carry out a wide variety of functions and are found in a variety of cellular locations. Many proteoglycans are components of extracellular matrix, where they participate in the assembly of cells and effect the attachment of cells to the matrix.

Decorin, also known as PG-II or PG-40, is a small proteoglycan produced by fibroblasts. Its core protein has a molecular weight of about 40,000 daltons. The core has been sequenced (Krusius and Ruoslahti, Proc. Natl. Acad. Sci. USA 83:7683 (1986); Day et al. Biochem. J. 248:801 (1987), both of which are incorporated herein by reference) and it is known to carry a single glycosaminoglycan chain of a chondroitin sulfate/dermatan sulfate type (Pearson, et al., J. Biol. Chem. 258:15101 (1983), which is incorporated herein by reference). The only previously known function for decorin is binding to type I and type II collagen and its effect on the fibril formation by these collagens (Vogel, et al., Biochem. J. 223:587 (1984); Schmidt et al., J. Cell Biol. 104:1683, (1987)). Two proteoglycans, biglycan (Fisher et al., J. Biol. Chem. 264:4571 (1989)) and fibromodulin, (Oldberg et al., EMBO J. 8:2601, (1989) have core proteins the amino acid sequences of which are closely related to that of decorin and they, together with decorin, can be considered a protein family. Each of their sequences is characterized by the presence of a leucine-rich repeat of about 24 amino acids. Several other proteins contain similar repeats. Together all of these proteins form a superfamily of proteins (Ruoslahti, Ann. Rev. Cell Biol. 4:229, (1988); McFarland et al., Science 245:494 (1989)).

Transforming growth factor .beta.'s (TGF.beta.) are a family of multi-functional cell regulatory factors produced in various forms by many types of cells (for review see Sporn et al., J. Cell Biol. 105:1039, (1987)). Five different TGF.beta.'s are known, but the functions of only two, TGF.beta.-1 and TGF.beta.-2, have been characterized in any detail. TGF.beta.'s are the subject of U.S. Pat. Nos. 4,863,899; 4,816,561; and 4,742,003 which are incorporated by reference. TGF.beta.-1 and TGF.beta.-2 are publicly available through many commercial sources (e.g. R & D Systems, Inc., Minneapolis, Minn.). These two proteins have similar functions and will be here collectively referred to as TGF.beta.. TGF.beta. binds to cell surface receptors possessed by essentially all types of cells, causing profound changes in them. In some cells, TGF.beta. promotes cell proliferation, in others it suppresses proliferation. A marked effect of TGF.beta. is that it promotes the production of extracellular matrix proteins and their receptors by cells (for review see Keski-Oja et al., J. Cell Biochem 33:95 (1987); Massague, Cell 49:437 (1987); Roberts and Sporn in "Peptides Growth Factors and Their Receptors" (Springer-Verlag, Heidelberg (1989)).

While TGF.beta. has many essential cell regulatory functions, improper TGF.beta. activity can be detrimental to an organism. Since the growth of mesenchyme and proliferation of mesenchymal cells is stimulated by TGF.beta., some tumor cells may use TGF.beta. as an autocrine growth factor. Therefore, if the growth factor activity of TGF.beta. could be prevented, tumor growth could be controlled. In other cases the inhibition of cell proliferation by TGF.beta. may be detrimental, in that it may prevent healing of injured tissues. The stimulation of extracellular matrix production by TGF.beta. is important in situations such as wound healing. However, in some cases the body takes this response too far and an excessive accumulation of extracellular matrix ensues. An example of excessive accumulation of extracellular matrix is glomerulonephritis, a disease with a detrimental involvement of TGF.beta..

Thus, a need exists to develop compounds that can modulate the effects of cell regulatory factors such as TGF.beta.. The present invention satisfies this need and provides related advantages.

The present invention provides active fragments of proteins having a cell regulatory factor binding domain. The invention further provides a method of inhibiting an activity of a cell regulatory factor comprising contacting the cell regulatory factor with a purified polypeptide, wherein the polypeptide comprises a cell regulatory factor binding domain of a protein and wherein the protein is characterized by a leucine-rich repeat of about 24 amino acids. In a specific embodiment, the present invention relates to the ability of decorin, a 40,000 dalton protein that usually carries a glycosaminoglycan chain, and more specifically to active fragments of decorin or a functional equivalent of decorin to bind TGF.beta. or other cell regulatory factors. The invention also provides a novel cell regulatory factor designated Morphology Restoring Factor, (MRF). Also provided are methods of identifying, detecting and purifying cell regulatory factors and proteins that bind and affect the activity of cell regulatory factors.

FIGS. 1A and 1B show expression of decorin cDNA containing a mutation of the serine acceptor site to alanine. COS-1 cultures were transfected with cDNA coding for wild-type decorin (lane 1), decorin in which the serine-4 residue was replaced by an alanine (lane 2), or decorin in which the serine-4 residue was replaced by a threonine (lane 3). Immunoprecipitations were performed with an anti-decorin antibody and medium which was labeled with .sup.35 S-sulfate (A) or -.sup.3 H-leucine (B). Lane 4 shows an immunoprecipitate from mock transfected COS-1 cultures. Arrow indicates top of gel. The numbers indicate M.sub.r .times.10.sup.-3 for molecular weight standards.

FIGS. 2A and 2B show binding of ›.sup.125 I!TGF.beta.1 to decorin-Sepharose: (FIG. 2A) Fractionation of ›.sup.125 I!-TGF.beta.1 to decorin-Sepharose affinity chromatography. ›.sup.125 I!TGF.beta.1 (5.times.10 .sup.5 cpm) was incubated in BSA-coated polypropylene tubes with 0.2 ml of packed decorin-Sepharose (.largecircle.) or gelatin-Sepharose (.circle-solid.) in 2 ml of PBS pH 7.4, containing 1M NaCl and 0.05% Tween 20. After overnight incubation, the affinity matrices were transferred into BSA-coated disposable columns (Bio Rad) and washed with the binding buffer. Elution was effected first with 3M NaCl in the binding buffer and then with 8M urea in the same buffer: (FIG. 2B) Analysis of eluents of decorin-Sepharose affinity chromatography by SDS-polyacrylamide gel under nonreducing conditions. Lane 1: the original ›.sup.125 I!-labeled TGF.beta.1 sample; lanes 2-7: flow through and wash fractions; lanes 8-10: 3M NaCl fractions; lanes 11-14: 8M urea fractions. Arrows indicate the top and bottom of the 12% separating gel.

FIGS. 3A and 3B show the inhibition of binding of ›.sup.125 I!TGF.beta.1 to decorin by proteoglycans and their core proteins: (FIG. 3A) Competition of ›.sup.125 I!TGF.beta.1 binding to decorin-coated microtiter wells by recombinant decorin (.circle-solid.), decorin isolated from bovine skin (PGII) (.box-solid.), biglycan isolated from bovine articular cartilage (PGI) (.rect-ver-solid.), chicken cartilage proteoglycan (.largecircle.), and BSA (.quadrature.). Each point represents the mean of duplicate determinants. (FIG 3B) Competition of ›.sup.125 I!TGF.beta.1 binding with chondroitinase ABC-treated proteoglycans and BSA. The concentrations of competitors were expressed as intact proteoglycan. The symbols are the same as in FIG. 3A.

FIGS. 4A and 4B show neutralization of the growth regulating activity of TGF.beta.1 by decorin: (FIG. 4A) Shows inhibition of TGF.beta.1-induced proliferation of CHO cells by decorin. ›.sup.3 H!Thymidine incorporation assay was performed in the presence of 5 ng/ml of TGF.beta.-1 and the indicated concentrations of purified decorin (.largecircle.) or BSA (.circle-solid.). At the concentration used, TGF.beta.-1 induced a 50% increase of ›.sup.3 H!thymidine incorporation in the CHO cells. The data represent percent neutralization of this growth stimulation; i.e. ›.sup.3 H!thymidine incorporation in the absence of either TGF.beta.1 or decorin=0%, incorporation in the presence of TGF.beta. but not decorin=100%. Each point shows the mean .+-. standard deviation of triplicate samples. (FIG. 4B) Shows neutralization of TGF.beta.1-induced growth inhibition in Mv1Lu cells by decorin. The assay was performed as in A except that TGF.beta.-1 was added at 0.5 ng/ml. This concentration of TGF.beta.-1 induces 50% reduction of ›.sup.3 H!thymidine incorporation in the Mv1Lu cells. The data represent neutralization of TGF.beta.-induced growth inhibition; i.e. ›.sup.3 H!thymidine incorporation in the presence of neither TGF.beta. or decorin=100%; incorporation in the presence of TGF.beta. but not decorin=0%.

FIG. 5A shows separation of growth inhibitory activity from decorin-expressing CHO cells by gel filtration. Serum-free conditioned medium of decorin overexpressor cells was fractionated by DEAE-Sepharose chromatography in a neutral Tris-HCl buffer and fractions containing growth inhibitory activity were pooled, made 4M with guanidine-HCl and fractionated on a Sepharose CL-6B column equilibrated with the same guanidine-HCl solution. The fractions were analyzed for protein content, decorin content, and growth regulatory activities. Elution positions of marker proteins are indicated by arrows. BSA: bovine serum albumin (Mr=66,000); CA: carbonic anhydrase (Mr=29,000); Cy:cytochrome c (Mr=12,400); Ap:aprotinin (Mr=6,500); TGF: ›.sup.125 I!TGF.beta.1(Mr=25,000).

FIG. 5B shows identification of the growth stimulatory material from gel filtration as TGF.beta.1. The growth stimulatory activity from the late fractions from Sepharose 6B (bar in panel A) was identified by inhibiting the activity with protein A-purified IgG from an anti-TGF.beta. antiserum. Data represent percent inhibition of growth stimulatory activity in a ›.sup.3 H!thymidine incorporation assay. Each point shows the mean .+-.standard deviation of triplicate determinations. Anti-TGF.beta.1 (.largecircle.), normal rabbit IgG (.circle-solid.).

FIG. 6 is a schematic diagram of MBP-decorin fragment fusion proteins. LRR is a leucine rich repeat. MBP is maltose binding protein.

FIG. 7 shows the results of binding studies of .sup.125 I-TGF.beta. to immobilized recombinant decorin (DC13) and MBP-decorin fragments PT-65, PT-71, PT-72 and PT-73.

FIG. 8 shows the results of binding studies of .sup.125 I-TGF.beta. to immobilized decorin (DC-18v) and MBP-decorin fragments PT-71, PT-72, PT-84, PT-85, PT-86 and PT-87.

FIG. 9 shows the results of binding studies of .sup.125 I-TGF.beta.1 to HepG2 cells in the presence of decorin fragments PT-65, PT-71, PT-72 and PT-78.

FIG. 10 shows the results of binding studies of .sup.125 I-TGF.beta. to L-M(tk-) cells in the presence of decorin and decorin fragments PT-71, PT-72, PT-84 and PT-85.

FIG. 11 shows the results of binding studies of .sup.125 I-TGF.beta.1 to L-M(tk-) cells in the presence of decorin and recombinant decorin fragments PT-71, PT-72, PT-86 and PT-87.

FIG. 12 shows the results of binding studies of .sup.125 I-TGF.beta.1 to L-M(tk-) cells in the presence of synthetic decorin peptide fragments P.sub.25 -Q.sub.36, H.sub.31 -S.sub.37 and H.sub.31 -L.sub.42 and a control peptide corresponding to the N-terminal 15-mer.

FIG. 13 shows the results of .sup.125 I-TGF.beta. binding to immobilized decorin with or without the presence of synthetic decorin peptide fragments 16D, 16E, 16G and 16H as well as a control peptide corresponding to the N-terminal 15-mer.

The invention provides a method of inhibiting an activity of a cell regulatory factor comprising contacting the cell regulatory factor with a purified polypeptide, wherein the polypeptide comprises the cell regulatory factor binding domain of a protein. The protein can be characterized by a leucine-rich repeat of about 24 amino acids. Since diseases such as cancer result from uncontrolled cell proliferation, the invention can be used to treat such diseases.

By "cell regulatory factor" is meant a molecule which can regulate an activity of a cell. The cell regulatory factors are generally proteins which bind cell surface receptors and include growth factors. Examples of cell regulatory factors include the five TGF.beta.'s, platelet-derived growth factor (PDGF), epidermal growth factor, insulin like growth factor I and II, fibroblast growth factor, interleukin-2, nerve growth factor, hemopoietic cell growth factors (IL-3, GM-CSF, M-CSF, G-CSF, erythropoietin) and the newly discovered Morphology Restoring Factor, hereinafter "MRF". Different regulatory factors can be bound by different proteins which can affect the regulatory factor's activity. For example, TGF.beta.-1 is bound by decorin and biglycan, and MRF by decorin.

By "cell regulatory factor binding domain" is meant a fragment of a protein which binds to the cell regulatory factor. A protein fragment that retains the binding activity is included within the scope of the invention and is referred to herein as an active fragment. Fragments that retain such activity, such as active fragments of decorin or biglycan, can be recognized by their ability to competitively inhibit the binding of, for example, decorin to TGF.beta., or of other polypeptides to their cognate growth factors.

Active fragments can be obtained by proteolytic digestion of the native polypeptide according to methods known in the art or as described, for example, in Example VIII. Alternatively, active fragments can be synthesized based on the known amino acid sequence by methods known to those skilled in the art or as described in Example VIII. The fragments can also be produced recombinantly by methods known in the art or as described in Example V. Examples of active fragments are included in Tables 4-15.

Such fragments can then be used in a competitive assay to determine whether they retain binding activity. For example, decorin can be attached to an affinity matrix, as by the method of Example II. Labelled TGF.beta. and an active fragment can then be contacted with the affinity matrix and the amount of TGF.beta. bound thereto determined.

As used herein, "decorin" refers to a proteoglycan having substantially the structural characteristics attributed to it in Krusius and Ruoslahti, supra. Human fibroblast decorin has substantially the amino acid sequence presented in Krusius and Ruoslahti, supra. "Decorin" refers both to the native composition and to modifications thereof which substantially retain the functional characteristics. Decorin core protein refers to decorin that no longer is substantially substituted with glycosaminoglycan and is included in the definition of decorin. Decorin can be rendered glycosaminoglycan-free by mutation or other means, such as by producing recombinant decorin in cells incapable of attaching glycosaminoglycan chains to a core protein.

Functional equivalents of decorin include modifications of decorin that retain its functional characteristics and molecules that are homologous to decorin, such as the decorin family members biglycan and fibromodulin, for example, that have the similar functional activity of decorin. Modifications can include, for example, the addition of one or more side chains that do not interfere with the functional activity of the decorin core protein.

Since the regulatory factor binding proteins each contain leucine-rich repeats of about 24 amino acids which can constitute 80% of the protein, it is likely that the fragments which retain the binding activity occur in the leucine-rich repeats. However, it is possible the binding activity resides elsewhere such as in the carboxy terminal amino acids or the junction of the repeats and the carboxy terminal amino acids.

The invention teaches a general method whereby one skilled in the art can identify proteins that can bind to cell regulatory factors or identify cell regulatory factors that bind to a certain family of proteins. The invention also teaches a general method in which these novel proteins or known existing proteins can be assayed to determine if they affect an activity of a cell regulatory factor. Specifically, the invention teaches the discovery that decorin and biglycan bind TGF.beta.-1 and MRF and that such binding can inhibit the cell regulatory functions of TGF.beta.-1. Further, both decorin and biglycan are about 80% homologous and contain a leucine-rich repeat of about 24 amino acids in which the arrangement of the leucine residues is conserved. As defined, each repeat generally contains at least two leucine residues and can contain five or more. These proteoglycans are thus considered members of the same protein family. See Ruoslahti, supra, Fisher et al., J. Biol. Chem., 264:4571-4576 (1989) and Patthy, J. Mol. Biol., 198:567-577 (1987), all of which are incorporated by reference. Other known or later discovered proteins having this leucine-rich repeat, i.e., fibromodulin, would be expected to have a similar cell regulatory activity. The ability of such proteins to bind cell regulatory factors could easily be tested, for example by affinity chromatography or microtiter assay as set forth in Example II, using known cell regulatory factors, such as TGF.beta.-1. Alternatively, any later discovered cell regulatory factor could be tested, for example by affinity chromatography using one or more regulatory factor binding proteins. Once it is determined that such binding occurs, the effect of the binding on the activity of all regulatory factors can be determined by methods such as growth assays as set forth in Example III. Moreover, one skilled in the art could simply substitute a novel cell regulatory factor for TGF.beta.-1 or a novel leucine-rich repeat protein for decorin or biglycan in the Examples to determine their activities. Thus, the invention provides general methods to identify and test novel cell regulatory factors and proteins which affect the activity of these factors.

The invention also provides a novel purified compound comprising a cell regulatory factor attached to a purified polypeptide wherein the polypeptide comprises the cell regulatory factor binding domain of a protein and the protein is characterized by a leucine-rich repeat of about 24 amino acids.

The invention further provides a novel purified protein, designated MRF, having a molecular weight of about 20 kd, which can be isolated from CHO cells, copurifies with decorin under nondissociating conditions, separates from decorin under dissociating conditions, changes the morphology of transformed 3T3 cells, and has an activity which is not inhibited with anti-TGF.beta.-1 antibody. Additionally, MRF separates from TGF.beta.-1 in HPLC.

The invention still further provides a method of purifying a cell regulatory factor comprising contacting the regulatory factor with a protein which binds the cell regulatory factor and has a leucine-rich repeat of about 24 amino acids and to purify the regulatory factor which becomes bound to the protein. The method can be used, for example, to purify TGF.beta.-1 by using decorin.

The invention additionally provides a method of treating a pathology caused by a TGF.beta.-regulated activity comprising contacting the TGF.beta. with a purified polypeptide, wherein the polypeptide comprises the TGF.beta. binding domain of a protein and wherein the protein is characterized by a leucine-rich repeat of about 24 amino acids, whereby the pathology-causing activity is prevented or reduced. While the method is generally applicable, specific examples of pathologies which can be treated include a cancer, a fibrotic disease, and glomerulonephritis. In cancer, for example, decorin can be used to bind TGF.beta.-1, destroying TGF.beta.-1's growth stimulating activity on the cancer cell.

Finally, a method of preventing the inhibition of a cell regulatory factor is provided. The method comprises contacting a protein which inhibits an activity of a cell regulator factor with a molecule which inhibits the activity of the protein. For example, decorin could be bound by a molecule, such as an antibody, which prevents decorin from binding TGF.beta.-1, thus preventing decorin from inhibiting the TGF.beta.-1 activity. Thus, the TGF.beta.-1 wound healing activity could be promoted by binding TGF.beta.-1 inhibitors.

It is understood that modifications which do not substantially affect the activity of the various molecules of this invention including TGF.beta., MRF, decorin, biglycan and fibromodulin are also included within the definition of those molecules. It is also understood that the core proteins of decorin, biglycan and fibromodulin are also included within the definition of those molecules.

The following examples are intended to illustrate but not limit the invention.

Expression system

The 1.8 kb full-length decorin cDNA described in Krusius and Ruoslahti, Proc. Natl. Acad. Sci. USA 83:7683 (1986), which is incorporated herein by reference, was used for the construction of decorin expression vectors. For the expression of decorin core protein, cDNA was mutagenized so the fourth codon, TCT, coding for serine, was changed to ACT coding for threonine, or GCT coding for alanine. This was engineered by site-directed mutagenesis according to the method of Kunkel, Proc. Natl. Acad. Sci USA 82:488 (1985), which is incorporated herein by reference. The presence of the appropriate mutation was verified by DNA sequencing.

The mammalian expression vectors pSV2-decorin and pSV2-decorin/CP-thr4 core protein were constructed by ligating the decorin cDNA or the mutagenized decorin cDNA into 3.4 kb HindIII-Bam HI fragment of pSV2 (Mulligan and Berg, Science 209:1423 (1980), which is incorporated herein by reference).

Dihydrofolate reductase (dhfr)-negative CHO cells (CHO-DG44) were cotransfected with pSV2-decorin or pSV2-decorin/CP and pSV2dhfr by the calcium phosphate coprecipitation method. The CHO-DG44 cells transfected with pSV2-decorin are deposited with the American Type Culture Collection under Accession Number ATCC No. CRL 10332. The transfected cells were cultured in nucleoside-minus alpha-modified minimal essential medium (.alpha.-MEM), (GIBCO, Long Island) supplemented with 9% dialyzed fetal calf serum, 2 mM glutamine, 100 units/ml penicillin and 100 .mu.g/ml streptomycin. Colonies arising from transfected cells were picked using cloning cylinders, expanded and checked for the expression of decorin by immunoprecipitation from .sup.35 SO.sub.4 -labeled culture supernatants. Clones expressing a substantial amount of decorin were then subjected to gene amplification by stepwise increasing concentration of methotrexate (MTX) up to 0.64 .mu.M (Kaufman and Sharp, J. Mol. Biol. 159:601 (1982), which is incorporated herein by reference). All the amplified cell lines were cloned either by limiting dilution or by picking single MTX resistant colonies. Stock cultures of these established cell lines were kept in MTX-containing medium. Before use in protein production, cells were subcultured in MTX-minus medium from stock cultures and passed at least once in this medium to eliminate the possible MTX effects.

Alternatively, the core protein was expressed in COS-1 cells as described in Adams and Rose, Cell 41:1007, (1985), which is incorporated herein by reference. Briefly, 6-well multiwell plates were seeded with 3-5.times.10.sup.5 cells per 9.6 cm.sup.2 growth area and allowed to attach and grow for 24 hours. Cultures were transfected with plasmid DNA when they were 50-70% confluent. Cell layers were washed briefly with Tris buffered saline (TBS) containing 50 mM Tris, 150 mM NaCl pH 7.2, supplemented with 1 mM CaCl.sub. and 0.5 mM MgCl.sub.2 at 37.degree. C. to prevent detachment. The wells were incubated for 30 minutes at 37.degree. C. with 1 ml of the above solution containing 2 .mu.g of closed circular plasmid DNA and 0.5 mg/ml DEAE-Dextran (Sigma) of average molecular mass of 500,000. As a control, cultures were transfected with the pSV2 expression plasmid lacking any decorin insert or mock transfected with no DNA. Culture were then incubated for 3 hours at 37.degree. C. with Dulbecco's Modified Eagle's medium (Irvine Scientific) containing 10% fetal calf serum and 100 .mu.M chloroquine (Sigma), after removing the DNA/TBS/DEAE-Dextran solution and rinsing the wells with TBS. The cell layers were then rinsed twice and cultured in the above medium, lacking any chloroquine, for approximately 36 hours. WI38 human embryonic lung fibroblasts were routinely cultured in the same medium.

COS-1 cultures were radiolabeled 36-48 hours after transfection with the plasmid DNAs. All radiolabeled metabolic precursors were purchased from New England Nuclear (Boston, Mass.). The isotopes used were .sup.35 S-sulfate (460 mCi/ml), L-›3,4,5-.sup.3 H(N)!-leucine (140 Ci/ml) and L-›.sup.14 C(U) !--amino acid mixture (product number 445E). Cultures were labeled for 24 hours in Ham's F-12 medium (GIBCO Labs), supplemented with 10% dialyzed fetal calf serum, 2 mM glutamine and 1 mM pyruvic acid, and containing 200 .mu.Ci/ml .sup.35 S-sulfate or .sup.3 H-leucine, or 10 .mu.Ci/ml of the .sup.14 C-amino acid mixture. The medium was collected, supplemented with 5 mM EDTA, 0.5 mM phenylmethylsulfonylfluoride, 0.04 mg/ml aprotinin and 1 .mu.g/ml pepstatin to inhibit protease activity, freed of cellular debris by centrifugation for 20 minutes at 2,000 .times.G and stored at -20.degree. C. Cell extracts were prepared by rinsing the cell layers with TBS and then scraping with a rubber policeman into 1 ml/well of ice cold cell lysis buffer: 0.05M Tris-HCl, 0.5M NaCl 0.1% BSA, 1% NP-40, 0.5% Triton X-100, 0.1% SDS, pH 8.3. The cell extracts were clarified by centrifugation for 1.5 hours at 13,000.times.G at 4.degree. C.

Rabbit antiserum was prepared against a synthetic peptide based on the first 15 residues of the mature form of the human decorin core protein (Asp-Glu-Ala-Ser-Gly-Ile-Gly-Pro-Glu-Val-Pro-Asp-Asp-Arg-Asp(SEQ ID No. 1)). The synthetic peptide and the antiserum against it have been described elsewhere (Krusius and Ruoslahti, 1986 supra.) Briefly, the peptide was synthesized with a solid phase peptide synthesizer (Applied Biosystems, Foster City, Calif.) by using the chemistry suggested by the manufacturer. The peptide was coupled to keyhole limpet hemocyanin by using N-succinimidyl 3-(2-pyridyldithio) propionate (Pharmacia Fine Chemicals, Piscataway, N.J.) according to the manufacturer's instructions. The resulting conjugates were emulsified in Freund's complete adjuvant and injected into rabbits. Further injections of conjugate in Freund's incomplete adjuvant were given after one, two and three months. The dose of each injection was equivalent to 0.6 mg of peptide. Blood was collected 10 days after the third and fourth injection. The antisera were tested against the glutaraldehyde-cross linked peptides and isolated decorin in ELISA (Engvall, Meth. Enzymol. 70:419-439 (1980)), in immunoprecipitation and immunoblotting, and by staining cells in immunofluorescence, as is well known in the art.

Immunoprecipitations were performed by adding 20 .mu.l of antiserum to the conditioned medium or cell extract collected from duplicate wells and then mixing overnight at 4.degree. C. Immunocomplexes were isolated by incubations for 2 hours at 4.degree. C. with 20 .mu.l of packed Protein A-agarose (Sigma). The beads were washed with the cell lysis buffer, with three tube changes, and then washed twice with phosphate-buffered saline prior to boiling in gel electrophoresis sample buffer containing 10% mercaptoethanol. Immunoprecipitated proteins were separated by SDS-PAGE in 7.5-20% gradient gels or 7.5% non-gradient gels as is well known in the art. Fluorography was performed by using Enlightning (New England Nuclear) with intensification screens. Typical exposure times were for 7-10 days at -70.degree. C. Autoradiographs were scanned with an LKB Ultroscan XL Enhanced Laser Densitometer to compare the relative intensities and mobilities of the proteoglycan bands.

SDS-PAGE analysis of cell extracts and culture medium from COS-1 cells transfected with the decorin-pSV2 construct and metabolically radiolabeled with .sup.35 S-sulfate revealed a sulfated band that was not present in mock-transfected cells. Immunoprecipitation with the antiserum raised against a synthetic peptide derived from the decorin core protein showed that the new band was decorin.

Expression of the construct mutated such that the serine residue which is normally substituted with a glycosaminoglycan (serine-4) was replaced by a threonine residue by SDS-PAGE revealed only about 10% of the level of proteoglycan obtained with the wild-type construct. The rest of the immunoreactive material migrated at the position of free core protein.

The alanine-mutated cDNA construct when expressed and analyzed in a similar manner yielded only core protein and no proteoglycan form of decorin. FIGS. 1A and 1B show the expression of decorin (lanes 1) and its threonine-4 (lanes 3) and alanine-4 (lanes 2) mutated core proteins expressed in COS cell transfectants. .sup.35 SO.sub.4 -labeled (FIG. 1A) and .sup.3 H-leucine labeled (FIG. 1 and B) culture supernatants were immunoprecipitated with rabbit antipeptide antiserum prepared against the NH.sub.2 -terminus of human decorin.

Purification of Decorin and Decorin Core Protein from Spent Culture Media

Cells transfected with pSV2-decorin vector and amplified as described above and in Yamaguchi and Ruoslahti, Nature 36:244-246 (1988), which is incorporated herein by reference, were grown to 90% confluence in eight culture flasks (175 cm.sup.2) in nucleoside minus .alpha.-MEM supplemented with 9% dialyzed fetal calf serum, 2 mM glutamine, 100 units/ml penicillin and 100 .mu.g/ml streptomycin. At 90% confluence culture media was changed to 25 ml per flask of nucleoside-free .alpha.-MEM supplemented with 6% dialyzed fetal calf serum which had been passed through a DEAE Sepharose Fast Flow column (Pharmacia) equilibrated with 0.25M NaCl in 0.05M phosphate buffer, pH 7.4. Cells were cultured for 3 days, spent media was collected and immediately made to 0.5 mM phenylmethylsulfonyl fluoride, 1 .mu.g/ml pepstatin, 0.04 mg/ml aprotinin and 5 mM EDTA.

Four hundred milliliters of the spent media were first passed through gelatin-Sepharose to remove fibronectin and materials which would bind to Sepharose. The flow-through fraction was then mixed with DEAE-Sepharose pre-equilibrated in 50 mM Tris/HCl, pH 7.4, plus 0.2M NaCl and batch absorbed overnight at 4.degree. C. with gentle mixing. The slurry was poured into a 1.6.times.24 cm column, washed extensively with 50 mM Tris/HCl, pH 7.4, containing 0.2M NaCl and eluted with 0.2M-0.8M linear gradient of NaCl in 50 mM Tris/HCl pH 7.4. Decorin concentration was determined by competitive ELISA as described in Yamaguchi and Ruoslahti, supra. The fractions containing decorin were pooled and further fractionated on a Sephadex gel filtration column equilibrated with 8M urea in the Tris-HCl buffer. Fractions containing decorin were collected.

The core protein is purified from cloned cell lines transfected with the pSV2-decorin/CP vector or the vector containing the alanine-mutated cDNA and amplified as described above. These cells are grown to confluency as described above. At confluency the cell monolayer is washed four times with serum-free medium and incubated in .alpha. MEM supplemented with 2 mM glutamine for 2 hours. This spent medium is discarded. Cells are then incubated with .alpha. MEM supplemented with 2 mM glutamine for 24 hours and the spent media are collected and immediately made to 0.5 mM phenylmethylsulfonyl fluoride, 1 .mu.g/ml pepstatin, 0.04 mg/ml aprotinin and 5 mM EDTA as serum-free spent media. The spent media are first passed through gelatin-Sepharose and the flow-through fraction is then batch-absorbed to CM-Sepharose Fast Flow (Pharmacia Fine Chemicals, Piscataway, N.J.) preequilibrated in 50 mM Tris/HCl, pH 7.4 containing 0.1M NaCl. After overnight incubation at 4.degree. C., the slurry is poured into a column, washed extensively with the preequilibration buffer and eluted with 0.1M-1M linear gradient of NaCl in 50 mM Tris/HCl, pH 7.4. The fractions containing decorin are pooled, dialyzed against 50 mM NH.sub.4 HCO.sub.3 and lyophilized. The lyophilized material is dissolved in 50 mM Tris, pH 7.4, containing 8M urea and applied to a Sephacryl S-200 column (1.5.times.110 cm). Fractions containing decorin core proteins as revealed by SDS-polyacrylamide electrophoresis are collected and represent purified decorin core protein.

A. Affinity Chromatography of TGF.beta. on Decorin-Sepharose

Decorin and gelatin were coupled to cyanogen bromide-activated Sepharose (Sigma) by using 1 mg of protein per ml of Sepharose matrix according to the manufacturer's instructions. Commercially obtained TGF.beta.-1 (Calbiochem, La Jolla, Calif.) was .sup.125 I-labelled by the chloramine T method (Frolik et al., J. Biol. Chem. 259:10995-11000 (1984)) which is incorporated herein by reference and the labeled TGF.beta. was separated from the unreacted iodine by gel filtration on Sephadex G-25, equilibrated with phosphate buffered saline (PBS) containing 0.1% bovine serum albumin (BSA) (FIG. 2). ›.sup.125 I!-TGF.beta.1 (5.times.10.sup.5 cpm) was incubated in BSA-coated polypropylene tubes with 0.2 ml of packed decorin-Sepharose (.largecircle.) or gelatin-Sepharose (.circle-solid.) in 2 ml of PBS pH 7.4, containing 1M NaCl and 0.05% Tween 20. After overnight incubation, the affinity matrices were transferred into BSA-coated disposable columns (Bio Rad) and washed with the binding buffer. Elution was effected first with 3M NaCl in the binding buffer and then with 8M urea in the same buffer. Fractions were collected, counted for radioactivity in a gamma counter and analyzed by SDS-PAGE under nonreducing condition using 12% gels.

FIG. 2A shows the radioactivity profile from the two columns and the SDS-PAGE analysis of the fractions is shown in FIG. 2B. The TGF.beta.-1 starting material contains a major band at 25 kd. This band represents the native TGF.beta.-1 dimer. In addition, there are numerous minor bands in the preparation. About 20-30% of the radioactivity binds to the decorin column and elutes with 8M urea, whereas only about 2% of the radioactivity is present in the urea-eluted fraction in the control fractionation performed on gelatin-Sepharose (FIG. 2A). The decorin-Sepharose nonbound fraction contains all of the minor components and some of the 25 kd TGF.beta.-1, whereas the bound, urea-eluted fraction contains only TGF.beta.-1 (FIG. 2B). These results show that TGF.beta.-1 binds specifically to decorin, since among the various components present in the original TGF.beta.-1 preparation, only TGF.beta.-1 bound to the decorin-Sepharose affinity matrix and since there was very little binding to the control gelatin-Sepharose affinity matrix. The TGF.beta.-1 that did not bind to the decorin-Sepharose column may have been denatured by the iodination. Evidence for this possibility was provided by affinity chromatography of unlabeled TGF.beta.-1 as described below.

In a second experiment, unlabeled TGF.beta.-1 180 ng was fractionated on decorin-Sepharose as described above for .sup.125 I-TGF.beta..

TGF.beta.-1 (180 ng) was incubated with decorin-Sepharose or BSA-agarose (0.2 ml packed volume) in PBS (pH 7.4) containing 1% BSA. After overnight incubation at 4.degree. C., the resins were washed with 15 ml of the buffer and eluted first with 5 ml of 3M NaCl in PBS then with 5 ml of PBS containing 8M urea. Aliquots of each pool were dialyzed against culture medium without serum and assayed for the inhibition of ›.sup.3 H!thymidine incorporation in Mv1Lu cells (Example III). The amounts of TGF.beta.-1 in each pool were calculated from the standard curve of ›.sup.3 H!thymidine incorporation obtained from a parallel experiment with known concentration of TGF.beta.-1. The results show that the TGF.beta.-1 bound essentially quantitatively to the decorin column, whereas there was little binding to the control column (Table 1). The partial recovery of the TGF.beta.-1 activity may be due to loss of TGF.beta.-1 in the dialyses.

B. Binding of TGF.beta.-1 to Decorin in a Microtiter Assay: Inhibition by Core Protein and Biglycan

The binding of TGF.beta.-1 to decorin was also examined in a microtiter binding assay. To perform the assay, the wells of a 96-well microtiter plate were coated overnight with 2.mu.g/ml of recombinant decorin in 0.1M sodium carbonate buffer, pH 9.5. The wells were washed with PBS containing 0.05% Tween (PBS/Tween) and samples containing 5.times.10.sup.4 cpm of ›.sup.125 I!-TGF.beta.-1 and various concentrations of competitors in PBS/Tween were added to each well. The plates were then incubated at 37.degree. C. for 4 hours (at 4.degree. C. overnight in experiments with chondroitinase ABC-digested proteoglycans), washed with PBS/Tween and the bound radioactivity was solubilized with 1% SDS in 0.2M NaOH. Total binding without competitors was about 4% under the conditions used. Nonspecific binding, determined by adding 100-fold molar excess of unlabeled TGF.beta.-1 over the labeled TGF.beta.-1 to the incubation mixture, was about 13% of total binding. This assay was also used to study the ability of other decorin preparations and related proteins to compete with the interaction.

Completion of the decorin binding was examined with the following proteins (FIG. 3A and 3B; symbols are indicated in the section of BRIEF DESCRIPTION OF THE FIGURES):

Decorin isolated from bovine skin and biglycan isolated from bovine articular cartilage (PGI and PGII, obtained from Dr. Lawrence Rosenberg, Monteflore Medical Center, N.Y.; and described in Rosenberg et al., J. Biol. Chem. 250:6304-6313, (1985), incorporated by reference herein), chicken cartilage proteoglycan (provided by Dr. Paul Goetinck, La Jolla Cancer Research Foundation, La Jolla, Calif., and described in Goetinck, P. F., in THE GLYCOCONJUGATES, Vol. III, Horwitz, M. I., Editor, pp. 197-217, Academic Press, NY). For the preparation of core proteins, proteoglycans were digested with chondroitinase ABC (Seikagaku, Tokyo, Japan) by incubating 500 .mu.g of proteoglycan with 0.8 units of chondroitinase ABC in 250 .mu.l of 0.1M Tris/Cl, pH 8.0, 30 mM sodium acetate, 2 mM PMSF, 10 mM N-ethylmalelmide, 10 mM EDTA, and 0.36 mM pepstatin for 1 hour at 37.degree. C. Recombinant decorin and decorin isolated from bovine skin (PGII) inhibited the binding of ›.sup.125 I!-TGF.beta.-1, as expected (FIG. 3A). Biglycan isolated from bovine articular cartilage was as effective an inhibitor as decorin. Since chicken cartilage proteoglycan, which carries many chondroitin sulfate chains, did not show any inhibition, the effect of decorin and biglycan is unlikely to be due to glycosaminoglycans. Bovine serum albumin did not shown any inhibition. This notion was further supported by competition experiments with the mutated decorin core protein (not shown) and chondroitinase ABC-digested decorin and biglycan (FIG.3B). Each of these proteins was inhibitory, whereas cartilage proteoglycan core protein was not. The decorin and biglycan core proteins were somewhat more active than the intact proteoglycans. Bovine serum albumin treated with chondroitinase ABC did not shown any inhibition. Additional binding experiments showed that ›.sup.125 I!-TGF.beta.-1 bound to microtiter wells coated with biglycan or its chondroitinase-treated core protein. These results show that TGF.beta.-1 binds to the core protein of decorin and biglycan and implicates the leucine-rich repeats these proteins share as the potential binding sites.

The ability of decorin to modulate the activity of TGF.beta.-1 was examined in ›.sup.3 H!thymidine incorporation assays. In one assay, an unamplified CHO cell line transfected only with pSV2dhfr (control cell line A in reference 1, called CHO cells here) was used. The cells were maintained in nucleoside-free alpha-modified minimal essential medium (e-MEM, GIBCO, Long Island, N.Y.) supplemented with 9% dialyzed fetal calf serum (dFCS) and ›.sup.3 H!thymidine incorporation was assayed as described (Cheifetz et al., Cell 48:409-415 (1987)). TGF.beta.-1 was added to the CHO cell cultures at 5 ng/ml. At this concentration, it induced a 50% increase of ›.sup.3 H!thymidine incorporation in these cells. Decorin or BSA was added to the medium at different concentrations. The results are shown in FIG. 4A. The data represent percent neutralization of the TGF.beta.-1-induced growth stimulation, i.e., ›.sup.3 H!thymidine incorporation, in the absence of either TGF.beta.-1 or decorin=0%, incorporation in the presence of TGF.beta.-1 but not decorin=100%. Each point shows the mean .+-. standard deviation of triplicate samples. Decorin (.circle-solid.) BSA (.largecircle.).

Decorin neutralized the growth stimulatory activity of TGF.beta.-1 with a half maximal activity at about 5 .mu.g/ml. Moreover, additional decorin suppressed the ›.sup.3 H!-thymidine incorporation below the level observed without any added TGF.beta.-1, demonstrating that decorin also inhibited TGF.beta. made by the CHO cells themselves. Both the decorin-expressor and control CHO cells produced an apparently active TGF.beta. concentration of about 0.25 ng/ml concentration into their conditioned media as determined by the inhibition of growth of the mink lung epithelial cells. (The assay could be performed without interference from the decorin in the culture media because, as shown below, the effect of TGF.beta. on the mink cells was not substantially inhibited at the decorin concentrations present in the decorin-producer media.)

Experiments in MvLu mink lung epithelial cells (American Type Culture Collection CCL64) also revealed an effect by decorin on the activity of TGF.beta.-1. FIG. 4B shows that in these cells, the growth of which is measured by thymidine incorporation, had been suppressed by TGF.beta.-1. Assay was performed as in FIG. 4A, except that TGF.beta.-1 was added at 0.5 ng/ml. This concentration of TGF.beta. induces 50% reduction of ›.sup.3 H!-thymidine incorporation in the Mv1Lu cells. The data represent neutralization of TGF.beta.-induced growth inhibition; i.e., ›.sup.3 H!-thymidine incorporation in the presence of neither TGF.beta. or decorin=100%; incorporation in the presence of TGF.beta. but not decorin=0%.

Analysis of the decorin contained in the overexpressor culture media not only uncovered the activities of decorin described above, but also revealed the presence of other decorin-associated growth regulatory activities. The overexpressor media were found to contain a TGF.beta.-like growth inhibitory activity. This was shown by gel filtration of the DEAE-isolated decorin under dissociating conditions. Serum-free conditioned medium of decorin overexpressor CHO-DG44 cells transfected with decorin cDNA was fractionated by DEAE-Sepharose chromatography in a neutral Tris-HCl buffer and fractions containing growth inhibitory activity dialyzed against 50 mM NH.sub.4 HCO.sub.3, lyophilized and dissolved in 4M with guanidine- HCl in a sodium acetate buffer, pH 5.9. The dissolved material was fractionated on a 1.5.times.70 cm Sepharose CL-6B column equilibrated with the same guanidine-HCl solution. The fractions were analyzed by SDS-PAGE, decorin ELISA and cell growth assays, all described above. Three protein peaks were obtained. One contained high molecular weight proteins such as fibronectin (m.w. 500,000) and no detectable growth regulatory activities, the second was decorin with the activities described under Example III and the third was a low molecular weight (10,000-30,000-dalton) fraction that had a growth inhibitory activity in the mink cell assay and stimulated the growth of the CHO cells. FIG. 5A summarizes these results. Shown are the ability of the gel filtration fractions to affect ›.sup.3 H!-thymidine incorporation by the CHO cells and the concentration of decorin as determined by enzyme immunoassay. Shown also (arrows) are the elution positions of molecular size markers: BSA, bovine serum albumin (Mr=66,000); CA, carbonic anhydrase (Mr=29,000); Cy, cytochrome c (Mr=12,400); AP, aprotinin (Mr=6,500); TGF, ›.sup.125 I!TGF.beta.-1 (Mr=25,000).

The nature of the growth regulatory activity detected in the low molecular weight fraction was examined with an anti-TGF.beta.-1 antiserum. The antiserum was prepared against a synthetic peptide from residues 78-109 of the human mature TGF.beta.-1. Antisera raised by others against a cyclic form of the same peptide, the terminal cysteine residues of which were disulfide-linked, have previously been shown to inhibit the binding of TGF.beta.-1 to its receptors (Flanders et al., Biochemistry 27:739-746 (1988), incorporated by reference herein). The peptide was synthesized in an Applied Biosystems solid phase peptide synthesizer and purified by HPLC. A rabbit was immunized subcutaneously with 2 mg per injection of the peptide which was mixed with 0.5 mg of methylated BSA (Sigma, St. Louis, Mo.) and emulsified in Freund's complete adjuvant. The injections were generally given four weeks apart and the rabbit was bled approximately one week after the second and every successive injection. The antisera used in this work has a titer (50% binding) of 1:6,000 in radioimmunoassay, bound to TGF.beta.-1 in immunoblots.

This antiserum was capable of inhibiting the activity of purified TGF.beta.-1 on the CHO cells. Moreover, as shown in FIG. 5B, the antiserum also inhibited the growth stimulatory activity of the low molecular weight fraction as determined by the ›.sup.3 H!-thymidine incorporation assay on the CHO cells. Increasing concentrations of an IgG fraction prepared from the anti-TGF.beta.-1 antiserum suppressed the stimulatory effect of the low molecular weight fraction in a concentration-dependent manner (.circle-solid.). IgG from a normal rabbit serum had no effect in the assay (.largecircle.).

The above result identified the stimulatory factor in the low molecular weight fraction as TGF.beta.-1. However, TGF.beta.-1 is not the only active compound in that fraction. Despite the restoration of thymidine incorporation by the anti-TGF.beta.-1 antibody shown in FIG. 5B, the cells treated with the low molecular weight fraction were morphologically different from the cells treated with the control IgG or cells treated with antibody alone. This effect was particularly clear when the antibody-treated, low molecular weight fraction was added to cultures of H-ras transformed NIH 3T3 cells (Der et al., Proc. Natl. Acad. Sci. USA 79:3637-3640 (1982)). Cells treated with the low molecular weight fraction and antibody appeared more spread and contact inhibited than the control cells. This result shows that the CHO cell-derived recombinant decorin is associated with a cell regulatory factor, MRF, distinct from the well characterized TGF.beta.'s.

Additional evidence that the new factor is distinct from TGF.beta.-1 came from HPLC experiments. Further separations of the low molecular weight from the Sepharose CL-6B column was done on a Vydac C4 reverse phase column (1 .times.25 cm, 5 .mu.m particle size, the Separations Group, Hesperia, Calif.) in 0.1% trifluoroacetic acid. Bound proteins were eluted with a gradient of acetonitrite (22-40%) and the factions were assayed for growth-inhibitory activity in the mink lung epithelial cells and MRF activity in H-ras 3T3 cells. The result showed that the TGF.beta.-1 activity eluted at the beginning of the gradient, whereas the MRF activity eluted toward the end of the gradient.

MBP-Decorin fragment fusion proteins of varying lengths were engineered such that the Maltose Binding Protein (MBP) was attached to the amino terminus of the gene encoding mature decorin as shown in FIG. 6. The techniques incorporated for such construction are described in F. M. Ausubel et al., Current Protocols in Molecular Biology, John Wiley and Sons (1987) and Maniatis et al., Molecular Cloning: A Laboratory Manual, Cold Spring Harbor Laboratory (1982), which are incorporated herein by reference.

The decorin-encoding DNA fragments were generated by polymerase chain reaction (PCR), Scharf et al., Science 233:1076-1078 (1986), which is incorporated herein by reference. The primers, synthetic oligonucleotides obtained from Genosys (Houston, Tex.), incorporate an Eco RI restriction site at the 5' end and an Xba I restriction site at the 3' end of the PCR product. In some instances, the primers also included a base change to code for a different amino acid. The primers used to generate specific inserts are identified in Table 2, while the primer sequences are identified in Table 3(SEQ ID Nos. 2 through 16). The template DNA was a large scale CsCl prep of pPG-40 described in Krusius and Ruoslahti, Proc. Natl. Acad. Sci. USA 83:7683-7687 (1986), incorporated herein by reference. The DNA amplification reaction was done in a thermal cycler according to manufacturer's recommendations (Perkin-Elmer Cetus; Norwalk, Conn.) using the Vent.TM. DNA Polymerase (New England Biolabs; Beverly, Mass.). The decorin-encoding DNA fragments cycled 35-40 times at 94.degree., 40.degree., and 72.degree. C.

The PCR products were analyzed by agarose gel electrophoresis, Ausubel et al., supra, and Maniatis et al., supra, to identify and determine the decorin-encoding DNA fragments (see Table 2 under "Insert Size"). The PCR products less than 200 base pairs (bp) in size were purified by electrophoresis onto DEAE-cellulose paper, Ausubel et al., supra,; the PCR products greater than 200 bp were purified by using Prep-A-Gene.TM. DNA Purification Kit (Bio-Rad; Richmond, Calif.) according to manufacturer's instructions.

The decorin-encoding DNA fragments (insert) were ligated between the Eco RI and Xba I restriction sites of the polylinker in the vector pMAL-p (Protein Fusion and Purification Sytem; New England Biolabs). The ligation had a total of 500 ng of DNA and the molar ratio of insert:vector was 3:1. The ligations were then transformed into Escherichia coli (E. coli) DH5.alpha. cells (Gibco BRL; Gaithersburg, Md.), genotype: F.sup.- .phi.80dlacZ.DELTA.M15, .DELTA.(lacZYA-argF)U169, deoR, recA1, endA1, hsdR17(r.sub.k.sup.-, m.sub.k.sup.+), supE44 .lambda..sup.- ' thi-1, gyrA96, relA1, or E. coli Sure.TM. cells (Stratagene, La Jolla, Calif.), genotype: e14-(mcrA), .DELTA.(mcrCB-hsdSMR-mrr)171, sbcC, recB, recJ, umuC::Tn5 (kan.sup.r), uvrC, supE44, lac, gyrA96, relA1, thi-1, endA1 ›F'proAB, lacI.sup.q Z.DELTA.M15, Tn10, (tet.sup.r)!. The .phi.80dlacZ.DELTA.M15 marker of the E. coli DH5.alpha. strain provides .alpha.-complementation of the .beta.-galactosidase gene from pMAL-p. Colonies containing pMAL-p with the decorin-encoding DNA fragments were colorless on plates containing 5-Bromo-4-chloro-3-indolyl-.beta.-D-galactoside (X-gal) due to the interruption of the .beta.-galatosidase gene. Host cells containing pMAL-p only produces blue colonies.

Minipreps of colorless colonies were then made as described in Ausubel et al., supra, and Maniatis et al., supra. Plasmids encoding MBP-decorin fragment fusion proteins PT-73, -74, -75, -77, and -78, were then digested with restriction endonucleases Eco RI and Xba I (both from Promega; Madison, Wis.) and the presence of specific inserts confirmed by agarose gel electrophoresis. The other plasmids encoding fusion proteins, PT-72, -76, -84, -85, -86, and -87, had inserts confirmed by sequencing using Sequenase.RTM. Version 2.0 DNA Sequencing Kit (U. S. Biochemical; Cleveland, Ohio) according to manufacturer's instructions.

Test expression of MBP-Decorin fragment fusion proteins were performed in the host bacterial strain (see Table 2). An overnight culture of E. coli DH5.beta. or E. coli Sure.TM. cells containing the MBP-decorin fragment fusion protein plasmids were made by taking a stab of a frozen stock and inoculating L-Broth, Ausubel et al., supra, containing 100 .mu.g/ml ampicillin at 37.degree. C. with rapid shaking. The following morning, 1 ml was used to inoculate 10 ml of prewarmed medium (L-Broth containing ampicillin). After 1 hour at 37.degree. C., 100 .mu.l of 0.1M IPTG were added per culture and the induced cultures were allowed to incubate for an additional 2-3 hours. The cells were lysed by resuspending in PAGE Sample Buffer (Novex Experimental Technology; Encinitas, Calif.) with 0.8% .beta.-mercaptoethanol and shearing 10 times with a 1 cc tuberculin syringe. The sample was run on an 8-16% gradient SDS-PAGE gel (Novex Experimental Technology) and a Western Blot (Novex Experimental Technology) was performed according to manufacturer's recommendations. The blot was developed, Ausubel et al., supra, with Rabbit anti-PG40 serum (Telios Pharmaceuticals, Inc.; La Jolla, Calif.) to test for PT-65, -73, -74, -75, -76, -77, and -78, and Rabbit anti-MBP serum (made in-house) to test for PT-72, -84, -85, -86, and -87. The results are indicated in Table 2 under "MW."

Large scale CsCl preps, Ausubel, et al., supra, and Maniatis, et al, supra, of the plasmids encoding MBP-Decorin fragment fusion proteins PT-84, -85, -86, and -87 were made and used to transform E. coli. DH5.beta.. Expression of the fusion proteins was confirmed by doing a test expression as described above.

Production batches of the fusion proteins were prepared as follows. An overnight culture of E. coli DH5.beta. cells containing the MBP-Decorin fragment fusion protein plasmids was made by taking a stab of the frozen stock and inoculating L-Broth containing 100 .mu.g/ml ampicillin at 37.degree. C. with rapid shaking. From this culture, 5 ml were used to inoculate a larger 50 ml overnight culture. The following morning, 50 ml of the larger culture were added to 500 ml of pre-warmed media. Typically 1-4 liters were prepared for each batch. After 1 hour at 37.degree. C., 5 ml of 0.1M IPTG were added per flask and the induced cultures were allowed to incubate for an additional 2-3 hours. The cells were harvested by centrifugation at 5,000 rpm for 10 minutes at 10.degree. C. using either a GSA or GS-3 rotor in an RC5B centrifuge (DuPont Instruments; Wilmington, Del.). The pellets were resuspended in 0.1 volume of lysis buffer (50mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.1M PMSF, and 0.25 mg/ml lysozyme) and incubated for 10-15 minutes on ice. The suspension was freeze/thawed three times by repeated cycling through a dry ice/ethanol bath and a room temperature shaking water bath. The suspension was sheared by homogenization using a dounce homogenizer. The lysate was pre-cleared by centrifugation at 12,000 rpm for 30 minutes in a SA-600 rotor (DuPont). The cleared supernatant was decanted and saved. A final clarification step was done by centrifuging for 30 minutes at 4.degree. C. in an RC-80 ultracentrifuge using an AH-629 rotor (DuPont). The final cleared lysates were stored either at 4.degree. C. or -20.degree. C. until ready to be purified.

Affinity purifications of the MBP-Decorin fusion proteins were done using an amylose resin (New England Biolabs). Briefly, six to seven ml of resin were packed into a 2.5.times.10 cm glass column in MBP column buffer (10mM Tris-HCl, pH 8.4, 1 mM EDTA, 0.5M NaCl). The resin was pre-equilibrated with at least 3 column volumes of MBP column buffer containing 0.25% Tween 20. Cleared lysate as prepared above was diluted 1 part lysate to 1 part 2.times.MBP column buffer containing 0.5% Tween 20 and added to the column at a flow rate of 50-100 ml/hr. Typically, 100-150 ml of diluted lysate were passed over each column. Non-specific material was removed by washing with at least 3 column volumes each of MBP column buffer containing 0.25% Tween 20 and MBP column buffer. The purified MBP-Decorin fragment fusion protein was eluted with 5.times.4 ml aliquots of MBP column buffer containing 10 mM maltose. Peak fractions containing the fusion protein were pooled, assayed to determine protein quantity (Bio-Rad Protein Kit; Richmond, Calif. or Pierce BCA Protein Kit; Rockford, Ill.), run on an 8-16% SDS-PAGE gel (Novex Experimental Technology), and stained with Coomasie Blue (Novex Experimental Technology) to check for purity. The results of the fusion protein are in Table I under "MW". The purified fusion protein was stored at 4.degree. C. or -20.degree. C. in aliquots until ready to be tested for activity.

The pMAL-p vector also was engineered such that a termination codon was incorporated between the Eco RI and Xba I sites. During this process, the original Eco RI site in the vector was destroyed and replaced at a position downstream from a second Factor Xa cleavage site. The second Factor Xa site was incorporated to facilitate subsequent cleavage of the decorin fusion protein from the MBP carrier. The construction involved annealing complimentary oligos (OT-98 and OT-99; sequences in Table 3) and ligating into pMAL-p at the Eco RI and Xba I sites, Ausubel, et al., supra, and Maniatis, et al., supra. The ligation (PT-71) was transformed into E. coli DH5.alpha. cells, mini-preps were made from colorless colonies and the clones were sequenced for insert. Expression of the protein followed the same procedure as the MBP-Decorin fragment fusion proteins above. The results are indicated in Table 2 under "MW."

Tables 4-15 below provide the nucleotide and corresponding amino acid sequences of the decorin fragment fusion proteins prepared as described above. Each table also identifies the Eco RI and Xba I ligation sites. - (SEQ ID NOS. 26 through 28) PT-65 Sequence ##STR1## ##STR2## ##STR3## ##STR4## ##STR5## ##STR6## ##STR7## ##STR8## ##STR9## ##STR10## ##STR11##

Immulon wells were coated with 0.5 .mu.g/ml recombinant decorin at 50 .mu.l/well. The wells were placed in a 37.degree. C. incubator overnight and thereafter washed 3 times with 200 .mu.l PBS (0.15M NaCl ) per well to remove unbound decorin. TGF-.beta. labeled with .sup.125 I (400 pM, New England Nuclear, Bolton-Hunter Labeled) was pre-incubated with or without competitors in 200 .mu.l PBS/0.05% Tween-20 for 1 hour and 45 minutes at room temperature. Competitors included recombinant human decorin preparations (DC-13 and DC-18v), decorin fragments, and MBP as a negative control. DC-13 and DC-18v are different preparations of recombinant human decorin; PT-71 or MBP (maltose-binding protein) is a negative control; PT-65 is MBP-whole decorin; PT-72 is MBP-decorin N-terminus; PT-73 is PT-72 +2 LRR; PT-84 and PT-85 are cysteine to serine mutant of PT-72; PT-86 is decorin C terminus; PT-87 is cysteine to serine mutant of PT-86.

Fifty .mu.l/well of the pre-incubated .sup.125 I-TGF-.beta. mixture or control were added and incubated overnight at 0.degree. C. Following the incubation, 50 .mu.l of the free TGF-.beta. supernatants were transferred to labeled tubes. The plate was washed 3 times with 0.05% Tween-20 in PBS (200 .mu.l/well). The wells were then transferred into tubes for counting in a gamma counter. The results of the binding studies with immobilized decorin are summarized in FIGS. 7 and 8.

Recombinant human decorin (DC-13), MBP-whole decorin (PT-65), MBP-decorin N-terminus (PT-72) and MBP-decorin N-terminus+2 leucine rich repeats (PT-73) inhibited .sup.125 I-TGF-.beta. binding to immobilized decorin as shown in FIG. 7. MBP alone (PT-71) had no effect on .sup.125 I-TGF-.beta. binding to immobilized decorin. These results demonstrate that the N-terminus of decorin is capable of binding TGF-.beta. in solution and preventing it from binding to immobilized decorin. Thus, two portions of the molecule appear to contain part of the binding site in decorin for TGF-.beta..

As shown in FIG. 8, recombinant human decorin (DC-18V)+MBP-decorin N-terminus (PT-72) inhibited .sup.125 I-TGF-.beta. binding to immobilized decorin. In addition, cysteine to serine mutants of PT-72, (C-24, C-28, C-30, C-37 to serine, PT-84; C-28 and C-30 to serine, PT-85) did not inhibit .sup.125 I- TGF-.beta. binding to decorin. MBP-decorin C-terminus (PT-86) and a cysteine to serine mutant (PT-87) of PT-86 also inhibited .sup.125 I-TGF-.beta. binding to decorin. These results demonstrate that the C-terminus of decorin is also capable of binding TGF-.beta. and that the first cysteine residue in the C-terminus is not required for TGF-.beta. binding.

About 2.5.times.10.sup.4 HepG2 cells or L-M(tk-) cells were incubated with 250 pM›.sup.125 I!TGF-.beta. in the presence of recombinant human decorin (DC-12), PT-71 (MBP), decorin fragments (PT-72, -73, -84, -85, -86 and -87) or anti-TGF-.beta. antibodies for 2 hours at room temperature. Cells were washed with washing buffer (128 mM NaCl, 5 mM KCl, 5 mM Mg.sub.2 SO.sub.4, 1.2 mM CaCl.sub.2, 50 mM HEPES, pH 7.5) four times before determination of bound CPM. The results are summarized in FIGS. 9, 10 and 11.

Recombinant human decorin, MBP-whole decorin (PT-65), MBP-decorin N-terminus (PT-72) and MBP-decorin N-terminus +2 leucine rich repeats (PT-78) inhibited .sup.125 I-TGF-.beta. binding to HepG2 cells. MBP alone (PT-71) had no effect on .sup.125 I-TGF-.beta. binding to HepG2. These results shown in FIG. 9 demonstrate that the N-terminus of decorin is capable of preventing TGF-.beta. from binding to its receptor on HepG2 cells.

As shown in FIG. 10, recombinant human decorin and MBP-decorin N-terminus (PT-72) inhibited .sup.125 I-TGF-.beta. binding to L-M(tk-) cells. An anti-TGF-.beta.1 antibody also inhibited TGF-.beta. binding to these cells. Cysteine to serine mutants of PT-72 (C-24, C-28, C-30, C-37 to serine, PT-84; C-28 and C-30 to serine, PT-85) did not inhibit .sup.125 I-TGF-.beta. binding to L-M(tk-) cells.

Recombinant human decorin, MBP-decorin N-terminus (PT-72) and anti-TGF-.beta. antibodies inhibited .sup.125 I-TGF-.beta. binding to L-M(tk-) cells. MBP-decorin C-terminus (PT-86) and a cysteine to serine mutant (PT-87) of PT-86 also inhibited .sup.125 I-TGF-.beta. binding to L-M(tk-) cells. As shown in FIG. 11, these results demonstrate that the C-terminus of decorin also is capable of inhibiting TGF-.beta. binding to its receptor, and that the first cysteine residue of the C-terminus is not required for inhibition.

The following peptides were synthesized and tested for their ability to inhibit binding of TGF.beta.1 to L-M(tk-) cells:

Peptide H.sub.31 - S.sub.37 corresponds to the same decorin sequence between His-31 and Cys-37 except the Cysteine at position 37 is replaced with a serine. Peptide P.sub.25 -Q.sub.36 corresponds to the sequence reported in Krusius and Ruoslahti, supra, from position Pro-25 through Gln-36, except the native Cysteine residues at positions 28 and 30 are each replaced with a serine. Peptide H.sub.31 - L.sub.42 also corresponds to decorin between His-31 and Leu-42 except the cysteine at position 37 is replaced with a serine. Peptide nP.sub.25 -Q.sub.36 corresponds exactly with decorin from position Pro-25 through Gln-36.

The peptides were synthesized using the applied Biosystems, Inc. Model 430A or 431A automatic peptide synthesizer and the chemistry provided by the manufacturer.

The activity of the peptides was evaluated using the L-M(tk-) TGF.beta.1 binding inhibition assay described in Example VII, except various concentrations of peptide were incubated with the cells and TGF.beta.1 instead of decorin and recombinant decorin fragments. The negative control was a synthetic peptide corresponding to the first 15 amino acids of decorin, which has the sequence DEASGIGPEVPDDRD(SEQ ID NO. 21).

FIG. 12 provides the binding data for peptides P.sub.25 -Q.sub.36, H.sub.31 -S.sub.37, and H.sub.31 -L.sub.42 and the control peptide. All three test peptides inhibited binding of TGF.beta.1 to L-M(tk-) cells. Peptide nP.sub.25 -Q.sub.36, in which the native Cys residues remain, also demonstrated inhibitory activity, albeit to a lesser extent. Table 16 lists the test peptides in the order of decreasing inhibitory activity, i.e., peptide H.sub.31 -S.sub.37 was found to show the highest inhibitory activity.

Further binding studies were conducted with soluble N-terminal decorin peptide fragments synthesized and tested for their ability to inhibit TGF-.beta.1 binding to immobilized decorin as described above. The N-terminal peptide fragments are listed in Table 17.

The results of the binding studies are shown in FIG. 13, which shows that the peptide 16G inhibited TGF-.beta. binding to immobilized decorin.

Although the invention has been described with reference to the presently-preferred embodiments, it should be understood that various modifications can be made without departing from the spirit of the invention. Accordingly, the invention is limited only by the following claims.